patch: Use pg_mbcliplen for truncation in text-to-name conversion
Started by Karl Schnaitterover 13 years ago2 messages
The text_name function was truncating its input string to the first
NAMEDATALEN-1 bytes, which is wrong if the string has multi-byte
characters. This patch changes it to use pg_mbcliplen, following
the namein function.
Attachments:
text_to_name.patchtext/plain; name=text_to_name.patchDownload
diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c
index a5592d5..02fe9b4 100644
--- a/src/backend/utils/adt/varlena.c
+++ b/src/backend/utils/adt/varlena.c
@@ -2255,8 +2255,7 @@ text_name(PG_FUNCTION_ARGS)
len = VARSIZE_ANY_EXHDR(s);
/* Truncate oversize input */
- if (len >= NAMEDATALEN)
- len = NAMEDATALEN - 1;
+ len = pg_mbcliplen(VARDATA_ANY(s), len, NAMEDATALEN - 1);
result = (Name) palloc(NAMEDATALEN);
memcpy(NameStr(*result), VARDATA_ANY(s), len);
diff --git a/src/test/regress/expected/name.out b/src/test/regress/expected/name.out
index b359d52..f4b58f1 100644
--- a/src/test/regress/expected/name.out
+++ b/src/test/regress/expected/name.out
@@ -15,6 +15,19 @@ SELECT name 'name string' = name 'name string ' AS "False";
f
(1 row)
+-- name truncation with unicode characters
+SELECT length(repeat(U&'\0400', 32)::unknown::name::bytea, 'utf8') as length_from_unknown;
+ length_from_unknown
+---------------------
+ 31
+(1 row)
+
+SELECT length(repeat(U&'\0400', 32)::text::name::bytea, 'utf8') as length_from_text;
+ length_from_text
+------------------
+ 31
+(1 row)
+
--
--
--
diff --git a/src/test/regress/sql/name.sql b/src/test/regress/sql/name.sql
index 1c7a671..9f7a5f0 100644
--- a/src/test/regress/sql/name.sql
+++ b/src/test/regress/sql/name.sql
@@ -8,6 +8,10 @@ SELECT name 'name string' = name 'name string' AS "True";
SELECT name 'name string' = name 'name string ' AS "False";
+-- name truncation with unicode characters
+SELECT length(repeat(U&'\0400', 32)::unknown::name::bytea, 'utf8') as length_from_unknown;
+SELECT length(repeat(U&'\0400', 32)::text::name::bytea, 'utf8') as length_from_text;
+
--
--
--
Re: patch: Use pg_mbcliplen for truncation in text-to-name conversion
Karl Schnaitter <karlsch@gmail.com> writes:
The text_name function was truncating its input string to the first
NAMEDATALEN-1 bytes, which is wrong if the string has multi-byte
characters. This patch changes it to use pg_mbcliplen, following
the namein function.
Good catch, but poking around I note that bpchar_name has the same
disease. Will fix, thanks for the report!
regards, tom lane