C11: should we use char32_t for unicode code points?

From a289df81514b1f784f9242a1c214bcc0749b6f9c Mon Sep 17 00:00:00 2001
From: Jeff Davis <jeff@j-davis.com>
Date: Tue, 21 Oct 2025 13:16:47 -0700
Subject: [PATCH v2] Use C11 char16_t and char32_t for Unicode code points.

Reviewed-by: Tatsuo Ishii <ishii@postgresql.org>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Discussion: https://postgr.es/m/bedcc93d06203dfd89815b10f815ca2de8626e85.camel%40j-davis.com
---
 configure.ac                                  |  1 +
 meson.build                                   |  1 +
 src/backend/parser/parser.c                   |  8 +--
 src/backend/parser/scan.l                     |  8 +--
 src/backend/utils/adt/jsonpath_scan.l         |  6 +-
 src/backend/utils/adt/pg_locale_builtin.c     | 44 ++++++++++-----
 src/backend/utils/adt/varlena.c               | 40 ++++++-------
 src/backend/utils/mb/mbutils.c                |  4 +-
 src/common/saslprep.c                         | 48 ++++++++--------
 src/common/unicode/case_test.c                | 23 ++++----
 src/common/unicode/category_test.c            |  3 +-
 .../unicode/generate-norm_test_table.pl       |  4 +-
 .../unicode/generate-unicode_case_table.pl    |  7 +--
 .../generate-unicode_category_table.pl        |  8 +--
 src/common/unicode/norm_test.c                |  6 +-
 src/common/unicode_case.c                     | 56 +++++++++----------
 src/common/unicode_category.c                 | 50 ++++++++---------
 src/common/unicode_norm.c                     | 56 +++++++++----------
 src/fe_utils/mbprint.c                        | 10 ++--
 src/include/c.h                               | 19 +++++++
 src/include/common/unicode_case.h             | 10 ++--
 src/include/common/unicode_case_table.h       | 13 ++---
 src/include/common/unicode_category.h         | 46 ++++++++-------
 src/include/common/unicode_category_table.h   |  8 +--
 src/include/common/unicode_norm.h             |  6 +-
 src/include/mb/pg_wchar.h                     | 32 +++++------
 src/tools/pgindent/typedefs.list              |  2 +
 27 files changed, 276 insertions(+), 243 deletions(-)

diff --git a/configure.ac b/configure.ac
index e44943aa6fe..6ab2e157531 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1513,6 +1513,7 @@ AC_CHECK_HEADERS(m4_normalize([
 	sys/signalfd.h
 	sys/ucred.h
 	termios.h
+	uchar.h
 	ucred.h
 	xlocale.h
 ]))
diff --git a/meson.build b/meson.build
index 395416a6060..c3128c7554f 100644
--- a/meson.build
+++ b/meson.build
@@ -2613,6 +2613,7 @@ header_checks = [
   'sys/signalfd.h',
   'sys/ucred.h',
   'termios.h',
+  'uchar.h',
   'ucred.h',
   'xlocale.h',
 ]
diff --git a/src/backend/parser/parser.c b/src/backend/parser/parser.c
index 33a040506b4..a3679f8e86c 100644
--- a/src/backend/parser/parser.c
+++ b/src/backend/parser/parser.c
@@ -339,7 +339,7 @@ hexval(unsigned char c)
 
 /* is Unicode code point acceptable? */
 static void
-check_unicode_value(pg_wchar c)
+check_unicode_value(char32_t c)
 {
 	if (!is_valid_unicode_codepoint(c))
 		ereport(ERROR,
@@ -376,7 +376,7 @@ str_udeescape(const char *str, char escape,
 	char	   *new,
 			   *out;
 	size_t		new_len;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	ScannerCallbackState scbstate;
 
 	/*
@@ -420,7 +420,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[3]) &&
 					 isxdigit((unsigned char) in[4]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[1]) << 12) +
 					(hexval(in[2]) << 8) +
@@ -457,7 +457,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[6]) &&
 					 isxdigit((unsigned char) in[7]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[2]) << 20) +
 					(hexval(in[3]) << 16) +
diff --git a/src/backend/parser/scan.l b/src/backend/parser/scan.l
index 08990831fe8..a67815339b7 100644
--- a/src/backend/parser/scan.l
+++ b/src/backend/parser/scan.l
@@ -121,7 +121,7 @@ static void addlitchar(unsigned char ychar, core_yyscan_t yyscanner);
 static char *litbufdup(core_yyscan_t yyscanner);
 static unsigned char unescape_single_char(unsigned char c, core_yyscan_t yyscanner);
 static int	process_integer_literal(const char *token, YYSTYPE *lval, int base);
-static void addunicode(pg_wchar c, yyscan_t yyscanner);
+static void addunicode(char32_t c, yyscan_t yyscanner);
 
 #define yyerror(msg)  scanner_yyerror(msg, yyscanner)
 
@@ -640,7 +640,7 @@ other			.
 					addlit(yytext, yyleng, yyscanner);
 				}
 <xe>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/*
 					 * For consistency with other productions, issue any
@@ -668,7 +668,7 @@ other			.
 					POP_YYLLOC();
 				}
 <xeu>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/* Remember start of overall string token ... */
 					PUSH_YYLLOC();
@@ -1376,7 +1376,7 @@ process_integer_literal(const char *token, YYSTYPE *lval, int base)
 }
 
 static void
-addunicode(pg_wchar c, core_yyscan_t yyscanner)
+addunicode(char32_t c, core_yyscan_t yyscanner)
 {
 	ScannerCallbackState scbstate;
 	char		buf[MAX_UNICODE_EQUIVALENT_STRING + 1];
diff --git a/src/backend/utils/adt/jsonpath_scan.l b/src/backend/utils/adt/jsonpath_scan.l
index c7aab83eeb4..8c3a0a9c642 100644
--- a/src/backend/utils/adt/jsonpath_scan.l
+++ b/src/backend/utils/adt/jsonpath_scan.l
@@ -574,7 +574,7 @@ hexval(char c, int *result, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add given unicode character to scanstring */
 static bool
-addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
+addUnicodeChar(char32_t ch, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (ch == 0)
 	{
@@ -607,7 +607,7 @@ addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add unicode character, processing any surrogate pairs */
 static bool
-addUnicode(int ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
+addUnicode(char32_t ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (is_utf16_surrogate_first(ch))
 	{
@@ -655,7 +655,7 @@ parseUnicode(char *s, int l, struct Node *escontext, yyscan_t yyscanner)
 
 	for (i = 2; i < l; i += 2)	/* skip '\u' */
 	{
-		int			ch = 0;
+		char32_t		ch = 0;
 		int			j,
 					si;
 
diff --git a/src/backend/utils/adt/pg_locale_builtin.c b/src/backend/utils/adt/pg_locale_builtin.c
index 3dc611b50e1..1021e0d129b 100644
--- a/src/backend/utils/adt/pg_locale_builtin.c
+++ b/src/backend/utils/adt/pg_locale_builtin.c
@@ -15,7 +15,6 @@
 #include "catalog/pg_collation.h"
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
-#include "mb/pg_wchar.h"
 #include "miscadmin.h"
 #include "utils/builtins.h"
 #include "utils/pg_locale.h"
@@ -35,6 +34,23 @@ struct WordBoundaryState
 	bool		prev_alnum;
 };
 
+/*
+ * In UTF-8, pg_wchar is guaranteed to be the code point value.
+ */
+static inline char32_t
+to_char32(pg_wchar wc)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (char32_t) wc;
+}
+
+static inline pg_wchar
+to_pg_wchar(char32_t c32)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (pg_wchar) c32;
+}
+
 /*
  * Simple word boundary iterator that draws boundaries each time the result of
  * pg_u_isalnum() changes.
@@ -47,7 +63,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -112,61 +128,61 @@ strfold_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen,
 static bool
 wc_isdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isalpha_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalpha(wc);
+	return pg_u_isalpha(to_char32(wc));
 }
 
 static bool
 wc_isalnum_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalnum(wc, !locale->builtin.casemap_full);
+	return pg_u_isalnum(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isupper(wc);
+	return pg_u_isupper(to_char32(wc));
 }
 
 static bool
 wc_islower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_islower(wc);
+	return pg_u_islower(to_char32(wc));
 }
 
 static bool
 wc_isgraph_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isgraph(wc);
+	return pg_u_isgraph(to_char32(wc));
 }
 
 static bool
 wc_isprint_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isprint(wc);
+	return pg_u_isprint(to_char32(wc));
 }
 
 static bool
 wc_ispunct_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_ispunct(wc, !locale->builtin.casemap_full);
+	return pg_u_ispunct(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isspace_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isspace(wc);
+	return pg_u_isspace(to_char32(wc));
 }
 
 static bool
 wc_isxdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isxdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isxdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
@@ -179,13 +195,13 @@ char_is_cased_builtin(char ch, pg_locale_t locale)
 static pg_wchar
 wc_toupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_uppercase_simple(wc);
+	return to_pg_wchar(unicode_uppercase_simple(to_char32(wc)));
 }
 
 static pg_wchar
 wc_tolower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_lowercase_simple(wc);
+	return to_pg_wchar(unicode_lowercase_simple(to_char32(wc)));
 }
 
 static const struct ctype_methods ctype_methods_builtin = {
diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c
index 2c398cd9e5c..8d735786e51 100644
--- a/src/backend/utils/adt/varlena.c
+++ b/src/backend/utils/adt/varlena.c
@@ -5419,12 +5419,12 @@ unicode_assigned(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errmsg("Unicode categorization can only be performed if server encoding is UTF8")));
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (int i = 0; i < size; i++)
 	{
-		pg_wchar	uchar = utf8_to_unicode(p);
+		char32_t	uchar = utf8_to_unicode(p);
 		int			category = unicode_category(uchar);
 
 		if (category == PG_U_UNASSIGNED)
@@ -5443,24 +5443,24 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	text	   *result;
 	int			i;
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* action */
@@ -5468,7 +5468,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 
 	/* convert back to UTF-8 string */
 	size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unsigned char buf[4];
 
@@ -5480,7 +5480,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	SET_VARSIZE(result, size + VARHDRSZ);
 
 	p = (unsigned char *) VARDATA_ANY(result);
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unicode_to_utf8(*wp, p);
 		p += pg_utf_mblen(p);
@@ -5509,8 +5509,8 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	int			i;
 	UnicodeNormalizationQC quickcheck;
@@ -5519,16 +5519,16 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* quick check (see UAX #15) */
@@ -5542,11 +5542,11 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	output_chars = unicode_normalize(form, input_chars);
 
 	output_size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 		output_size++;
 
 	result = (size == output_size) &&
-		(memcmp(input_chars, output_chars, size * sizeof(pg_wchar)) == 0);
+		(memcmp(input_chars, output_chars, size * sizeof(char32_t)) == 0);
 
 	PG_RETURN_BOOL(result);
 }
@@ -5602,7 +5602,7 @@ unistr(PG_FUNCTION_ARGS)
 	int			len;
 	StringInfoData str;
 	text	   *result;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	char		cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
 
 	instr = VARDATA_ANY(input_text);
@@ -5626,7 +5626,7 @@ unistr(PG_FUNCTION_ARGS)
 			else if ((len >= 5 && isxdigits_n(instr + 1, 4)) ||
 					 (len >= 6 && instr[1] == 'u' && isxdigits_n(instr + 2, 4)))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 				int			offset = instr[1] == 'u' ? 2 : 1;
 
 				unicode = hexval_n(instr + offset, 4);
@@ -5662,7 +5662,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 8 && instr[1] == '+' && isxdigits_n(instr + 2, 6))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 6);
 
@@ -5697,7 +5697,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 10 && instr[1] == 'U' && isxdigits_n(instr + 2, 8))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 8);
 
diff --git a/src/backend/utils/mb/mbutils.c b/src/backend/utils/mb/mbutils.c
index 886ecbad871..fb629ed5c8f 100644
--- a/src/backend/utils/mb/mbutils.c
+++ b/src/backend/utils/mb/mbutils.c
@@ -862,7 +862,7 @@ perform_default_encoding_conversion(const char *src, int len,
  * may call this outside any transaction, or in an aborted transaction.
  */
 void
-pg_unicode_to_server(pg_wchar c, unsigned char *s)
+pg_unicode_to_server(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
@@ -924,7 +924,7 @@ pg_unicode_to_server(pg_wchar c, unsigned char *s)
  * but simply return false on conversion failure.
  */
 bool
-pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s)
+pg_unicode_to_server_noerror(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
diff --git a/src/common/saslprep.c b/src/common/saslprep.c
index 97beb47940b..101e8d65a4d 100644
--- a/src/common/saslprep.c
+++ b/src/common/saslprep.c
@@ -47,7 +47,7 @@
 
 /* Prototypes for local functions */
 static int	codepoint_range_cmp(const void *a, const void *b);
-static bool is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize);
+static bool is_code_in_table(char32_t code, const char32_t *map, int mapsize);
 static int	pg_utf8_string_len(const char *source);
 
 /*
@@ -64,7 +64,7 @@ static int	pg_utf8_string_len(const char *source);
  *
  * These are all mapped to the ASCII space character (U+00A0).
  */
-static const pg_wchar non_ascii_space_ranges[] =
+static const char32_t non_ascii_space_ranges[] =
 {
 	0x00A0, 0x00A0,
 	0x1680, 0x1680,
@@ -79,7 +79,7 @@ static const pg_wchar non_ascii_space_ranges[] =
  *
  * If any of these appear in the input, they are removed.
  */
-static const pg_wchar commonly_mapped_to_nothing_ranges[] =
+static const char32_t commonly_mapped_to_nothing_ranges[] =
 {
 	0x00AD, 0x00AD,
 	0x034F, 0x034F,
@@ -114,7 +114,7 @@ static const pg_wchar commonly_mapped_to_nothing_ranges[] =
  * tables, so one code might originate from multiple source tables.
  * Adjacent ranges have also been merged together, to save space.
  */
-static const pg_wchar prohibited_output_ranges[] =
+static const char32_t prohibited_output_ranges[] =
 {
 	0x0000, 0x001F,				/* C.2.1 */
 	0x007F, 0x00A0,				/* C.1.2, C.2.1, C.2.2 */
@@ -155,7 +155,7 @@ static const pg_wchar prohibited_output_ranges[] =
 };
 
 /* A.1 Unassigned code points in Unicode 3.2 */
-static const pg_wchar unassigned_codepoint_ranges[] =
+static const char32_t unassigned_codepoint_ranges[] =
 {
 	0x0221, 0x0221,
 	0x0234, 0x024F,
@@ -556,7 +556,7 @@ static const pg_wchar unassigned_codepoint_ranges[] =
 };
 
 /* D.1 Characters with bidirectional property "R" or "AL" */
-static const pg_wchar RandALCat_codepoint_ranges[] =
+static const char32_t RandALCat_codepoint_ranges[] =
 {
 	0x05BE, 0x05BE,
 	0x05C0, 0x05C0,
@@ -595,7 +595,7 @@ static const pg_wchar RandALCat_codepoint_ranges[] =
 };
 
 /* D.2 Characters with bidirectional property "L" */
-static const pg_wchar LCat_codepoint_ranges[] =
+static const char32_t LCat_codepoint_ranges[] =
 {
 	0x0041, 0x005A,
 	0x0061, 0x007A,
@@ -968,8 +968,8 @@ static const pg_wchar LCat_codepoint_ranges[] =
 static int
 codepoint_range_cmp(const void *a, const void *b)
 {
-	const pg_wchar *key = (const pg_wchar *) a;
-	const pg_wchar *range = (const pg_wchar *) b;
+	const char32_t *key = (const char32_t *) a;
+	const char32_t *range = (const char32_t *) b;
 
 	if (*key < range[0])
 		return -1;				/* less than lower bound */
@@ -980,14 +980,14 @@ codepoint_range_cmp(const void *a, const void *b)
 }
 
 static bool
-is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize)
+is_code_in_table(char32_t code, const char32_t *map, int mapsize)
 {
 	Assert(mapsize % 2 == 0);
 
 	if (code < map[0] || code > map[mapsize - 1])
 		return false;
 
-	if (bsearch(&code, map, mapsize / 2, sizeof(pg_wchar) * 2,
+	if (bsearch(&code, map, mapsize / 2, sizeof(char32_t) * 2,
 				codepoint_range_cmp))
 		return true;
 	else
@@ -1046,8 +1046,8 @@ pg_utf8_string_len(const char *source)
 pg_saslprep_rc
 pg_saslprep(const char *input, char **output)
 {
-	pg_wchar   *input_chars = NULL;
-	pg_wchar   *output_chars = NULL;
+	char32_t   *input_chars = NULL;
+	char32_t   *output_chars = NULL;
 	int			input_size;
 	char	   *result;
 	int			result_size;
@@ -1055,7 +1055,7 @@ pg_saslprep(const char *input, char **output)
 	int			i;
 	bool		contains_RandALCat;
 	unsigned char *p;
-	pg_wchar   *wp;
+	char32_t   *wp;
 
 	/* Ensure we return *output as NULL on failure */
 	*output = NULL;
@@ -1080,10 +1080,10 @@ pg_saslprep(const char *input, char **output)
 	input_size = pg_utf8_string_len(input);
 	if (input_size < 0)
 		return SASLPREP_INVALID_UTF8;
-	if (input_size >= MaxAllocSize / sizeof(pg_wchar))
+	if (input_size >= MaxAllocSize / sizeof(char32_t))
 		goto oom;
 
-	input_chars = ALLOC((input_size + 1) * sizeof(pg_wchar));
+	input_chars = ALLOC((input_size + 1) * sizeof(char32_t));
 	if (!input_chars)
 		goto oom;
 
@@ -1093,7 +1093,7 @@ pg_saslprep(const char *input, char **output)
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 
 	/*
 	 * The steps below correspond to the steps listed in [RFC3454], Section
@@ -1107,7 +1107,7 @@ pg_saslprep(const char *input, char **output)
 	count = 0;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, non_ascii_space_ranges))
 			input_chars[count++] = 0x0020;
@@ -1118,7 +1118,7 @@ pg_saslprep(const char *input, char **output)
 		else
 			input_chars[count++] = code;
 	}
-	input_chars[count] = (pg_wchar) '\0';
+	input_chars[count] = (char32_t) '\0';
 	input_size = count;
 
 	if (input_size == 0)
@@ -1138,7 +1138,7 @@ pg_saslprep(const char *input, char **output)
 	 */
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, prohibited_output_ranges))
 			goto prohibited;
@@ -1170,7 +1170,7 @@ pg_saslprep(const char *input, char **output)
 	contains_RandALCat = false;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, RandALCat_codepoint_ranges))
 		{
@@ -1181,12 +1181,12 @@ pg_saslprep(const char *input, char **output)
 
 	if (contains_RandALCat)
 	{
-		pg_wchar	first = input_chars[0];
-		pg_wchar	last = input_chars[input_size - 1];
+		char32_t	first = input_chars[0];
+		char32_t	last = input_chars[input_size - 1];
 
 		for (i = 0; i < input_size; i++)
 		{
-			pg_wchar	code = input_chars[i];
+			char32_t	code = input_chars[i];
 
 			if (IS_CODE_IN_TABLE(code, LCat_codepoint_ranges))
 				goto prohibited;
diff --git a/src/common/unicode/case_test.c b/src/common/unicode/case_test.c
index fdfb62e8552..00d4f85e5a5 100644
--- a/src/common/unicode/case_test.c
+++ b/src/common/unicode/case_test.c
@@ -24,6 +24,7 @@
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 /* enough to hold largest source or result string, including NUL */
 #define BUFSZ 256
@@ -54,7 +55,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -77,16 +78,16 @@ initcap_wbnext(void *state)
 #ifdef USE_ICU
 
 static void
-icu_test_simple(pg_wchar code)
+icu_test_simple(char32_t code)
 {
-	pg_wchar	lower = unicode_lowercase_simple(code);
-	pg_wchar	title = unicode_titlecase_simple(code);
-	pg_wchar	upper = unicode_uppercase_simple(code);
-	pg_wchar	fold = unicode_casefold_simple(code);
-	pg_wchar	iculower = u_tolower(code);
-	pg_wchar	icutitle = u_totitle(code);
-	pg_wchar	icuupper = u_toupper(code);
-	pg_wchar	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
+	char32_t	lower = unicode_lowercase_simple(code);
+	char32_t	title = unicode_titlecase_simple(code);
+	char32_t	upper = unicode_uppercase_simple(code);
+	char32_t	fold = unicode_casefold_simple(code);
+	char32_t	iculower = u_tolower(code);
+	char32_t	icutitle = u_totitle(code);
+	char32_t	icuupper = u_toupper(code);
+	char32_t	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
 
 	if (lower != iculower || title != icutitle || upper != icuupper ||
 		fold != icufold)
@@ -172,7 +173,7 @@ test_icu(void)
 	int			successful = 0;
 	int			skipped_mismatch = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		pg_unicode_category category = unicode_category(code);
 
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index 5d37ba39196..1e8c1f7905f 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -22,6 +22,7 @@
 
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 static int	pg_unicode_version = 0;
 #ifdef USE_ICU
@@ -59,7 +60,7 @@ icu_test()
 	int			pg_skipped_codepoints = 0;
 	int			icu_skipped_codepoints = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		uint8_t		pg_category = unicode_category(code);
 		uint8_t		icu_category = u_charType(code);
diff --git a/src/common/unicode/generate-norm_test_table.pl b/src/common/unicode/generate-norm_test_table.pl
index 1b401be9409..1a8b908ff33 100644
--- a/src/common/unicode/generate-norm_test_table.pl
+++ b/src/common/unicode/generate-norm_test_table.pl
@@ -47,8 +47,8 @@ print $OUTPUT <<HEADER;
 typedef struct
 {
 	int			linenum;
-	pg_wchar	input[50];
-	pg_wchar	output[4][50];
+	char32_t	input[50];
+	char32_t	output[4][50];
 } pg_unicode_test;
 
 /* test table */
diff --git a/src/common/unicode/generate-unicode_case_table.pl b/src/common/unicode/generate-unicode_case_table.pl
index 5d9ddd62803..f71eb25c94e 100644
--- a/src/common/unicode/generate-unicode_case_table.pl
+++ b/src/common/unicode/generate-unicode_case_table.pl
@@ -270,7 +270,6 @@ print $OT <<"EOS";
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -297,7 +296,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -430,7 +429,7 @@ foreach my $kind ('lower', 'title', 'upper', 'fold')
  * The entry case_map_${kind}[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_$kind\[$index\] =
+static const char32_t case_map_$kind\[$index\] =
 {
 EOS
 
@@ -502,7 +501,7 @@ print $OT <<"EOS";
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < $fastpath_limit */
 	if (cp < $fastpath_limit)
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index abab5cd9696..7e094b13720 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -366,15 +366,15 @@ print $OT <<"EOS";
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/common/unicode/norm_test.c b/src/common/unicode/norm_test.c
index 25bc59463f2..058817f1719 100644
--- a/src/common/unicode/norm_test.c
+++ b/src/common/unicode/norm_test.c
@@ -20,7 +20,7 @@
 #include "norm_test_table.h"
 
 static char *
-print_wchar_str(const pg_wchar *s)
+print_wchar_str(const char32_t *s)
 {
 #define BUF_DIGITS 50
 	static char buf[BUF_DIGITS * 11 + 1];
@@ -41,7 +41,7 @@ print_wchar_str(const pg_wchar *s)
 }
 
 static int
-pg_wcscmp(const pg_wchar *s1, const pg_wchar *s2)
+pg_wcscmp(const char32_t *s1, const char32_t *s2)
 {
 	for (;;)
 	{
@@ -65,7 +65,7 @@ main(int argc, char **argv)
 	{
 		for (int form = 0; form < 4; form++)
 		{
-			pg_wchar   *result;
+			char32_t   *result;
 
 			result = unicode_normalize(form, test->input);
 
diff --git a/src/common/unicode_case.c b/src/common/unicode_case.c
index 073faf6a0d5..e5e494db43c 100644
--- a/src/common/unicode_case.c
+++ b/src/common/unicode_case.c
@@ -30,7 +30,7 @@ enum CaseMapResult
 /*
  * Map for each case kind.
  */
-static const pg_wchar *const casekind_map[NCaseKind] =
+static const char32_t *const casekind_map[NCaseKind] =
 {
 	[CaseLower] = case_map_lower,
 	[CaseTitle] = case_map_title,
@@ -38,42 +38,42 @@ static const pg_wchar *const casekind_map[NCaseKind] =
 	[CaseFold] = case_map_fold,
 };
 
-static pg_wchar find_case_map(pg_wchar ucs, const pg_wchar *map);
+static char32_t find_case_map(char32_t ucs, const char32_t *map);
 static size_t convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 						   CaseKind str_casekind, bool full, WordBoundaryNext wbnext,
 						   void *wbstate);
-static enum CaseMapResult casemap(pg_wchar u1, CaseKind casekind, bool full,
+static enum CaseMapResult casemap(char32_t u1, CaseKind casekind, bool full,
 								  const char *src, size_t srclen, size_t srcoff,
-								  pg_wchar *simple, const pg_wchar **special);
+								  char32_t *simple, const char32_t **special);
 
-pg_wchar
-unicode_lowercase_simple(pg_wchar code)
+char32_t
+unicode_lowercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_lower);
+	char32_t	cp = find_case_map(code, case_map_lower);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_titlecase_simple(pg_wchar code)
+char32_t
+unicode_titlecase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_title);
+	char32_t	cp = find_case_map(code, case_map_title);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_uppercase_simple(pg_wchar code)
+char32_t
+unicode_uppercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_upper);
+	char32_t	cp = find_case_map(code, case_map_upper);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_casefold_simple(pg_wchar code)
+char32_t
+unicode_casefold_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_fold);
+	char32_t	cp = find_case_map(code, case_map_fold);
 
 	return cp != 0 ? cp : code;
 }
@@ -231,10 +231,10 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 
 	while ((srclen < 0 || srcoff < srclen) && src[srcoff] != '\0')
 	{
-		pg_wchar	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
+		char32_t	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
 		int			u1len = unicode_utf8len(u1);
-		pg_wchar	simple = 0;
-		const pg_wchar *special = NULL;
+		char32_t	simple = 0;
+		const char32_t *special = NULL;
 		enum CaseMapResult casemap_result;
 
 		if (str_casekind == CaseTitle)
@@ -265,8 +265,8 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 			case CASEMAP_SIMPLE:
 				{
 					/* replace with single character */
-					pg_wchar	u2 = simple;
-					pg_wchar	u2len = unicode_utf8len(u2);
+					char32_t	u2 = simple;
+					char32_t	u2len = unicode_utf8len(u2);
 
 					Assert(special == NULL);
 					if (result_len + u2len <= dstsize)
@@ -280,7 +280,7 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 				Assert(simple == 0);
 				for (int i = 0; i < MAX_CASE_EXPANSION && special[i]; i++)
 				{
-					pg_wchar	u2 = special[i];
+					char32_t	u2 = special[i];
 					size_t		u2len = unicode_utf8len(u2);
 
 					if (result_len + u2len <= dstsize)
@@ -320,7 +320,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -344,7 +344,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -394,9 +394,9 @@ check_special_conditions(int conditions, const char *str, size_t len,
  * character without modification.
  */
 static enum CaseMapResult
-casemap(pg_wchar u1, CaseKind casekind, bool full,
+casemap(char32_t u1, CaseKind casekind, bool full,
 		const char *src, size_t srclen, size_t srcoff,
-		pg_wchar *simple, const pg_wchar **special)
+		char32_t *simple, const char32_t **special)
 {
 	uint16		idx;
 
@@ -434,8 +434,8 @@ casemap(pg_wchar u1, CaseKind casekind, bool full,
  * Find entry in simple case map.
  * If the entry does not exist, 0 will be returned.
  */
-static pg_wchar
-find_case_map(pg_wchar ucs, const pg_wchar *map)
+static char32_t
+find_case_map(char32_t ucs, const char32_t *map)
 {
 	/* Fast path for codepoints < 0x80 */
 	if (ucs < 0x80)
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index 4136c4d4f92..aab667a7bb4 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -1,7 +1,7 @@
 /*-------------------------------------------------------------------------
  * unicode_category.c
  *		Determine general category and character properties of Unicode
- *		characters. Encoding must be UTF8, where we assume that the pg_wchar
+ *		characters. Encoding must be UTF8, where we assume that the char32_t
  *		representation is a code point.
  *
  * Portions Copyright (c) 2017-2025, PostgreSQL Global Development Group
@@ -76,13 +76,13 @@
 #define PG_U_CHARACTER_TAB	0x09
 
 static bool range_search(const pg_unicode_range *tbl, size_t size,
-						 pg_wchar code);
+						 char32_t code);
 
 /*
  * Unicode general category for the given codepoint.
  */
 pg_unicode_category
-unicode_category(pg_wchar code)
+unicode_category(char32_t code)
 {
 	int			min = 0;
 	int			mid;
@@ -108,7 +108,7 @@ unicode_category(pg_wchar code)
 }
 
 bool
-pg_u_prop_alphabetic(pg_wchar code)
+pg_u_prop_alphabetic(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_ALPHABETIC;
@@ -119,7 +119,7 @@ pg_u_prop_alphabetic(pg_wchar code)
 }
 
 bool
-pg_u_prop_lowercase(pg_wchar code)
+pg_u_prop_lowercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_LOWERCASE;
@@ -130,7 +130,7 @@ pg_u_prop_lowercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_uppercase(pg_wchar code)
+pg_u_prop_uppercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_UPPERCASE;
@@ -141,7 +141,7 @@ pg_u_prop_uppercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_cased(pg_wchar code)
+pg_u_prop_cased(char32_t code)
 {
 	uint32		category_mask;
 
@@ -156,7 +156,7 @@ pg_u_prop_cased(pg_wchar code)
 }
 
 bool
-pg_u_prop_case_ignorable(pg_wchar code)
+pg_u_prop_case_ignorable(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_CASE_IGNORABLE;
@@ -167,7 +167,7 @@ pg_u_prop_case_ignorable(pg_wchar code)
 }
 
 bool
-pg_u_prop_white_space(pg_wchar code)
+pg_u_prop_white_space(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_WHITE_SPACE;
@@ -178,7 +178,7 @@ pg_u_prop_white_space(pg_wchar code)
 }
 
 bool
-pg_u_prop_hex_digit(pg_wchar code)
+pg_u_prop_hex_digit(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_HEX_DIGIT;
@@ -189,7 +189,7 @@ pg_u_prop_hex_digit(pg_wchar code)
 }
 
 bool
-pg_u_prop_join_control(pg_wchar code)
+pg_u_prop_join_control(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_JOIN_CONTROL;
@@ -208,7 +208,7 @@ pg_u_prop_join_control(pg_wchar code)
  */
 
 bool
-pg_u_isdigit(pg_wchar code, bool posix)
+pg_u_isdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return ('0' <= code && code <= '9');
@@ -217,19 +217,19 @@ pg_u_isdigit(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isalpha(pg_wchar code)
+pg_u_isalpha(char32_t code)
 {
 	return pg_u_prop_alphabetic(code);
 }
 
 bool
-pg_u_isalnum(pg_wchar code, bool posix)
+pg_u_isalnum(char32_t code, bool posix)
 {
 	return pg_u_isalpha(code) || pg_u_isdigit(code, posix);
 }
 
 bool
-pg_u_isword(pg_wchar code)
+pg_u_isword(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -240,32 +240,32 @@ pg_u_isword(pg_wchar code)
 }
 
 bool
-pg_u_isupper(pg_wchar code)
+pg_u_isupper(char32_t code)
 {
 	return pg_u_prop_uppercase(code);
 }
 
 bool
-pg_u_islower(pg_wchar code)
+pg_u_islower(char32_t code)
 {
 	return pg_u_prop_lowercase(code);
 }
 
 bool
-pg_u_isblank(pg_wchar code)
+pg_u_isblank(char32_t code)
 {
 	return code == PG_U_CHARACTER_TAB ||
 		unicode_category(code) == PG_U_SPACE_SEPARATOR;
 }
 
 bool
-pg_u_iscntrl(pg_wchar code)
+pg_u_iscntrl(char32_t code)
 {
 	return unicode_category(code) == PG_U_CONTROL;
 }
 
 bool
-pg_u_isgraph(pg_wchar code)
+pg_u_isgraph(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -276,7 +276,7 @@ pg_u_isgraph(pg_wchar code)
 }
 
 bool
-pg_u_isprint(pg_wchar code)
+pg_u_isprint(char32_t code)
 {
 	pg_unicode_category category = unicode_category(code);
 
@@ -287,7 +287,7 @@ pg_u_isprint(pg_wchar code)
 }
 
 bool
-pg_u_ispunct(pg_wchar code, bool posix)
+pg_u_ispunct(char32_t code, bool posix)
 {
 	uint32		category_mask;
 
@@ -308,13 +308,13 @@ pg_u_ispunct(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isspace(pg_wchar code)
+pg_u_isspace(char32_t code)
 {
 	return pg_u_prop_white_space(code);
 }
 
 bool
-pg_u_isxdigit(pg_wchar code, bool posix)
+pg_u_isxdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return (('0' <= code && code <= '9') ||
@@ -478,7 +478,7 @@ unicode_category_abbrev(pg_unicode_category category)
  * given table.
  */
 static bool
-range_search(const pg_unicode_range *tbl, size_t size, pg_wchar code)
+range_search(const pg_unicode_range *tbl, size_t size, char32_t code)
 {
 	int			min = 0;
 	int			mid;
diff --git a/src/common/unicode_norm.c b/src/common/unicode_norm.c
index 6654b4cbc49..489d99cd5ab 100644
--- a/src/common/unicode_norm.c
+++ b/src/common/unicode_norm.c
@@ -69,7 +69,7 @@ conv_compare(const void *p1, const void *p2)
  * lookup, while the frontend version uses a binary search.
  */
 static const pg_unicode_decomposition *
-get_code_entry(pg_wchar code)
+get_code_entry(char32_t code)
 {
 #ifndef FRONTEND
 	int			h;
@@ -109,7 +109,7 @@ get_code_entry(pg_wchar code)
  * Get the combining class of the given codepoint.
  */
 static uint8
-get_canonical_class(pg_wchar code)
+get_canonical_class(char32_t code)
 {
 	const pg_unicode_decomposition *entry = get_code_entry(code);
 
