Commit
1e7fe06c10c0a8da9dd6261a6be8d405dc17c728 changed
pg_mbstrlen_with_len() to ereport(ERROR) if the input ends in an
incomplete character. Most callers want that. text_substring() does
not. It detoasts the most bytes it could possibly need to get the
requested number of characters. For example, to extract up to 2 chars
from UTF8, it needs to detoast 8 bytes. In a string of 3-byte UTF8
chars, 8 bytes spans 2 complete chars and 1 partial char.
Fix this by replacing this pg_mbstrlen_with_len() call with a string
traversal that differs by stopping upon finding as many chars as the
substring could need. This also makes SUBSTRING() stop raising an
encoding error if the incomplete char is past the end of the substring.
This is consistent with the general philosophy of the above commit,
which was to raise errors on a just-in-time basis. Before the above
commit, SUBSTRING() never raised an encoding error.
SUBSTRING() has long been detoasting enough for one more char than
needed, because it did not distinguish exclusive and inclusive end
position. For avoidance of doubt, stop detoasting extra.
Back-patch to v14, like the above commit. For applications using
SUBSTRING() on non-ASCII column values, consider applying this to your
copy of any of the February 12, 2026 releases.
Reported-by: SATŌ Kentarō <ranvis@gmail.com>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Bug: #19406
Discussion: https://postgr.es/m/19406-
9867fddddd724fca@postgresql.org
Backpatch-through: 14
int32 start,
int32 length,
bool length_not_specified);
+static int pg_mbcharcliplen_chars(const char *mbstr, int len, int limit);
static text *text_overlay(text *t1, text *t2, int sp, int sl);
static int text_position(text *t1, text *t2, Oid collid);
static void text_position_setup(text *t1, text *t2, Oid collid, TextPositionState *state);
int32 S = start; /* start position */
int32 S1; /* adjusted start position */
int32 L1; /* adjusted substring length */
- int32 E; /* end position */
+ int32 E; /* end position, exclusive */
/*
* SQL99 says S can be zero or negative (which we don't document), but we
else
{
/*
- * A zero or negative value for the end position can happen if the
- * start was negative or one. SQL99 says to return a zero-length
- * string.
+ * Ending at position 1, exclusive, obviously yields an empty
+ * string. A zero or negative value can happen if the start was
+ * negative or one. SQL99 says to return a zero-length string.
*/
- if (E < 1)
+ if (E <= 1)
return cstring_to_text("");
/*
L1 = E - S1;
/*
- * Total slice size in bytes can't be any longer than the start
- * position plus substring length times the encoding max length.
- * If that overflows, we can just use -1.
+ * Total slice size in bytes can't be any longer than the
+ * inclusive end position times the encoding max length. If that
+ * overflows, we can just use -1.
*/
- if (pg_mul_s32_overflow(E, eml, &slice_size))
+ if (pg_mul_s32_overflow(E - 1, eml, &slice_size))
slice_size = -1;
}
return cstring_to_text("");
}
- /* Now we can get the actual length of the slice in MB characters */
- slice_strlen = pg_mbstrlen_with_len(VARDATA_ANY(slice),
- slice_len);
+ /*
+ * Now we can get the actual length of the slice in MB characters,
+ * stopping at the end of the substring. Continuing beyond the
+ * substring end could find an incomplete character attributable
+ * solely to DatumGetTextPSlice() chopping in the middle of a
+ * character, and it would be superfluous work at best.
+ */
+ slice_strlen =
+ (slice_size == -1 ?
+ pg_mbstrlen_with_len(VARDATA_ANY(slice), slice_len) :
+ pg_mbcharcliplen_chars(VARDATA_ANY(slice), slice_len, E - 1));
/*
* Check that the start position wasn't > slice_strlen. If so, SQL99
return NULL;
}
+/*
+ * pg_mbcharcliplen_chars -
+ * Mirror pg_mbcharcliplen(), except return value unit is chars, not bytes.
+ *
+ * This mirrors all the dubious historical behavior, so it's static to
+ * discourage proliferation. The assertions are specific to the one caller.