@@ -130,15 +130,15 @@ get_canonical_class(pg_wchar code)
  * Note: the returned pointer can point to statically allocated buffer, and
  * is only valid until next call to this function!
  */
-static const pg_wchar *
+static const char32_t *
 get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
 {
-	static pg_wchar x;
+	static char32_t x;
 
 	if (DECOMPOSITION_IS_INLINE(entry))
 	{
 		Assert(DECOMPOSITION_SIZE(entry) == 1);
-		x = (pg_wchar) entry->dec_index;
+		x = (char32_t) entry->dec_index;
 		*dec_size = 1;
 		return &x;
 	}
@@ -156,7 +156,7 @@ get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
  * are, in turn, decomposable.
  */
 static int
-get_decomposed_size(pg_wchar code, bool compat)
+get_decomposed_size(char32_t code, bool compat)
 {
 	const pg_unicode_decomposition *entry;
 	int			size = 0;
@@ -318,7 +318,7 @@ recompose_code(uint32 start, uint32 code, uint32 *result)
  * in the array result.
  */
 static void
-decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
+decompose_code(char32_t code, bool compat, char32_t **result, int *current)
 {
 	const pg_unicode_decomposition *entry;
 	int			i;
@@ -337,7 +337,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 					v,
 					tindex,
 					sindex;
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		sindex = code - SBASE;
 		l = LBASE + sindex / (VCOUNT * TCOUNT);
@@ -369,7 +369,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	if (entry == NULL || DECOMPOSITION_SIZE(entry) == 0 ||
 		(!compat && DECOMPOSITION_IS_COMPAT(entry)))
 	{
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		res[*current] = code;
 		(*current)++;
@@ -382,7 +382,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	decomp = get_code_decomposition(entry, &dec_size);
 	for (i = 0; i < dec_size; i++)
 	{
-		pg_wchar	lcode = (pg_wchar) decomp[i];
+		char32_t	lcode = (char32_t) decomp[i];
 
 		/* Leave if no more decompositions */
 		decompose_code(lcode, compat, result, current);
@@ -398,17 +398,17 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
  * malloc. Or NULL if we run out of memory. In backend, the returned
  * string is palloc'd instead, and OOM is reported with ereport().
  */
-pg_wchar *
-unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
+char32_t *
+unicode_normalize(UnicodeNormalizationForm form, const char32_t *input)
 {
 	bool		compat = (form == UNICODE_NFKC || form == UNICODE_NFKD);
 	bool		recompose = (form == UNICODE_NFC || form == UNICODE_NFKC);
-	pg_wchar   *decomp_chars;
-	pg_wchar   *recomp_chars;
+	char32_t   *decomp_chars;
+	char32_t   *recomp_chars;
 	int			decomp_size,
 				current_size;
 	int			count;
-	const pg_wchar *p;
+	const char32_t *p;
 
 	/* variables for recomposition */
 	int			last_class;
@@ -425,7 +425,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	for (p = input; *p; p++)
 		decomp_size += get_decomposed_size(*p, compat);
 
-	decomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	decomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (decomp_chars == NULL)
 		return NULL;
 
@@ -448,9 +448,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 */
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	prev = decomp_chars[count - 1];
-		pg_wchar	next = decomp_chars[count];
-		pg_wchar	tmp;
+		char32_t	prev = decomp_chars[count - 1];
+		char32_t	next = decomp_chars[count];
+		char32_t	tmp;
 		const uint8 prevClass = get_canonical_class(prev);
 		const uint8 nextClass = get_canonical_class(next);
 
@@ -487,7 +487,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 * longer than the decomposed one, so make the allocation of the output
 	 * string based on that assumption.
 	 */
-	recomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	recomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (!recomp_chars)
 	{
 		FREE(decomp_chars);
@@ -501,9 +501,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	ch = decomp_chars[count];
+		char32_t	ch = decomp_chars[count];
 		int			ch_class = get_canonical_class(ch);
-		pg_wchar	composite;
+		char32_t	composite;
 
 		if (last_class < ch_class &&
 			recompose_code(starter_ch, ch, &composite))
@@ -524,7 +524,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 			recomp_chars[target_pos++] = ch;
 		}
 	}
-	recomp_chars[target_pos] = (pg_wchar) '\0';
+	recomp_chars[target_pos] = (char32_t) '\0';
 
 	FREE(decomp_chars);
 
@@ -540,7 +540,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 #ifndef FRONTEND
 
 static const pg_unicode_normprops *
-qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
+qc_hash_lookup(char32_t ch, const pg_unicode_norminfo *norminfo)
 {
 	int			h;
 	uint32		hashkey;
@@ -571,7 +571,7 @@ qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
  * Look up the normalization quick check character property
  */
 static UnicodeNormalizationQC
-qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
+qc_is_allowed(UnicodeNormalizationForm form, char32_t ch)
 {
 	const pg_unicode_normprops *found = NULL;
 
@@ -595,7 +595,7 @@ qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
 }
 
 UnicodeNormalizationQC
-unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input)
+unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input)
 {
 	uint8		lastCanonicalClass = 0;
 	UnicodeNormalizationQC result = UNICODE_NORM_QC_YES;
@@ -610,9 +610,9 @@ unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *
 	if (form == UNICODE_NFD || form == UNICODE_NFKD)
 		return UNICODE_NORM_QC_MAYBE;
 
-	for (const pg_wchar *p = input; *p; p++)
+	for (const char32_t *p = input; *p; p++)
 	{
-		pg_wchar	ch = *p;
+		char32_t	ch = *p;
 		uint8		canonicalClass;
 		UnicodeNormalizationQC check;
 
diff --git a/src/fe_utils/mbprint.c b/src/fe_utils/mbprint.c
index eb3eeee9925..abffdbe18a2 100644
--- a/src/fe_utils/mbprint.c
+++ b/src/fe_utils/mbprint.c
@@ -49,20 +49,20 @@ pg_get_utf8_id(void)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static pg_wchar
+static char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
diff --git a/src/include/c.h b/src/include/c.h
index 9ab5e617995..54fa20b0e83 100644
--- a/src/include/c.h
+++ b/src/include/c.h
@@ -513,6 +513,25 @@ typedef void (*pg_funcptr_t) (void);
 
 #include <stdbool.h>
 
+/*
+ * char16_t and char32_t
+ *      Unicode code points.
+ */
+#ifndef __cplusplus
+#ifdef HAVE_UCHAR_H
+#include <uchar.h>
+#ifndef __STDC_UTF_16__
+#error "char16_t must use UTF-16 encoding"
+#endif
+#ifndef __STDC_UTF_32__
+#error "char32_t must use UTF-32 encoding"
+#endif
+#else
+typedef uint16_t char16_t;
+typedef uint32_t char32_t;
+#endif
+#endif
+
 
 /* ----------------------------------------------------------------
  *				Section 3:	standard system types
diff --git a/src/include/common/unicode_case.h b/src/include/common/unicode_case.h
index 41e2c1f4b33..6bcffd349c2 100644
--- a/src/include/common/unicode_case.h
+++ b/src/include/common/unicode_case.h
@@ -14,14 +14,12 @@
 #ifndef UNICODE_CASE_H
 #define UNICODE_CASE_H
 
-#include "mb/pg_wchar.h"
-
 typedef size_t (*WordBoundaryNext) (void *wbstate);
 
-pg_wchar	unicode_lowercase_simple(pg_wchar code);
-pg_wchar	unicode_titlecase_simple(pg_wchar code);
-pg_wchar	unicode_uppercase_simple(pg_wchar code);
-pg_wchar	unicode_casefold_simple(pg_wchar code);
+char32_t	unicode_lowercase_simple(char32_t code);
+char32_t	unicode_titlecase_simple(char32_t code);
+char32_t	unicode_uppercase_simple(char32_t code);
+char32_t	unicode_casefold_simple(char32_t code);
 size_t		unicode_strlower(char *dst, size_t dstsize, const char *src,
 							 ssize_t srclen, bool full);
 size_t		unicode_strtitle(char *dst, size_t dstsize, const char *src,
diff --git a/src/include/common/unicode_case_table.h b/src/include/common/unicode_case_table.h
index d5311786582..0a14fb2d97b 100644
--- a/src/include/common/unicode_case_table.h
+++ b/src/include/common/unicode_case_table.h
@@ -18,7 +18,6 @@
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -45,7 +44,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -166,7 +165,7 @@ static const pg_special_case special_case[106] =
  * The entry case_map_lower[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_lower[1704] =
+static const char32_t case_map_lower[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -1879,7 +1878,7 @@ static const pg_wchar case_map_lower[1704] =
  * The entry case_map_title[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_title[1704] =
+static const char32_t case_map_title[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -3592,7 +3591,7 @@ static const pg_wchar case_map_title[1704] =
  * The entry case_map_upper[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_upper[1704] =
+static const char32_t case_map_upper[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -5305,7 +5304,7 @@ static const pg_wchar case_map_upper[1704] =
  * The entry case_map_fold[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_fold[1704] =
+static const char32_t case_map_fold[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -13522,7 +13521,7 @@ static const uint16 case_map[4778] =
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < 0x0588 */
 	if (cp < 0x0588)
diff --git a/src/include/common/unicode_category.h b/src/include/common/unicode_category.h
index 8fd8b67a416..684143d3c8a 100644
--- a/src/include/common/unicode_category.h
+++ b/src/include/common/unicode_category.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_CATEGORY_H
 #define UNICODE_CATEGORY_H
 
-#include "mb/pg_wchar.h"
-
 /*
  * Unicode General Category Values
  *
@@ -61,31 +59,31 @@ typedef enum pg_unicode_category
 	PG_U_FINAL_PUNCTUATION = 29 /* Pf */
 } pg_unicode_category;
 
-extern pg_unicode_category unicode_category(pg_wchar code);
+extern pg_unicode_category unicode_category(char32_t code);
 extern const char *unicode_category_string(pg_unicode_category category);
 extern const char *unicode_category_abbrev(pg_unicode_category category);
 
-extern bool pg_u_prop_alphabetic(pg_wchar code);
-extern bool pg_u_prop_lowercase(pg_wchar code);
-extern bool pg_u_prop_uppercase(pg_wchar code);
-extern bool pg_u_prop_cased(pg_wchar code);
-extern bool pg_u_prop_case_ignorable(pg_wchar code);
-extern bool pg_u_prop_white_space(pg_wchar code);
-extern bool pg_u_prop_hex_digit(pg_wchar code);
-extern bool pg_u_prop_join_control(pg_wchar code);
+extern bool pg_u_prop_alphabetic(char32_t code);
+extern bool pg_u_prop_lowercase(char32_t code);
+extern bool pg_u_prop_uppercase(char32_t code);
+extern bool pg_u_prop_cased(char32_t code);
+extern bool pg_u_prop_case_ignorable(char32_t code);
+extern bool pg_u_prop_white_space(char32_t code);
+extern bool pg_u_prop_hex_digit(char32_t code);
+extern bool pg_u_prop_join_control(char32_t code);
 
-extern bool pg_u_isdigit(pg_wchar code, bool posix);
-extern bool pg_u_isalpha(pg_wchar code);
-extern bool pg_u_isalnum(pg_wchar code, bool posix);
-extern bool pg_u_isword(pg_wchar code);
-extern bool pg_u_isupper(pg_wchar code);
-extern bool pg_u_islower(pg_wchar code);
-extern bool pg_u_isblank(pg_wchar code);
-extern bool pg_u_iscntrl(pg_wchar code);
-extern bool pg_u_isgraph(pg_wchar code);
-extern bool pg_u_isprint(pg_wchar code);
-extern bool pg_u_ispunct(pg_wchar code, bool posix);
-extern bool pg_u_isspace(pg_wchar code);
-extern bool pg_u_isxdigit(pg_wchar code, bool posix);
+extern bool pg_u_isdigit(char32_t code, bool posix);
+extern bool pg_u_isalpha(char32_t code);
+extern bool pg_u_isalnum(char32_t code, bool posix);
+extern bool pg_u_isword(char32_t code);
+extern bool pg_u_isupper(char32_t code);
+extern bool pg_u_islower(char32_t code);
+extern bool pg_u_isblank(char32_t code);
+extern bool pg_u_iscntrl(char32_t code);
+extern bool pg_u_isgraph(char32_t code);
+extern bool pg_u_isprint(char32_t code);
+extern bool pg_u_ispunct(char32_t code, bool posix);
+extern bool pg_u_isspace(char32_t code);
+extern bool pg_u_isxdigit(char32_t code, bool posix);
 
 #endif							/* UNICODE_CATEGORY_H */
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 95a1c65da7e..466a41b72b0 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -20,15 +20,15 @@
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/include/common/unicode_norm.h b/src/include/common/unicode_norm.h
index 5bc3b79e78e..516c192cc4c 100644
--- a/src/include/common/unicode_norm.h
+++ b/src/include/common/unicode_norm.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_NORM_H
 #define UNICODE_NORM_H
 
-#include "mb/pg_wchar.h"
-
 typedef enum
 {
 	UNICODE_NFC = 0,
@@ -32,8 +30,8 @@ typedef enum
 	UNICODE_NORM_QC_MAYBE = -1,
 } UnicodeNormalizationQC;
 
-extern pg_wchar *unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input);
+extern char32_t *unicode_normalize(UnicodeNormalizationForm form, const char32_t *input);
 
-extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input);
+extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input);
 
 #endif							/* UNICODE_NORM_H */
diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h
index 4b4a9974b75..4d84bdc81e4 100644
--- a/src/include/mb/pg_wchar.h
+++ b/src/include/mb/pg_wchar.h
@@ -532,25 +532,25 @@ typedef uint32 (*utf_local_conversion_func) (uint32 code);
  * Some handy functions for Unicode-specific tests.
  */
 static inline bool
-is_valid_unicode_codepoint(pg_wchar c)
+is_valid_unicode_codepoint(char32_t c)
 {
 	return (c > 0 && c <= 0x10FFFF);
 }
 
 static inline bool
-is_utf16_surrogate_first(pg_wchar c)
+is_utf16_surrogate_first(char32_t c)
 {
 	return (c >= 0xD800 && c <= 0xDBFF);
 }
 
 static inline bool
-is_utf16_surrogate_second(pg_wchar c)
+is_utf16_surrogate_second(char32_t c)
 {
 	return (c >= 0xDC00 && c <= 0xDFFF);
 }
 
-static inline pg_wchar
-surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
+static inline char32_t
+surrogate_pair_to_codepoint(char16_t first, char16_t second)
 {
 	return ((first & 0x3FF) << 10) + 0x10000 + (second & 0x3FF);
 }
@@ -561,20 +561,20 @@ surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static inline pg_wchar
+static inline char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
@@ -588,7 +588,7 @@ utf8_to_unicode(const unsigned char *c)
  * unicode_utf8len(c) bytes available.
  */
 static inline unsigned char *
-unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
+unicode_to_utf8(char32_t c, unsigned char *utf8string)
 {
 	if (c <= 0x7F)
 	{
@@ -620,7 +620,7 @@ unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
  * Number of bytes needed to represent the given char in UTF8.
  */
 static inline int
-unicode_utf8len(pg_wchar c)
+unicode_utf8len(char32_t c)
 {
 	if (c <= 0x7F)
 		return 1;
@@ -676,8 +676,8 @@ extern int	pg_valid_server_encoding(const char *name);
 extern bool is_encoding_supported_by_icu(int encoding);
 extern const char *get_encoding_name_for_icu(int encoding);
 
-extern unsigned char *unicode_to_utf8(pg_wchar c, unsigned char *utf8string);
-extern pg_wchar utf8_to_unicode(const unsigned char *c);
+extern unsigned char *unicode_to_utf8(char32_t c, unsigned char *utf8string);
+extern char32_t utf8_to_unicode(const unsigned char *c);
 extern bool pg_utf8_islegal(const unsigned char *source, int length);
 extern int	pg_utf_mblen(const unsigned char *s);
 extern int	pg_mule_mblen(const unsigned char *s);
@@ -739,8 +739,8 @@ extern char *pg_server_to_client(const char *s, int len);
 extern char *pg_any_to_server(const char *s, int len, int encoding);
 extern char *pg_server_to_any(const char *s, int len, int encoding);
 
-extern void pg_unicode_to_server(pg_wchar c, unsigned char *s);
-extern bool pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s);
+extern void pg_unicode_to_server(char32_t c, unsigned char *s);
+extern bool pg_unicode_to_server_noerror(char32_t c, unsigned char *s);
 
 extern unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc);
 extern unsigned short CNStoBIG5(unsigned short cns, unsigned char lc);
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index 43fe3bcd593..ce382c546e2 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -3506,6 +3506,8 @@ cb_cleanup_dir
 cb_options
 cb_tablespace
 cb_tablespace_mapping
+char16_t
+char32_t
 check_agg_arguments_context
 check_function_callback
 check_network_data
-- 
2.43.0

From ef398d17fe36afdecc55390cf0f35d24a57e7fdf Mon Sep 17 00:00:00 2001
From: Jeff Davis <jeff@j-davis.com>
Date: Tue, 21 Oct 2025 13:16:47 -0700
Subject: [PATCH v3] Use C11 char16_t and char32_t for Unicode code points.

Reviewed-by: Tatsuo Ishii <ishii@postgresql.org>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Reviewed-by: Peter Eisentraut <peter@eisentraut.org>
Discussion: https://postgr.es/m/bedcc93d06203dfd89815b10f815ca2de8626e85.camel%40j-davis.com
---
 configure                                     | 22 +++++++-
 configure.ac                                  |  3 +
 meson.build                                   | 11 ++++
 src/backend/parser/parser.c                   |  8 +--
 src/backend/parser/scan.l                     |  8 +--
 src/backend/utils/adt/jsonpath_scan.l         |  6 +-
 src/backend/utils/adt/pg_locale_builtin.c     | 44 ++++++++++-----
 src/backend/utils/adt/varlena.c               | 40 ++++++-------
 src/backend/utils/mb/mbutils.c                |  4 +-
 src/common/saslprep.c                         | 48 ++++++++--------
 src/common/unicode/case_test.c                | 23 ++++----
 src/common/unicode/category_test.c            |  3 +-
 .../unicode/generate-norm_test_table.pl       |  4 +-
 .../unicode/generate-unicode_case_table.pl    |  7 +--
 .../generate-unicode_category_table.pl        |  8 +--
 src/common/unicode/norm_test.c                |  6 +-
 src/common/unicode_case.c                     | 56 +++++++++----------
 src/common/unicode_category.c                 | 50 ++++++++---------
 src/common/unicode_norm.c                     | 56 +++++++++----------
 src/fe_utils/mbprint.c                        | 10 ++--
 src/include/c.h                               | 13 +++++
 src/include/common/unicode_case.h             | 10 ++--
 src/include/common/unicode_case_table.h       | 13 ++---
 src/include/common/unicode_category.h         | 46 ++++++++-------
 src/include/common/unicode_category_table.h   |  8 +--
 src/include/common/unicode_norm.h             |  6 +-
 src/include/mb/pg_wchar.h                     | 32 +++++------
 src/include/pg_config.h.in                    |  9 +++
 src/tools/pgindent/typedefs.list              |  2 +
 29 files changed, 312 insertions(+), 244 deletions(-)

diff --git a/configure b/configure
index 22cd866147b..0f03b92ee37 100755
--- a/configure
+++ b/configure
@@ -13627,7 +13627,7 @@ fi
 ## Header files
 ##
 
-for ac_header in atomic.h copyfile.h execinfo.h getopt.h ifaddrs.h mbarrier.h sys/epoll.h sys/event.h sys/personality.h sys/prctl.h sys/procctl.h sys/signalfd.h sys/ucred.h termios.h ucred.h xlocale.h
+for ac_header in atomic.h copyfile.h execinfo.h getopt.h ifaddrs.h mbarrier.h sys/epoll.h sys/event.h sys/personality.h sys/prctl.h sys/procctl.h sys/signalfd.h sys/ucred.h termios.h uchar.h ucred.h xlocale.h
 do :
   as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
 ac_fn_c_check_header_mongrel "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default"
@@ -14996,6 +14996,26 @@ cat >>confdefs.h <<_ACEOF
 _ACEOF
 
 
+fi
+
+ac_fn_c_check_type "$LINENO" "char16_t" "ac_cv_type_char16_t" "$ac_includes_default"
+if test "x$ac_cv_type_char16_t" = xyes; then :
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_CHAR16_T 1
+_ACEOF
+
+
+fi
+
+ac_fn_c_check_type "$LINENO" "char32_t" "ac_cv_type_char32_t" "$ac_includes_default"
+if test "x$ac_cv_type_char32_t" = xyes; then :
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_CHAR32_T 1
+_ACEOF
+
+
 fi
 
 
diff --git a/configure.ac b/configure.ac
index e44943aa6fe..9e677d37dce 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1513,6 +1513,7 @@ AC_CHECK_HEADERS(m4_normalize([
 	sys/signalfd.h
 	sys/ucred.h
 	termios.h
+	uchar.h
 	ucred.h
 	xlocale.h
 ]))
@@ -1684,6 +1685,8 @@ PGAC_STRUCT_TIMEZONE
 PGAC_UNION_SEMUN
 AC_CHECK_TYPES(socklen_t, [], [], [#include <sys/socket.h>])
 PGAC_STRUCT_SOCKADDR_SA_LEN
+AC_CHECK_TYPES(char16_t, [], [], [])
+AC_CHECK_TYPES(char32_t, [], [], [])
 
 # MSVC doesn't cope well with defining restrict to __restrict, the
 # spelling it understands, because it conflicts with
diff --git a/meson.build b/meson.build
index 395416a6060..a60635ff6a0 100644
--- a/meson.build
+++ b/meson.build
@@ -2613,6 +2613,7 @@ header_checks = [
   'sys/signalfd.h',
   'sys/ucred.h',
   'termios.h',
+  'uchar.h',
   'ucred.h',
   'xlocale.h',
 ]
@@ -2720,6 +2721,16 @@ if cc.has_type('socklen_t',
   cdata.set('HAVE_SOCKLEN_T', 1)
 endif
 
+if cc.has_type('char16_t',
+    args: test_c_args, include_directories: postgres_inc)
+  cdata.set('HAVE_CHAR16_T', 1)
+endif
+
+if cc.has_type('char32_t',
+    args: test_c_args, include_directories: postgres_inc)
+  cdata.set('HAVE_CHAR32_T', 1)
+endif
+
 if cc.has_member('struct sockaddr', 'sa_len',
     args: test_c_args, include_directories: postgres_inc,
     prefix: '''
diff --git a/src/backend/parser/parser.c b/src/backend/parser/parser.c
index 33a040506b4..a3679f8e86c 100644
--- a/src/backend/parser/parser.c
+++ b/src/backend/parser/parser.c
@@ -339,7 +339,7 @@ hexval(unsigned char c)
 
 /* is Unicode code point acceptable? */
 static void
-check_unicode_value(pg_wchar c)
+check_unicode_value(char32_t c)
 {
 	if (!is_valid_unicode_codepoint(c))
 		ereport(ERROR,
@@ -376,7 +376,7 @@ str_udeescape(const char *str, char escape,
 	char	   *new,
 			   *out;
 	size_t		new_len;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	ScannerCallbackState scbstate;
 
 	/*
@@ -420,7 +420,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[3]) &&
 					 isxdigit((unsigned char) in[4]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[1]) << 12) +
 					(hexval(in[2]) << 8) +
@@ -457,7 +457,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[6]) &&
 					 isxdigit((unsigned char) in[7]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[2]) << 20) +
 					(hexval(in[3]) << 16) +
diff --git a/src/backend/parser/scan.l b/src/backend/parser/scan.l
index 08990831fe8..a67815339b7 100644
--- a/src/backend/parser/scan.l
+++ b/src/backend/parser/scan.l
@@ -121,7 +121,7 @@ static void addlitchar(unsigned char ychar, core_yyscan_t yyscanner);
 static char *litbufdup(core_yyscan_t yyscanner);
 static unsigned char unescape_single_char(unsigned char c, core_yyscan_t yyscanner);
 static int	process_integer_literal(const char *token, YYSTYPE *lval, int base);
-static void addunicode(pg_wchar c, yyscan_t yyscanner);
+static void addunicode(char32_t c, yyscan_t yyscanner);
 
 #define yyerror(msg)  scanner_yyerror(msg, yyscanner)
 
@@ -640,7 +640,7 @@ other			.
 					addlit(yytext, yyleng, yyscanner);
 				}
 <xe>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/*
 					 * For consistency with other productions, issue any
@@ -668,7 +668,7 @@ other			.
 					POP_YYLLOC();
 				}
 <xeu>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/* Remember start of overall string token ... */
 					PUSH_YYLLOC();
@@ -1376,7 +1376,7 @@ process_integer_literal(const char *token, YYSTYPE *lval, int base)
 }
 
 static void
-addunicode(pg_wchar c, core_yyscan_t yyscanner)
+addunicode(char32_t c, core_yyscan_t yyscanner)
 {
 	ScannerCallbackState scbstate;
 	char		buf[MAX_UNICODE_EQUIVALENT_STRING + 1];
diff --git a/src/backend/utils/adt/jsonpath_scan.l b/src/backend/utils/adt/jsonpath_scan.l
index c7aab83eeb4..8c3a0a9c642 100644
--- a/src/backend/utils/adt/jsonpath_scan.l
+++ b/src/backend/utils/adt/jsonpath_scan.l
@@ -574,7 +574,7 @@ hexval(char c, int *result, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add given unicode character to scanstring */
 static bool
-addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
+addUnicodeChar(char32_t ch, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (ch == 0)
 	{
@@ -607,7 +607,7 @@ addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add unicode character, processing any surrogate pairs */
 static bool
-addUnicode(int ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
+addUnicode(char32_t ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (is_utf16_surrogate_first(ch))
 	{
@@ -655,7 +655,7 @@ parseUnicode(char *s, int l, struct Node *escontext, yyscan_t yyscanner)
 
 	for (i = 2; i < l; i += 2)	/* skip '\u' */
 	{
-		int			ch = 0;
+		char32_t		ch = 0;
 		int			j,
 					si;
 
diff --git a/src/backend/utils/adt/pg_locale_builtin.c b/src/backend/utils/adt/pg_locale_builtin.c
index 3dc611b50e1..1021e0d129b 100644
--- a/src/backend/utils/adt/pg_locale_builtin.c
+++ b/src/backend/utils/adt/pg_locale_builtin.c
@@ -15,7 +15,6 @@
 #include "catalog/pg_collation.h"
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
-#include "mb/pg_wchar.h"
 #include "miscadmin.h"
 #include "utils/builtins.h"
 #include "utils/pg_locale.h"
@@ -35,6 +34,23 @@ struct WordBoundaryState
 	bool		prev_alnum;
 };
 
+/*
+ * In UTF-8, pg_wchar is guaranteed to be the code point value.
+ */
+static inline char32_t
+to_char32(pg_wchar wc)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (char32_t) wc;
+}
+
+static inline pg_wchar
+to_pg_wchar(char32_t c32)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (pg_wchar) c32;
+}
+
 /*
  * Simple word boundary iterator that draws boundaries each time the result of
  * pg_u_isalnum() changes.
@@ -47,7 +63,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -112,61 +128,61 @@ strfold_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen,
 static bool
 wc_isdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isalpha_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalpha(wc);
+	return pg_u_isalpha(to_char32(wc));
 }
 
 static bool
 wc_isalnum_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalnum(wc, !locale->builtin.casemap_full);
+	return pg_u_isalnum(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isupper(wc);
+	return pg_u_isupper(to_char32(wc));
 }
 
 static bool
 wc_islower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_islower(wc);
+	return pg_u_islower(to_char32(wc));
 }
 
 static bool
 wc_isgraph_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isgraph(wc);
+	return pg_u_isgraph(to_char32(wc));
 }
 
 static bool
 wc_isprint_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isprint(wc);
+	return pg_u_isprint(to_char32(wc));
 }
 
 static bool
 wc_ispunct_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_ispunct(wc, !locale->builtin.casemap_full);
+	return pg_u_ispunct(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isspace_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isspace(wc);
+	return pg_u_isspace(to_char32(wc));
 }
 
 static bool
 wc_isxdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isxdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isxdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
@@ -179,13 +195,13 @@ char_is_cased_builtin(char ch, pg_locale_t locale)
 static pg_wchar
 wc_toupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_uppercase_simple(wc);
+	return to_pg_wchar(unicode_uppercase_simple(to_char32(wc)));
 }
 
 static pg_wchar
 wc_tolower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_lowercase_simple(wc);
+	return to_pg_wchar(unicode_lowercase_simple(to_char32(wc)));
 }
 
 static const struct ctype_methods ctype_methods_builtin = {
diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c
index 2c398cd9e5c..8d735786e51 100644
--- a/src/backend/utils/adt/varlena.c
+++ b/src/backend/utils/adt/varlena.c
@@ -5419,12 +5419,12 @@ unicode_assigned(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errmsg("Unicode categorization can only be performed if server encoding is UTF8")));
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (int i = 0; i < size; i++)
 	{
-		pg_wchar	uchar = utf8_to_unicode(p);
+		char32_t	uchar = utf8_to_unicode(p);
 		int			category = unicode_category(uchar);
 
 		if (category == PG_U_UNASSIGNED)
@@ -5443,24 +5443,24 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	text	   *result;
 	int			i;
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* action */
@@ -5468,7 +5468,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 
 	/* convert back to UTF-8 string */
 	size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unsigned char buf[4];
 
@@ -5480,7 +5480,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	SET_VARSIZE(result, size + VARHDRSZ);
 
 	p = (unsigned char *) VARDATA_ANY(result);
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unicode_to_utf8(*wp, p);
 		p += pg_utf_mblen(p);
@@ -5509,8 +5509,8 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	int			i;
 	UnicodeNormalizationQC quickcheck;
@@ -5519,16 +5519,16 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* quick check (see UAX #15) */
@@ -5542,11 +5542,11 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	output_chars = unicode_normalize(form, input_chars);
 
 	output_size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 		output_size++;
 
 	result = (size == output_size) &&
-		(memcmp(input_chars, output_chars, size * sizeof(pg_wchar)) == 0);
+		(memcmp(input_chars, output_chars, size * sizeof(char32_t)) == 0);
 
 	PG_RETURN_BOOL(result);
 }
@@ -5602,7 +5602,7 @@ unistr(PG_FUNCTION_ARGS)
 	int			len;
 	StringInfoData str;
 	text	   *result;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	char		cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
 
 	instr = VARDATA_ANY(input_text);
@@ -5626,7 +5626,7 @@ unistr(PG_FUNCTION_ARGS)
 			else if ((len >= 5 && isxdigits_n(instr + 1, 4)) ||
 					 (len >= 6 && instr[1] == 'u' && isxdigits_n(instr + 2, 4)))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 				int			offset = instr[1] == 'u' ? 2 : 1;
 
 				unicode = hexval_n(instr + offset, 4);
@@ -5662,7 +5662,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 8 && instr[1] == '+' && isxdigits_n(instr + 2, 6))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 6);
 
@@ -5697,7 +5697,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 10 && instr[1] == 'U' && isxdigits_n(instr + 2, 8))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 8);
 
diff --git a/src/backend/utils/mb/mbutils.c b/src/backend/utils/mb/mbutils.c
index 886ecbad871..fb629ed5c8f 100644
--- a/src/backend/utils/mb/mbutils.c
+++ b/src/backend/utils/mb/mbutils.c
@@ -862,7 +862,7 @@ perform_default_encoding_conversion(const char *src, int len,
  * may call this outside any transaction, or in an aborted transaction.
  */
 void
-pg_unicode_to_server(pg_wchar c, unsigned char *s)
+pg_unicode_to_server(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
@@ -924,7 +924,7 @@ pg_unicode_to_server(pg_wchar c, unsigned char *s)
  * but simply return false on conversion failure.
  */
 bool
-pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s)
+pg_unicode_to_server_noerror(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
diff --git a/src/common/saslprep.c b/src/common/saslprep.c
index 97beb47940b..101e8d65a4d 100644
--- a/src/common/saslprep.c
+++ b/src/common/saslprep.c
@@ -47,7 +47,7 @@
 
 /* Prototypes for local functions */
 static int	codepoint_range_cmp(const void *a, const void *b);
-static bool is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize);
+static bool is_code_in_table(char32_t code, const char32_t *map, int mapsize);
 static int	pg_utf8_string_len(const char *source);
 
 /*
@@ -64,7 +64,7 @@ static int	pg_utf8_string_len(const char *source);
  *
  * These are all mapped to the ASCII space character (U+00A0).
  */
-static const pg_wchar non_ascii_space_ranges[] =
+static const char32_t non_ascii_space_ranges[] =
 {
 	0x00A0, 0x00A0,
 	0x1680, 0x1680,
@@ -79,7 +79,7 @@ static const pg_wchar non_ascii_space_ranges[] =
  *
  * If any of these appear in the input, they are removed.
  */
-static const pg_wchar commonly_mapped_to_nothing_ranges[] =
+static const char32_t commonly_mapped_to_nothing_ranges[] =
 {
 	0x00AD, 0x00AD,
 	0x034F, 0x034F,
@@ -114,7 +114,7 @@ static const pg_wchar commonly_mapped_to_nothing_ranges[] =
  * tables, so one code might originate from multiple source tables.
  * Adjacent ranges have also been merged together, to save space.
  */
-static const pg_wchar prohibited_output_ranges[] =
+static const char32_t prohibited_output_ranges[] =
 {
 	0x0000, 0x001F,				/* C.2.1 */
 	0x007F, 0x00A0,				/* C.1.2, C.2.1, C.2.2 */
@@ -155,7 +155,7 @@ static const pg_wchar prohibited_output_ranges[] =
 };
 
 /* A.1 Unassigned code points in Unicode 3.2 */
-static const pg_wchar unassigned_codepoint_ranges[] =
+static const char32_t unassigned_codepoint_ranges[] =
 {
 	0x0221, 0x0221,
 	0x0234, 0x024F,
@@ -556,7 +556,7 @@ static const pg_wchar unassigned_codepoint_ranges[] =
 };
 
 /* D.1 Characters with bidirectional property "R" or "AL" */
-static const pg_wchar RandALCat_codepoint_ranges[] =
+static const char32_t RandALCat_codepoint_ranges[] =
 {
 	0x05BE, 0x05BE,
 	0x05C0, 0x05C0,
@@ -595,7 +595,7 @@ static const pg_wchar RandALCat_codepoint_ranges[] =
 };
 
 /* D.2 Characters with bidirectional property "L" */
-static const pg_wchar LCat_codepoint_ranges[] =
+static const char32_t LCat_codepoint_ranges[] =
 {
 	0x0041, 0x005A,
 	0x0061, 0x007A,
@@ -968,8 +968,8 @@ static const pg_wchar LCat_codepoint_ranges[] =
 static int
 codepoint_range_cmp(const void *a, const void *b)
 {
-	const pg_wchar *key = (const pg_wchar *) a;
-	const pg_wchar *range = (const pg_wchar *) b;
+	const char32_t *key = (const char32_t *) a;
+	const char32_t *range = (const char32_t *) b;
 
 	if (*key < range[0])
 		return -1;				/* less than lower bound */
@@ -980,14 +980,14 @@ codepoint_range_cmp(const void *a, const void *b)
 }
 
 static bool
-is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize)
+is_code_in_table(char32_t code, const char32_t *map, int mapsize)
 {
 	Assert(mapsize % 2 == 0);
 
 	if (code < map[0] || code > map[mapsize - 1])
 		return false;
 
-	if (bsearch(&code, map, mapsize / 2, sizeof(pg_wchar) * 2,
+	if (bsearch(&code, map, mapsize / 2, sizeof(char32_t) * 2,
 				codepoint_range_cmp))
 		return true;
 	else
@@ -1046,8 +1046,8 @@ pg_utf8_string_len(const char *source)
 pg_saslprep_rc
 pg_saslprep(const char *input, char **output)
 {
-	pg_wchar   *input_chars = NULL;
-	pg_wchar   *output_chars = NULL;
+	char32_t   *input_chars = NULL;
+	char32_t   *output_chars = NULL;
 	int			input_size;
 	char	   *result;
 	int			result_size;
@@ -1055,7 +1055,7 @@ pg_saslprep(const char *input, char **output)
 	int			i;
 	bool		contains_RandALCat;
 	unsigned char *p;
-	pg_wchar   *wp;
+	char32_t   *wp;
 
 	/* Ensure we return *output as NULL on failure */
 	*output = NULL;
@@ -1080,10 +1080,10 @@ pg_saslprep(const char *input, char **output)
 	input_size = pg_utf8_string_len(input);
 	if (input_size < 0)
 		return SASLPREP_INVALID_UTF8;
-	if (input_size >= MaxAllocSize / sizeof(pg_wchar))
+	if (input_size >= MaxAllocSize / sizeof(char32_t))
 		goto oom;
 
-	input_chars = ALLOC((input_size + 1) * sizeof(pg_wchar));
+	input_chars = ALLOC((input_size + 1) * sizeof(char32_t));
 	if (!input_chars)
 		goto oom;
 
@@ -1093,7 +1093,7 @@ pg_saslprep(const char *input, char **output)
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 
 	/*
 	 * The steps below correspond to the steps listed in [RFC3454], Section
@@ -1107,7 +1107,7 @@ pg_saslprep(const char *input, char **output)
 	count = 0;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, non_ascii_space_ranges))
 			input_chars[count++] = 0x0020;
@@ -1118,7 +1118,7 @@ pg_saslprep(const char *input, char **output)
 		else
 			input_chars[count++] = code;
 	}
-	input_chars[count] = (pg_wchar) '\0';
+	input_chars[count] = (char32_t) '\0';
 	input_size = count;
 
 	if (input_size == 0)
@@ -1138,7 +1138,7 @@ pg_saslprep(const char *input, char **output)
 	 */
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, prohibited_output_ranges))
 			goto prohibited;
@@ -1170,7 +1170,7 @@ pg_saslprep(const char *input, char **output)
 	contains_RandALCat = false;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, RandALCat_codepoint_ranges))
 		{
@@ -1181,12 +1181,12 @@ pg_saslprep(const char *input, char **output)
 
 	if (contains_RandALCat)
 	{
-		pg_wchar	first = input_chars[0];
-		pg_wchar	last = input_chars[input_size - 1];
+		char32_t	first = input_chars[0];
+		char32_t	last = input_chars[input_size - 1];
 
 		for (i = 0; i < input_size; i++)
 		{
-			pg_wchar	code = input_chars[i];
+			char32_t	code = input_chars[i];
 
 			if (IS_CODE_IN_TABLE(code, LCat_codepoint_ranges))
 				goto prohibited;
diff --git a/src/common/unicode/case_test.c b/src/common/unicode/case_test.c
index fdfb62e8552..00d4f85e5a5 100644
--- a/src/common/unicode/case_test.c
+++ b/src/common/unicode/case_test.c
@@ -24,6 +24,7 @@
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 /* enough to hold largest source or result string, including NUL */
 #define BUFSZ 256
@@ -54,7 +55,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -77,16 +78,16 @@ initcap_wbnext(void *state)
 #ifdef USE_ICU
 
 static void
-icu_test_simple(pg_wchar code)
+icu_test_simple(char32_t code)
 {
-	pg_wchar	lower = unicode_lowercase_simple(code);
-	pg_wchar	title = unicode_titlecase_simple(code);
-	pg_wchar	upper = unicode_uppercase_simple(code);
-	pg_wchar	fold = unicode_casefold_simple(code);
-	pg_wchar	iculower = u_tolower(code);
-	pg_wchar	icutitle = u_totitle(code);
-	pg_wchar	icuupper = u_toupper(code);
-	pg_wchar	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
+	char32_t	lower = unicode_lowercase_simple(code);
+	char32_t	title = unicode_titlecase_simple(code);
+	char32_t	upper = unicode_uppercase_simple(code);
+	char32_t	fold = unicode_casefold_simple(code);
+	char32_t	iculower = u_tolower(code);
+	char32_t	icutitle = u_totitle(code);
+	char32_t	icuupper = u_toupper(code);
+	char32_t	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
 
 	if (lower != iculower || title != icutitle || upper != icuupper ||
 		fold != icufold)
@@ -172,7 +173,7 @@ test_icu(void)
 	int			successful = 0;
 	int			skipped_mismatch = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		pg_unicode_category category = unicode_category(code);
 
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index 5d37ba39196..1e8c1f7905f 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -22,6 +22,7 @@
 
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 static int	pg_unicode_version = 0;
 #ifdef USE_ICU
@@ -59,7 +60,7 @@ icu_test()
 	int			pg_skipped_codepoints = 0;
 	int			icu_skipped_codepoints = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		uint8_t		pg_category = unicode_category(code);
 		uint8_t		icu_category = u_charType(code);
diff --git a/src/common/unicode/generate-norm_test_table.pl b/src/common/unicode/generate-norm_test_table.pl
index 1b401be9409..1a8b908ff33 100644
--- a/src/common/unicode/generate-norm_test_table.pl
+++ b/src/common/unicode/generate-norm_test_table.pl
@@ -47,8 +47,8 @@ print $OUTPUT <<HEADER;
 typedef struct
 {
 	int			linenum;
-	pg_wchar	input[50];
-	pg_wchar	output[4][50];
+	char32_t	input[50];
+	char32_t	output[4][50];
 } pg_unicode_test;
 
 /* test table */
diff --git a/src/common/unicode/generate-unicode_case_table.pl b/src/common/unicode/generate-unicode_case_table.pl
index 5d9ddd62803..f71eb25c94e 100644
--- a/src/common/unicode/generate-unicode_case_table.pl
+++ b/src/common/unicode/generate-unicode_case_table.pl
@@ -270,7 +270,6 @@ print $OT <<"EOS";
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -297,7 +296,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -430,7 +429,7 @@ foreach my $kind ('lower', 'title', 'upper', 'fold')
  * The entry case_map_${kind}[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_$kind\[$index\] =
+static const char32_t case_map_$kind\[$index\] =
 {
 EOS
 
@@ -502,7 +501,7 @@ print $OT <<"EOS";
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < $fastpath_limit */
 	if (cp < $fastpath_limit)
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index abab5cd9696..7e094b13720 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -366,15 +366,15 @@ print $OT <<"EOS";
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/common/unicode/norm_test.c b/src/common/unicode/norm_test.c
index 25bc59463f2..058817f1719 100644
--- a/src/common/unicode/norm_test.c
+++ b/src/common/unicode/norm_test.c
@@ -20,7 +20,7 @@
 #include "norm_test_table.h"
 
 static char *
-print_wchar_str(const pg_wchar *s)
+print_wchar_str(const char32_t *s)
 {
 #define BUF_DIGITS 50
 	static char buf[BUF_DIGITS * 11 + 1];
@@ -41,7 +41,7 @@ print_wchar_str(const pg_wchar *s)
 }
 
 static int
-pg_wcscmp(const pg_wchar *s1, const pg_wchar *s2)
+pg_wcscmp(const char32_t *s1, const char32_t *s2)
 {
 	for (;;)
 	{
@@ -65,7 +65,7 @@ main(int argc, char **argv)
 	{
 		for (int form = 0; form < 4; form++)
 		{
-			pg_wchar   *result;
+			char32_t   *result;
 
 			result = unicode_normalize(form, test->input);
 
diff --git a/src/common/unicode_case.c b/src/common/unicode_case.c
index 073faf6a0d5..e5e494db43c 100644
--- a/src/common/unicode_case.c
+++ b/src/common/unicode_case.c
@@ -30,7 +30,7 @@ enum CaseMapResult
 /*
  * Map for each case kind.
  */
-static const pg_wchar *const casekind_map[NCaseKind] =
+static const char32_t *const casekind_map[NCaseKind] =
 {
 	[CaseLower] = case_map_lower,
 	[CaseTitle] = case_map_title,
@@ -38,42 +38,42 @@ static const pg_wchar *const casekind_map[NCaseKind] =
 	[CaseFold] = case_map_fold,
 };
 
-static pg_wchar find_case_map(pg_wchar ucs, const pg_wchar *map);
+static char32_t find_case_map(char32_t ucs, const char32_t *map);
 static size_t convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 						   CaseKind str_casekind, bool full, WordBoundaryNext wbnext,
 						   void *wbstate);
-static enum CaseMapResult casemap(pg_wchar u1, CaseKind casekind, bool full,
+static enum CaseMapResult casemap(char32_t u1, CaseKind casekind, bool full,
 								  const char *src, size_t srclen, size_t srcoff,
-								  pg_wchar *simple, const pg_wchar **special);
+								  char32_t *simple, const char32_t **special);
 
-pg_wchar
-unicode_lowercase_simple(pg_wchar code)
+char32_t
+unicode_lowercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_lower);
+	char32_t	cp = find_case_map(code, case_map_lower);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_titlecase_simple(pg_wchar code)
+char32_t
+unicode_titlecase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_title);
+	char32_t	cp = find_case_map(code, case_map_title);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_uppercase_simple(pg_wchar code)
+char32_t
+unicode_uppercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_upper);
+	char32_t	cp = find_case_map(code, case_map_upper);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_casefold_simple(pg_wchar code)
+char32_t
+unicode_casefold_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_fold);
+	char32_t	cp = find_case_map(code, case_map_fold);
 
 	return cp != 0 ? cp : code;
 }
@@ -231,10 +231,10 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 
 	while ((srclen < 0 || srcoff < srclen) && src[srcoff] != '\0')
 	{
-		pg_wchar	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
+		char32_t	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
 		int			u1len = unicode_utf8len(u1);
-		pg_wchar	simple = 0;
-		const pg_wchar *special = NULL;
+		char32_t	simple = 0;
+		const char32_t *special = NULL;
 		enum CaseMapResult casemap_result;
 
 		if (str_casekind == CaseTitle)
@@ -265,8 +265,8 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 			case CASEMAP_SIMPLE:
 				{
 					/* replace with single character */
-					pg_wchar	u2 = simple;
-					pg_wchar	u2len = unicode_utf8len(u2);
+					char32_t	u2 = simple;
+					char32_t	u2len = unicode_utf8len(u2);
 
 					Assert(special == NULL);
 					if (result_len + u2len <= dstsize)
@@ -280,7 +280,7 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 				Assert(simple == 0);
 				for (int i = 0; i < MAX_CASE_EXPANSION && special[i]; i++)
 				{
-					pg_wchar	u2 = special[i];
+					char32_t	u2 = special[i];
 					size_t		u2len = unicode_utf8len(u2);
 
 					if (result_len + u2len <= dstsize)
@@ -320,7 +320,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -344,7 +344,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -394,9 +394,9 @@ check_special_conditions(int conditions, const char *str, size_t len,
  * character without modification.
  */
 static enum CaseMapResult
-casemap(pg_wchar u1, CaseKind casekind, bool full,
+casemap(char32_t u1, CaseKind casekind, bool full,
 		const char *src, size_t srclen, size_t srcoff,
-		pg_wchar *simple, const pg_wchar **special)
+		char32_t *simple, const char32_t **special)
 {
 	uint16		idx;
 
@@ -434,8 +434,8 @@ casemap(pg_wchar u1, CaseKind casekind, bool full,
  * Find entry in simple case map.
  * If the entry does not exist, 0 will be returned.
  */
-static pg_wchar
-find_case_map(pg_wchar ucs, const pg_wchar *map)
+static char32_t
+find_case_map(char32_t ucs, const char32_t *map)
 {
 	/* Fast path for codepoints < 0x80 */
 	if (ucs < 0x80)
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index 4136c4d4f92..aab667a7bb4 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -1,7 +1,7 @@
 /*-------------------------------------------------------------------------
  * unicode_category.c
  *		Determine general category and character properties of Unicode
- *		characters. Encoding must be UTF8, where we assume that the pg_wchar
+ *		characters. Encoding must be UTF8, where we assume that the char32_t
  *		representation is a code point.
  *
  * Portions Copyright (c) 2017-2025, PostgreSQL Global Development Group
@@ -76,13 +76,13 @@
 #define PG_U_CHARACTER_TAB	0x09
 
 static bool range_search(const pg_unicode_range *tbl, size_t size,
-						 pg_wchar code);
+						 char32_t code);
 
 /*
  * Unicode general category for the given codepoint.
  */
 pg_unicode_category
-unicode_category(pg_wchar code)
+unicode_category(char32_t code)
 {
 	int			min = 0;
 	int			mid;
@@ -108,7 +108,7 @@ unicode_category(pg_wchar code)
 }
 
 bool
-pg_u_prop_alphabetic(pg_wchar code)
+pg_u_prop_alphabetic(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_ALPHABETIC;
@@ -119,7 +119,7 @@ pg_u_prop_alphabetic(pg_wchar code)
 }
 
 bool
-pg_u_prop_lowercase(pg_wchar code)
+pg_u_prop_lowercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_LOWERCASE;
@@ -130,7 +130,7 @@ pg_u_prop_lowercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_uppercase(pg_wchar code)
+pg_u_prop_uppercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_UPPERCASE;
@@ -141,7 +141,7 @@ pg_u_prop_uppercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_cased(pg_wchar code)
+pg_u_prop_cased(char32_t code)
 {
 	uint32		category_mask;
 
@@ -156,7 +156,7 @@ pg_u_prop_cased(pg_wchar code)
 }
 
 bool
-pg_u_prop_case_ignorable(pg_wchar code)
+pg_u_prop_case_ignorable(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_CASE_IGNORABLE;
@@ -167,7 +167,7 @@ pg_u_prop_case_ignorable(pg_wchar code)
 }
 
 bool
-pg_u_prop_white_space(pg_wchar code)
+pg_u_prop_white_space(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_WHITE_SPACE;
@@ -178,7 +178,7 @@ pg_u_prop_white_space(pg_wchar code)
 }
 
 bool
-pg_u_prop_hex_digit(pg_wchar code)
+pg_u_prop_hex_digit(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_HEX_DIGIT;
@@ -189,7 +189,7 @@ pg_u_prop_hex_digit(pg_wchar code)
 }
 
 bool
-pg_u_prop_join_control(pg_wchar code)
+pg_u_prop_join_control(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_JOIN_CONTROL;
@@ -208,7 +208,7 @@ pg_u_prop_join_control(pg_wchar code)
  */
 
 bool
-pg_u_isdigit(pg_wchar code, bool posix)
+pg_u_isdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return ('0' <= code && code <= '9');
@@ -217,19 +217,19 @@ pg_u_isdigit(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isalpha(pg_wchar code)
+pg_u_isalpha(char32_t code)
 {
 	return pg_u_prop_alphabetic(code);
 }
 
 bool
-pg_u_isalnum(pg_wchar code, bool posix)
+pg_u_isalnum(char32_t code, bool posix)
 {
 	return pg_u_isalpha(code) || pg_u_isdigit(code, posix);
 }
 
 bool
-pg_u_isword(pg_wchar code)
+pg_u_isword(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -240,32 +240,32 @@ pg_u_isword(pg_wchar code)
 }
 
 bool
-pg_u_isupper(pg_wchar code)
+pg_u_isupper(char32_t code)
 {
 	return pg_u_prop_uppercase(code);
 }
 
 bool
-pg_u_islower(pg_wchar code)
+pg_u_islower(char32_t code)
 {
 	return pg_u_prop_lowercase(code);
 }
 
 bool
-pg_u_isblank(pg_wchar code)
+pg_u_isblank(char32_t code)
 {
 	return code == PG_U_CHARACTER_TAB ||
 		unicode_category(code) == PG_U_SPACE_SEPARATOR;
 }
 
 bool
-pg_u_iscntrl(pg_wchar code)
+pg_u_iscntrl(char32_t code)
 {
 	return unicode_category(code) == PG_U_CONTROL;
 }
 
 bool
-pg_u_isgraph(pg_wchar code)
+pg_u_isgraph(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -276,7 +276,7 @@ pg_u_isgraph(pg_wchar code)
 }
 
 bool
-pg_u_isprint(pg_wchar code)
+pg_u_isprint(char32_t code)
 {
 	pg_unicode_category category = unicode_category(code);
 
@@ -287,7 +287,7 @@ pg_u_isprint(pg_wchar code)
 }
 
 bool
-pg_u_ispunct(pg_wchar code, bool posix)
+pg_u_ispunct(char32_t code, bool posix)
 {
 	uint32		category_mask;
 
@@ -308,13 +308,13 @@ pg_u_ispunct(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isspace(pg_wchar code)
+pg_u_isspace(char32_t code)
 {
 	return pg_u_prop_white_space(code);
 }
 
 bool
-pg_u_isxdigit(pg_wchar code, bool posix)
+pg_u_isxdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return (('0' <= code && code <= '9') ||
@@ -478,7 +478,7 @@ unicode_category_abbrev(pg_unicode_category category)
  * given table.
  */
 static bool
-range_search(const pg_unicode_range *tbl, size_t size, pg_wchar code)
+range_search(const pg_unicode_range *tbl, size_t size, char32_t code)
 {
 	int			min = 0;
 	int			mid;
diff --git a/src/common/unicode_norm.c b/src/common/unicode_norm.c
index 6654b4cbc49..489d99cd5ab 100644
--- a/src/common/unicode_norm.c
+++ b/src/common/unicode_norm.c
@@ -69,7 +69,7 @@ conv_compare(const void *p1, const void *p2)
  * lookup, while the frontend version uses a binary search.
  */
 static const pg_unicode_decomposition *
-get_code_entry(pg_wchar code)
+get_code_entry(char32_t code)
 {
 #ifndef FRONTEND
 	int			h;
@@ -109,7 +109,7 @@ get_code_entry(pg_wchar code)
  * Get the combining class of the given codepoint.
  */
 static uint8
-get_canonical_class(pg_wchar code)
+get_canonical_class(char32_t code)
 {
 	const pg_unicode_decomposition *entry = get_code_entry(code);
 
@@ -130,15 +130,15 @@ get_canonical_class(pg_wchar code)
  * Note: the returned pointer can point to statically allocated buffer, and
  * is only valid until next call to this function!
  */
-static const pg_wchar *
+static const char32_t *
 get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
 {
-	static pg_wchar x;
+	static char32_t x;
 
 	if (DECOMPOSITION_IS_INLINE(entry))
 	{
 		Assert(DECOMPOSITION_SIZE(entry) == 1);
-		x = (pg_wchar) entry->dec_index;
+		x = (char32_t) entry->dec_index;
 		*dec_size = 1;
 		return &x;
 	}
@@ -156,7 +156,7 @@ get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
  * are, in turn, decomposable.
  */
 static int
-get_decomposed_size(pg_wchar code, bool compat)
+get_decomposed_size(char32_t code, bool compat)
 {
 	const pg_unicode_decomposition *entry;
 	int			size = 0;
@@ -318,7 +318,7 @@ recompose_code(uint32 start, uint32 code, uint32 *result)
  * in the array result.
  */
 static void
-decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
+decompose_code(char32_t code, bool compat, char32_t **result, int *current)
 {
 	const pg_unicode_decomposition *entry;
 	int			i;
@@ -337,7 +337,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 					v,
 					tindex,
 					sindex;
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		sindex = code - SBASE;
 		l = LBASE + sindex / (VCOUNT * TCOUNT);
@@ -369,7 +369,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	if (entry == NULL || DECOMPOSITION_SIZE(entry) == 0 ||
 		(!compat && DECOMPOSITION_IS_COMPAT(entry)))
 	{
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		res[*current] = code;
 		(*current)++;
@@ -382,7 +382,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	decomp = get_code_decomposition(entry, &dec_size);
 	for (i = 0; i < dec_size; i++)
 	{
-		pg_wchar	lcode = (pg_wchar) decomp[i];
+		char32_t	lcode = (char32_t) decomp[i];
 
 		/* Leave if no more decompositions */
 		decompose_code(lcode, compat, result, current);
@@ -398,17 +398,17 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
  * malloc. Or NULL if we run out of memory. In backend, the returned
  * string is palloc'd instead, and OOM is reported with ereport().
  */
-pg_wchar *
-unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
+char32_t *
+unicode_normalize(UnicodeNormalizationForm form, const char32_t *input)
 {
 	bool		compat = (form == UNICODE_NFKC || form == UNICODE_NFKD);
 	bool		recompose = (form == UNICODE_NFC || form == UNICODE_NFKC);
-	pg_wchar   *decomp_chars;
-	pg_wchar   *recomp_chars;
+	char32_t   *decomp_chars;
+	char32_t   *recomp_chars;
 	int			decomp_size,
 				current_size;
 	int			count;
-	const pg_wchar *p;
+	const char32_t *p;
 
 	/* variables for recomposition */
 	int			last_class;
@@ -425,7 +425,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	for (p = input; *p; p++)
 		decomp_size += get_decomposed_size(*p, compat);
 
-	decomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	decomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (decomp_chars == NULL)
 		return NULL;
 
@@ -448,9 +448,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 */
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	prev = decomp_chars[count - 1];
-		pg_wchar	next = decomp_chars[count];
-		pg_wchar	tmp;
+		char32_t	prev = decomp_chars[count - 1];
+		char32_t	next = decomp_chars[count];
+		char32_t	tmp;
 		const uint8 prevClass = get_canonical_class(prev);
 		const uint8 nextClass = get_canonical_class(next);
 
@@ -487,7 +487,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 * longer than the decomposed one, so make the allocation of the output
 	 * string based on that assumption.
 	 */
-	recomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	recomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (!recomp_chars)
 	{
 		FREE(decomp_chars);
@@ -501,9 +501,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	ch = decomp_chars[count];
+		char32_t	ch = decomp_chars[count];
 		int			ch_class = get_canonical_class(ch);
-		pg_wchar	composite;
+		char32_t	composite;
 
 		if (last_class < ch_class &&
 			recompose_code(starter_ch, ch, &composite))
@@ -524,7 +524,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 			recomp_chars[target_pos++] = ch;
 		}
 	}
-	recomp_chars[target_pos] = (pg_wchar) '\0';
+	recomp_chars[target_pos] = (char32_t) '\0';
 
 	FREE(decomp_chars);
 
@@ -540,7 +540,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 #ifndef FRONTEND
 
 static const pg_unicode_normprops *
-qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
+qc_hash_lookup(char32_t ch, const pg_unicode_norminfo *norminfo)
 {
 	int			h;
 	uint32		hashkey;
@@ -571,7 +571,7 @@ qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
  * Look up the normalization quick check character property
  */
 static UnicodeNormalizationQC
-qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
+qc_is_allowed(UnicodeNormalizationForm form, char32_t ch)
 {
 	const pg_unicode_normprops *found = NULL;
 
@@ -595,7 +595,7 @@ qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
 }
 
 UnicodeNormalizationQC
-unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input)
+unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input)
 {
 	uint8		lastCanonicalClass = 0;
 	UnicodeNormalizationQC result = UNICODE_NORM_QC_YES;
@@ -610,9 +610,9 @@ unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *
 	if (form == UNICODE_NFD || form == UNICODE_NFKD)
 		return UNICODE_NORM_QC_MAYBE;
 
-	for (const pg_wchar *p = input; *p; p++)
+	for (const char32_t *p = input; *p; p++)
 	{
-		pg_wchar	ch = *p;
+		char32_t	ch = *p;
 		uint8		canonicalClass;
 		UnicodeNormalizationQC check;
 
diff --git a/src/fe_utils/mbprint.c b/src/fe_utils/mbprint.c
index eb3eeee9925..abffdbe18a2 100644
--- a/src/fe_utils/mbprint.c
+++ b/src/fe_utils/mbprint.c
@@ -49,20 +49,20 @@ pg_get_utf8_id(void)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static pg_wchar
+static char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
diff --git a/src/include/c.h b/src/include/c.h
index 9ab5e617995..15e33ae650d 100644
--- a/src/include/c.h
+++ b/src/include/c.h
@@ -513,6 +513,19 @@ typedef void (*pg_funcptr_t) (void);
 
 #include <stdbool.h>
 
+/*
+ * char16_t and char32_t
+ *      Unicode code points.
+ */
+#if !defined(HAVE_CHAR16_T) || !defined(HAVE_CHAR32_T)
+#ifdef HAVE_UCHAR_H
+#include <uchar.h>
+#else
+typedef uint16_t char16_t;
+typedef uint32_t char32_t;
+#endif
+#endif
+
 
 /* ----------------------------------------------------------------
  *				Section 3:	standard system types
diff --git a/src/include/common/unicode_case.h b/src/include/common/unicode_case.h
index 41e2c1f4b33..6bcffd349c2 100644
--- a/src/include/common/unicode_case.h
+++ b/src/include/common/unicode_case.h
@@ -14,14 +14,12 @@
 #ifndef UNICODE_CASE_H
 #define UNICODE_CASE_H
 
-#include "mb/pg_wchar.h"
-
 typedef size_t (*WordBoundaryNext) (void *wbstate);
 
-pg_wchar	unicode_lowercase_simple(pg_wchar code);
-pg_wchar	unicode_titlecase_simple(pg_wchar code);
-pg_wchar	unicode_uppercase_simple(pg_wchar code);
-pg_wchar	unicode_casefold_simple(pg_wchar code);
+char32_t	unicode_lowercase_simple(char32_t code);
+char32_t	unicode_titlecase_simple(char32_t code);
+char32_t	unicode_uppercase_simple(char32_t code);
+char32_t	unicode_casefold_simple(char32_t code);
 size_t		unicode_strlower(char *dst, size_t dstsize, const char *src,
 							 ssize_t srclen, bool full);
 size_t		unicode_strtitle(char *dst, size_t dstsize, const char *src,
diff --git a/src/include/common/unicode_case_table.h b/src/include/common/unicode_case_table.h
index d5311786582..0a14fb2d97b 100644
--- a/src/include/common/unicode_case_table.h
+++ b/src/include/common/unicode_case_table.h
@@ -18,7 +18,6 @@
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -45,7 +44,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -166,7 +165,7 @@ static const pg_special_case special_case[106] =
  * The entry case_map_lower[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_lower[1704] =
+static const char32_t case_map_lower[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -1879,7 +1878,7 @@ static const pg_wchar case_map_lower[1704] =
  * The entry case_map_title[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_title[1704] =
+static const char32_t case_map_title[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -3592,7 +3591,7 @@ static const pg_wchar case_map_title[1704] =
  * The entry case_map_upper[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_upper[1704] =
+static const char32_t case_map_upper[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -5305,7 +5304,7 @@ static const pg_wchar case_map_upper[1704] =
  * The entry case_map_fold[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_fold[1704] =
+static const char32_t case_map_fold[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -13522,7 +13521,7 @@ static const uint16 case_map[4778] =
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < 0x0588 */
 	if (cp < 0x0588)
diff --git a/src/include/common/unicode_category.h b/src/include/common/unicode_category.h
index 8fd8b67a416..684143d3c8a 100644
--- a/src/include/common/unicode_category.h
+++ b/src/include/common/unicode_category.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_CATEGORY_H
 #define UNICODE_CATEGORY_H
 
-#include "mb/pg_wchar.h"
-
 /*
  * Unicode General Category Values
  *
@@ -61,31 +59,31 @@ typedef enum pg_unicode_category
 	PG_U_FINAL_PUNCTUATION = 29 /* Pf */
 } pg_unicode_category;
 
-extern pg_unicode_category unicode_category(pg_wchar code);
+extern pg_unicode_category unicode_category(char32_t code);
 extern const char *unicode_category_string(pg_unicode_category category);
 extern const char *unicode_category_abbrev(pg_unicode_category category);
 
-extern bool pg_u_prop_alphabetic(pg_wchar code);
-extern bool pg_u_prop_lowercase(pg_wchar code);
-extern bool pg_u_prop_uppercase(pg_wchar code);
-extern bool pg_u_prop_cased(pg_wchar code);
-extern bool pg_u_prop_case_ignorable(pg_wchar code);
-extern bool pg_u_prop_white_space(pg_wchar code);
-extern bool pg_u_prop_hex_digit(pg_wchar code);
-extern bool pg_u_prop_join_control(pg_wchar code);
+extern bool pg_u_prop_alphabetic(char32_t code);
+extern bool pg_u_prop_lowercase(char32_t code);
+extern bool pg_u_prop_uppercase(char32_t code);
+extern bool pg_u_prop_cased(char32_t code);
+extern bool pg_u_prop_case_ignorable(char32_t code);
+extern bool pg_u_prop_white_space(char32_t code);
+extern bool pg_u_prop_hex_digit(char32_t code);
+extern bool pg_u_prop_join_control(char32_t code);
 
-extern bool pg_u_isdigit(pg_wchar code, bool posix);
-extern bool pg_u_isalpha(pg_wchar code);
-extern bool pg_u_isalnum(pg_wchar code, bool posix);
-extern bool pg_u_isword(pg_wchar code);
-extern bool pg_u_isupper(pg_wchar code);
-extern bool pg_u_islower(pg_wchar code);
-extern bool pg_u_isblank(pg_wchar code);
-extern bool pg_u_iscntrl(pg_wchar code);
-extern bool pg_u_isgraph(pg_wchar code);
-extern bool pg_u_isprint(pg_wchar code);
-extern bool pg_u_ispunct(pg_wchar code, bool posix);
-extern bool pg_u_isspace(pg_wchar code);
-extern bool pg_u_isxdigit(pg_wchar code, bool posix);
+extern bool pg_u_isdigit(char32_t code, bool posix);
+extern bool pg_u_isalpha(char32_t code);
+extern bool pg_u_isalnum(char32_t code, bool posix);
+extern bool pg_u_isword(char32_t code);
+extern bool pg_u_isupper(char32_t code);
+extern bool pg_u_islower(char32_t code);
+extern bool pg_u_isblank(char32_t code);
+extern bool pg_u_iscntrl(char32_t code);
+extern bool pg_u_isgraph(char32_t code);
+extern bool pg_u_isprint(char32_t code);
+extern bool pg_u_ispunct(char32_t code, bool posix);
+extern bool pg_u_isspace(char32_t code);
+extern bool pg_u_isxdigit(char32_t code, bool posix);
 
 #endif							/* UNICODE_CATEGORY_H */
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 95a1c65da7e..466a41b72b0 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -20,15 +20,15 @@
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/include/common/unicode_norm.h b/src/include/common/unicode_norm.h
index 5bc3b79e78e..516c192cc4c 100644
--- a/src/include/common/unicode_norm.h
+++ b/src/include/common/unicode_norm.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_NORM_H
 #define UNICODE_NORM_H
 
-#include "mb/pg_wchar.h"
-
 typedef enum
 {
 	UNICODE_NFC = 0,
@@ -32,8 +30,8 @@ typedef enum
 	UNICODE_NORM_QC_MAYBE = -1,
 } UnicodeNormalizationQC;
 
-extern pg_wchar *unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input);
+extern char32_t *unicode_normalize(UnicodeNormalizationForm form, const char32_t *input);
 
-extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input);
+extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input);
 
 #endif							/* UNICODE_NORM_H */
diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h
index 4b4a9974b75..4d84bdc81e4 100644
--- a/src/include/mb/pg_wchar.h
+++ b/src/include/mb/pg_wchar.h
@@ -532,25 +532,25 @@ typedef uint32 (*utf_local_conversion_func) (uint32 code);
  * Some handy functions for Unicode-specific tests.
  */
 static inline bool
-is_valid_unicode_codepoint(pg_wchar c)
+is_valid_unicode_codepoint(char32_t c)
 {
 	return (c > 0 && c <= 0x10FFFF);
 }
 
 static inline bool
-is_utf16_surrogate_first(pg_wchar c)
+is_utf16_surrogate_first(char32_t c)
 {
 	return (c >= 0xD800 && c <= 0xDBFF);
 }
 
 static inline bool
-is_utf16_surrogate_second(pg_wchar c)
+is_utf16_surrogate_second(char32_t c)
 {
 	return (c >= 0xDC00 && c <= 0xDFFF);
 }
 
-static inline pg_wchar
-surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
+static inline char32_t
+surrogate_pair_to_codepoint(char16_t first, char16_t second)
 {
 	return ((first & 0x3FF) << 10) + 0x10000 + (second & 0x3FF);
 }
@@ -561,20 +561,20 @@ surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static inline pg_wchar
+static inline char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
@@ -588,7 +588,7 @@ utf8_to_unicode(const unsigned char *c)
  * unicode_utf8len(c) bytes available.
  */
 static inline unsigned char *
-unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
+unicode_to_utf8(char32_t c, unsigned char *utf8string)
 {
 	if (c <= 0x7F)
 	{
@@ -620,7 +620,7 @@ unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
  * Number of bytes needed to represent the given char in UTF8.
  */
 static inline int
-unicode_utf8len(pg_wchar c)
+unicode_utf8len(char32_t c)
 {
 	if (c <= 0x7F)
 		return 1;
@@ -676,8 +676,8 @@ extern int	pg_valid_server_encoding(const char *name);
 extern bool is_encoding_supported_by_icu(int encoding);
 extern const char *get_encoding_name_for_icu(int encoding);
 
-extern unsigned char *unicode_to_utf8(pg_wchar c, unsigned char *utf8string);
-extern pg_wchar utf8_to_unicode(const unsigned char *c);
+extern unsigned char *unicode_to_utf8(char32_t c, unsigned char *utf8string);
+extern char32_t utf8_to_unicode(const unsigned char *c);
 extern bool pg_utf8_islegal(const unsigned char *source, int length);
 extern int	pg_utf_mblen(const unsigned char *s);
 extern int	pg_mule_mblen(const unsigned char *s);
@@ -739,8 +739,8 @@ extern char *pg_server_to_client(const char *s, int len);
 extern char *pg_any_to_server(const char *s, int len, int encoding);
 extern char *pg_server_to_any(const char *s, int len, int encoding);
 
-extern void pg_unicode_to_server(pg_wchar c, unsigned char *s);
-extern bool pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s);
+extern void pg_unicode_to_server(char32_t c, unsigned char *s);
+extern bool pg_unicode_to_server_noerror(char32_t c, unsigned char *s);
 
 extern unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc);
 extern unsigned short CNStoBIG5(unsigned short cns, unsigned char lc);
diff --git a/src/include/pg_config.h.in b/src/include/pg_config.h.in
index c4dc5d72bdb..4a150fbd551 100644
--- a/src/include/pg_config.h.in
+++ b/src/include/pg_config.h.in
@@ -373,6 +373,12 @@
 /* Define to 1 if the system has the type `socklen_t'. */
 #undef HAVE_SOCKLEN_T
 
+/* Define to 1 if the system has the type `char16_t'. */
+#undef HAVE_CHAR16_T
+
+/* Define to 1 if the system has the type `char32_t'. */
+#undef HAVE_CHAR32_T
+
 /* Define to 1 if you have the `SSL_CTX_set_cert_cb' function. */
 #undef HAVE_SSL_CTX_SET_CERT_CB
 
@@ -463,6 +469,9 @@
 /* Define to 1 if you have the <termios.h> header file. */
 #undef HAVE_TERMIOS_H
 
+/* Define to 1 if you have the <uchar.h> header file. */
+#undef HAVE_UCHAR_H
+
 /* Define to 1 if curl_global_init() is guaranteed to be thread-safe. */
 #undef HAVE_THREADSAFE_CURL_GLOBAL_INIT
 
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index bb4e1b37005..790be386d75 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -3505,6 +3505,8 @@ cb_cleanup_dir
 cb_options
 cb_tablespace
 cb_tablespace_mapping
+char16_t
+char32_t
 check_agg_arguments_context
 check_function_callback
 check_network_data
-- 
2.43.0

From ceecbc970e7fcd026edbbe7cc324ea8fe74c571c Mon Sep 17 00:00:00 2001
From: Jeff Davis <jeff@j-davis.com>
Date: Tue, 21 Oct 2025 13:16:47 -0700
Subject: [PATCH v4] Use C11 char16_t and char32_t for Unicode code points.