+ */
+static int
+pg_mbcharcliplen_chars(const char *mbstr, int len, int limit)
+{
+ int nch = 0;
+ int l;
+
+ Assert(len > 0);
+ Assert(limit > 0);
+ Assert(pg_database_encoding_max_length() > 1);
+
+ while (len > 0 && *mbstr)
+ {
+ l = pg_mblen_with_len(mbstr, len);
+ nch++;
+ if (nch == limit)
+ break;
+ len -= l;
+ mbstr += l;
+ }
+ return nch;
+}
+
/*
* textoverlay
* Replace specified substring of first string with second
-- invalid short mb character = error
SELECT length(truncated) FROM regress_encoding;
ERROR: invalid byte sequence for encoding "UTF8": 0xc3
-SELECT substring(truncated, 1, 1) FROM regress_encoding;
+SELECT substring(truncated, 1, 3) FROM regress_encoding;
+ substring
+-----------
+ caf
+(1 row)
+
+SELECT substring(truncated, 1, 4) FROM regress_encoding;
ERROR: invalid byte sequence for encoding "UTF8": 0xc3
SELECT reverse(truncated) FROM regress_encoding;
ERROR: invalid byte sequence for encoding "UTF8": 0xc3
t
(1 row)
+-- substring fetches a slice of a toasted value; unused tail of that slice is
+-- an incomplete char (bug #19406)
+CREATE TABLE toast_3b_utf8 (c text);
+INSERT INTO toast_3b_utf8 VALUES (repeat(U&'\2026', 4000));
+SELECT SUBSTRING(c FROM 1 FOR 1) FROM toast_3b_utf8;
+ substring
+-----------
+ …
+(1 row)
+
+SELECT SUBSTRING(c FROM 4001 FOR 1) FROM toast_3b_utf8;
+ substring
+-----------
+
+(1 row)
+
+-- diagnose incomplete char iff within the substring
+UPDATE toast_3b_utf8 SET c = c || test_bytea_to_text('\xe280');
+SELECT SUBSTRING(c FROM 4000 FOR 1) FROM toast_3b_utf8;
+ substring
+-----------
+ …
+(1 row)
+
+SELECT SUBSTRING(c FROM 4001 FOR 1) FROM toast_3b_utf8;
+ERROR: invalid byte sequence for encoding "UTF8": 0xe2 0x80
+-- substring needing last byte of its slice_size
+ALTER TABLE toast_3b_utf8 RENAME TO toast_4b_utf8;
+UPDATE toast_4b_utf8 SET c = repeat(U&'\+01F680', 3000);
+SELECT SUBSTRING(c FROM 3000 FOR 1) FROM toast_4b_utf8;
+ substring
+-----------
+ 🚀
+(1 row)
+
DROP TABLE encoding_tests;
+DROP TABLE toast_4b_utf8;
DROP FUNCTION test_encoding;
DROP FUNCTION test_text_to_wchars;
DROP FUNCTION test_mblen_func;
-- invalid short mb character = error
SELECT length(truncated) FROM regress_encoding;
-SELECT substring(truncated, 1, 1) FROM regress_encoding;
+SELECT substring(truncated, 1, 3) FROM regress_encoding;
+SELECT substring(truncated, 1, 4) FROM regress_encoding;
SELECT reverse(truncated) FROM regress_encoding;
-- invalid short mb character = silently dropped
SELECT regexp_replace(truncated, '^caf(.)$', '\1') FROM regress_encoding;
SELECT COUNT(test_encoding(encoding, description, input)) > 0
FROM encoding_tests;
+-- substring fetches a slice of a toasted value; unused tail of that slice is
+-- an incomplete char (bug #19406)
+CREATE TABLE toast_3b_utf8 (c text);
+INSERT INTO toast_3b_utf8 VALUES (repeat(U&'\2026', 4000));
+SELECT SUBSTRING(c FROM 1 FOR 1) FROM toast_3b_utf8;
+SELECT SUBSTRING(c FROM 4001 FOR 1) FROM toast_3b_utf8;
+-- diagnose incomplete char iff within the substring
+UPDATE toast_3b_utf8 SET c = c || test_bytea_to_text('\xe280');
+SELECT SUBSTRING(c FROM 4000 FOR 1) FROM toast_3b_utf8;
+SELECT SUBSTRING(c FROM 4001 FOR 1) FROM toast_3b_utf8;
+-- substring needing last byte of its slice_size
+ALTER TABLE toast_3b_utf8 RENAME TO toast_4b_utf8;
+UPDATE toast_4b_utf8 SET c = repeat(U&'\+01F680', 3000);
+SELECT SUBSTRING(c FROM 3000 FOR 1) FROM toast_4b_utf8;
+
DROP TABLE encoding_tests;
+DROP TABLE toast_4b_utf8;
DROP FUNCTION test_encoding;
DROP FUNCTION test_text_to_wchars;
DROP FUNCTION test_mblen_func;
projects / thirdparty / postgresql.git / commitdiff