Reviewed-by: Tatsuo Ishii <ishii@postgresql.org>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Reviewed-by: Peter Eisentraut <peter@eisentraut.org>
Discussion: https://postgr.es/m/bedcc93d06203dfd89815b10f815ca2de8626e85.camel%40j-davis.com
---
 configure                                     |  2 +-
 configure.ac                                  |  1 +
 meson.build                                   |  1 +
 src/backend/parser/parser.c                   |  8 +--
 src/backend/parser/scan.l                     |  8 +--
 src/backend/utils/adt/jsonpath_scan.l         |  6 +-
 src/backend/utils/adt/pg_locale_builtin.c     | 44 ++++++++++-----
 src/backend/utils/adt/varlena.c               | 40 ++++++-------
 src/backend/utils/mb/mbutils.c                |  4 +-
 src/common/saslprep.c                         | 48 ++++++++--------
 src/common/unicode/case_test.c                | 23 ++++----
 src/common/unicode/category_test.c            |  3 +-
 .../unicode/generate-norm_test_table.pl       |  4 +-
 .../unicode/generate-unicode_case_table.pl    |  7 +--
 .../generate-unicode_category_table.pl        |  8 +--
 src/common/unicode/norm_test.c                |  6 +-
 src/common/unicode_case.c                     | 56 +++++++++----------
 src/common/unicode_category.c                 | 50 ++++++++---------
 src/common/unicode_norm.c                     | 56 +++++++++----------
 src/fe_utils/mbprint.c                        | 10 ++--
 src/include/c.h                               | 18 ++++++
 src/include/common/unicode_case.h             | 10 ++--
 src/include/common/unicode_case_table.h       | 13 ++---
 src/include/common/unicode_category.h         | 46 ++++++++-------
 src/include/common/unicode_category_table.h   |  8 +--
 src/include/common/unicode_norm.h             |  6 +-
 src/include/mb/pg_wchar.h                     | 32 +++++------
 src/include/pg_config.h.in                    |  3 +
 src/tools/pgindent/typedefs.list              |  2 +
 29 files changed, 279 insertions(+), 244 deletions(-)

diff --git a/configure b/configure
index 22cd866147b..600a56f91a3 100755
--- a/configure
+++ b/configure
@@ -13627,7 +13627,7 @@ fi
 ## Header files
 ##
 
-for ac_header in atomic.h copyfile.h execinfo.h getopt.h ifaddrs.h mbarrier.h sys/epoll.h sys/event.h sys/personality.h sys/prctl.h sys/procctl.h sys/signalfd.h sys/ucred.h termios.h ucred.h xlocale.h
+for ac_header in atomic.h copyfile.h execinfo.h getopt.h ifaddrs.h mbarrier.h sys/epoll.h sys/event.h sys/personality.h sys/prctl.h sys/procctl.h sys/signalfd.h sys/ucred.h termios.h uchar.h ucred.h xlocale.h
 do :
   as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
 ac_fn_c_check_header_mongrel "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default"
diff --git a/configure.ac b/configure.ac
index e44943aa6fe..6ab2e157531 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1513,6 +1513,7 @@ AC_CHECK_HEADERS(m4_normalize([
 	sys/signalfd.h
 	sys/ucred.h
 	termios.h
+	uchar.h
 	ucred.h
 	xlocale.h
 ]))
diff --git a/meson.build b/meson.build
index 395416a6060..c3128c7554f 100644
--- a/meson.build
+++ b/meson.build
@@ -2613,6 +2613,7 @@ header_checks = [
   'sys/signalfd.h',
   'sys/ucred.h',
   'termios.h',
+  'uchar.h',
   'ucred.h',
   'xlocale.h',
 ]
diff --git a/src/backend/parser/parser.c b/src/backend/parser/parser.c
index 33a040506b4..a3679f8e86c 100644
--- a/src/backend/parser/parser.c
+++ b/src/backend/parser/parser.c
@@ -339,7 +339,7 @@ hexval(unsigned char c)
 
 /* is Unicode code point acceptable? */
 static void
-check_unicode_value(pg_wchar c)
+check_unicode_value(char32_t c)
 {
 	if (!is_valid_unicode_codepoint(c))
 		ereport(ERROR,
@@ -376,7 +376,7 @@ str_udeescape(const char *str, char escape,
 	char	   *new,
 			   *out;
 	size_t		new_len;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	ScannerCallbackState scbstate;
 
 	/*
@@ -420,7 +420,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[3]) &&
 					 isxdigit((unsigned char) in[4]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[1]) << 12) +
 					(hexval(in[2]) << 8) +
@@ -457,7 +457,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[6]) &&
 					 isxdigit((unsigned char) in[7]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[2]) << 20) +
 					(hexval(in[3]) << 16) +
diff --git a/src/backend/parser/scan.l b/src/backend/parser/scan.l
index 08990831fe8..a67815339b7 100644
--- a/src/backend/parser/scan.l
+++ b/src/backend/parser/scan.l
@@ -121,7 +121,7 @@ static void addlitchar(unsigned char ychar, core_yyscan_t yyscanner);
 static char *litbufdup(core_yyscan_t yyscanner);
 static unsigned char unescape_single_char(unsigned char c, core_yyscan_t yyscanner);
 static int	process_integer_literal(const char *token, YYSTYPE *lval, int base);
-static void addunicode(pg_wchar c, yyscan_t yyscanner);
+static void addunicode(char32_t c, yyscan_t yyscanner);
 
 #define yyerror(msg)  scanner_yyerror(msg, yyscanner)
 
@@ -640,7 +640,7 @@ other			.
 					addlit(yytext, yyleng, yyscanner);
 				}
 <xe>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/*
 					 * For consistency with other productions, issue any
@@ -668,7 +668,7 @@ other			.
 					POP_YYLLOC();
 				}
 <xeu>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/* Remember start of overall string token ... */
 					PUSH_YYLLOC();
@@ -1376,7 +1376,7 @@ process_integer_literal(const char *token, YYSTYPE *lval, int base)
 }
 
 static void
-addunicode(pg_wchar c, core_yyscan_t yyscanner)
+addunicode(char32_t c, core_yyscan_t yyscanner)
 {
 	ScannerCallbackState scbstate;
 	char		buf[MAX_UNICODE_EQUIVALENT_STRING + 1];
diff --git a/src/backend/utils/adt/jsonpath_scan.l b/src/backend/utils/adt/jsonpath_scan.l
index c7aab83eeb4..8c3a0a9c642 100644
--- a/src/backend/utils/adt/jsonpath_scan.l
+++ b/src/backend/utils/adt/jsonpath_scan.l
@@ -574,7 +574,7 @@ hexval(char c, int *result, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add given unicode character to scanstring */
 static bool
-addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
+addUnicodeChar(char32_t ch, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (ch == 0)
 	{
@@ -607,7 +607,7 @@ addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add unicode character, processing any surrogate pairs */
 static bool
-addUnicode(int ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
+addUnicode(char32_t ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (is_utf16_surrogate_first(ch))
 	{
@@ -655,7 +655,7 @@ parseUnicode(char *s, int l, struct Node *escontext, yyscan_t yyscanner)
 
 	for (i = 2; i < l; i += 2)	/* skip '\u' */
 	{
-		int			ch = 0;
+		char32_t		ch = 0;
 		int			j,
 					si;
 
diff --git a/src/backend/utils/adt/pg_locale_builtin.c b/src/backend/utils/adt/pg_locale_builtin.c
index 3dc611b50e1..1021e0d129b 100644
--- a/src/backend/utils/adt/pg_locale_builtin.c
+++ b/src/backend/utils/adt/pg_locale_builtin.c
@@ -15,7 +15,6 @@
 #include "catalog/pg_collation.h"
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
-#include "mb/pg_wchar.h"
 #include "miscadmin.h"
 #include "utils/builtins.h"
 #include "utils/pg_locale.h"
@@ -35,6 +34,23 @@ struct WordBoundaryState
 	bool		prev_alnum;
 };
 
+/*
+ * In UTF-8, pg_wchar is guaranteed to be the code point value.
+ */
+static inline char32_t
+to_char32(pg_wchar wc)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (char32_t) wc;
+}
+
+static inline pg_wchar
+to_pg_wchar(char32_t c32)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (pg_wchar) c32;
+}
+
 /*
  * Simple word boundary iterator that draws boundaries each time the result of
  * pg_u_isalnum() changes.
@@ -47,7 +63,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -112,61 +128,61 @@ strfold_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen,
 static bool
 wc_isdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isalpha_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalpha(wc);
+	return pg_u_isalpha(to_char32(wc));
 }
 
 static bool
 wc_isalnum_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalnum(wc, !locale->builtin.casemap_full);
+	return pg_u_isalnum(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isupper(wc);
+	return pg_u_isupper(to_char32(wc));
 }
 
 static bool
 wc_islower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_islower(wc);
+	return pg_u_islower(to_char32(wc));
 }
 
 static bool
 wc_isgraph_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isgraph(wc);
+	return pg_u_isgraph(to_char32(wc));
 }
 
 static bool
 wc_isprint_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isprint(wc);
+	return pg_u_isprint(to_char32(wc));
 }
 
 static bool
 wc_ispunct_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_ispunct(wc, !locale->builtin.casemap_full);
+	return pg_u_ispunct(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isspace_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isspace(wc);
+	return pg_u_isspace(to_char32(wc));
 }
 
 static bool
 wc_isxdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isxdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isxdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
@@ -179,13 +195,13 @@ char_is_cased_builtin(char ch, pg_locale_t locale)
 static pg_wchar
 wc_toupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_uppercase_simple(wc);
+	return to_pg_wchar(unicode_uppercase_simple(to_char32(wc)));
 }
 
 static pg_wchar
 wc_tolower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_lowercase_simple(wc);
+	return to_pg_wchar(unicode_lowercase_simple(to_char32(wc)));
 }
 
 static const struct ctype_methods ctype_methods_builtin = {
diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c
index 2c398cd9e5c..8d735786e51 100644
--- a/src/backend/utils/adt/varlena.c
+++ b/src/backend/utils/adt/varlena.c
@@ -5419,12 +5419,12 @@ unicode_assigned(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errmsg("Unicode categorization can only be performed if server encoding is UTF8")));
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (int i = 0; i < size; i++)
 	{
-		pg_wchar	uchar = utf8_to_unicode(p);
+		char32_t	uchar = utf8_to_unicode(p);
 		int			category = unicode_category(uchar);
 
 		if (category == PG_U_UNASSIGNED)
@@ -5443,24 +5443,24 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	text	   *result;
 	int			i;
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* action */
@@ -5468,7 +5468,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 
 	/* convert back to UTF-8 string */
 	size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unsigned char buf[4];
 
@@ -5480,7 +5480,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	SET_VARSIZE(result, size + VARHDRSZ);
 
 	p = (unsigned char *) VARDATA_ANY(result);
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unicode_to_utf8(*wp, p);
 		p += pg_utf_mblen(p);
@@ -5509,8 +5509,8 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	int			i;
 	UnicodeNormalizationQC quickcheck;
@@ -5519,16 +5519,16 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* quick check (see UAX #15) */
@@ -5542,11 +5542,11 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	output_chars = unicode_normalize(form, input_chars);
 
 	output_size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 		output_size++;
 
 	result = (size == output_size) &&
-		(memcmp(input_chars, output_chars, size * sizeof(pg_wchar)) == 0);
+		(memcmp(input_chars, output_chars, size * sizeof(char32_t)) == 0);
 
 	PG_RETURN_BOOL(result);
 }
@@ -5602,7 +5602,7 @@ unistr(PG_FUNCTION_ARGS)
 	int			len;
 	StringInfoData str;
 	text	   *result;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	char		cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
 
 	instr = VARDATA_ANY(input_text);
@@ -5626,7 +5626,7 @@ unistr(PG_FUNCTION_ARGS)
 			else if ((len >= 5 && isxdigits_n(instr + 1, 4)) ||
 					 (len >= 6 && instr[1] == 'u' && isxdigits_n(instr + 2, 4)))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 				int			offset = instr[1] == 'u' ? 2 : 1;
 
 				unicode = hexval_n(instr + offset, 4);
@@ -5662,7 +5662,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 8 && instr[1] == '+' && isxdigits_n(instr + 2, 6))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 6);
 
@@ -5697,7 +5697,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 10 && instr[1] == 'U' && isxdigits_n(instr + 2, 8))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 8);
 
diff --git a/src/backend/utils/mb/mbutils.c b/src/backend/utils/mb/mbutils.c
index 886ecbad871..fb629ed5c8f 100644
--- a/src/backend/utils/mb/mbutils.c
+++ b/src/backend/utils/mb/mbutils.c
@@ -862,7 +862,7 @@ perform_default_encoding_conversion(const char *src, int len,
  * may call this outside any transaction, or in an aborted transaction.
  */
 void
-pg_unicode_to_server(pg_wchar c, unsigned char *s)
+pg_unicode_to_server(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
@@ -924,7 +924,7 @@ pg_unicode_to_server(pg_wchar c, unsigned char *s)
  * but simply return false on conversion failure.
  */
 bool
-pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s)
+pg_unicode_to_server_noerror(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
diff --git a/src/common/saslprep.c b/src/common/saslprep.c
index 97beb47940b..101e8d65a4d 100644
--- a/src/common/saslprep.c
+++ b/src/common/saslprep.c
@@ -47,7 +47,7 @@
 
 /* Prototypes for local functions */
 static int	codepoint_range_cmp(const void *a, const void *b);
-static bool is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize);
+static bool is_code_in_table(char32_t code, const char32_t *map, int mapsize);
 static int	pg_utf8_string_len(const char *source);
 
 /*
@@ -64,7 +64,7 @@ static int	pg_utf8_string_len(const char *source);
  *
  * These are all mapped to the ASCII space character (U+00A0).
  */
-static const pg_wchar non_ascii_space_ranges[] =
+static const char32_t non_ascii_space_ranges[] =
 {
 	0x00A0, 0x00A0,
 	0x1680, 0x1680,
@@ -79,7 +79,7 @@ static const pg_wchar non_ascii_space_ranges[] =
  *
  * If any of these appear in the input, they are removed.
  */
-static const pg_wchar commonly_mapped_to_nothing_ranges[] =
+static const char32_t commonly_mapped_to_nothing_ranges[] =
 {
 	0x00AD, 0x00AD,
 	0x034F, 0x034F,
@@ -114,7 +114,7 @@ static const pg_wchar commonly_mapped_to_nothing_ranges[] =
  * tables, so one code might originate from multiple source tables.
  * Adjacent ranges have also been merged together, to save space.
  */
-static const pg_wchar prohibited_output_ranges[] =
+static const char32_t prohibited_output_ranges[] =
 {
 	0x0000, 0x001F,				/* C.2.1 */
 	0x007F, 0x00A0,				/* C.1.2, C.2.1, C.2.2 */
@@ -155,7 +155,7 @@ static const pg_wchar prohibited_output_ranges[] =
 };
 
 /* A.1 Unassigned code points in Unicode 3.2 */
-static const pg_wchar unassigned_codepoint_ranges[] =
+static const char32_t unassigned_codepoint_ranges[] =
 {
 	0x0221, 0x0221,
 	0x0234, 0x024F,
@@ -556,7 +556,7 @@ static const pg_wchar unassigned_codepoint_ranges[] =
 };
 
 /* D.1 Characters with bidirectional property "R" or "AL" */
-static const pg_wchar RandALCat_codepoint_ranges[] =
+static const char32_t RandALCat_codepoint_ranges[] =
 {
 	0x05BE, 0x05BE,
 	0x05C0, 0x05C0,
@@ -595,7 +595,7 @@ static const pg_wchar RandALCat_codepoint_ranges[] =
 };
 
 /* D.2 Characters with bidirectional property "L" */
-static const pg_wchar LCat_codepoint_ranges[] =
+static const char32_t LCat_codepoint_ranges[] =
 {
 	0x0041, 0x005A,
 	0x0061, 0x007A,
@@ -968,8 +968,8 @@ static const pg_wchar LCat_codepoint_ranges[] =
 static int
 codepoint_range_cmp(const void *a, const void *b)
 {
-	const pg_wchar *key = (const pg_wchar *) a;
-	const pg_wchar *range = (const pg_wchar *) b;
+	const char32_t *key = (const char32_t *) a;
+	const char32_t *range = (const char32_t *) b;
 
 	if (*key < range[0])
 		return -1;				/* less than lower bound */
@@ -980,14 +980,14 @@ codepoint_range_cmp(const void *a, const void *b)
 }
 
 static bool
-is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize)
+is_code_in_table(char32_t code, const char32_t *map, int mapsize)
 {
 	Assert(mapsize % 2 == 0);
 
 	if (code < map[0] || code > map[mapsize - 1])
 		return false;
 
-	if (bsearch(&code, map, mapsize / 2, sizeof(pg_wchar) * 2,
+	if (bsearch(&code, map, mapsize / 2, sizeof(char32_t) * 2,
 				codepoint_range_cmp))
 		return true;
 	else
@@ -1046,8 +1046,8 @@ pg_utf8_string_len(const char *source)
 pg_saslprep_rc
 pg_saslprep(const char *input, char **output)
 {
-	pg_wchar   *input_chars = NULL;
-	pg_wchar   *output_chars = NULL;
+	char32_t   *input_chars = NULL;
+	char32_t   *output_chars = NULL;
 	int			input_size;
 	char	   *result;
 	int			result_size;
@@ -1055,7 +1055,7 @@ pg_saslprep(const char *input, char **output)
 	int			i;
 	bool		contains_RandALCat;
 	unsigned char *p;
-	pg_wchar   *wp;
+	char32_t   *wp;
 
 	/* Ensure we return *output as NULL on failure */
 	*output = NULL;
@@ -1080,10 +1080,10 @@ pg_saslprep(const char *input, char **output)
 	input_size = pg_utf8_string_len(input);
 	if (input_size < 0)
 		return SASLPREP_INVALID_UTF8;
-	if (input_size >= MaxAllocSize / sizeof(pg_wchar))
+	if (input_size >= MaxAllocSize / sizeof(char32_t))
 		goto oom;
 
-	input_chars = ALLOC((input_size + 1) * sizeof(pg_wchar));
+	input_chars = ALLOC((input_size + 1) * sizeof(char32_t));
 	if (!input_chars)
 		goto oom;
 
@@ -1093,7 +1093,7 @@ pg_saslprep(const char *input, char **output)
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 
 	/*
 	 * The steps below correspond to the steps listed in [RFC3454], Section
@@ -1107,7 +1107,7 @@ pg_saslprep(const char *input, char **output)
 	count = 0;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, non_ascii_space_ranges))
 			input_chars[count++] = 0x0020;
@@ -1118,7 +1118,7 @@ pg_saslprep(const char *input, char **output)
 		else
 			input_chars[count++] = code;
 	}
-	input_chars[count] = (pg_wchar) '\0';
+	input_chars[count] = (char32_t) '\0';
 	input_size = count;
 
 	if (input_size == 0)
@@ -1138,7 +1138,7 @@ pg_saslprep(const char *input, char **output)
 	 */
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, prohibited_output_ranges))
 			goto prohibited;
@@ -1170,7 +1170,7 @@ pg_saslprep(const char *input, char **output)
 	contains_RandALCat = false;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, RandALCat_codepoint_ranges))
 		{
@@ -1181,12 +1181,12 @@ pg_saslprep(const char *input, char **output)
 
 	if (contains_RandALCat)
 	{
-		pg_wchar	first = input_chars[0];
-		pg_wchar	last = input_chars[input_size - 1];
+		char32_t	first = input_chars[0];
+		char32_t	last = input_chars[input_size - 1];
 
 		for (i = 0; i < input_size; i++)
 		{
-			pg_wchar	code = input_chars[i];
+			char32_t	code = input_chars[i];
 
 			if (IS_CODE_IN_TABLE(code, LCat_codepoint_ranges))
 				goto prohibited;
diff --git a/src/common/unicode/case_test.c b/src/common/unicode/case_test.c
index fdfb62e8552..00d4f85e5a5 100644
--- a/src/common/unicode/case_test.c
+++ b/src/common/unicode/case_test.c
@@ -24,6 +24,7 @@
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 /* enough to hold largest source or result string, including NUL */
 #define BUFSZ 256
@@ -54,7 +55,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -77,16 +78,16 @@ initcap_wbnext(void *state)
 #ifdef USE_ICU
 
 static void
-icu_test_simple(pg_wchar code)
+icu_test_simple(char32_t code)
 {
-	pg_wchar	lower = unicode_lowercase_simple(code);
-	pg_wchar	title = unicode_titlecase_simple(code);
-	pg_wchar	upper = unicode_uppercase_simple(code);
-	pg_wchar	fold = unicode_casefold_simple(code);
-	pg_wchar	iculower = u_tolower(code);
-	pg_wchar	icutitle = u_totitle(code);
-	pg_wchar	icuupper = u_toupper(code);
-	pg_wchar	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
+	char32_t	lower = unicode_lowercase_simple(code);
+	char32_t	title = unicode_titlecase_simple(code);
+	char32_t	upper = unicode_uppercase_simple(code);
+	char32_t	fold = unicode_casefold_simple(code);
+	char32_t	iculower = u_tolower(code);
+	char32_t	icutitle = u_totitle(code);
+	char32_t	icuupper = u_toupper(code);
+	char32_t	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
 
 	if (lower != iculower || title != icutitle || upper != icuupper ||
 		fold != icufold)
@@ -172,7 +173,7 @@ test_icu(void)
 	int			successful = 0;
 	int			skipped_mismatch = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		pg_unicode_category category = unicode_category(code);
 
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index 5d37ba39196..1e8c1f7905f 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -22,6 +22,7 @@
 
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 static int	pg_unicode_version = 0;
 #ifdef USE_ICU
@@ -59,7 +60,7 @@ icu_test()
 	int			pg_skipped_codepoints = 0;
 	int			icu_skipped_codepoints = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		uint8_t		pg_category = unicode_category(code);
 		uint8_t		icu_category = u_charType(code);
diff --git a/src/common/unicode/generate-norm_test_table.pl b/src/common/unicode/generate-norm_test_table.pl
index 1b401be9409..1a8b908ff33 100644
--- a/src/common/unicode/generate-norm_test_table.pl
+++ b/src/common/unicode/generate-norm_test_table.pl
@@ -47,8 +47,8 @@ print $OUTPUT <<HEADER;
 typedef struct
 {
 	int			linenum;
-	pg_wchar	input[50];
-	pg_wchar	output[4][50];
+	char32_t	input[50];
+	char32_t	output[4][50];
 } pg_unicode_test;
 
 /* test table */
diff --git a/src/common/unicode/generate-unicode_case_table.pl b/src/common/unicode/generate-unicode_case_table.pl
index 5d9ddd62803..f71eb25c94e 100644
--- a/src/common/unicode/generate-unicode_case_table.pl
+++ b/src/common/unicode/generate-unicode_case_table.pl
@@ -270,7 +270,6 @@ print $OT <<"EOS";
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -297,7 +296,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -430,7 +429,7 @@ foreach my $kind ('lower', 'title', 'upper', 'fold')
  * The entry case_map_${kind}[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_$kind\[$index\] =
+static const char32_t case_map_$kind\[$index\] =
 {
 EOS
 
@@ -502,7 +501,7 @@ print $OT <<"EOS";
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < $fastpath_limit */
 	if (cp < $fastpath_limit)
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index abab5cd9696..7e094b13720 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -366,15 +366,15 @@ print $OT <<"EOS";
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/common/unicode/norm_test.c b/src/common/unicode/norm_test.c
index 25bc59463f2..058817f1719 100644
--- a/src/common/unicode/norm_test.c
+++ b/src/common/unicode/norm_test.c
@@ -20,7 +20,7 @@
 #include "norm_test_table.h"
 
 static char *
-print_wchar_str(const pg_wchar *s)
+print_wchar_str(const char32_t *s)
 {
 #define BUF_DIGITS 50
 	static char buf[BUF_DIGITS * 11 + 1];
@@ -41,7 +41,7 @@ print_wchar_str(const pg_wchar *s)
 }
 
 static int
-pg_wcscmp(const pg_wchar *s1, const pg_wchar *s2)
+pg_wcscmp(const char32_t *s1, const char32_t *s2)
 {
 	for (;;)
 	{
@@ -65,7 +65,7 @@ main(int argc, char **argv)
 	{
 		for (int form = 0; form < 4; form++)
 		{
-			pg_wchar   *result;
+			char32_t   *result;
 
 			result = unicode_normalize(form, test->input);
 
diff --git a/src/common/unicode_case.c b/src/common/unicode_case.c
index 073faf6a0d5..e5e494db43c 100644
--- a/src/common/unicode_case.c
+++ b/src/common/unicode_case.c
@@ -30,7 +30,7 @@ enum CaseMapResult
 /*
  * Map for each case kind.
  */
-static const pg_wchar *const casekind_map[NCaseKind] =
+static const char32_t *const casekind_map[NCaseKind] =
 {
 	[CaseLower] = case_map_lower,
 	[CaseTitle] = case_map_title,
@@ -38,42 +38,42 @@ static const pg_wchar *const casekind_map[NCaseKind] =
 	[CaseFold] = case_map_fold,
 };
 
-static pg_wchar find_case_map(pg_wchar ucs, const pg_wchar *map);
+static char32_t find_case_map(char32_t ucs, const char32_t *map);
 static size_t convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 						   CaseKind str_casekind, bool full, WordBoundaryNext wbnext,
 						   void *wbstate);
-static enum CaseMapResult casemap(pg_wchar u1, CaseKind casekind, bool full,
+static enum CaseMapResult casemap(char32_t u1, CaseKind casekind, bool full,
 								  const char *src, size_t srclen, size_t srcoff,
-								  pg_wchar *simple, const pg_wchar **special);
+								  char32_t *simple, const char32_t **special);
 
-pg_wchar
-unicode_lowercase_simple(pg_wchar code)
+char32_t
+unicode_lowercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_lower);
+	char32_t	cp = find_case_map(code, case_map_lower);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_titlecase_simple(pg_wchar code)
+char32_t
+unicode_titlecase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_title);
+	char32_t	cp = find_case_map(code, case_map_title);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_uppercase_simple(pg_wchar code)
+char32_t
+unicode_uppercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_upper);
+	char32_t	cp = find_case_map(code, case_map_upper);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_casefold_simple(pg_wchar code)
+char32_t
+unicode_casefold_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_fold);
+	char32_t	cp = find_case_map(code, case_map_fold);
 
 	return cp != 0 ? cp : code;
 }
@@ -231,10 +231,10 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 
 	while ((srclen < 0 || srcoff < srclen) && src[srcoff] != '\0')
 	{
-		pg_wchar	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
+		char32_t	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
 		int			u1len = unicode_utf8len(u1);
-		pg_wchar	simple = 0;
-		const pg_wchar *special = NULL;
+		char32_t	simple = 0;
+		const char32_t *special = NULL;
 		enum CaseMapResult casemap_result;
 
 		if (str_casekind == CaseTitle)
@@ -265,8 +265,8 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 			case CASEMAP_SIMPLE:
 				{
 					/* replace with single character */
-					pg_wchar	u2 = simple;
-					pg_wchar	u2len = unicode_utf8len(u2);
+					char32_t	u2 = simple;
+					char32_t	u2len = unicode_utf8len(u2);
 
 					Assert(special == NULL);
 					if (result_len + u2len <= dstsize)
@@ -280,7 +280,7 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 				Assert(simple == 0);
 				for (int i = 0; i < MAX_CASE_EXPANSION && special[i]; i++)
 				{
-					pg_wchar	u2 = special[i];
+					char32_t	u2 = special[i];
 					size_t		u2len = unicode_utf8len(u2);
 
 					if (result_len + u2len <= dstsize)
@@ -320,7 +320,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -344,7 +344,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -394,9 +394,9 @@ check_special_conditions(int conditions, const char *str, size_t len,
  * character without modification.
  */
 static enum CaseMapResult
-casemap(pg_wchar u1, CaseKind casekind, bool full,
+casemap(char32_t u1, CaseKind casekind, bool full,
 		const char *src, size_t srclen, size_t srcoff,
-		pg_wchar *simple, const pg_wchar **special)
+		char32_t *simple, const char32_t **special)
 {
 	uint16		idx;
 
@@ -434,8 +434,8 @@ casemap(pg_wchar u1, CaseKind casekind, bool full,
  * Find entry in simple case map.
  * If the entry does not exist, 0 will be returned.
  */
-static pg_wchar
-find_case_map(pg_wchar ucs, const pg_wchar *map)
+static char32_t
+find_case_map(char32_t ucs, const char32_t *map)
 {
 	/* Fast path for codepoints < 0x80 */
 	if (ucs < 0x80)
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index 4136c4d4f92..aab667a7bb4 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -1,7 +1,7 @@
 /*-------------------------------------------------------------------------
  * unicode_category.c
  *		Determine general category and character properties of Unicode
- *		characters. Encoding must be UTF8, where we assume that the pg_wchar
+ *		characters. Encoding must be UTF8, where we assume that the char32_t
  *		representation is a code point.
  *
  * Portions Copyright (c) 2017-2025, PostgreSQL Global Development Group
@@ -76,13 +76,13 @@
 #define PG_U_CHARACTER_TAB	0x09
 
 static bool range_search(const pg_unicode_range *tbl, size_t size,
-						 pg_wchar code);
+						 char32_t code);
 
 /*
  * Unicode general category for the given codepoint.
  */
 pg_unicode_category
-unicode_category(pg_wchar code)
+unicode_category(char32_t code)
 {
 	int			min = 0;
 	int			mid;
@@ -108,7 +108,7 @@ unicode_category(pg_wchar code)
 }
 
 bool
-pg_u_prop_alphabetic(pg_wchar code)
+pg_u_prop_alphabetic(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_ALPHABETIC;
@@ -119,7 +119,7 @@ pg_u_prop_alphabetic(pg_wchar code)
 }
 
 bool
-pg_u_prop_lowercase(pg_wchar code)
+pg_u_prop_lowercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_LOWERCASE;
@@ -130,7 +130,7 @@ pg_u_prop_lowercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_uppercase(pg_wchar code)
+pg_u_prop_uppercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_UPPERCASE;
@@ -141,7 +141,7 @@ pg_u_prop_uppercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_cased(pg_wchar code)
+pg_u_prop_cased(char32_t code)
 {
 	uint32		category_mask;
 
@@ -156,7 +156,7 @@ pg_u_prop_cased(pg_wchar code)
 }
 
 bool
-pg_u_prop_case_ignorable(pg_wchar code)
+pg_u_prop_case_ignorable(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_CASE_IGNORABLE;
@@ -167,7 +167,7 @@ pg_u_prop_case_ignorable(pg_wchar code)
 }
 
 bool
-pg_u_prop_white_space(pg_wchar code)
+pg_u_prop_white_space(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_WHITE_SPACE;
@@ -178,7 +178,7 @@ pg_u_prop_white_space(pg_wchar code)
 }
 
 bool
-pg_u_prop_hex_digit(pg_wchar code)
+pg_u_prop_hex_digit(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_HEX_DIGIT;
@@ -189,7 +189,7 @@ pg_u_prop_hex_digit(pg_wchar code)
 }
 
 bool
-pg_u_prop_join_control(pg_wchar code)
+pg_u_prop_join_control(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_JOIN_CONTROL;
@@ -208,7 +208,7 @@ pg_u_prop_join_control(pg_wchar code)
  */
 
 bool
-pg_u_isdigit(pg_wchar code, bool posix)
+pg_u_isdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return ('0' <= code && code <= '9');
@@ -217,19 +217,19 @@ pg_u_isdigit(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isalpha(pg_wchar code)
+pg_u_isalpha(char32_t code)
 {
 	return pg_u_prop_alphabetic(code);
 }
 
 bool
-pg_u_isalnum(pg_wchar code, bool posix)
+pg_u_isalnum(char32_t code, bool posix)
 {
 	return pg_u_isalpha(code) || pg_u_isdigit(code, posix);
 }
 
 bool
-pg_u_isword(pg_wchar code)
+pg_u_isword(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -240,32 +240,32 @@ pg_u_isword(pg_wchar code)
 }
 
 bool
-pg_u_isupper(pg_wchar code)
+pg_u_isupper(char32_t code)
 {
 	return pg_u_prop_uppercase(code);
 }
 
 bool
-pg_u_islower(pg_wchar code)
+pg_u_islower(char32_t code)
 {
 	return pg_u_prop_lowercase(code);
 }
 
 bool
-pg_u_isblank(pg_wchar code)
+pg_u_isblank(char32_t code)
 {
 	return code == PG_U_CHARACTER_TAB ||
 		unicode_category(code) == PG_U_SPACE_SEPARATOR;
 }
 
 bool
-pg_u_iscntrl(pg_wchar code)
+pg_u_iscntrl(char32_t code)
 {
 	return unicode_category(code) == PG_U_CONTROL;
 }
 
 bool
-pg_u_isgraph(pg_wchar code)
+pg_u_isgraph(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -276,7 +276,7 @@ pg_u_isgraph(pg_wchar code)
 }
 
 bool
-pg_u_isprint(pg_wchar code)
+pg_u_isprint(char32_t code)
 {
 	pg_unicode_category category = unicode_category(code);
 
@@ -287,7 +287,7 @@ pg_u_isprint(pg_wchar code)
 }
 
 bool
-pg_u_ispunct(pg_wchar code, bool posix)
+pg_u_ispunct(char32_t code, bool posix)
 {
 	uint32		category_mask;
 
@@ -308,13 +308,13 @@ pg_u_ispunct(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isspace(pg_wchar code)
+pg_u_isspace(char32_t code)
 {
 	return pg_u_prop_white_space(code);
 }
 
 bool
-pg_u_isxdigit(pg_wchar code, bool posix)
+pg_u_isxdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return (('0' <= code && code <= '9') ||
@@ -478,7 +478,7 @@ unicode_category_abbrev(pg_unicode_category category)
  * given table.
  */
 static bool
-range_search(const pg_unicode_range *tbl, size_t size, pg_wchar code)
+range_search(const pg_unicode_range *tbl, size_t size, char32_t code)
 {
 	int			min = 0;
 	int			mid;
diff --git a/src/common/unicode_norm.c b/src/common/unicode_norm.c
index 6654b4cbc49..489d99cd5ab 100644
--- a/src/common/unicode_norm.c
+++ b/src/common/unicode_norm.c
@@ -69,7 +69,7 @@ conv_compare(const void *p1, const void *p2)
  * lookup, while the frontend version uses a binary search.
  */
 static const pg_unicode_decomposition *
-get_code_entry(pg_wchar code)
+get_code_entry(char32_t code)
 {
 #ifndef FRONTEND
 	int			h;
@@ -109,7 +109,7 @@ get_code_entry(pg_wchar code)
  * Get the combining class of the given codepoint.
  */
 static uint8
-get_canonical_class(pg_wchar code)
+get_canonical_class(char32_t code)
 {
 	const pg_unicode_decomposition *entry = get_code_entry(code);
 
@@ -130,15 +130,15 @@ get_canonical_class(pg_wchar code)
  * Note: the returned pointer can point to statically allocated buffer, and
  * is only valid until next call to this function!
  */
-static const pg_wchar *
+static const char32_t *
 get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
 {
-	static pg_wchar x;
+	static char32_t x;
 
 	if (DECOMPOSITION_IS_INLINE(entry))
 	{
 		Assert(DECOMPOSITION_SIZE(entry) == 1);
-		x = (pg_wchar) entry->dec_index;
+		x = (char32_t) entry->dec_index;
 		*dec_size = 1;
 		return &x;
 	}
@@ -156,7 +156,7 @@ get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
  * are, in turn, decomposable.
  */
 static int
-get_decomposed_size(pg_wchar code, bool compat)
+get_decomposed_size(char32_t code, bool compat)
 {
 	const pg_unicode_decomposition *entry;
 	int			size = 0;
@@ -318,7 +318,7 @@ recompose_code(uint32 start, uint32 code, uint32 *result)
  * in the array result.
  */
 static void
-decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
+decompose_code(char32_t code, bool compat, char32_t **result, int *current)
 {
 	const pg_unicode_decomposition *entry;
 	int			i;
@@ -337,7 +337,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 					v,
 					tindex,
 					sindex;
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		sindex = code - SBASE;
 		l = LBASE + sindex / (VCOUNT * TCOUNT);
@@ -369,7 +369,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	if (entry == NULL || DECOMPOSITION_SIZE(entry) == 0 ||
 		(!compat && DECOMPOSITION_IS_COMPAT(entry)))
 	{
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		res[*current] = code;
 		(*current)++;
@@ -382,7 +382,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	decomp = get_code_decomposition(entry, &dec_size);
 	for (i = 0; i < dec_size; i++)
 	{
-		pg_wchar	lcode = (pg_wchar) decomp[i];
+		char32_t	lcode = (char32_t) decomp[i];
 
 		/* Leave if no more decompositions */
 		decompose_code(lcode, compat, result, current);
@@ -398,17 +398,17 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
  * malloc. Or NULL if we run out of memory. In backend, the returned
  * string is palloc'd instead, and OOM is reported with ereport().
  */
-pg_wchar *
-unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
+char32_t *
+unicode_normalize(UnicodeNormalizationForm form, const char32_t *input)
 {
 	bool		compat = (form == UNICODE_NFKC || form == UNICODE_NFKD);
 	bool		recompose = (form == UNICODE_NFC || form == UNICODE_NFKC);
-	pg_wchar   *decomp_chars;
-	pg_wchar   *recomp_chars;
+	char32_t   *decomp_chars;
+	char32_t   *recomp_chars;
 	int			decomp_size,
 				current_size;
 	int			count;
-	const pg_wchar *p;
+	const char32_t *p;
 
 	/* variables for recomposition */
 	int			last_class;
@@ -425,7 +425,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	for (p = input; *p; p++)
 		decomp_size += get_decomposed_size(*p, compat);
 
-	decomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	decomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (decomp_chars == NULL)
 		return NULL;
 
@@ -448,9 +448,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 */
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	prev = decomp_chars[count - 1];
-		pg_wchar	next = decomp_chars[count];
-		pg_wchar	tmp;
+		char32_t	prev = decomp_chars[count - 1];
+		char32_t	next = decomp_chars[count];
+		char32_t	tmp;
 		const uint8 prevClass = get_canonical_class(prev);
 		const uint8 nextClass = get_canonical_class(next);
 
@@ -487,7 +487,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 * longer than the decomposed one, so make the allocation of the output
 	 * string based on that assumption.
 	 */
-	recomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	recomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (!recomp_chars)
 	{
 		FREE(decomp_chars);
@@ -501,9 +501,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	ch = decomp_chars[count];
+		char32_t	ch = decomp_chars[count];
 		int			ch_class = get_canonical_class(ch);
-		pg_wchar	composite;
+		char32_t	composite;
 
 		if (last_class < ch_class &&
 			recompose_code(starter_ch, ch, &composite))
@@ -524,7 +524,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 			recomp_chars[target_pos++] = ch;
 		}
 	}
-	recomp_chars[target_pos] = (pg_wchar) '\0';
+	recomp_chars[target_pos] = (char32_t) '\0';
 
 	FREE(decomp_chars);
 
@@ -540,7 +540,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 #ifndef FRONTEND
 
 static const pg_unicode_normprops *
-qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
+qc_hash_lookup(char32_t ch, const pg_unicode_norminfo *norminfo)
 {
 	int			h;
 	uint32		hashkey;
@@ -571,7 +571,7 @@ qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
  * Look up the normalization quick check character property
  */
 static UnicodeNormalizationQC
-qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
+qc_is_allowed(UnicodeNormalizationForm form, char32_t ch)
 {
 	const pg_unicode_normprops *found = NULL;
 
@@ -595,7 +595,7 @@ qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
 }
 
 UnicodeNormalizationQC
-unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input)
+unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input)
 {
 	uint8		lastCanonicalClass = 0;
 	UnicodeNormalizationQC result = UNICODE_NORM_QC_YES;
@@ -610,9 +610,9 @@ unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *
 	if (form == UNICODE_NFD || form == UNICODE_NFKD)
 		return UNICODE_NORM_QC_MAYBE;
 
-	for (const pg_wchar *p = input; *p; p++)
+	for (const char32_t *p = input; *p; p++)
 	{
-		pg_wchar	ch = *p;
+		char32_t	ch = *p;
 		uint8		canonicalClass;
 		UnicodeNormalizationQC check;
 
diff --git a/src/fe_utils/mbprint.c b/src/fe_utils/mbprint.c
index eb3eeee9925..abffdbe18a2 100644
--- a/src/fe_utils/mbprint.c
+++ b/src/fe_utils/mbprint.c
@@ -49,20 +49,20 @@ pg_get_utf8_id(void)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static pg_wchar
+static char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
diff --git a/src/include/c.h b/src/include/c.h
index 9ab5e617995..ce48732311e 100644
--- a/src/include/c.h
+++ b/src/include/c.h
@@ -513,6 +513,24 @@ typedef void (*pg_funcptr_t) (void);
 
 #include <stdbool.h>
 
+/*
+ * char16_t and char32_t
+ *      Unicode code points.
+ *
+ * uchar.h should always be available in C11, but it's not available on
+ * Mac. However, these types are keywords in C++11, so when using C++, we
+ * can't redefine the types. XXX: when uchar.h is available everywhere, we can
+ * remove this check and just include uchar.h unconditionally.
+ */
+#ifdef HAVE_UCHAR_H
+#include <uchar.h>
+#else
+#ifndef __cplusplus
+typedef uint16_t char16_t;
+typedef uint32_t char32_t;
+#endif
+#endif
+
 
 /* ----------------------------------------------------------------
  *				Section 3:	standard system types
diff --git a/src/include/common/unicode_case.h b/src/include/common/unicode_case.h
index 41e2c1f4b33..6bcffd349c2 100644
--- a/src/include/common/unicode_case.h
+++ b/src/include/common/unicode_case.h
@@ -14,14 +14,12 @@
 #ifndef UNICODE_CASE_H
 #define UNICODE_CASE_H
 
-#include "mb/pg_wchar.h"
-
 typedef size_t (*WordBoundaryNext) (void *wbstate);
 
-pg_wchar	unicode_lowercase_simple(pg_wchar code);
-pg_wchar	unicode_titlecase_simple(pg_wchar code);
-pg_wchar	unicode_uppercase_simple(pg_wchar code);
-pg_wchar	unicode_casefold_simple(pg_wchar code);
+char32_t	unicode_lowercase_simple(char32_t code);
+char32_t	unicode_titlecase_simple(char32_t code);
+char32_t	unicode_uppercase_simple(char32_t code);
+char32_t	unicode_casefold_simple(char32_t code);
 size_t		unicode_strlower(char *dst, size_t dstsize, const char *src,
 							 ssize_t srclen, bool full);
 size_t		unicode_strtitle(char *dst, size_t dstsize, const char *src,
diff --git a/src/include/common/unicode_case_table.h b/src/include/common/unicode_case_table.h
index d5311786582..0a14fb2d97b 100644
--- a/src/include/common/unicode_case_table.h
+++ b/src/include/common/unicode_case_table.h
@@ -18,7 +18,6 @@
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -45,7 +44,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -166,7 +165,7 @@ static const pg_special_case special_case[106] =
  * The entry case_map_lower[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_lower[1704] =
+static const char32_t case_map_lower[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -1879,7 +1878,7 @@ static const pg_wchar case_map_lower[1704] =
  * The entry case_map_title[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_title[1704] =
+static const char32_t case_map_title[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -3592,7 +3591,7 @@ static const pg_wchar case_map_title[1704] =
  * The entry case_map_upper[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_upper[1704] =
+static const char32_t case_map_upper[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -5305,7 +5304,7 @@ static const pg_wchar case_map_upper[1704] =
  * The entry case_map_fold[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_fold[1704] =
+static const char32_t case_map_fold[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -13522,7 +13521,7 @@ static const uint16 case_map[4778] =
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < 0x0588 */
 	if (cp < 0x0588)
diff --git a/src/include/common/unicode_category.h b/src/include/common/unicode_category.h
index 8fd8b67a416..684143d3c8a 100644
--- a/src/include/common/unicode_category.h
+++ b/src/include/common/unicode_category.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_CATEGORY_H
 #define UNICODE_CATEGORY_H
 
-#include "mb/pg_wchar.h"
-
 /*
  * Unicode General Category Values
  *
@@ -61,31 +59,31 @@ typedef enum pg_unicode_category
 	PG_U_FINAL_PUNCTUATION = 29 /* Pf */
 } pg_unicode_category;
 
-extern pg_unicode_category unicode_category(pg_wchar code);
+extern pg_unicode_category unicode_category(char32_t code);
 extern const char *unicode_category_string(pg_unicode_category category);
 extern const char *unicode_category_abbrev(pg_unicode_category category);
 
-extern bool pg_u_prop_alphabetic(pg_wchar code);
-extern bool pg_u_prop_lowercase(pg_wchar code);
-extern bool pg_u_prop_uppercase(pg_wchar code);
-extern bool pg_u_prop_cased(pg_wchar code);
-extern bool pg_u_prop_case_ignorable(pg_wchar code);
-extern bool pg_u_prop_white_space(pg_wchar code);
-extern bool pg_u_prop_hex_digit(pg_wchar code);
-extern bool pg_u_prop_join_control(pg_wchar code);
+extern bool pg_u_prop_alphabetic(char32_t code);
+extern bool pg_u_prop_lowercase(char32_t code);
+extern bool pg_u_prop_uppercase(char32_t code);
+extern bool pg_u_prop_cased(char32_t code);
+extern bool pg_u_prop_case_ignorable(char32_t code);
+extern bool pg_u_prop_white_space(char32_t code);
+extern bool pg_u_prop_hex_digit(char32_t code);
+extern bool pg_u_prop_join_control(char32_t code);
 
-extern bool pg_u_isdigit(pg_wchar code, bool posix);
-extern bool pg_u_isalpha(pg_wchar code);
-extern bool pg_u_isalnum(pg_wchar code, bool posix);
-extern bool pg_u_isword(pg_wchar code);
-extern bool pg_u_isupper(pg_wchar code);
-extern bool pg_u_islower(pg_wchar code);
-extern bool pg_u_isblank(pg_wchar code);
-extern bool pg_u_iscntrl(pg_wchar code);
-extern bool pg_u_isgraph(pg_wchar code);
-extern bool pg_u_isprint(pg_wchar code);
-extern bool pg_u_ispunct(pg_wchar code, bool posix);
-extern bool pg_u_isspace(pg_wchar code);
-extern bool pg_u_isxdigit(pg_wchar code, bool posix);
+extern bool pg_u_isdigit(char32_t code, bool posix);
+extern bool pg_u_isalpha(char32_t code);
+extern bool pg_u_isalnum(char32_t code, bool posix);
+extern bool pg_u_isword(char32_t code);
+extern bool pg_u_isupper(char32_t code);
+extern bool pg_u_islower(char32_t code);
+extern bool pg_u_isblank(char32_t code);
+extern bool pg_u_iscntrl(char32_t code);
+extern bool pg_u_isgraph(char32_t code);
+extern bool pg_u_isprint(char32_t code);
+extern bool pg_u_ispunct(char32_t code, bool posix);
+extern bool pg_u_isspace(char32_t code);
+extern bool pg_u_isxdigit(char32_t code, bool posix);
 
 #endif							/* UNICODE_CATEGORY_H */
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 95a1c65da7e..466a41b72b0 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -20,15 +20,15 @@
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/include/common/unicode_norm.h b/src/include/common/unicode_norm.h
index 5bc3b79e78e..516c192cc4c 100644
--- a/src/include/common/unicode_norm.h
+++ b/src/include/common/unicode_norm.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_NORM_H
 #define UNICODE_NORM_H
 
-#include "mb/pg_wchar.h"
-
 typedef enum
 {
 	UNICODE_NFC = 0,
@@ -32,8 +30,8 @@ typedef enum
 	UNICODE_NORM_QC_MAYBE = -1,
 } UnicodeNormalizationQC;
 
-extern pg_wchar *unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input);
+extern char32_t *unicode_normalize(UnicodeNormalizationForm form, const char32_t *input);
 
-extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input);
+extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input);
 
 #endif							/* UNICODE_NORM_H */
diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h
index 4b4a9974b75..4d84bdc81e4 100644
--- a/src/include/mb/pg_wchar.h
+++ b/src/include/mb/pg_wchar.h
@@ -532,25 +532,25 @@ typedef uint32 (*utf_local_conversion_func) (uint32 code);
  * Some handy functions for Unicode-specific tests.
  */
 static inline bool
-is_valid_unicode_codepoint(pg_wchar c)
+is_valid_unicode_codepoint(char32_t c)
 {
 	return (c > 0 && c <= 0x10FFFF);
 }
 
 static inline bool
-is_utf16_surrogate_first(pg_wchar c)
+is_utf16_surrogate_first(char32_t c)
 {
 	return (c >= 0xD800 && c <= 0xDBFF);
 }
 
 static inline bool
-is_utf16_surrogate_second(pg_wchar c)
+is_utf16_surrogate_second(char32_t c)
 {
 	return (c >= 0xDC00 && c <= 0xDFFF);
 }
 
-static inline pg_wchar
-surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
+static inline char32_t
+surrogate_pair_to_codepoint(char16_t first, char16_t second)
 {
 	return ((first & 0x3FF) << 10) + 0x10000 + (second & 0x3FF);
 }
@@ -561,20 +561,20 @@ surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static inline pg_wchar
+static inline char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
@@ -588,7 +588,7 @@ utf8_to_unicode(const unsigned char *c)
  * unicode_utf8len(c) bytes available.
  */
 static inline unsigned char *
-unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
+unicode_to_utf8(char32_t c, unsigned char *utf8string)
 {
 	if (c <= 0x7F)
 	{
@@ -620,7 +620,7 @@ unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
  * Number of bytes needed to represent the given char in UTF8.
  */
 static inline int
-unicode_utf8len(pg_wchar c)
+unicode_utf8len(char32_t c)
 {
 	if (c <= 0x7F)
 		return 1;
@@ -676,8 +676,8 @@ extern int	pg_valid_server_encoding(const char *name);
 extern bool is_encoding_supported_by_icu(int encoding);
 extern const char *get_encoding_name_for_icu(int encoding);
 
-extern unsigned char *unicode_to_utf8(pg_wchar c, unsigned char *utf8string);
-extern pg_wchar utf8_to_unicode(const unsigned char *c);
+extern unsigned char *unicode_to_utf8(char32_t c, unsigned char *utf8string);
+extern char32_t utf8_to_unicode(const unsigned char *c);
 extern bool pg_utf8_islegal(const unsigned char *source, int length);
 extern int	pg_utf_mblen(const unsigned char *s);
 extern int	pg_mule_mblen(const unsigned char *s);
@@ -739,8 +739,8 @@ extern char *pg_server_to_client(const char *s, int len);
 extern char *pg_any_to_server(const char *s, int len, int encoding);
 extern char *pg_server_to_any(const char *s, int len, int encoding);
 
-extern void pg_unicode_to_server(pg_wchar c, unsigned char *s);
-extern bool pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s);
+extern void pg_unicode_to_server(char32_t c, unsigned char *s);
+extern bool pg_unicode_to_server_noerror(char32_t c, unsigned char *s);
 
 extern unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc);
 extern unsigned short CNStoBIG5(unsigned short cns, unsigned char lc);
diff --git a/src/include/pg_config.h.in b/src/include/pg_config.h.in
index c4dc5d72bdb..fab8ba4aa6a 100644
--- a/src/include/pg_config.h.in
+++ b/src/include/pg_config.h.in
@@ -463,6 +463,9 @@
 /* Define to 1 if you have the <termios.h> header file. */
 #undef HAVE_TERMIOS_H
 
+/* Define to 1 if you have the <uchar.h> header file. */
+#undef HAVE_UCHAR_H
+
 /* Define to 1 if curl_global_init() is guaranteed to be thread-safe. */
 #undef HAVE_THREADSAFE_CURL_GLOBAL_INIT
 
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index bb4e1b37005..790be386d75 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -3505,6 +3505,8 @@ cb_cleanup_dir
 cb_options
 cb_tablespace
 cb_tablespace_mapping
+char16_t
+char32_t
 check_agg_arguments_context
 check_function_callback
 check_network_data
-- 
2.43.0

From 5525b5e35121bdfd5eb566b7a08916fe90822422 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 15:53:46 +1300
Subject: [PATCH 1/8] Fix Windows wctype.h usage for codepoints outside Unicode
 BMP.

Windows' wchar_t is only 16 bits wide.  As established by the
towupper_l()/towlower_l() wrapper functions, we should avoid truncating
overflowing code points when calling wctype.h functions, and just return
false.  Windows just can't answer that question, but it didn't make
sense to return the answer for a totally different character.
---
 src/backend/utils/adt/pg_locale_libc.c | 27 +++++++++++++++++---------
 1 file changed, 18 insertions(+), 9 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index 9c7fcd1fc7a..761ed1a0603 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -187,55 +187,64 @@ wc_isxdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
 static bool
 wc_isdigit_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswdigit_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswdigit_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_isalpha_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswalpha_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswalpha_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_isalnum_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswalnum_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswalnum_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_isupper_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswupper_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswupper_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_islower_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswlower_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswlower_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_isgraph_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswgraph_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswgraph_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_isprint_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswprint_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswprint_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_ispunct_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswpunct_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswpunct_l((wint_t) wc, locale->lt);
 }
 
 static bool
 wc_isspace_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-	return iswspace_l((wint_t) wc, locale->lt);
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
+		iswspace_l((wint_t) wc, locale->lt);
 }
 
 static bool
-- 
2.50.1 (Apple Git-155)

From 9a9026c29f3e9cd3c1b7fd92e053bcb5ecc5f6ae Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 15:14:13 +1300
Subject: [PATCH 2/8] Formalize pg_wchar encoding schemes.

Create a bit more clarity about the different ways that pg_wchar can be
encoded, by naming the three schemes in use.  This also allows a
dispatch-table format in pg_locale_libc.c.

Discussion: https://www.postgresql.org/message-id/flat/CA%2BhUKG%2BhDkp1etcfy%3DtaxJ8ybf8KapyOjqdBRPF7yaoSoSj1_w%40mail.gmail.com
---
 src/backend/utils/adt/pg_locale_libc.c | 163 +++++++++++++------------
 src/common/wchar.c                     |  94 +++++++-------
 src/include/mb/pg_wchar.h              |  51 ++++++++
 src/tools/pgindent/typedefs.list       |   1 +
 4 files changed, 187 insertions(+), 122 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index 761ed1a0603..1892ed3c5ce 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -43,20 +43,25 @@
  * the <ctype.h> functions since those will obey LC_CTYPE.  Note that these
  * collations don't give a fig about multibyte characters.
  *
- * 2. When working in UTF8 encoding, we use the <wctype.h> functions.
+ * 2. PG_WCHAR_UTF32 encoding scheme:
+ *
+ * When working in UTF8 encoding, we use the <wctype.h> functions.
  * This assumes that every platform uses Unicode codepoints directly
  * as the wchar_t representation of Unicode.  On some platforms
  * wchar_t is only 16 bits wide, so we have to punt for codepoints > 0xFFFF.
  *
- * 3. In all other encodings, we use the <ctype.h> functions for pg_wchar
+ * 3. PG_WCHAR_CHAR and PG_WCHAR_CUSTOM encoding schemes:
+ *
+ * In all other encodings, we use the <ctype.h> functions for pg_wchar
  * values up to 255, and punt for values above that.  This is 100% correct
- * only in single-byte encodings such as LATINn.  However, non-Unicode
- * multibyte encodings are mostly Far Eastern character sets for which the
- * properties being tested here aren't very relevant for higher code values
- * anyway.  The difficulty with using the <wctype.h> functions with
- * non-Unicode multibyte encodings is that we can have no certainty that
- * the platform's wchar_t representation matches what we do in pg_wchar
- * conversions.
+ * only in single-byte encodings such as LATINn (PG_WCHAR_CHAR).  However,
+ * non-Unicode multibyte encodings (PG_WCHAR_CUSTOM) are all Far Eastern
+ * character sets for which the properties being tested here aren't very
+ * relevant for higher code values anyway.  The difficulty with using the
+ * <wctype.h> functions with non-Unicode multibyte encodings is that we can
+ * have no certainty that the platform's wchar_t representation matches what we
+ * do in pg_wchar conversions.  (MULE is also declared PG_WCHAR_CUSTOM but is
+ * not available as a multi-byte encoding in any known libc.)
  *
  * As a special case, in the "default" collation, (2) and (3) force ASCII
  * letters to follow ASCII upcase/downcase rules, while in a non-default
@@ -331,70 +336,75 @@ tolower_libc_mb(pg_wchar wc, pg_locale_t locale)
 		return wc;
 }
 
-static const struct ctype_methods ctype_methods_libc_sb = {
-	.strlower = strlower_libc_sb,
-	.strtitle = strtitle_libc_sb,
-	.strupper = strupper_libc_sb,
-	.wc_isdigit = wc_isdigit_libc_sb,
-	.wc_isalpha = wc_isalpha_libc_sb,
-	.wc_isalnum = wc_isalnum_libc_sb,
-	.wc_isupper = wc_isupper_libc_sb,
-	.wc_islower = wc_islower_libc_sb,
-	.wc_isgraph = wc_isgraph_libc_sb,
-	.wc_isprint = wc_isprint_libc_sb,
-	.wc_ispunct = wc_ispunct_libc_sb,
-	.wc_isspace = wc_isspace_libc_sb,
-	.wc_isxdigit = wc_isxdigit_libc_sb,
-	.char_is_cased = char_is_cased_libc,
-	.char_tolower = char_tolower_libc,
-	.wc_toupper = toupper_libc_sb,
-	.wc_tolower = tolower_libc_sb,
-	.max_chr = UCHAR_MAX,
-};
-
-/*
- * Non-UTF8 multibyte encodings use multibyte semantics for case mapping, but
- * single-byte semantics for pattern matching.
- */
-static const struct ctype_methods ctype_methods_libc_other_mb = {
-	.strlower = strlower_libc_mb,
-	.strtitle = strtitle_libc_mb,
-	.strupper = strupper_libc_mb,
-	.wc_isdigit = wc_isdigit_libc_sb,
-	.wc_isalpha = wc_isalpha_libc_sb,
-	.wc_isalnum = wc_isalnum_libc_sb,
-	.wc_isupper = wc_isupper_libc_sb,
-	.wc_islower = wc_islower_libc_sb,
-	.wc_isgraph = wc_isgraph_libc_sb,
-	.wc_isprint = wc_isprint_libc_sb,
-	.wc_ispunct = wc_ispunct_libc_sb,
-	.wc_isspace = wc_isspace_libc_sb,
-	.wc_isxdigit = wc_isxdigit_libc_sb,
-	.char_is_cased = char_is_cased_libc,
-	.char_tolower = char_tolower_libc,
-	.wc_toupper = toupper_libc_sb,
-	.wc_tolower = tolower_libc_sb,
-	.max_chr = UCHAR_MAX,
-};
+static const struct ctype_methods ctype_methods_libc[] = {
+	[PG_WCHAR_CHAR] = {
+		.strlower = strlower_libc_sb,
+		.strtitle = strtitle_libc_sb,
+		.strupper = strupper_libc_sb,
+		.wc_isdigit = wc_isdigit_libc_sb,
+		.wc_isalpha = wc_isalpha_libc_sb,
+		.wc_isalnum = wc_isalnum_libc_sb,
+		.wc_isupper = wc_isupper_libc_sb,
+		.wc_islower = wc_islower_libc_sb,
+		.wc_isgraph = wc_isgraph_libc_sb,
+		.wc_isprint = wc_isprint_libc_sb,
+		.wc_ispunct = wc_ispunct_libc_sb,
+		.wc_isspace = wc_isspace_libc_sb,
+		.wc_isxdigit = wc_isxdigit_libc_sb,
+		.char_is_cased = char_is_cased_libc,
+		.char_tolower = char_tolower_libc,
+		.wc_toupper = toupper_libc_sb,
+		.wc_tolower = tolower_libc_sb,
+		.max_chr = UCHAR_MAX,
+	},
+	[PG_WCHAR_UTF32] = {
+		.strlower = strlower_libc_mb,
+		.strtitle = strtitle_libc_mb,
+		.strupper = strupper_libc_mb,
+		.wc_isdigit = wc_isdigit_libc_mb,
+		.wc_isalpha = wc_isalpha_libc_mb,
+		.wc_isalnum = wc_isalnum_libc_mb,
+		.wc_isupper = wc_isupper_libc_mb,
+		.wc_islower = wc_islower_libc_mb,
+		.wc_isgraph = wc_isgraph_libc_mb,
+		.wc_isprint = wc_isprint_libc_mb,
+		.wc_ispunct = wc_ispunct_libc_mb,
+		.wc_isspace = wc_isspace_libc_mb,
+		.wc_isxdigit = wc_isxdigit_libc_mb,
+		.char_is_cased = char_is_cased_libc,
+		.char_tolower = char_tolower_libc,
+		.wc_toupper = toupper_libc_mb,
+		.wc_tolower = tolower_libc_mb,
+	},
 
-static const struct ctype_methods ctype_methods_libc_utf8 = {
-	.strlower = strlower_libc_mb,
-	.strtitle = strtitle_libc_mb,
-	.strupper = strupper_libc_mb,
-	.wc_isdigit = wc_isdigit_libc_mb,
-	.wc_isalpha = wc_isalpha_libc_mb,
-	.wc_isalnum = wc_isalnum_libc_mb,
-	.wc_isupper = wc_isupper_libc_mb,
-	.wc_islower = wc_islower_libc_mb,
-	.wc_isgraph = wc_isgraph_libc_mb,
-	.wc_isprint = wc_isprint_libc_mb,
-	.wc_ispunct = wc_ispunct_libc_mb,
-	.wc_isspace = wc_isspace_libc_mb,
-	.wc_isxdigit = wc_isxdigit_libc_mb,
-	.char_is_cased = char_is_cased_libc,
-	.char_tolower = char_tolower_libc,
-	.wc_toupper = toupper_libc_mb,
-	.wc_tolower = tolower_libc_mb,
+	/*
+	 * Custom pg_wchar format converted from non-UTF8 multibyte encodings use
+	 * multibyte semantics for case mapping, but single-byte semantics for
+	 * pattern matching.
+	 *
+	 * XXX Therefore this gives incorrect results for pattern matching outside
+	 * the ASCII range.  Could be fixed.
+	 */
+	[PG_WCHAR_CUSTOM] = {
+		.strlower = strlower_libc_mb,
+		.strtitle = strtitle_libc_mb,
+		.strupper = strupper_libc_mb,
+		.wc_isdigit = wc_isdigit_libc_sb,
+		.wc_isalpha = wc_isalpha_libc_sb,
+		.wc_isalnum = wc_isalnum_libc_sb,
+		.wc_isupper = wc_isupper_libc_sb,
+		.wc_islower = wc_islower_libc_sb,
+		.wc_isgraph = wc_isgraph_libc_sb,
+		.wc_isprint = wc_isprint_libc_sb,
+		.wc_ispunct = wc_ispunct_libc_sb,
+		.wc_isspace = wc_isspace_libc_sb,
+		.wc_isxdigit = wc_isxdigit_libc_sb,
+		.char_is_cased = char_is_cased_libc,
+		.char_tolower = char_tolower_libc,
+		.wc_toupper = toupper_libc_sb,
+		.wc_tolower = tolower_libc_sb,
+		.max_chr = UCHAR_MAX,
+	},
 };
 
 static const struct collate_methods collate_methods_libc = {
@@ -763,14 +773,7 @@ create_pg_locale_libc(Oid collid, MemoryContext context)
 			result->collate = &collate_methods_libc;
 	}
 	if (!result->ctype_is_c)
-	{
-		if (GetDatabaseEncoding() == PG_UTF8)
-			result->ctype = &ctype_methods_libc_utf8;
-		else if (pg_database_encoding_max_length() > 1)
-			result->ctype = &ctype_methods_libc_other_mb;
-		else
-			result->ctype = &ctype_methods_libc_sb;
-	}
+		result->ctype = &ctype_methods_libc[pg_wchar_encoding_scheme(GetDatabaseEncoding())];
 
 	return result;
 }
diff --git a/src/common/wchar.c b/src/common/wchar.c
index a4bc29921de..f453587749a 100644
--- a/src/common/wchar.c
+++ b/src/common/wchar.c
@@ -2062,50 +2062,60 @@ pg_encoding_set_invalid(int encoding, char *dst)
  *-------------------------------------------------------------------
  */
 const pg_wchar_tbl pg_wchar_table[] = {
-	[PG_SQL_ASCII] = {pg_ascii2wchar_with_len, pg_wchar2single_with_len, pg_ascii_mblen, pg_ascii_dsplen, pg_ascii_verifychar, pg_ascii_verifystr, 1},
-	[PG_EUC_JP] = {pg_eucjp2wchar_with_len, pg_wchar2euc_with_len, pg_eucjp_mblen, pg_eucjp_dsplen, pg_eucjp_verifychar, pg_eucjp_verifystr, 3},
-	[PG_EUC_CN] = {pg_euccn2wchar_with_len, pg_wchar2euc_with_len, pg_euccn_mblen, pg_euccn_dsplen, pg_euccn_verifychar, pg_euccn_verifystr, 2},
-	[PG_EUC_KR] = {pg_euckr2wchar_with_len, pg_wchar2euc_with_len, pg_euckr_mblen, pg_euckr_dsplen, pg_euckr_verifychar, pg_euckr_verifystr, 3},
-	[PG_EUC_TW] = {pg_euctw2wchar_with_len, pg_wchar2euc_with_len, pg_euctw_mblen, pg_euctw_dsplen, pg_euctw_verifychar, pg_euctw_verifystr, 4},
-	[PG_EUC_JIS_2004] = {pg_eucjp2wchar_with_len, pg_wchar2euc_with_len, pg_eucjp_mblen, pg_eucjp_dsplen, pg_eucjp_verifychar, pg_eucjp_verifystr, 3},
-	[PG_UTF8] = {pg_utf2wchar_with_len, pg_wchar2utf_with_len, pg_utf_mblen, pg_utf_dsplen, pg_utf8_verifychar, pg_utf8_verifystr, 4},
-	[PG_MULE_INTERNAL] = {pg_mule2wchar_with_len, pg_wchar2mule_with_len, pg_mule_mblen, pg_mule_dsplen, pg_mule_verifychar, pg_mule_verifystr, 4},
-	[PG_LATIN1] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN2] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN3] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN4] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN5] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN6] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN7] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN8] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN9] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_LATIN10] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1256] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1258] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN866] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN874] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_KOI8R] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1251] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1252] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_ISO_8859_5] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_ISO_8859_6] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_ISO_8859_7] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_ISO_8859_8] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1250] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1253] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1254] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1255] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_WIN1257] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_KOI8U] = {pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
-	[PG_SJIS] = {0, 0, pg_sjis_mblen, pg_sjis_dsplen, pg_sjis_verifychar, pg_sjis_verifystr, 2},
-	[PG_BIG5] = {0, 0, pg_big5_mblen, pg_big5_dsplen, pg_big5_verifychar, pg_big5_verifystr, 2},
-	[PG_GBK] = {0, 0, pg_gbk_mblen, pg_gbk_dsplen, pg_gbk_verifychar, pg_gbk_verifystr, 2},
-	[PG_UHC] = {0, 0, pg_uhc_mblen, pg_uhc_dsplen, pg_uhc_verifychar, pg_uhc_verifystr, 2},
-	[PG_GB18030] = {0, 0, pg_gb18030_mblen, pg_gb18030_dsplen, pg_gb18030_verifychar, pg_gb18030_verifystr, 4},
-	[PG_JOHAB] = {0, 0, pg_johab_mblen, pg_johab_dsplen, pg_johab_verifychar, pg_johab_verifystr, 3},
-	[PG_SHIFT_JIS_2004] = {0, 0, pg_sjis_mblen, pg_sjis_dsplen, pg_sjis_verifychar, pg_sjis_verifystr, 2},
+	[PG_SQL_ASCII] = {PG_WCHAR_CHAR, pg_ascii2wchar_with_len, pg_wchar2single_with_len, pg_ascii_mblen, pg_ascii_dsplen, pg_ascii_verifychar, pg_ascii_verifystr, 1},
+	[PG_EUC_JP] = {PG_WCHAR_CUSTOM, pg_eucjp2wchar_with_len, pg_wchar2euc_with_len, pg_eucjp_mblen, pg_eucjp_dsplen, pg_eucjp_verifychar, pg_eucjp_verifystr, 3},
+	[PG_EUC_CN] = {PG_WCHAR_CUSTOM, pg_euccn2wchar_with_len, pg_wchar2euc_with_len, pg_euccn_mblen, pg_euccn_dsplen, pg_euccn_verifychar, pg_euccn_verifystr, 2},
+	[PG_EUC_KR] = {PG_WCHAR_CUSTOM, pg_euckr2wchar_with_len, pg_wchar2euc_with_len, pg_euckr_mblen, pg_euckr_dsplen, pg_euckr_verifychar, pg_euckr_verifystr, 3},
+	[PG_EUC_TW] = {PG_WCHAR_CUSTOM, pg_euctw2wchar_with_len, pg_wchar2euc_with_len, pg_euctw_mblen, pg_euctw_dsplen, pg_euctw_verifychar, pg_euctw_verifystr, 4},
+	[PG_EUC_JIS_2004] = {PG_WCHAR_CUSTOM, pg_eucjp2wchar_with_len, pg_wchar2euc_with_len, pg_eucjp_mblen, pg_eucjp_dsplen, pg_eucjp_verifychar, pg_eucjp_verifystr, 3},
+	[PG_UTF8] = {PG_WCHAR_UTF32, pg_utf2wchar_with_len, pg_wchar2utf_with_len, pg_utf_mblen, pg_utf_dsplen, pg_utf8_verifychar, pg_utf8_verifystr, 4},
+	[PG_MULE_INTERNAL] = {PG_WCHAR_CUSTOM, pg_mule2wchar_with_len, pg_wchar2mule_with_len, pg_mule_mblen, pg_mule_dsplen, pg_mule_verifychar, pg_mule_verifystr, 4},
+	[PG_LATIN1] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN2] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN3] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN4] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN5] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN6] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN7] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN8] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN9] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_LATIN10] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1256] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1258] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN866] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN874] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_KOI8R] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1251] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1252] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_ISO_8859_5] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_ISO_8859_6] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_ISO_8859_7] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_ISO_8859_8] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1250] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1253] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1254] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1255] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_WIN1257] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_KOI8U] = {PG_WCHAR_CHAR, pg_latin12wchar_with_len, pg_wchar2single_with_len, pg_latin1_mblen, pg_latin1_dsplen, pg_latin1_verifychar, pg_latin1_verifystr, 1},
+	[PG_SJIS] = {PG_WCHAR_NONE, 0, 0, pg_sjis_mblen, pg_sjis_dsplen, pg_sjis_verifychar, pg_sjis_verifystr, 2},
+	[PG_BIG5] = {PG_WCHAR_NONE, 0, 0, pg_big5_mblen, pg_big5_dsplen, pg_big5_verifychar, pg_big5_verifystr, 2},
+	[PG_GBK] = {PG_WCHAR_NONE, 0, 0, pg_gbk_mblen, pg_gbk_dsplen, pg_gbk_verifychar, pg_gbk_verifystr, 2},
+	[PG_UHC] = {PG_WCHAR_NONE, 0, 0, pg_uhc_mblen, pg_uhc_dsplen, pg_uhc_verifychar, pg_uhc_verifystr, 2},
+	[PG_GB18030] = {PG_WCHAR_NONE, 0, 0, pg_gb18030_mblen, pg_gb18030_dsplen, pg_gb18030_verifychar, pg_gb18030_verifystr, 4},
+	[PG_JOHAB] = {PG_WCHAR_NONE, 0, 0, pg_johab_mblen, pg_johab_dsplen, pg_johab_verifychar, pg_johab_verifystr, 3},
+	[PG_SHIFT_JIS_2004] = {PG_WCHAR_NONE, 0, 0, pg_sjis_mblen, pg_sjis_dsplen, pg_sjis_verifychar, pg_sjis_verifystr, 2},
 };
 
+/*
+ * Returns the encoding scheme for pg_wchar values in the current database
+ * encoding.
+ */
+PgWcharEncodingScheme
+pg_wchar_encoding_scheme(int encoding)
+{
+	return pg_wchar_table[encoding].encoding_scheme;
+}
+
 /*
  * Returns the byte length of a multibyte character.
  *
diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h
index 4b4a9974b75..5db00cebcef 100644
--- a/src/include/mb/pg_wchar.h
+++ b/src/include/mb/pg_wchar.h
@@ -27,6 +27,55 @@
  */
 typedef unsigned int pg_wchar;
 
+/*
+ * Encoding schemes that pg_wchar might hold.
+ *
+ * Each multi-byte encoding has a corresponding wide encoding scheme,
+ * conceptually like wchar_t in C.  Conversions to and from char should be
+ * performed by pg_mb2wchar*() and pg_wchar2mb*() functions.  In all encoding
+ * schemes, values 0-127 represent ASCII.  For higher values, see below.
+ *
+ * Locale providers make use of the known properties of these encoding schemes
+ * to implement ctype/wctype functionality.
+ */
+typedef enum PgWcharEncodingScheme
+{
+	/*
+	 * 8-bit characters in the database encoding, zero-extended to pg_wchar
+	 * width.
+	 */
+	PG_WCHAR_CHAR,
+
+	/*
+	 * 32-bit Unicode code points.  PostgreSQL assumes that all libc
+	 * implementations use UTF-32 or at least UTF-16 if wchar_t is narrow for
+	 * locales that use UTF-8 encoding for char strings, so it has a special
+	 * case for this.
+	 */
+	PG_WCHAR_UTF32,
+
+	/*
+	 * For multi-byte database encodings other than UTF-8, the encoding is
+	 * unspecified outside the ASCII range.
+	 */
+	PG_WCHAR_CUSTOM,
+
+	/*
+	 * This scheme is not currently used by any of the supported encodings,
+	 * but is included here for completeness, providing terminology.  In a few
+	 * places, pg_wchar is used to transport wchar_t in whatever unknown
+	 * encoding libc uses for the database encoding.  This is second from last
+	 * so that lookup arrays don't have to waste an entry.
+	 */
+	PG_WCHAR_SYSTEM_WCHAR_T,
+
+	/*
+	 * pg_wchar conversion is not available for the database encoding.  This
+	 * is last so that lookup arrays don't have to waste an entry.
+	 */
+	PG_WCHAR_NONE,
+} PgWcharEncodingScheme;
+
 /*
  * Maximum byte length of multibyte characters in any backend encoding
  */
@@ -391,6 +440,7 @@ typedef int (*mbstr_verifier) (const unsigned char *mbstr, int len);
 
 typedef struct
 {
+	PgWcharEncodingScheme encoding_scheme;	/* pg_wchar representation */
 	mb2wchar_with_len_converter mb2wchar_with_len;	/* convert a multibyte
 													 * string to a wchar */
 	wchar2mb_with_len_converter wchar2mb_with_len;	/* convert a wchar string
@@ -713,6 +763,7 @@ extern int	SetClientEncoding(int encoding);
 extern void InitializeClientEncoding(void);
 extern int	pg_get_client_encoding(void);
 extern const char *pg_get_client_encoding_name(void);
+extern PgWcharEncodingScheme pg_wchar_encoding_scheme(int encoding);
 
 extern void SetDatabaseEncoding(int encoding);
 extern int	GetDatabaseEncoding(void);
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index ac2da4c98cf..d6973751f12 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -2264,6 +2264,7 @@ PgStat_WalCounters
 PgStat_WalStats
 PgXmlErrorContext
 PgXmlStrictness
+PgWcharEncodingScheme
 Pg_abi_values
 Pg_finfo_record
 Pg_magic_struct
-- 
2.50.1 (Apple Git-155)

From 304dc61ed765f1a57a9b9f9cf32a6342f0b15e6a Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 19:25:40 +1300
Subject: [PATCH 3/8] Fix corrupted ctype.h handling for non-ASCII in EUC
 encodings.

Previously we treated PG_WCHAR_CUSTOM encodings the same way as
PG_WCHAR_CHAR, by passing the lower 8 bits of pg_wchar to ctype.h
functions.  That was OK for 7-bit ASCII, but arbitrary junk for any
higher values.  New functions are provided that just return false for
non-ASCII values.

A more correct fix would convert to libc wchar_t format and use
wctype.h, but that isn't done here.
---
 src/backend/utils/adt/pg_locale_libc.c | 103 ++++++++++++++++---------
 1 file changed, 68 insertions(+), 35 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index 1892ed3c5ce..1d6e8be3a82 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -33,6 +33,11 @@
 #include <shlwapi.h>
 #endif
 
+#if defined(WIN32)
+#define isxdigit_l _isxdigit_l
+#define iswxdigit_l _iswxdigit_l
+#endif
+
 /*
  * For the libc provider, to provide as much functionality as possible on a
  * variety of platforms without going so far as to implement everything from
@@ -50,20 +55,20 @@
  * as the wchar_t representation of Unicode.  On some platforms
  * wchar_t is only 16 bits wide, so we have to punt for codepoints > 0xFFFF.
  *
- * 3. PG_WCHAR_CHAR and PG_WCHAR_CUSTOM encoding schemes:
+ * 3. PG_WCHAR_CUSTOM encoding scheme:
+ *
+ * When working with the EUC_* family of encodings (and technically MULE
+ * internal too, but no libc systems are known to support that encoding), we
+ * convert to wchar_t on the fly and use the <wctype.h> functions, except in
+ * the ASCII range where we use the <ctype.h> functions.
+ *
+ * 4. PG_WCHAR_CHAR encoding scheme:
  *
  * In all other encodings, we use the <ctype.h> functions for pg_wchar
- * values up to 255, and punt for values above that.  This is 100% correct
- * only in single-byte encodings such as LATINn (PG_WCHAR_CHAR).  However,
- * non-Unicode multibyte encodings (PG_WCHAR_CUSTOM) are all Far Eastern
- * character sets for which the properties being tested here aren't very
- * relevant for higher code values anyway.  The difficulty with using the
- * <wctype.h> functions with non-Unicode multibyte encodings is that we can
- * have no certainty that the platform's wchar_t representation matches what we
- * do in pg_wchar conversions.  (MULE is also declared PG_WCHAR_CUSTOM but is
- * not available as a multi-byte encoding in any known libc.)
+ * values up to 255.  This is 100% correct since the values originated as char
+ * and were just widened to pg_wchar without change.
  *
- * As a special case, in the "default" collation, (2) and (3) force ASCII
+ * As a special case, in the "default" collation, (2), (3) and (4) force ASCII
  * letters to follow ASCII upcase/downcase rules, while in a non-default
  * collation we just let the library functions do what they will.  The case
  * where this matters is treatment of I/i in Turkish, and the behavior is
@@ -125,6 +130,30 @@ static size_t strupper_libc_mb(char *dest, size_t destsize,
 							   const char *src, ssize_t srclen,
 							   pg_locale_t locale);
 
+/*
+ * Generate a function that passes single-byte characters directly to <ctype.h>
+ * functions, but only if they are in the ASCII range.  This is suitable for
+ * PG_WCHAR_CUSTOM pg_wchar encoding (used with EUC_* encodings).  Values
+ * outside ASCII have an unknown encoding, so we just return false.
+ */
+#define DEFINE_WC_CTYPE_LIBC_ASCII(ctype) \
+static bool \
+wc_is##ctype##_libc_ascii(pg_wchar wc, pg_locale_t locale) \
+{ \
+   return is##ctype##_l((unsigned char) wc, locale->lt); \
+}
+
+DEFINE_WC_CTYPE_LIBC_ASCII(digit);
+DEFINE_WC_CTYPE_LIBC_ASCII(alpha);
+DEFINE_WC_CTYPE_LIBC_ASCII(alnum);
+DEFINE_WC_CTYPE_LIBC_ASCII(upper);
+DEFINE_WC_CTYPE_LIBC_ASCII(lower);
+DEFINE_WC_CTYPE_LIBC_ASCII(graph);
+DEFINE_WC_CTYPE_LIBC_ASCII(print);
+DEFINE_WC_CTYPE_LIBC_ASCII(punct);
+DEFINE_WC_CTYPE_LIBC_ASCII(space);
+DEFINE_WC_CTYPE_LIBC_ASCII(xdigit);
+
 static bool
 wc_isdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
 {
@@ -182,11 +211,7 @@ wc_isspace_libc_sb(pg_wchar wc, pg_locale_t locale)
 static bool
 wc_isxdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
 {
-#ifndef WIN32
 	return isxdigit_l((unsigned char) wc, locale->lt);
-#else
-	return _isxdigit_l((unsigned char) wc, locale->lt);
-#endif
 }
 
 static bool
@@ -255,11 +280,7 @@ wc_isspace_libc_mb(pg_wchar wc, pg_locale_t locale)
 static bool
 wc_isxdigit_libc_mb(pg_wchar wc, pg_locale_t locale)
 {
-#ifndef WIN32
 	return iswxdigit_l((wint_t) wc, locale->lt);
-#else
-	return _iswxdigit_l((wint_t) wc, locale->lt);
-#endif
 }
 
 static char
@@ -280,6 +301,12 @@ char_is_cased_libc(char ch, pg_locale_t locale)
 		return isalpha_l((unsigned char) ch, locale->lt);
 }
 
+static pg_wchar
+toupper_libc_ascii(pg_wchar wc, pg_locale_t locale)
+{
+	return wc < 128 ? toupper_l((unsigned char) wc, locale->lt) : wc;
+}
+
 static pg_wchar
 toupper_libc_sb(pg_wchar wc, pg_locale_t locale)
 {
@@ -308,6 +335,12 @@ toupper_libc_mb(pg_wchar wc, pg_locale_t locale)
 		return wc;
 }
 
+static pg_wchar
+tolower_libc_ascii(pg_wchar wc, pg_locale_t locale)
+{
+	return wc < 128 ? tolower_l((unsigned char) wc, locale->lt) : wc;
+}
+
 static pg_wchar
 tolower_libc_sb(pg_wchar wc, pg_locale_t locale)
 {
@@ -379,30 +412,30 @@ static const struct ctype_methods ctype_methods_libc[] = {
 
 	/*
 	 * Custom pg_wchar format converted from non-UTF8 multibyte encodings use
-	 * multibyte semantics for case mapping, but single-byte semantics for
-	 * pattern matching.
+	 * multibyte semantics for case mapping, but ASCII-only semantics for
+	 * pattern matching, since libc doesn't understand custom encoding of
+	 * higher values.
 	 *
-	 * XXX Therefore this gives incorrect results for pattern matching outside
-	 * the ASCII range.  Could be fixed.
+	 * XXX We could convert to wchar_t to fix that, at considerable cost.
 	 */
 	[PG_WCHAR_CUSTOM] = {
 		.strlower = strlower_libc_mb,
 		.strtitle = strtitle_libc_mb,
 		.strupper = strupper_libc_mb,
-		.wc_isdigit = wc_isdigit_libc_sb,
-		.wc_isalpha = wc_isalpha_libc_sb,
-		.wc_isalnum = wc_isalnum_libc_sb,
-		.wc_isupper = wc_isupper_libc_sb,
-		.wc_islower = wc_islower_libc_sb,
-		.wc_isgraph = wc_isgraph_libc_sb,
-		.wc_isprint = wc_isprint_libc_sb,
-		.wc_ispunct = wc_ispunct_libc_sb,
-		.wc_isspace = wc_isspace_libc_sb,
-		.wc_isxdigit = wc_isxdigit_libc_sb,
+		.wc_isdigit = wc_isdigit_libc_ascii,
+		.wc_isalpha = wc_isalpha_libc_ascii,
+		.wc_isalnum = wc_isalnum_libc_ascii,
+		.wc_isupper = wc_isupper_libc_ascii,
+		.wc_islower = wc_islower_libc_ascii,
+		.wc_isgraph = wc_isgraph_libc_ascii,
+		.wc_isprint = wc_isprint_libc_ascii,
+		.wc_ispunct = wc_ispunct_libc_ascii,
+		.wc_isspace = wc_isspace_libc_ascii,
+		.wc_isxdigit = wc_isxdigit_libc_ascii,
 		.char_is_cased = char_is_cased_libc,
 		.char_tolower = char_tolower_libc,
-		.wc_toupper = toupper_libc_sb,
-		.wc_tolower = tolower_libc_sb,
+		.wc_toupper = toupper_libc_ascii,
+		.wc_tolower = tolower_libc_ascii,
 		.max_chr = UCHAR_MAX,
 	},
 };
-- 
2.50.1 (Apple Git-155)

From 39ebd5e689a458508b2762b84beb197f7dc6fd92 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 17:37:03 +1300
Subject: [PATCH 4/8] Support wctype.h classification for EUC encodings.

Instead of giving up on non-ASCII characters, convert pg_wchar values
using the PG_WCHAR_CUSTOM encoding scheme to wchar_t so that we can use
wctype.h functions on any character.

XXX This replaces the _ascii() version from the previous patch, to
experiment with a different approach

XXX Is this too expensive?
---
 src/backend/utils/adt/pg_locale_libc.c | 148 ++++++++++++++++++-------
 1 file changed, 109 insertions(+), 39 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index 1d6e8be3a82..e6724880f1b 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -130,29 +130,81 @@ static size_t strupper_libc_mb(char *dest, size_t destsize,
 							   const char *src, ssize_t srclen,
 							   pg_locale_t locale);
 
+static wint_t
+pg_wchar_to_wchar_t(pg_wchar wc, pg_locale_t locale)
+{
+	wchar_t		out[MAX_CONVERSION_GROWTH + 1];
+	char		mb[MAX_CONVERSION_GROWTH + 1];
+	size_t		mb_len;
+	size_t		wchar_t_len;
+
+	/* pg_wchar -> multibyte using PostgreSQL pg_wchar encoding */
+	mb_len = pg_wchar2mb_with_len(&wc, mb, 1);
+
+	/* multibyte -> wchar_t using libc */
+	wchar_t_len = char2wchar(out, lengthof(out), mb, mb_len, locale->lt);
+
+	/* reject surrogates and combining characters */
+	if (wchar_t_len != 1)
+		return WEOF;
+
+	return out[0];
+}
+
+static pg_wchar
+wchar_t_to_pg_wchar(wchar_t wc, pg_locale_t locale)
+{
+	wchar_t		in[] = {wc, 0};
+	pg_wchar	out[MAX_CONVERSION_GROWTH + 1];
+	char		mb[MAX_CONVERSION_GROWTH + 1];
+	size_t		mb_len;
+	size_t		pg_wchar_len;
+
+	/* wchar_t -> multibyte using libc */
+	mb_len = wchar2char(mb, in, lengthof(mb), locale->lt);
+
+	/* multibyte -> pg_wchar using PostgreSQL pg_wchar encoding */
+	pg_wchar_len = pg_mb2wchar_with_len(mb, out, mb_len);
+
+	/* should be exactly one character */
+	if (pg_wchar_len != 1)
+		elog(ERROR, "unexpected number of output characters: %zu", pg_wchar_len);
+
+	return out[0];
+}
+
 /*
- * Generate a function that passes single-byte characters directly to <ctype.h>
- * functions, but only if they are in the ASCII range.  This is suitable for
- * PG_WCHAR_CUSTOM pg_wchar encoding (used with EUC_* encodings).  Values
- * outside ASCII have an unknown encoding, so we just return false.
+ * Generate a function that handles the ASCII range with <ctype.h>, and
+ * otherwise converts pg_wchar to libc's wchar_t to be able to use <wctype.h>
+ * routines.  This extra conversion is only required when using encodings that
+ * declare a PG_WCHAR_CUSTOM encoding scheme (EUC encodings for CJK).
+ *
+ * XXX If wchar.c had a function that could identify pg_wchar values that
+ * definitely won't return true (eg the big kanji/hanzi ranges), then we could
+ * skip the expensive conversion but still give correct answers for other
+ * characters.
  */
-#define DEFINE_WC_CTYPE_LIBC_ASCII(ctype) \
+#define DEFINE_WC_CTYPE_LIBC_CUSTOM(ctype) \
 static bool \
-wc_is##ctype##_libc_ascii(pg_wchar wc, pg_locale_t locale) \
+wc_is##ctype##_libc_custom(pg_wchar wc, pg_locale_t locale) \
 { \
-   return is##ctype##_l((unsigned char) wc, locale->lt); \
+	wint_t wint; \
+	if (wc < 128) \
+		return is##ctype##_l(wc, locale->lt); \
+	wint = pg_wchar_to_wchar_t(wc, locale); \
+	return wint != WEOF && isw##ctype##_l(wint, locale->lt); \
 }
 
-DEFINE_WC_CTYPE_LIBC_ASCII(digit);
-DEFINE_WC_CTYPE_LIBC_ASCII(alpha);
-DEFINE_WC_CTYPE_LIBC_ASCII(alnum);
-DEFINE_WC_CTYPE_LIBC_ASCII(upper);
-DEFINE_WC_CTYPE_LIBC_ASCII(lower);
-DEFINE_WC_CTYPE_LIBC_ASCII(graph);
-DEFINE_WC_CTYPE_LIBC_ASCII(print);
-DEFINE_WC_CTYPE_LIBC_ASCII(punct);
-DEFINE_WC_CTYPE_LIBC_ASCII(space);
-DEFINE_WC_CTYPE_LIBC_ASCII(xdigit);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(digit);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(alpha);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(alnum);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(upper);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(lower);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(graph);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(print);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(punct);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(space);
+DEFINE_WC_CTYPE_LIBC_CUSTOM(xdigit);
 
 static bool
 wc_isdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
@@ -302,9 +354,19 @@ char_is_cased_libc(char ch, pg_locale_t locale)
 }
 
 static pg_wchar
-toupper_libc_ascii(pg_wchar wc, pg_locale_t locale)
+toupper_libc_custom(pg_wchar wc, pg_locale_t locale)
 {
-	return wc < 128 ? toupper_l((unsigned char) wc, locale->lt) : wc;
+	wint_t		wint;
+
+	/* force C behavior for ASCII characters, per comments above */
+	if (locale->is_default && wc <= (pg_wchar) 127)
+		return pg_ascii_toupper((unsigned char) wc);
+	if (wc <= 127)
+		return towupper_l((wint_t) wc, locale->lt);
+	wint = pg_wchar_to_wchar_t(wc, locale);
+	if ((wint = pg_wchar_to_wchar_t(wc, locale)) != WEOF)
+		return wchar_t_to_pg_wchar(towupper_l(wint, locale->lt), locale);
+	return wc;
 }
 
 static pg_wchar
@@ -336,9 +398,19 @@ toupper_libc_mb(pg_wchar wc, pg_locale_t locale)
 }
 
 static pg_wchar
-tolower_libc_ascii(pg_wchar wc, pg_locale_t locale)
+tolower_libc_custom(pg_wchar wc, pg_locale_t locale)
 {
-	return wc < 128 ? tolower_l((unsigned char) wc, locale->lt) : wc;
+	wint_t		wint;
+
+	/* force C behavior for ASCII characters, per comments above */
+	if (locale->is_default && wc <= (pg_wchar) 127)
+		return pg_ascii_tolower((unsigned char) wc);
+	if (wc <= 127)
+		return towlower_l((wint_t) wc, locale->lt);
+	wint = pg_wchar_to_wchar_t(wc, locale);
+	if ((wint = pg_wchar_to_wchar_t(wc, locale)) != WEOF)
+		return wchar_t_to_pg_wchar(towlower_l(wint, locale->lt), locale);
+	return wc;
 }
 
 static pg_wchar
@@ -412,31 +484,29 @@ static const struct ctype_methods ctype_methods_libc[] = {
 
 	/*
 	 * Custom pg_wchar format converted from non-UTF8 multibyte encodings use
-	 * multibyte semantics for case mapping, but ASCII-only semantics for
-	 * pattern matching, since libc doesn't understand custom encoding of
-	 * higher values.
-	 *
-	 * XXX We could convert to wchar_t to fix that, at considerable cost.
+	 * multibyte semantics for case mapping, and conversions to libc's wchar_t
+	 * except for the ASCII range that can be handled directly by ctype
+	 * functions.
 	 */
 	[PG_WCHAR_CUSTOM] = {
 		.strlower = strlower_libc_mb,
 		.strtitle = strtitle_libc_mb,
 		.strupper = strupper_libc_mb,
-		.wc_isdigit = wc_isdigit_libc_ascii,
-		.wc_isalpha = wc_isalpha_libc_ascii,
-		.wc_isalnum = wc_isalnum_libc_ascii,
-		.wc_isupper = wc_isupper_libc_ascii,
-		.wc_islower = wc_islower_libc_ascii,
-		.wc_isgraph = wc_isgraph_libc_ascii,
-		.wc_isprint = wc_isprint_libc_ascii,
-		.wc_ispunct = wc_ispunct_libc_ascii,
-		.wc_isspace = wc_isspace_libc_ascii,
-		.wc_isxdigit = wc_isxdigit_libc_ascii,
+		.wc_isdigit = wc_isdigit_libc_custom,
+		.wc_isalpha = wc_isalpha_libc_custom,
+		.wc_isalnum = wc_isalnum_libc_custom,
+		.wc_isupper = wc_isupper_libc_custom,
+		.wc_islower = wc_islower_libc_custom,
+		.wc_isgraph = wc_isgraph_libc_custom,
+		.wc_isprint = wc_isprint_libc_custom,
+		.wc_ispunct = wc_ispunct_libc_custom,
+		.wc_isspace = wc_isspace_libc_custom,
+		.wc_isxdigit = wc_isxdigit_libc_custom,
 		.char_is_cased = char_is_cased_libc,
 		.char_tolower = char_tolower_libc,
-		.wc_toupper = toupper_libc_ascii,
-		.wc_tolower = tolower_libc_ascii,
-		.max_chr = UCHAR_MAX,
+		.wc_toupper = toupper_libc_custom,
+		.wc_tolower = tolower_libc_custom,
+		.max_chr = 127,		/* values outside ASCII may be illegal to probe */
 	},
 };
 
-- 
2.50.1 (Apple Git-155)

From 50dc5a57bcd31be893071da62c4ba6b0537695cc Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Thu, 30 Oct 2025 00:52:51 +1300
Subject: [PATCH 5/8] XXX work around regc_pg_locale.c's probing logic

---
 src/backend/regex/regc_pg_locale.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/src/backend/regex/regc_pg_locale.c b/src/backend/regex/regc_pg_locale.c
index e0c892db713..fdc3fac0bbe 100644
--- a/src/backend/regex/regc_pg_locale.c
+++ b/src/backend/regex/regc_pg_locale.c
@@ -352,7 +352,15 @@ regc_ctype_get_cache(regc_wc_probefunc probefunc, int cclasscode)
 			pg_regex_locale->ctype->max_chr <= MAX_SIMPLE_CHR)
 		{
 			max_chr = pg_regex_locale->ctype->max_chr;
-			pcc->cv.cclasscode = -1;
+
+			/*
+			 * XXX TODO: don't turn off locales just because pg_locale_libc.c
+			 * told us it's not cool to probe arbitrary pg_wchar values over
+			 * 127!  Without this, re-encoding fails at pg_wchar 0x80, which
+			 * can't be converted back to mb (the EUC pg_wchar encoding has
+			 * holes in it)
+			 */
+			//pcc->cv.cclasscode = -1;
 		}
 		else
 			max_chr = (pg_wchar) MAX_SIMPLE_CHR;
-- 
2.50.1 (Apple Git-155)

From 553e2d5a2923d582a0f1a3ef72033b40149c08c7 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 16:01:32 +1300
Subject: [PATCH 6/8] Improve naming of libc collation functions.

The functions that expect pg_wchar to hold a UTF-32-encoded code point
because the encoding scheme is PG_WCHAR_UTF32 had names ending _mb, but
_utf32 makes more sense.  The remaining _mb functions really do work
with multibyte input.

It might be tempting to rename the _sb functions to _char to match
PG_WCHAR_CHAR, but since the _sb and _mb functions both work with chars (one
or variable), that would probably just be more confusing.
---
 src/backend/utils/adt/pg_locale_libc.c | 48 +++++++++++++-------------
 1 file changed, 24 insertions(+), 24 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index e6724880f1b..b33897c683e 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -267,70 +267,70 @@ wc_isxdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
 }
 
 static bool
-wc_isdigit_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isdigit_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswdigit_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isalpha_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isalpha_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswalpha_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isalnum_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isalnum_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswalnum_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isupper_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isupper_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswupper_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_islower_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_islower_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswlower_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isgraph_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isgraph_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswgraph_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isprint_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isprint_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswprint_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_ispunct_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_ispunct_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswpunct_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isspace_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isspace_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
 		iswspace_l((wint_t) wc, locale->lt);
 }
 
 static bool
-wc_isxdigit_libc_mb(pg_wchar wc, pg_locale_t locale)
+wc_isxdigit_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	return iswxdigit_l((wint_t) wc, locale->lt);
 }
@@ -384,7 +384,7 @@ toupper_libc_sb(pg_wchar wc, pg_locale_t locale)
 }
 
 static pg_wchar
-toupper_libc_mb(pg_wchar wc, pg_locale_t locale)
+toupper_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	Assert(GetDatabaseEncoding() == PG_UTF8);
 
@@ -428,7 +428,7 @@ tolower_libc_sb(pg_wchar wc, pg_locale_t locale)
 }
 
 static pg_wchar
-tolower_libc_mb(pg_wchar wc, pg_locale_t locale)
+tolower_libc_utf32(pg_wchar wc, pg_locale_t locale)
 {
 	Assert(GetDatabaseEncoding() == PG_UTF8);
 
@@ -466,20 +466,20 @@ static const struct ctype_methods ctype_methods_libc[] = {
 		.strlower = strlower_libc_mb,
 		.strtitle = strtitle_libc_mb,
 		.strupper = strupper_libc_mb,
-		.wc_isdigit = wc_isdigit_libc_mb,
-		.wc_isalpha = wc_isalpha_libc_mb,
-		.wc_isalnum = wc_isalnum_libc_mb,
-		.wc_isupper = wc_isupper_libc_mb,
-		.wc_islower = wc_islower_libc_mb,
-		.wc_isgraph = wc_isgraph_libc_mb,
-		.wc_isprint = wc_isprint_libc_mb,
-		.wc_ispunct = wc_ispunct_libc_mb,
-		.wc_isspace = wc_isspace_libc_mb,
-		.wc_isxdigit = wc_isxdigit_libc_mb,
+		.wc_isdigit = wc_isdigit_libc_utf32,
+		.wc_isalpha = wc_isalpha_libc_utf32,
+		.wc_isalnum = wc_isalnum_libc_utf32,
+		.wc_isupper = wc_isupper_libc_utf32,
+		.wc_islower = wc_islower_libc_utf32,
+		.wc_isgraph = wc_isgraph_libc_utf32,
+		.wc_isprint = wc_isprint_libc_utf32,
+		.wc_ispunct = wc_ispunct_libc_utf32,
+		.wc_isspace = wc_isspace_libc_utf32,
+		.wc_isxdigit = wc_isxdigit_libc_utf32,
 		.char_is_cased = char_is_cased_libc,
 		.char_tolower = char_tolower_libc,
-		.wc_toupper = toupper_libc_mb,
-		.wc_tolower = tolower_libc_mb,
+		.wc_toupper = toupper_libc_utf32,
+		.wc_tolower = tolower_libc_utf32,
 	},
 
 	/*
-- 
2.50.1 (Apple Git-155)

From b2ddd005740f73c6d4a9e2441616926fcea94684 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 17:54:06 +1300
Subject: [PATCH 7/8] Use compact notation for isXXX_l() wrappers.

Instead of loads of repeating functions for the PG_WCHAR_CHAR and
PG_WCHAR_UTF32 handlers, make a macro to avoid typos, as was already
done for the new PG_WCHAR_CUSTOM handlers.
---
 src/backend/utils/adt/pg_locale_libc.c | 166 +++++++------------------
 1 file changed, 43 insertions(+), 123 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index b33897c683e..fc758e2607c 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -206,134 +206,54 @@ DEFINE_WC_CTYPE_LIBC_CUSTOM(punct);
 DEFINE_WC_CTYPE_LIBC_CUSTOM(space);
 DEFINE_WC_CTYPE_LIBC_CUSTOM(xdigit);
 
-static bool
-wc_isdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isdigit_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isalpha_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isalpha_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isalnum_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isalnum_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isupper_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isupper_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_islower_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return islower_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isgraph_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isgraph_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isprint_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isprint_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_ispunct_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return ispunct_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isspace_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isspace_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isxdigit_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	return isxdigit_l((unsigned char) wc, locale->lt);
-}
-
-static bool
-wc_isdigit_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswdigit_l((wint_t) wc, locale->lt);
-}
-
-static bool
-wc_isalpha_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswalpha_l((wint_t) wc, locale->lt);
-}
-
-static bool
-wc_isalnum_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswalnum_l((wint_t) wc, locale->lt);
-}
-
-static bool
-wc_isupper_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswupper_l((wint_t) wc, locale->lt);
-}
-
-static bool
-wc_islower_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswlower_l((wint_t) wc, locale->lt);
-}
-
-static bool
-wc_isgraph_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswgraph_l((wint_t) wc, locale->lt);
-}
-
-static bool
-wc_isprint_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswprint_l((wint_t) wc, locale->lt);
+/*
+ * Generate a function that passes single-byte characters directly to <ctype.h>
+ * functions.  This is suitable for PG_WCHAR_CHAR encodings, where pg_wchar
+ * holds a one byte.
+ */
+#define DEFINE_WC_CTYPE_LIBC_SB(ctype) \
+static bool \
+wc_is##ctype##_libc_sb(pg_wchar wc, pg_locale_t locale) \
+{ \
+	return is##ctype##_l((unsigned char) wc, locale->lt); \
 }
 
-static bool
-wc_ispunct_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswpunct_l((wint_t) wc, locale->lt);
-}
+DEFINE_WC_CTYPE_LIBC_SB(digit);
+DEFINE_WC_CTYPE_LIBC_SB(alpha);
+DEFINE_WC_CTYPE_LIBC_SB(alnum);
+DEFINE_WC_CTYPE_LIBC_SB(upper);
+DEFINE_WC_CTYPE_LIBC_SB(lower);
+DEFINE_WC_CTYPE_LIBC_SB(graph);
+DEFINE_WC_CTYPE_LIBC_SB(print);
+DEFINE_WC_CTYPE_LIBC_SB(punct);
+DEFINE_WC_CTYPE_LIBC_SB(space);
+DEFINE_WC_CTYPE_LIBC_SB(xdigit);
 
-static bool
-wc_isspace_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) &&
-		iswspace_l((wint_t) wc, locale->lt);
+/*
+ * Generate a function that passes UTF-32 characters directly to <wctype.h>
+ * functions, This is suitable for PG_WCHAR_UTF32 encodings, with the
+ * assumption that any libc locale that uses UTF-8 as its char encoding must
+ * use UTF-32 or UTF-16 for its wchar_t encoding.  For the UTF-16 case, just
+ * return false for codepoints outside the BMP.
+ */
+#define DEFINE_WC_CTYPE_LIBC_UTF32(ctype) \
+static bool \
+wc_is##ctype##_libc_utf32(pg_wchar wc, pg_locale_t locale) \
+{ \
+	return (sizeof(wchar_t) >= 4 || wc <= 0xffff) && \
+		isw##ctype##_l((wint_t) wc, locale->lt); \
 }
 
-static bool
-wc_isxdigit_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	return iswxdigit_l((wint_t) wc, locale->lt);
-}
+DEFINE_WC_CTYPE_LIBC_UTF32(digit);
+DEFINE_WC_CTYPE_LIBC_UTF32(alpha);
+DEFINE_WC_CTYPE_LIBC_UTF32(alnum);
+DEFINE_WC_CTYPE_LIBC_UTF32(upper);
+DEFINE_WC_CTYPE_LIBC_UTF32(lower);
+DEFINE_WC_CTYPE_LIBC_UTF32(graph);
+DEFINE_WC_CTYPE_LIBC_UTF32(print);
+DEFINE_WC_CTYPE_LIBC_UTF32(punct);
+DEFINE_WC_CTYPE_LIBC_UTF32(space);
+DEFINE_WC_CTYPE_LIBC_UTF32(xdigit);
 
 static char
 char_tolower_libc(unsigned char ch, pg_locale_t locale)
-- 
2.50.1 (Apple Git-155)

From c70c67492e6c1a9fffd97037abe49251e408cff8 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Wed, 29 Oct 2025 23:23:42 +1300
Subject: [PATCH 8/8] Use compact notation for toupper/tolower wrappers.

Extend the macro technique used for generating isalpha etc also to
toupper/tolower functions, removing some duplication.
---
 src/backend/utils/adt/pg_locale_libc.c | 118 ++++++++-----------------
 1 file changed, 39 insertions(+), 79 deletions(-)

diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c
index fc758e2607c..7591fb812ac 100644
--- a/src/backend/utils/adt/pg_locale_libc.c
+++ b/src/backend/utils/adt/pg_locale_libc.c
@@ -273,93 +273,53 @@ char_is_cased_libc(char ch, pg_locale_t locale)
 		return isalpha_l((unsigned char) ch, locale->lt);
 }
 
-static pg_wchar
-toupper_libc_custom(pg_wchar wc, pg_locale_t locale)
-{
-	wint_t		wint;
-
-	/* force C behavior for ASCII characters, per comments above */
-	if (locale->is_default && wc <= (pg_wchar) 127)
-		return pg_ascii_toupper((unsigned char) wc);
-	if (wc <= 127)
-		return towupper_l((wint_t) wc, locale->lt);
-	wint = pg_wchar_to_wchar_t(wc, locale);
-	if ((wint = pg_wchar_to_wchar_t(wc, locale)) != WEOF)
-		return wchar_t_to_pg_wchar(towupper_l(wint, locale->lt), locale);
-	return wc;
-}
-
-static pg_wchar
-toupper_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	Assert(GetDatabaseEncoding() != PG_UTF8);
-
-	/* force C behavior for ASCII characters, per comments above */
-	if (locale->is_default && wc <= (pg_wchar) 127)
-		return pg_ascii_toupper((unsigned char) wc);
-	if (wc <= (pg_wchar) UCHAR_MAX)
-		return toupper_l((unsigned char) wc, locale->lt);
-	else
-		return wc;
+#define DEFINE_WC_CASE_LIBC_CUSTOM(case) \
+static pg_wchar \
+to##case##_libc_custom(pg_wchar wc, pg_locale_t locale) \
+{ \
+	wint_t		wint; \
+	if (locale->is_default && wc <= (pg_wchar) 127) \
+		return pg_ascii_to##case((unsigned char) wc); \
+	if (wc <= 127) \
+		return to##case##_l((wint_t) wc, locale->lt); \
+	wint = pg_wchar_to_wchar_t(wc, locale); \
+	if ((wint = pg_wchar_to_wchar_t(wc, locale)) != WEOF) \
+		return wchar_t_to_pg_wchar(to##case##_l(wint, locale->lt), locale); \
+	return wc; \
 }
 
-static pg_wchar
-toupper_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	Assert(GetDatabaseEncoding() == PG_UTF8);
+DEFINE_WC_CASE_LIBC_CUSTOM(upper);
+DEFINE_WC_CASE_LIBC_CUSTOM(lower);
 
-	/* force C behavior for ASCII characters, per comments above */
-	if (locale->is_default && wc <= (pg_wchar) 127)
-		return pg_ascii_toupper((unsigned char) wc);
-	if (sizeof(wchar_t) >= 4 || wc <= (pg_wchar) 0xFFFF)
-		return towupper_l((wint_t) wc, locale->lt);
-	else
-		return wc;
-}
-
-static pg_wchar
-tolower_libc_custom(pg_wchar wc, pg_locale_t locale)
-{
-	wint_t		wint;
-
-	/* force C behavior for ASCII characters, per comments above */
-	if (locale->is_default && wc <= (pg_wchar) 127)
-		return pg_ascii_tolower((unsigned char) wc);
-	if (wc <= 127)
-		return towlower_l((wint_t) wc, locale->lt);
-	wint = pg_wchar_to_wchar_t(wc, locale);
-	if ((wint = pg_wchar_to_wchar_t(wc, locale)) != WEOF)
-		return wchar_t_to_pg_wchar(towlower_l(wint, locale->lt), locale);
-	return wc;
+#define DEFINE_WC_CASE_LIBC_SB(case) \
+static pg_wchar \
+to##case##_libc_sb(pg_wchar wc, pg_locale_t locale) \
+{ \
+	if (locale->is_default && wc <= (pg_wchar) 127) \
+		return pg_ascii_to##case((unsigned char) wc); \
+	if (wc <= (pg_wchar) UCHAR_MAX) \
+		return to##case##_l((unsigned char) wc, locale->lt); \
+	else \
+		return wc; \
 }
 
-static pg_wchar
-tolower_libc_sb(pg_wchar wc, pg_locale_t locale)
-{
-	Assert(GetDatabaseEncoding() != PG_UTF8);
+DEFINE_WC_CASE_LIBC_SB(upper);
+DEFINE_WC_CASE_LIBC_SB(lower);
 
-	/* force C behavior for ASCII characters, per comments above */
-	if (locale->is_default && wc <= (pg_wchar) 127)
-		return pg_ascii_tolower((unsigned char) wc);
-	if (wc <= (pg_wchar) UCHAR_MAX)
-		return tolower_l((unsigned char) wc, locale->lt);
-	else
-		return wc;
+#define DEFINE_WC_CASE_LIBC_UTF32(case) \
+static pg_wchar \
+to##case##_libc_utf32(pg_wchar wc, pg_locale_t locale) \
+{ \
+	if (locale->is_default && wc <= (pg_wchar) 127) \
+		return pg_ascii_to##case((unsigned char) wc); \
+	if (sizeof(wchar_t) >= 4 || wc <= (pg_wchar) 0xFFFF) \
+		return tow##case##_l((wint_t) wc, locale->lt); \
+	else \
+		return wc; \
 }
 
-static pg_wchar
-tolower_libc_utf32(pg_wchar wc, pg_locale_t locale)
-{
-	Assert(GetDatabaseEncoding() == PG_UTF8);
-
-	/* force C behavior for ASCII characters, per comments above */
-	if (locale->is_default && wc <= (pg_wchar) 127)
-		return pg_ascii_tolower((unsigned char) wc);
-	if (sizeof(wchar_t) >= 4 || wc <= (pg_wchar) 0xFFFF)
-		return towlower_l((wint_t) wc, locale->lt);
-	else
-		return wc;
-}
+DEFINE_WC_CASE_LIBC_UTF32(upper);
+DEFINE_WC_CASE_LIBC_UTF32(lower);
 
 static const struct ctype_methods ctype_methods_libc[] = {
 	[PG_WCHAR_CHAR] = {
-- 
2.50.1 (Apple Git-155)

From 87baaa9f4d6f6b230e082ea8c5667d11997bbeb1 Mon Sep 17 00:00:00 2001
From: Jeff Davis <jeff@j-davis.com>
Date: Tue, 21 Oct 2025 13:16:47 -0700
Subject: [PATCH v5] Use C11 char16_t and char32_t for Unicode code points.

Reviewed-by: Tatsuo Ishii <ishii@postgresql.org>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Reviewed-by: Peter Eisentraut <peter@eisentraut.org>
Discussion: https://postgr.es/m/bedcc93d06203dfd89815b10f815ca2de8626e85.camel%40j-davis.com
---
 configure                                     |  2 +-
 configure.ac                                  |  1 +
 meson.build                                   |  1 +
 src/backend/parser/parser.c                   |  8 +--
 src/backend/parser/scan.l                     |  8 +--
 src/backend/utils/adt/jsonpath_scan.l         |  6 +-
 src/backend/utils/adt/pg_locale_builtin.c     | 44 ++++++++++-----
 src/backend/utils/adt/varlena.c               | 40 ++++++-------
 src/backend/utils/mb/mbutils.c                |  4 +-
 src/common/saslprep.c                         | 48 ++++++++--------
 src/common/unicode/case_test.c                | 23 ++++----
 src/common/unicode/category_test.c            |  3 +-
 .../unicode/generate-norm_test_table.pl       |  4 +-
 .../unicode/generate-unicode_case_table.pl    |  7 +--
 .../generate-unicode_category_table.pl        |  8 +--
 src/common/unicode/norm_test.c                |  6 +-
 src/common/unicode_case.c                     | 56 +++++++++----------
 src/common/unicode_category.c                 | 50 ++++++++---------
 src/common/unicode_norm.c                     | 56 +++++++++----------
 src/fe_utils/mbprint.c                        | 10 ++--
 src/include/c.h                               | 23 ++++++++
 src/include/common/unicode_case.h             | 10 ++--
 src/include/common/unicode_case_table.h       | 13 ++---
 src/include/common/unicode_category.h         | 46 ++++++++-------
 src/include/common/unicode_category_table.h   |  8 +--
 src/include/common/unicode_norm.h             |  6 +-
 src/include/mb/pg_wchar.h                     | 32 +++++------
 src/include/pg_config.h.in                    |  3 +
 src/tools/pgindent/typedefs.list              |  2 +
 29 files changed, 284 insertions(+), 244 deletions(-)

diff --git a/configure b/configure
index 7ce52173dd8..f7c24c8f576 100755
--- a/configure
+++ b/configure
@@ -13627,7 +13627,7 @@ fi
 ## Header files
 ##
 
-for ac_header in atomic.h copyfile.h execinfo.h getopt.h ifaddrs.h mbarrier.h sys/epoll.h sys/event.h sys/personality.h sys/prctl.h sys/procctl.h sys/signalfd.h sys/ucred.h termios.h ucred.h xlocale.h
+for ac_header in atomic.h copyfile.h execinfo.h getopt.h ifaddrs.h mbarrier.h sys/epoll.h sys/event.h sys/personality.h sys/prctl.h sys/procctl.h sys/signalfd.h sys/ucred.h termios.h uchar.h ucred.h xlocale.h
 do :
   as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
 ac_fn_c_check_header_mongrel "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default"
diff --git a/configure.ac b/configure.ac
index 0842fd06259..6c802deaacb 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1513,6 +1513,7 @@ AC_CHECK_HEADERS(m4_normalize([
 	sys/signalfd.h
 	sys/ucred.h
 	termios.h
+	uchar.h
 	ucred.h
 	xlocale.h
 ]))
diff --git a/meson.build b/meson.build
index 1a123ce151a..0f61ff6a700 100644
--- a/meson.build
+++ b/meson.build
@@ -2613,6 +2613,7 @@ header_checks = [
   'sys/signalfd.h',
   'sys/ucred.h',
   'termios.h',
+  'uchar.h',
   'ucred.h',
   'xlocale.h',
 ]
diff --git a/src/backend/parser/parser.c b/src/backend/parser/parser.c
index 33a040506b4..a3679f8e86c 100644
--- a/src/backend/parser/parser.c
+++ b/src/backend/parser/parser.c
@@ -339,7 +339,7 @@ hexval(unsigned char c)
 
 /* is Unicode code point acceptable? */
 static void
-check_unicode_value(pg_wchar c)
+check_unicode_value(char32_t c)
 {
 	if (!is_valid_unicode_codepoint(c))
 		ereport(ERROR,
@@ -376,7 +376,7 @@ str_udeescape(const char *str, char escape,
 	char	   *new,
 			   *out;
 	size_t		new_len;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	ScannerCallbackState scbstate;
 
 	/*
@@ -420,7 +420,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[3]) &&
 					 isxdigit((unsigned char) in[4]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[1]) << 12) +
 					(hexval(in[2]) << 8) +
@@ -457,7 +457,7 @@ str_udeescape(const char *str, char escape,
 					 isxdigit((unsigned char) in[6]) &&
 					 isxdigit((unsigned char) in[7]))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = (hexval(in[2]) << 20) +
 					(hexval(in[3]) << 16) +
diff --git a/src/backend/parser/scan.l b/src/backend/parser/scan.l
index 08990831fe8..a67815339b7 100644
--- a/src/backend/parser/scan.l
+++ b/src/backend/parser/scan.l
@@ -121,7 +121,7 @@ static void addlitchar(unsigned char ychar, core_yyscan_t yyscanner);
 static char *litbufdup(core_yyscan_t yyscanner);
 static unsigned char unescape_single_char(unsigned char c, core_yyscan_t yyscanner);
 static int	process_integer_literal(const char *token, YYSTYPE *lval, int base);
-static void addunicode(pg_wchar c, yyscan_t yyscanner);
+static void addunicode(char32_t c, yyscan_t yyscanner);
 
 #define yyerror(msg)  scanner_yyerror(msg, yyscanner)
 
@@ -640,7 +640,7 @@ other			.
 					addlit(yytext, yyleng, yyscanner);
 				}
 <xe>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/*
 					 * For consistency with other productions, issue any
@@ -668,7 +668,7 @@ other			.
 					POP_YYLLOC();
 				}
 <xeu>{xeunicode} {
-					pg_wchar	c = strtoul(yytext + 2, NULL, 16);
+					char32_t	c = strtoul(yytext + 2, NULL, 16);
 
 					/* Remember start of overall string token ... */
 					PUSH_YYLLOC();
@@ -1376,7 +1376,7 @@ process_integer_literal(const char *token, YYSTYPE *lval, int base)
 }
 
 static void
-addunicode(pg_wchar c, core_yyscan_t yyscanner)
+addunicode(char32_t c, core_yyscan_t yyscanner)
 {
 	ScannerCallbackState scbstate;
 	char		buf[MAX_UNICODE_EQUIVALENT_STRING + 1];
diff --git a/src/backend/utils/adt/jsonpath_scan.l b/src/backend/utils/adt/jsonpath_scan.l
index c7aab83eeb4..8c3a0a9c642 100644
--- a/src/backend/utils/adt/jsonpath_scan.l
+++ b/src/backend/utils/adt/jsonpath_scan.l
@@ -574,7 +574,7 @@ hexval(char c, int *result, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add given unicode character to scanstring */
 static bool
-addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
+addUnicodeChar(char32_t ch, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (ch == 0)
 	{
@@ -607,7 +607,7 @@ addUnicodeChar(int ch, struct Node *escontext, yyscan_t yyscanner)
 
 /* Add unicode character, processing any surrogate pairs */
 static bool
-addUnicode(int ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
+addUnicode(char32_t ch, int *hi_surrogate, struct Node *escontext, yyscan_t yyscanner)
 {
 	if (is_utf16_surrogate_first(ch))
 	{
@@ -655,7 +655,7 @@ parseUnicode(char *s, int l, struct Node *escontext, yyscan_t yyscanner)
 
 	for (i = 2; i < l; i += 2)	/* skip '\u' */
 	{
-		int			ch = 0;
+		char32_t		ch = 0;
 		int			j,
 					si;
 
diff --git a/src/backend/utils/adt/pg_locale_builtin.c b/src/backend/utils/adt/pg_locale_builtin.c
index 3dc611b50e1..1021e0d129b 100644
--- a/src/backend/utils/adt/pg_locale_builtin.c
+++ b/src/backend/utils/adt/pg_locale_builtin.c
@@ -15,7 +15,6 @@
 #include "catalog/pg_collation.h"
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
-#include "mb/pg_wchar.h"
 #include "miscadmin.h"
 #include "utils/builtins.h"
 #include "utils/pg_locale.h"
@@ -35,6 +34,23 @@ struct WordBoundaryState
 	bool		prev_alnum;
 };
 
+/*
+ * In UTF-8, pg_wchar is guaranteed to be the code point value.
+ */
+static inline char32_t
+to_char32(pg_wchar wc)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (char32_t) wc;
+}
+
+static inline pg_wchar
+to_pg_wchar(char32_t c32)
+{
+	Assert(GetDatabaseEncoding() == PG_UTF8);
+	return (pg_wchar) c32;
+}
+
 /*
  * Simple word boundary iterator that draws boundaries each time the result of
  * pg_u_isalnum() changes.
@@ -47,7 +63,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -112,61 +128,61 @@ strfold_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen,
 static bool
 wc_isdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isalpha_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalpha(wc);
+	return pg_u_isalpha(to_char32(wc));
 }
 
 static bool
 wc_isalnum_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isalnum(wc, !locale->builtin.casemap_full);
+	return pg_u_isalnum(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isupper(wc);
+	return pg_u_isupper(to_char32(wc));
 }
 
 static bool
 wc_islower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_islower(wc);
+	return pg_u_islower(to_char32(wc));
 }
 
 static bool
 wc_isgraph_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isgraph(wc);
+	return pg_u_isgraph(to_char32(wc));
 }
 
 static bool
 wc_isprint_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isprint(wc);
+	return pg_u_isprint(to_char32(wc));
 }
 
 static bool
 wc_ispunct_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_ispunct(wc, !locale->builtin.casemap_full);
+	return pg_u_ispunct(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
 wc_isspace_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isspace(wc);
+	return pg_u_isspace(to_char32(wc));
 }
 
 static bool
 wc_isxdigit_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return pg_u_isxdigit(wc, !locale->builtin.casemap_full);
+	return pg_u_isxdigit(to_char32(wc), !locale->builtin.casemap_full);
 }
 
 static bool
@@ -179,13 +195,13 @@ char_is_cased_builtin(char ch, pg_locale_t locale)
 static pg_wchar
 wc_toupper_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_uppercase_simple(wc);
+	return to_pg_wchar(unicode_uppercase_simple(to_char32(wc)));
 }
 
 static pg_wchar
 wc_tolower_builtin(pg_wchar wc, pg_locale_t locale)
 {
-	return unicode_lowercase_simple(wc);
+	return to_pg_wchar(unicode_lowercase_simple(to_char32(wc)));
 }
 
 static const struct ctype_methods ctype_methods_builtin = {
diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c
index 2c398cd9e5c..8d735786e51 100644
--- a/src/backend/utils/adt/varlena.c
+++ b/src/backend/utils/adt/varlena.c
@@ -5419,12 +5419,12 @@ unicode_assigned(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errmsg("Unicode categorization can only be performed if server encoding is UTF8")));
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (int i = 0; i < size; i++)
 	{
-		pg_wchar	uchar = utf8_to_unicode(p);
+		char32_t	uchar = utf8_to_unicode(p);
 		int			category = unicode_category(uchar);
 
 		if (category == PG_U_UNASSIGNED)
@@ -5443,24 +5443,24 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	text	   *result;
 	int			i;
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* action */
@@ -5468,7 +5468,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 
 	/* convert back to UTF-8 string */
 	size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unsigned char buf[4];
 
@@ -5480,7 +5480,7 @@ unicode_normalize_func(PG_FUNCTION_ARGS)
 	SET_VARSIZE(result, size + VARHDRSZ);
 
 	p = (unsigned char *) VARDATA_ANY(result);
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 	{
 		unicode_to_utf8(*wp, p);
 		p += pg_utf_mblen(p);
@@ -5509,8 +5509,8 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	char	   *formstr = text_to_cstring(PG_GETARG_TEXT_PP(1));
 	UnicodeNormalizationForm form;
 	int			size;
-	pg_wchar   *input_chars;
-	pg_wchar   *output_chars;
+	char32_t   *input_chars;
+	char32_t   *output_chars;
 	unsigned char *p;
 	int			i;
 	UnicodeNormalizationQC quickcheck;
@@ -5519,16 +5519,16 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 
 	form = unicode_norm_form_from_string(formstr);
 
-	/* convert to pg_wchar */
+	/* convert to char32_t */
 	size = pg_mbstrlen_with_len(VARDATA_ANY(input), VARSIZE_ANY_EXHDR(input));
-	input_chars = palloc((size + 1) * sizeof(pg_wchar));
+	input_chars = palloc((size + 1) * sizeof(char32_t));
 	p = (unsigned char *) VARDATA_ANY(input);
 	for (i = 0; i < size; i++)
 	{
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 	Assert((char *) p == VARDATA_ANY(input) + VARSIZE_ANY_EXHDR(input));
 
 	/* quick check (see UAX #15) */
@@ -5542,11 +5542,11 @@ unicode_is_normalized(PG_FUNCTION_ARGS)
 	output_chars = unicode_normalize(form, input_chars);
 
 	output_size = 0;
-	for (pg_wchar *wp = output_chars; *wp; wp++)
+	for (char32_t *wp = output_chars; *wp; wp++)
 		output_size++;
 
 	result = (size == output_size) &&
-		(memcmp(input_chars, output_chars, size * sizeof(pg_wchar)) == 0);
+		(memcmp(input_chars, output_chars, size * sizeof(char32_t)) == 0);
 
 	PG_RETURN_BOOL(result);
 }
@@ -5602,7 +5602,7 @@ unistr(PG_FUNCTION_ARGS)
 	int			len;
 	StringInfoData str;
 	text	   *result;
-	pg_wchar	pair_first = 0;
+	char16_t	pair_first = 0;
 	char		cbuf[MAX_UNICODE_EQUIVALENT_STRING + 1];
 
 	instr = VARDATA_ANY(input_text);
@@ -5626,7 +5626,7 @@ unistr(PG_FUNCTION_ARGS)
 			else if ((len >= 5 && isxdigits_n(instr + 1, 4)) ||
 					 (len >= 6 && instr[1] == 'u' && isxdigits_n(instr + 2, 4)))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 				int			offset = instr[1] == 'u' ? 2 : 1;
 
 				unicode = hexval_n(instr + offset, 4);
@@ -5662,7 +5662,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 8 && instr[1] == '+' && isxdigits_n(instr + 2, 6))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 6);
 
@@ -5697,7 +5697,7 @@ unistr(PG_FUNCTION_ARGS)
 			}
 			else if (len >= 10 && instr[1] == 'U' && isxdigits_n(instr + 2, 8))
 			{
-				pg_wchar	unicode;
+				char32_t	unicode;
 
 				unicode = hexval_n(instr + 2, 8);
 
diff --git a/src/backend/utils/mb/mbutils.c b/src/backend/utils/mb/mbutils.c
index 886ecbad871..fb629ed5c8f 100644
--- a/src/backend/utils/mb/mbutils.c
+++ b/src/backend/utils/mb/mbutils.c
@@ -862,7 +862,7 @@ perform_default_encoding_conversion(const char *src, int len,
  * may call this outside any transaction, or in an aborted transaction.
  */
 void
-pg_unicode_to_server(pg_wchar c, unsigned char *s)
+pg_unicode_to_server(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
@@ -924,7 +924,7 @@ pg_unicode_to_server(pg_wchar c, unsigned char *s)
  * but simply return false on conversion failure.
  */
 bool
-pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s)
+pg_unicode_to_server_noerror(char32_t c, unsigned char *s)
 {
 	unsigned char c_as_utf8[MAX_MULTIBYTE_CHAR_LEN + 1];
 	int			c_as_utf8_len;
diff --git a/src/common/saslprep.c b/src/common/saslprep.c
index 97beb47940b..101e8d65a4d 100644
--- a/src/common/saslprep.c
+++ b/src/common/saslprep.c
@@ -47,7 +47,7 @@
 
 /* Prototypes for local functions */
 static int	codepoint_range_cmp(const void *a, const void *b);
-static bool is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize);
+static bool is_code_in_table(char32_t code, const char32_t *map, int mapsize);
 static int	pg_utf8_string_len(const char *source);
 
 /*
@@ -64,7 +64,7 @@ static int	pg_utf8_string_len(const char *source);
  *
  * These are all mapped to the ASCII space character (U+00A0).
  */
-static const pg_wchar non_ascii_space_ranges[] =
+static const char32_t non_ascii_space_ranges[] =
 {
 	0x00A0, 0x00A0,
 	0x1680, 0x1680,
@@ -79,7 +79,7 @@ static const pg_wchar non_ascii_space_ranges[] =
  *
  * If any of these appear in the input, they are removed.
  */
-static const pg_wchar commonly_mapped_to_nothing_ranges[] =
+static const char32_t commonly_mapped_to_nothing_ranges[] =
 {
 	0x00AD, 0x00AD,
 	0x034F, 0x034F,
@@ -114,7 +114,7 @@ static const pg_wchar commonly_mapped_to_nothing_ranges[] =
  * tables, so one code might originate from multiple source tables.
  * Adjacent ranges have also been merged together, to save space.
  */
-static const pg_wchar prohibited_output_ranges[] =
+static const char32_t prohibited_output_ranges[] =
 {
 	0x0000, 0x001F,				/* C.2.1 */
 	0x007F, 0x00A0,				/* C.1.2, C.2.1, C.2.2 */
@@ -155,7 +155,7 @@ static const pg_wchar prohibited_output_ranges[] =
 };
 
 /* A.1 Unassigned code points in Unicode 3.2 */
-static const pg_wchar unassigned_codepoint_ranges[] =
+static const char32_t unassigned_codepoint_ranges[] =
 {
 	0x0221, 0x0221,
 	0x0234, 0x024F,
@@ -556,7 +556,7 @@ static const pg_wchar unassigned_codepoint_ranges[] =
 };
 
 /* D.1 Characters with bidirectional property "R" or "AL" */
-static const pg_wchar RandALCat_codepoint_ranges[] =
+static const char32_t RandALCat_codepoint_ranges[] =
 {
 	0x05BE, 0x05BE,
 	0x05C0, 0x05C0,
@@ -595,7 +595,7 @@ static const pg_wchar RandALCat_codepoint_ranges[] =
 };
 
 /* D.2 Characters with bidirectional property "L" */
-static const pg_wchar LCat_codepoint_ranges[] =
+static const char32_t LCat_codepoint_ranges[] =
 {
 	0x0041, 0x005A,
 	0x0061, 0x007A,
@@ -968,8 +968,8 @@ static const pg_wchar LCat_codepoint_ranges[] =
 static int
 codepoint_range_cmp(const void *a, const void *b)
 {
-	const pg_wchar *key = (const pg_wchar *) a;
-	const pg_wchar *range = (const pg_wchar *) b;
+	const char32_t *key = (const char32_t *) a;
+	const char32_t *range = (const char32_t *) b;
 
 	if (*key < range[0])
 		return -1;				/* less than lower bound */
@@ -980,14 +980,14 @@ codepoint_range_cmp(const void *a, const void *b)
 }
 
 static bool
-is_code_in_table(pg_wchar code, const pg_wchar *map, int mapsize)
+is_code_in_table(char32_t code, const char32_t *map, int mapsize)
 {
 	Assert(mapsize % 2 == 0);
 
 	if (code < map[0] || code > map[mapsize - 1])
 		return false;
 
-	if (bsearch(&code, map, mapsize / 2, sizeof(pg_wchar) * 2,
+	if (bsearch(&code, map, mapsize / 2, sizeof(char32_t) * 2,
 				codepoint_range_cmp))
 		return true;
 	else
@@ -1046,8 +1046,8 @@ pg_utf8_string_len(const char *source)
 pg_saslprep_rc
 pg_saslprep(const char *input, char **output)
 {
-	pg_wchar   *input_chars = NULL;
-	pg_wchar   *output_chars = NULL;
+	char32_t   *input_chars = NULL;
+	char32_t   *output_chars = NULL;
 	int			input_size;
 	char	   *result;
 	int			result_size;
@@ -1055,7 +1055,7 @@ pg_saslprep(const char *input, char **output)
 	int			i;
 	bool		contains_RandALCat;
 	unsigned char *p;
-	pg_wchar   *wp;
+	char32_t   *wp;
 
 	/* Ensure we return *output as NULL on failure */
 	*output = NULL;
@@ -1080,10 +1080,10 @@ pg_saslprep(const char *input, char **output)
 	input_size = pg_utf8_string_len(input);
 	if (input_size < 0)
 		return SASLPREP_INVALID_UTF8;
-	if (input_size >= MaxAllocSize / sizeof(pg_wchar))
+	if (input_size >= MaxAllocSize / sizeof(char32_t))
 		goto oom;
 
-	input_chars = ALLOC((input_size + 1) * sizeof(pg_wchar));
+	input_chars = ALLOC((input_size + 1) * sizeof(char32_t));
 	if (!input_chars)
 		goto oom;
 
@@ -1093,7 +1093,7 @@ pg_saslprep(const char *input, char **output)
 		input_chars[i] = utf8_to_unicode(p);
 		p += pg_utf_mblen(p);
 	}
-	input_chars[i] = (pg_wchar) '\0';
+	input_chars[i] = (char32_t) '\0';
 
 	/*
 	 * The steps below correspond to the steps listed in [RFC3454], Section
@@ -1107,7 +1107,7 @@ pg_saslprep(const char *input, char **output)
 	count = 0;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, non_ascii_space_ranges))
 			input_chars[count++] = 0x0020;
@@ -1118,7 +1118,7 @@ pg_saslprep(const char *input, char **output)
 		else
 			input_chars[count++] = code;
 	}
-	input_chars[count] = (pg_wchar) '\0';
+	input_chars[count] = (char32_t) '\0';
 	input_size = count;
 
 	if (input_size == 0)
@@ -1138,7 +1138,7 @@ pg_saslprep(const char *input, char **output)
 	 */
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, prohibited_output_ranges))
 			goto prohibited;
@@ -1170,7 +1170,7 @@ pg_saslprep(const char *input, char **output)
 	contains_RandALCat = false;
 	for (i = 0; i < input_size; i++)
 	{
-		pg_wchar	code = input_chars[i];
+		char32_t	code = input_chars[i];
 
 		if (IS_CODE_IN_TABLE(code, RandALCat_codepoint_ranges))
 		{
@@ -1181,12 +1181,12 @@ pg_saslprep(const char *input, char **output)
 
 	if (contains_RandALCat)
 	{
-		pg_wchar	first = input_chars[0];
-		pg_wchar	last = input_chars[input_size - 1];
+		char32_t	first = input_chars[0];
+		char32_t	last = input_chars[input_size - 1];
 
 		for (i = 0; i < input_size; i++)
 		{
-			pg_wchar	code = input_chars[i];
+			char32_t	code = input_chars[i];
 
 			if (IS_CODE_IN_TABLE(code, LCat_codepoint_ranges))
 				goto prohibited;
diff --git a/src/common/unicode/case_test.c b/src/common/unicode/case_test.c
index fdfb62e8552..00d4f85e5a5 100644
--- a/src/common/unicode/case_test.c
+++ b/src/common/unicode/case_test.c
@@ -24,6 +24,7 @@
 #include "common/unicode_case.h"
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 /* enough to hold largest source or result string, including NUL */
 #define BUFSZ 256
@@ -54,7 +55,7 @@ initcap_wbnext(void *state)
 	while (wbstate->offset < wbstate->len &&
 		   wbstate->str[wbstate->offset] != '\0')
 	{
-		pg_wchar	u = utf8_to_unicode((unsigned char *) wbstate->str +
+		char32_t	u = utf8_to_unicode((unsigned char *) wbstate->str +
 										wbstate->offset);
 		bool		curr_alnum = pg_u_isalnum(u, wbstate->posix);
 
@@ -77,16 +78,16 @@ initcap_wbnext(void *state)
 #ifdef USE_ICU
 
 static void
-icu_test_simple(pg_wchar code)
+icu_test_simple(char32_t code)
 {
-	pg_wchar	lower = unicode_lowercase_simple(code);
-	pg_wchar	title = unicode_titlecase_simple(code);
-	pg_wchar	upper = unicode_uppercase_simple(code);
-	pg_wchar	fold = unicode_casefold_simple(code);
-	pg_wchar	iculower = u_tolower(code);
-	pg_wchar	icutitle = u_totitle(code);
-	pg_wchar	icuupper = u_toupper(code);
-	pg_wchar	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
+	char32_t	lower = unicode_lowercase_simple(code);
+	char32_t	title = unicode_titlecase_simple(code);
+	char32_t	upper = unicode_uppercase_simple(code);
+	char32_t	fold = unicode_casefold_simple(code);
+	char32_t	iculower = u_tolower(code);
+	char32_t	icutitle = u_totitle(code);
+	char32_t	icuupper = u_toupper(code);
+	char32_t	icufold = u_foldCase(code, U_FOLD_CASE_DEFAULT);
 
 	if (lower != iculower || title != icutitle || upper != icuupper ||
 		fold != icufold)
@@ -172,7 +173,7 @@ test_icu(void)
 	int			successful = 0;
 	int			skipped_mismatch = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		pg_unicode_category category = unicode_category(code);
 
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index 5d37ba39196..1e8c1f7905f 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -22,6 +22,7 @@
 
 #include "common/unicode_category.h"
 #include "common/unicode_version.h"
+#include "mb/pg_wchar.h"
 
 static int	pg_unicode_version = 0;
 #ifdef USE_ICU
@@ -59,7 +60,7 @@ icu_test()
 	int			pg_skipped_codepoints = 0;
 	int			icu_skipped_codepoints = 0;
 
-	for (pg_wchar code = 0; code <= 0x10ffff; code++)
+	for (char32_t code = 0; code <= 0x10ffff; code++)
 	{
 		uint8_t		pg_category = unicode_category(code);
 		uint8_t		icu_category = u_charType(code);
diff --git a/src/common/unicode/generate-norm_test_table.pl b/src/common/unicode/generate-norm_test_table.pl
index 1b401be9409..1a8b908ff33 100644
--- a/src/common/unicode/generate-norm_test_table.pl
+++ b/src/common/unicode/generate-norm_test_table.pl
@@ -47,8 +47,8 @@ print $OUTPUT <<HEADER;
 typedef struct
 {
 	int			linenum;
-	pg_wchar	input[50];
-	pg_wchar	output[4][50];
+	char32_t	input[50];
+	char32_t	output[4][50];
 } pg_unicode_test;
 
 /* test table */
diff --git a/src/common/unicode/generate-unicode_case_table.pl b/src/common/unicode/generate-unicode_case_table.pl
index 5d9ddd62803..f71eb25c94e 100644
--- a/src/common/unicode/generate-unicode_case_table.pl
+++ b/src/common/unicode/generate-unicode_case_table.pl
@@ -270,7 +270,6 @@ print $OT <<"EOS";
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -297,7 +296,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -430,7 +429,7 @@ foreach my $kind ('lower', 'title', 'upper', 'fold')
  * The entry case_map_${kind}[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_$kind\[$index\] =
+static const char32_t case_map_$kind\[$index\] =
 {
 EOS
 
@@ -502,7 +501,7 @@ print $OT <<"EOS";
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < $fastpath_limit */
 	if (cp < $fastpath_limit)
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index abab5cd9696..7e094b13720 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -366,15 +366,15 @@ print $OT <<"EOS";
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/common/unicode/norm_test.c b/src/common/unicode/norm_test.c
index 25bc59463f2..058817f1719 100644
--- a/src/common/unicode/norm_test.c
+++ b/src/common/unicode/norm_test.c
@@ -20,7 +20,7 @@
 #include "norm_test_table.h"
 
 static char *
-print_wchar_str(const pg_wchar *s)
+print_wchar_str(const char32_t *s)
 {
 #define BUF_DIGITS 50
 	static char buf[BUF_DIGITS * 11 + 1];
@@ -41,7 +41,7 @@ print_wchar_str(const pg_wchar *s)
 }
 
 static int
-pg_wcscmp(const pg_wchar *s1, const pg_wchar *s2)
+pg_wcscmp(const char32_t *s1, const char32_t *s2)
 {
 	for (;;)
 	{
@@ -65,7 +65,7 @@ main(int argc, char **argv)
 	{
 		for (int form = 0; form < 4; form++)
 		{
-			pg_wchar   *result;
+			char32_t   *result;
 
 			result = unicode_normalize(form, test->input);
 
diff --git a/src/common/unicode_case.c b/src/common/unicode_case.c
index 073faf6a0d5..e5e494db43c 100644
--- a/src/common/unicode_case.c
+++ b/src/common/unicode_case.c
@@ -30,7 +30,7 @@ enum CaseMapResult
 /*
  * Map for each case kind.
  */
-static const pg_wchar *const casekind_map[NCaseKind] =
+static const char32_t *const casekind_map[NCaseKind] =
 {
 	[CaseLower] = case_map_lower,
 	[CaseTitle] = case_map_title,
@@ -38,42 +38,42 @@ static const pg_wchar *const casekind_map[NCaseKind] =
 	[CaseFold] = case_map_fold,
 };
 
-static pg_wchar find_case_map(pg_wchar ucs, const pg_wchar *map);
+static char32_t find_case_map(char32_t ucs, const char32_t *map);
 static size_t convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 						   CaseKind str_casekind, bool full, WordBoundaryNext wbnext,
 						   void *wbstate);
-static enum CaseMapResult casemap(pg_wchar u1, CaseKind casekind, bool full,
+static enum CaseMapResult casemap(char32_t u1, CaseKind casekind, bool full,
 								  const char *src, size_t srclen, size_t srcoff,
-								  pg_wchar *simple, const pg_wchar **special);
+								  char32_t *simple, const char32_t **special);
 
-pg_wchar
-unicode_lowercase_simple(pg_wchar code)
+char32_t
+unicode_lowercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_lower);
+	char32_t	cp = find_case_map(code, case_map_lower);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_titlecase_simple(pg_wchar code)
+char32_t
+unicode_titlecase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_title);
+	char32_t	cp = find_case_map(code, case_map_title);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_uppercase_simple(pg_wchar code)
+char32_t
+unicode_uppercase_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_upper);
+	char32_t	cp = find_case_map(code, case_map_upper);
 
 	return cp != 0 ? cp : code;
 }
 
-pg_wchar
-unicode_casefold_simple(pg_wchar code)
+char32_t
+unicode_casefold_simple(char32_t code)
 {
-	pg_wchar	cp = find_case_map(code, case_map_fold);
+	char32_t	cp = find_case_map(code, case_map_fold);
 
 	return cp != 0 ? cp : code;
 }
@@ -231,10 +231,10 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 
 	while ((srclen < 0 || srcoff < srclen) && src[srcoff] != '\0')
 	{
-		pg_wchar	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
+		char32_t	u1 = utf8_to_unicode((unsigned char *) src + srcoff);
 		int			u1len = unicode_utf8len(u1);
-		pg_wchar	simple = 0;
-		const pg_wchar *special = NULL;
+		char32_t	simple = 0;
+		const char32_t *special = NULL;
 		enum CaseMapResult casemap_result;
 
 		if (str_casekind == CaseTitle)
@@ -265,8 +265,8 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 			case CASEMAP_SIMPLE:
 				{
 					/* replace with single character */
-					pg_wchar	u2 = simple;
-					pg_wchar	u2len = unicode_utf8len(u2);
+					char32_t	u2 = simple;
+					char32_t	u2len = unicode_utf8len(u2);
 
 					Assert(special == NULL);
 					if (result_len + u2len <= dstsize)
@@ -280,7 +280,7 @@ convert_case(char *dst, size_t dstsize, const char *src, ssize_t srclen,
 				Assert(simple == 0);
 				for (int i = 0; i < MAX_CASE_EXPANSION && special[i]; i++)
 				{
-					pg_wchar	u2 = special[i];
+					char32_t	u2 = special[i];
 					size_t		u2len = unicode_utf8len(u2);
 
 					if (result_len + u2len <= dstsize)
@@ -320,7 +320,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -344,7 +344,7 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
 	{
 		if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
 		{
-			pg_wchar	curr = utf8_to_unicode(str + i);
+			char32_t	curr = utf8_to_unicode(str + i);
 
 			if (pg_u_prop_case_ignorable(curr))
 				continue;
@@ -394,9 +394,9 @@ check_special_conditions(int conditions, const char *str, size_t len,
  * character without modification.
  */
 static enum CaseMapResult
-casemap(pg_wchar u1, CaseKind casekind, bool full,
+casemap(char32_t u1, CaseKind casekind, bool full,
 		const char *src, size_t srclen, size_t srcoff,
-		pg_wchar *simple, const pg_wchar **special)
+		char32_t *simple, const char32_t **special)
 {
 	uint16		idx;
 
@@ -434,8 +434,8 @@ casemap(pg_wchar u1, CaseKind casekind, bool full,
  * Find entry in simple case map.
  * If the entry does not exist, 0 will be returned.
  */
-static pg_wchar
-find_case_map(pg_wchar ucs, const pg_wchar *map)
+static char32_t
+find_case_map(char32_t ucs, const char32_t *map)
 {
 	/* Fast path for codepoints < 0x80 */
 	if (ucs < 0x80)
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index 4136c4d4f92..aab667a7bb4 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -1,7 +1,7 @@
 /*-------------------------------------------------------------------------
  * unicode_category.c
  *		Determine general category and character properties of Unicode
- *		characters. Encoding must be UTF8, where we assume that the pg_wchar
+ *		characters. Encoding must be UTF8, where we assume that the char32_t
  *		representation is a code point.
  *
  * Portions Copyright (c) 2017-2025, PostgreSQL Global Development Group
@@ -76,13 +76,13 @@
 #define PG_U_CHARACTER_TAB	0x09
 
 static bool range_search(const pg_unicode_range *tbl, size_t size,
-						 pg_wchar code);
+						 char32_t code);
 
 /*
  * Unicode general category for the given codepoint.
  */
 pg_unicode_category
-unicode_category(pg_wchar code)
+unicode_category(char32_t code)
 {
 	int			min = 0;
 	int			mid;
@@ -108,7 +108,7 @@ unicode_category(pg_wchar code)
 }
 
 bool
-pg_u_prop_alphabetic(pg_wchar code)
+pg_u_prop_alphabetic(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_ALPHABETIC;
@@ -119,7 +119,7 @@ pg_u_prop_alphabetic(pg_wchar code)
 }
 
 bool
-pg_u_prop_lowercase(pg_wchar code)
+pg_u_prop_lowercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_LOWERCASE;
@@ -130,7 +130,7 @@ pg_u_prop_lowercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_uppercase(pg_wchar code)
+pg_u_prop_uppercase(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_UPPERCASE;
@@ -141,7 +141,7 @@ pg_u_prop_uppercase(pg_wchar code)
 }
 
 bool
-pg_u_prop_cased(pg_wchar code)
+pg_u_prop_cased(char32_t code)
 {
 	uint32		category_mask;
 
@@ -156,7 +156,7 @@ pg_u_prop_cased(pg_wchar code)
 }
 
 bool
-pg_u_prop_case_ignorable(pg_wchar code)
+pg_u_prop_case_ignorable(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_CASE_IGNORABLE;
@@ -167,7 +167,7 @@ pg_u_prop_case_ignorable(pg_wchar code)
 }
 
 bool
-pg_u_prop_white_space(pg_wchar code)
+pg_u_prop_white_space(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_WHITE_SPACE;
@@ -178,7 +178,7 @@ pg_u_prop_white_space(pg_wchar code)
 }
 
 bool
-pg_u_prop_hex_digit(pg_wchar code)
+pg_u_prop_hex_digit(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_HEX_DIGIT;
@@ -189,7 +189,7 @@ pg_u_prop_hex_digit(pg_wchar code)
 }
 
 bool
-pg_u_prop_join_control(pg_wchar code)
+pg_u_prop_join_control(char32_t code)
 {
 	if (code < 0x80)
 		return unicode_opt_ascii[code].properties & PG_U_PROP_JOIN_CONTROL;
@@ -208,7 +208,7 @@ pg_u_prop_join_control(pg_wchar code)
  */
 
 bool
-pg_u_isdigit(pg_wchar code, bool posix)
+pg_u_isdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return ('0' <= code && code <= '9');
@@ -217,19 +217,19 @@ pg_u_isdigit(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isalpha(pg_wchar code)
+pg_u_isalpha(char32_t code)
 {
 	return pg_u_prop_alphabetic(code);
 }
 
 bool
-pg_u_isalnum(pg_wchar code, bool posix)
+pg_u_isalnum(char32_t code, bool posix)
 {
 	return pg_u_isalpha(code) || pg_u_isdigit(code, posix);
 }
 
 bool
-pg_u_isword(pg_wchar code)
+pg_u_isword(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -240,32 +240,32 @@ pg_u_isword(pg_wchar code)
 }
 
 bool
-pg_u_isupper(pg_wchar code)
+pg_u_isupper(char32_t code)
 {
 	return pg_u_prop_uppercase(code);
 }
 
 bool
-pg_u_islower(pg_wchar code)
+pg_u_islower(char32_t code)
 {
 	return pg_u_prop_lowercase(code);
 }
 
 bool
-pg_u_isblank(pg_wchar code)
+pg_u_isblank(char32_t code)
 {
 	return code == PG_U_CHARACTER_TAB ||
 		unicode_category(code) == PG_U_SPACE_SEPARATOR;
 }
 
 bool
-pg_u_iscntrl(pg_wchar code)
+pg_u_iscntrl(char32_t code)
 {
 	return unicode_category(code) == PG_U_CONTROL;
 }
 
 bool
-pg_u_isgraph(pg_wchar code)
+pg_u_isgraph(char32_t code)
 {
 	uint32		category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
 
@@ -276,7 +276,7 @@ pg_u_isgraph(pg_wchar code)
 }
 
 bool
-pg_u_isprint(pg_wchar code)
+pg_u_isprint(char32_t code)
 {
 	pg_unicode_category category = unicode_category(code);
 
@@ -287,7 +287,7 @@ pg_u_isprint(pg_wchar code)
 }
 
 bool
-pg_u_ispunct(pg_wchar code, bool posix)
+pg_u_ispunct(char32_t code, bool posix)
 {
 	uint32		category_mask;
 
@@ -308,13 +308,13 @@ pg_u_ispunct(pg_wchar code, bool posix)
 }
 
 bool
-pg_u_isspace(pg_wchar code)
+pg_u_isspace(char32_t code)
 {
 	return pg_u_prop_white_space(code);
 }
 
 bool
-pg_u_isxdigit(pg_wchar code, bool posix)
+pg_u_isxdigit(char32_t code, bool posix)
 {
 	if (posix)
 		return (('0' <= code && code <= '9') ||
@@ -478,7 +478,7 @@ unicode_category_abbrev(pg_unicode_category category)
  * given table.
  */
 static bool
-range_search(const pg_unicode_range *tbl, size_t size, pg_wchar code)
+range_search(const pg_unicode_range *tbl, size_t size, char32_t code)
 {
 	int			min = 0;
 	int			mid;
diff --git a/src/common/unicode_norm.c b/src/common/unicode_norm.c
index 6654b4cbc49..489d99cd5ab 100644
--- a/src/common/unicode_norm.c
+++ b/src/common/unicode_norm.c
@@ -69,7 +69,7 @@ conv_compare(const void *p1, const void *p2)
  * lookup, while the frontend version uses a binary search.
  */
 static const pg_unicode_decomposition *
-get_code_entry(pg_wchar code)
+get_code_entry(char32_t code)
 {
 #ifndef FRONTEND
 	int			h;
@@ -109,7 +109,7 @@ get_code_entry(pg_wchar code)
  * Get the combining class of the given codepoint.
  */
 static uint8
-get_canonical_class(pg_wchar code)
+get_canonical_class(char32_t code)
 {
 	const pg_unicode_decomposition *entry = get_code_entry(code);
 
@@ -130,15 +130,15 @@ get_canonical_class(pg_wchar code)
  * Note: the returned pointer can point to statically allocated buffer, and
  * is only valid until next call to this function!
  */
-static const pg_wchar *
+static const char32_t *
 get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
 {
-	static pg_wchar x;
+	static char32_t x;
 
 	if (DECOMPOSITION_IS_INLINE(entry))
 	{
 		Assert(DECOMPOSITION_SIZE(entry) == 1);
-		x = (pg_wchar) entry->dec_index;
+		x = (char32_t) entry->dec_index;
 		*dec_size = 1;
 		return &x;
 	}
@@ -156,7 +156,7 @@ get_code_decomposition(const pg_unicode_decomposition *entry, int *dec_size)
  * are, in turn, decomposable.
  */
 static int
-get_decomposed_size(pg_wchar code, bool compat)
+get_decomposed_size(char32_t code, bool compat)
 {
 	const pg_unicode_decomposition *entry;
 	int			size = 0;
@@ -318,7 +318,7 @@ recompose_code(uint32 start, uint32 code, uint32 *result)
  * in the array result.
  */
 static void
-decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
+decompose_code(char32_t code, bool compat, char32_t **result, int *current)
 {
 	const pg_unicode_decomposition *entry;
 	int			i;
@@ -337,7 +337,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 					v,
 					tindex,
 					sindex;
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		sindex = code - SBASE;
 		l = LBASE + sindex / (VCOUNT * TCOUNT);
@@ -369,7 +369,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	if (entry == NULL || DECOMPOSITION_SIZE(entry) == 0 ||
 		(!compat && DECOMPOSITION_IS_COMPAT(entry)))
 	{
-		pg_wchar   *res = *result;
+		char32_t   *res = *result;
 
 		res[*current] = code;
 		(*current)++;
@@ -382,7 +382,7 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
 	decomp = get_code_decomposition(entry, &dec_size);
 	for (i = 0; i < dec_size; i++)
 	{
-		pg_wchar	lcode = (pg_wchar) decomp[i];
+		char32_t	lcode = (char32_t) decomp[i];
 
 		/* Leave if no more decompositions */
 		decompose_code(lcode, compat, result, current);
@@ -398,17 +398,17 @@ decompose_code(pg_wchar code, bool compat, pg_wchar **result, int *current)
  * malloc. Or NULL if we run out of memory. In backend, the returned
  * string is palloc'd instead, and OOM is reported with ereport().
  */
-pg_wchar *
-unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
+char32_t *
+unicode_normalize(UnicodeNormalizationForm form, const char32_t *input)
 {
 	bool		compat = (form == UNICODE_NFKC || form == UNICODE_NFKD);
 	bool		recompose = (form == UNICODE_NFC || form == UNICODE_NFKC);
-	pg_wchar   *decomp_chars;
-	pg_wchar   *recomp_chars;
+	char32_t   *decomp_chars;
+	char32_t   *recomp_chars;
 	int			decomp_size,
 				current_size;
 	int			count;
-	const pg_wchar *p;
+	const char32_t *p;
 
 	/* variables for recomposition */
 	int			last_class;
@@ -425,7 +425,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	for (p = input; *p; p++)
 		decomp_size += get_decomposed_size(*p, compat);
 
-	decomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	decomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (decomp_chars == NULL)
 		return NULL;
 
@@ -448,9 +448,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 */
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	prev = decomp_chars[count - 1];
-		pg_wchar	next = decomp_chars[count];
-		pg_wchar	tmp;
+		char32_t	prev = decomp_chars[count - 1];
+		char32_t	next = decomp_chars[count];
+		char32_t	tmp;
 		const uint8 prevClass = get_canonical_class(prev);
 		const uint8 nextClass = get_canonical_class(next);
 
@@ -487,7 +487,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 	 * longer than the decomposed one, so make the allocation of the output
 	 * string based on that assumption.
 	 */
-	recomp_chars = (pg_wchar *) ALLOC((decomp_size + 1) * sizeof(pg_wchar));
+	recomp_chars = (char32_t *) ALLOC((decomp_size + 1) * sizeof(char32_t));
 	if (!recomp_chars)
 	{
 		FREE(decomp_chars);
@@ -501,9 +501,9 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 
 	for (count = 1; count < decomp_size; count++)
 	{
-		pg_wchar	ch = decomp_chars[count];
+		char32_t	ch = decomp_chars[count];
 		int			ch_class = get_canonical_class(ch);
-		pg_wchar	composite;
+		char32_t	composite;
 
 		if (last_class < ch_class &&
 			recompose_code(starter_ch, ch, &composite))
@@ -524,7 +524,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 			recomp_chars[target_pos++] = ch;
 		}
 	}
-	recomp_chars[target_pos] = (pg_wchar) '\0';
+	recomp_chars[target_pos] = (char32_t) '\0';
 
 	FREE(decomp_chars);
 
@@ -540,7 +540,7 @@ unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input)
 #ifndef FRONTEND
 
 static const pg_unicode_normprops *
-qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
+qc_hash_lookup(char32_t ch, const pg_unicode_norminfo *norminfo)
 {
 	int			h;
 	uint32		hashkey;
@@ -571,7 +571,7 @@ qc_hash_lookup(pg_wchar ch, const pg_unicode_norminfo *norminfo)
  * Look up the normalization quick check character property
  */
 static UnicodeNormalizationQC
-qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
+qc_is_allowed(UnicodeNormalizationForm form, char32_t ch)
 {
 	const pg_unicode_normprops *found = NULL;
 
@@ -595,7 +595,7 @@ qc_is_allowed(UnicodeNormalizationForm form, pg_wchar ch)
 }
 
 UnicodeNormalizationQC
-unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input)
+unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input)
 {
 	uint8		lastCanonicalClass = 0;
 	UnicodeNormalizationQC result = UNICODE_NORM_QC_YES;
@@ -610,9 +610,9 @@ unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *
 	if (form == UNICODE_NFD || form == UNICODE_NFKD)
 		return UNICODE_NORM_QC_MAYBE;
 
-	for (const pg_wchar *p = input; *p; p++)
+	for (const char32_t *p = input; *p; p++)
 	{
-		pg_wchar	ch = *p;
+		char32_t	ch = *p;
 		uint8		canonicalClass;
 		UnicodeNormalizationQC check;
 
diff --git a/src/fe_utils/mbprint.c b/src/fe_utils/mbprint.c
index eb3eeee9925..abffdbe18a2 100644
--- a/src/fe_utils/mbprint.c
+++ b/src/fe_utils/mbprint.c
@@ -49,20 +49,20 @@ pg_get_utf8_id(void)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static pg_wchar
+static char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
diff --git a/src/include/c.h b/src/include/c.h
index f4ec33e9b07..757dfff4782 100644
--- a/src/include/c.h
+++ b/src/include/c.h
@@ -1376,6 +1376,29 @@ typedef intptr_t sigjmp_buf[5];
 /* /port compatibility functions */
 #include "port.h"
 
+/*
+ * char16_t and char32_t
+ *      Unicode code points.
+ *
+ * uchar.h should always be available in C11, but it's not available on
+ * Mac. However, these types are keywords in C++11, so when using C++, we
+ * can't redefine the types.
+ *
+ * XXX: when uchar.h is available everywhere, we can remove this check and
+ * just include uchar.h unconditionally.
+ *
+ * XXX: this section is out of place because uchar.h needs to be included
+ * after port.h, due to an interaction with win32_port.h in some cases.
+ */
+#ifdef HAVE_UCHAR_H
+#include <uchar.h>
+#else
+#ifndef __cplusplus
+typedef uint16_t char16_t;
+typedef uint32_t char32_t;
+#endif
+#endif
+
 /* IWYU pragma: end_exports */
 
 #endif							/* C_H */
diff --git a/src/include/common/unicode_case.h b/src/include/common/unicode_case.h
index 41e2c1f4b33..6bcffd349c2 100644
--- a/src/include/common/unicode_case.h
+++ b/src/include/common/unicode_case.h
@@ -14,14 +14,12 @@
 #ifndef UNICODE_CASE_H
 #define UNICODE_CASE_H
 
-#include "mb/pg_wchar.h"
-
 typedef size_t (*WordBoundaryNext) (void *wbstate);
 
-pg_wchar	unicode_lowercase_simple(pg_wchar code);
-pg_wchar	unicode_titlecase_simple(pg_wchar code);
-pg_wchar	unicode_uppercase_simple(pg_wchar code);
-pg_wchar	unicode_casefold_simple(pg_wchar code);
+char32_t	unicode_lowercase_simple(char32_t code);
+char32_t	unicode_titlecase_simple(char32_t code);
+char32_t	unicode_uppercase_simple(char32_t code);
+char32_t	unicode_casefold_simple(char32_t code);
 size_t		unicode_strlower(char *dst, size_t dstsize, const char *src,
 							 ssize_t srclen, bool full);
 size_t		unicode_strtitle(char *dst, size_t dstsize, const char *src,
diff --git a/src/include/common/unicode_case_table.h b/src/include/common/unicode_case_table.h
index d5311786582..0a14fb2d97b 100644
--- a/src/include/common/unicode_case_table.h
+++ b/src/include/common/unicode_case_table.h
@@ -18,7 +18,6 @@
  */
 
 #include "common/unicode_case.h"
-#include "mb/pg_wchar.h"
 
 /*
  * The maximum number of codepoints that can result from case mapping
@@ -45,7 +44,7 @@ typedef enum
 typedef struct
 {
 	int16		conditions;
-	pg_wchar	map[NCaseKind][MAX_CASE_EXPANSION];
+	char32_t	map[NCaseKind][MAX_CASE_EXPANSION];
 } pg_special_case;
 
 /*
@@ -166,7 +165,7 @@ static const pg_special_case special_case[106] =
  * The entry case_map_lower[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_lower[1704] =
+static const char32_t case_map_lower[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -1879,7 +1878,7 @@ static const pg_wchar case_map_lower[1704] =
  * The entry case_map_title[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_title[1704] =
+static const char32_t case_map_title[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -3592,7 +3591,7 @@ static const pg_wchar case_map_title[1704] =
  * The entry case_map_upper[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_upper[1704] =
+static const char32_t case_map_upper[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -5305,7 +5304,7 @@ static const pg_wchar case_map_upper[1704] =
  * The entry case_map_fold[case_index(codepoint)] is the mapping for the
  * given codepoint.
  */
-static const pg_wchar case_map_fold[1704] =
+static const char32_t case_map_fold[1704] =
 {
 	0x000000,					/* reserved */
 	0x000000,					/* U+000000 */
@@ -13522,7 +13521,7 @@ static const uint16 case_map[4778] =
  * the offset into the mapping tables.
  */
 static inline uint16
-case_index(pg_wchar cp)
+case_index(char32_t cp)
 {
 	/* Fast path for codepoints < 0x0588 */
 	if (cp < 0x0588)
diff --git a/src/include/common/unicode_category.h b/src/include/common/unicode_category.h
index 8fd8b67a416..684143d3c8a 100644
--- a/src/include/common/unicode_category.h
+++ b/src/include/common/unicode_category.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_CATEGORY_H
 #define UNICODE_CATEGORY_H
 
-#include "mb/pg_wchar.h"
-
 /*
  * Unicode General Category Values
  *
@@ -61,31 +59,31 @@ typedef enum pg_unicode_category
 	PG_U_FINAL_PUNCTUATION = 29 /* Pf */
 } pg_unicode_category;
 
-extern pg_unicode_category unicode_category(pg_wchar code);
+extern pg_unicode_category unicode_category(char32_t code);
 extern const char *unicode_category_string(pg_unicode_category category);
 extern const char *unicode_category_abbrev(pg_unicode_category category);
 
-extern bool pg_u_prop_alphabetic(pg_wchar code);
-extern bool pg_u_prop_lowercase(pg_wchar code);
-extern bool pg_u_prop_uppercase(pg_wchar code);
-extern bool pg_u_prop_cased(pg_wchar code);
-extern bool pg_u_prop_case_ignorable(pg_wchar code);
-extern bool pg_u_prop_white_space(pg_wchar code);
-extern bool pg_u_prop_hex_digit(pg_wchar code);
-extern bool pg_u_prop_join_control(pg_wchar code);
+extern bool pg_u_prop_alphabetic(char32_t code);
+extern bool pg_u_prop_lowercase(char32_t code);
+extern bool pg_u_prop_uppercase(char32_t code);
+extern bool pg_u_prop_cased(char32_t code);
+extern bool pg_u_prop_case_ignorable(char32_t code);
+extern bool pg_u_prop_white_space(char32_t code);
+extern bool pg_u_prop_hex_digit(char32_t code);
+extern bool pg_u_prop_join_control(char32_t code);
 
-extern bool pg_u_isdigit(pg_wchar code, bool posix);
-extern bool pg_u_isalpha(pg_wchar code);
-extern bool pg_u_isalnum(pg_wchar code, bool posix);
-extern bool pg_u_isword(pg_wchar code);
-extern bool pg_u_isupper(pg_wchar code);
-extern bool pg_u_islower(pg_wchar code);
-extern bool pg_u_isblank(pg_wchar code);
-extern bool pg_u_iscntrl(pg_wchar code);
-extern bool pg_u_isgraph(pg_wchar code);
-extern bool pg_u_isprint(pg_wchar code);
-extern bool pg_u_ispunct(pg_wchar code, bool posix);
-extern bool pg_u_isspace(pg_wchar code);
-extern bool pg_u_isxdigit(pg_wchar code, bool posix);
+extern bool pg_u_isdigit(char32_t code, bool posix);
+extern bool pg_u_isalpha(char32_t code);
+extern bool pg_u_isalnum(char32_t code, bool posix);
+extern bool pg_u_isword(char32_t code);
+extern bool pg_u_isupper(char32_t code);
+extern bool pg_u_islower(char32_t code);
+extern bool pg_u_isblank(char32_t code);
+extern bool pg_u_iscntrl(char32_t code);
+extern bool pg_u_isgraph(char32_t code);
+extern bool pg_u_isprint(char32_t code);
+extern bool pg_u_ispunct(char32_t code, bool posix);
+extern bool pg_u_isspace(char32_t code);
+extern bool pg_u_isxdigit(char32_t code, bool posix);
 
 #endif							/* UNICODE_CATEGORY_H */
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 95a1c65da7e..466a41b72b0 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -20,15 +20,15 @@
  */
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 	uint8		category;		/* General Category */
 } pg_category_range;
 
 typedef struct
 {
-	uint32		first;			/* Unicode codepoint */
-	uint32		last;			/* Unicode codepoint */
+	char32_t	first;			/* Unicode codepoint */
+	char32_t	last;			/* Unicode codepoint */
 } pg_unicode_range;
 
 typedef struct
diff --git a/src/include/common/unicode_norm.h b/src/include/common/unicode_norm.h
index 5bc3b79e78e..516c192cc4c 100644
--- a/src/include/common/unicode_norm.h
+++ b/src/include/common/unicode_norm.h
@@ -14,8 +14,6 @@
 #ifndef UNICODE_NORM_H
 #define UNICODE_NORM_H
 
-#include "mb/pg_wchar.h"
-
 typedef enum
 {
 	UNICODE_NFC = 0,
@@ -32,8 +30,8 @@ typedef enum
 	UNICODE_NORM_QC_MAYBE = -1,
 } UnicodeNormalizationQC;
 
-extern pg_wchar *unicode_normalize(UnicodeNormalizationForm form, const pg_wchar *input);
+extern char32_t *unicode_normalize(UnicodeNormalizationForm form, const char32_t *input);
 
-extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const pg_wchar *input);
+extern UnicodeNormalizationQC unicode_is_normalized_quickcheck(UnicodeNormalizationForm form, const char32_t *input);
 
 #endif							/* UNICODE_NORM_H */
diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h
index 4b4a9974b75..4d84bdc81e4 100644
--- a/src/include/mb/pg_wchar.h
+++ b/src/include/mb/pg_wchar.h
@@ -532,25 +532,25 @@ typedef uint32 (*utf_local_conversion_func) (uint32 code);
  * Some handy functions for Unicode-specific tests.
  */
 static inline bool
-is_valid_unicode_codepoint(pg_wchar c)
+is_valid_unicode_codepoint(char32_t c)
 {
 	return (c > 0 && c <= 0x10FFFF);
 }
 
 static inline bool
-is_utf16_surrogate_first(pg_wchar c)
+is_utf16_surrogate_first(char32_t c)
 {
 	return (c >= 0xD800 && c <= 0xDBFF);
 }
 
 static inline bool
-is_utf16_surrogate_second(pg_wchar c)
+is_utf16_surrogate_second(char32_t c)
 {
 	return (c >= 0xDC00 && c <= 0xDFFF);
 }
 
-static inline pg_wchar
-surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
+static inline char32_t
+surrogate_pair_to_codepoint(char16_t first, char16_t second)
 {
 	return ((first & 0x3FF) << 10) + 0x10000 + (second & 0x3FF);
 }
@@ -561,20 +561,20 @@ surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
  *
  * No error checks here, c must point to a long-enough string.
  */
-static inline pg_wchar
+static inline char32_t
 utf8_to_unicode(const unsigned char *c)
 {
 	if ((*c & 0x80) == 0)
-		return (pg_wchar) c[0];
+		return (char32_t) c[0];
 	else if ((*c & 0xe0) == 0xc0)
-		return (pg_wchar) (((c[0] & 0x1f) << 6) |
+		return (char32_t) (((c[0] & 0x1f) << 6) |
 						   (c[1] & 0x3f));
 	else if ((*c & 0xf0) == 0xe0)
-		return (pg_wchar) (((c[0] & 0x0f) << 12) |
+		return (char32_t) (((c[0] & 0x0f) << 12) |
 						   ((c[1] & 0x3f) << 6) |
 						   (c[2] & 0x3f));
 	else if ((*c & 0xf8) == 0xf0)
-		return (pg_wchar) (((c[0] & 0x07) << 18) |
+		return (char32_t) (((c[0] & 0x07) << 18) |
 						   ((c[1] & 0x3f) << 12) |
 						   ((c[2] & 0x3f) << 6) |
 						   (c[3] & 0x3f));
@@ -588,7 +588,7 @@ utf8_to_unicode(const unsigned char *c)
  * unicode_utf8len(c) bytes available.
  */
 static inline unsigned char *
-unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
+unicode_to_utf8(char32_t c, unsigned char *utf8string)
 {
 	if (c <= 0x7F)
 	{
@@ -620,7 +620,7 @@ unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
  * Number of bytes needed to represent the given char in UTF8.
  */
 static inline int
-unicode_utf8len(pg_wchar c)
+unicode_utf8len(char32_t c)
 {
 	if (c <= 0x7F)
 		return 1;
@@ -676,8 +676,8 @@ extern int	pg_valid_server_encoding(const char *name);
 extern bool is_encoding_supported_by_icu(int encoding);
 extern const char *get_encoding_name_for_icu(int encoding);
 
-extern unsigned char *unicode_to_utf8(pg_wchar c, unsigned char *utf8string);
-extern pg_wchar utf8_to_unicode(const unsigned char *c);
+extern unsigned char *unicode_to_utf8(char32_t c, unsigned char *utf8string);
+extern char32_t utf8_to_unicode(const unsigned char *c);
 extern bool pg_utf8_islegal(const unsigned char *source, int length);
 extern int	pg_utf_mblen(const unsigned char *s);
 extern int	pg_mule_mblen(const unsigned char *s);
@@ -739,8 +739,8 @@ extern char *pg_server_to_client(const char *s, int len);
 extern char *pg_any_to_server(const char *s, int len, int encoding);
 extern char *pg_server_to_any(const char *s, int len, int encoding);
 
-extern void pg_unicode_to_server(pg_wchar c, unsigned char *s);
-extern bool pg_unicode_to_server_noerror(pg_wchar c, unsigned char *s);
+extern void pg_unicode_to_server(char32_t c, unsigned char *s);
+extern bool pg_unicode_to_server_noerror(char32_t c, unsigned char *s);
 
 extern unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc);
 extern unsigned short CNStoBIG5(unsigned short cns, unsigned char lc);
diff --git a/src/include/pg_config.h.in b/src/include/pg_config.h.in
index 08d7bfbee10..f52f14cc566 100644
--- a/src/include/pg_config.h.in
+++ b/src/include/pg_config.h.in
@@ -463,6 +463,9 @@
 /* Define to 1 if you have the <termios.h> header file. */
 #undef HAVE_TERMIOS_H
 
+/* Define to 1 if you have the <uchar.h> header file. */
+#undef HAVE_UCHAR_H
+
 /* Define to 1 if curl_global_init() is guaranteed to be thread-safe. */
 #undef HAVE_THREADSAFE_CURL_GLOBAL_INIT
 
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index ac2da4c98cf..df88c78fe3a 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -3505,6 +3505,8 @@ cb_cleanup_dir
 cb_options
 cb_tablespace
 cb_tablespace_mapping
+char16_t
+char32_t
 check_agg_arguments_context
 check_function_callback
 check_network_data
-- 
2.43.0