From d8288deb6b806deba7fa29f598e9a7e9bbaf759c Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Thu, 5 Feb 2026 01:04:24 +1300 Subject: [PATCH 01/11] Fix encoding length for EUC_CN. While EUC_CN supports only 1- and 2-byte sequences (CS0, CS1), the mb<->wchar conversion functions allow 3-byte sequences beginning SS2, SS3. Change pg_encoding_max_length() to return 3, not 2, to close a hypothesized buffer overrun if a corrupted string is converted to wchar and back again in a newly allocated buffer. We might reconsider that in master (ie harmonizing in a different direction), but this change seems better for the back-branches. Also change pg_euccn_mblen() to report SS2 and SS3 characters as having length 3 (following the example of EUC_KR). Even though such characters would not pass verification, it's remotely possible that invalid bytes could be used to compute a buffer size for use in wchar conversion. Security: CVE-2026-2006 Backpatch-through: 14 Author: Thomas Munro Reviewed-by: Noah Misch Reviewed-by: Heikki Linnakangas (cherry picked from commit b2c81ac8678ce2e570db0843f48f50a971af28c5) --- src/common/wchar.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) diff --git a/src/common/wchar.c b/src/common/wchar.c index 0d9eca700325c..461cd67a42a8c 100644 --- a/src/common/wchar.c +++ b/src/common/wchar.c @@ -267,12 +267,22 @@ pg_euccn2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) return cnt; } +/* + * mbverifychar does not accept SS2 or SS3 (CS2 and CS3 are not defined for + * EUC_CN), but mb2wchar_with_len does. Tell a coherent story for code that + * relies on agreement between mb2wchar_with_len and mblen. Invalid text + * datums (e.g. from shared catalogs) reach this. + */ static int pg_euccn_mblen(const unsigned char *s) { int len; - if (IS_HIGHBIT_SET(*s)) + if (*s == SS2) + len = 3; + else if (*s == SS3) + len = 3; + else if (IS_HIGHBIT_SET(*s)) len = 2; else len = 1; @@ -2125,7 +2135,7 @@ pg_encoding_set_invalid(int encoding, char *dst) const pg_wchar_tbl pg_wchar_table[] = { {pg_ascii2wchar_with_len, pg_wchar2single_with_len, pg_ascii_mblen, pg_ascii_dsplen, pg_ascii_verifychar, pg_ascii_verifystr, 1}, /* PG_SQL_ASCII */ {pg_eucjp2wchar_with_len, pg_wchar2euc_with_len, pg_eucjp_mblen, pg_eucjp_dsplen, pg_eucjp_verifychar, pg_eucjp_verifystr, 3}, /* PG_EUC_JP */ - {pg_euccn2wchar_with_len, pg_wchar2euc_with_len, pg_euccn_mblen, pg_euccn_dsplen, pg_euccn_verifychar, pg_euccn_verifystr, 2}, /* PG_EUC_CN */ + {pg_euccn2wchar_with_len, pg_wchar2euc_with_len, pg_euccn_mblen, pg_euccn_dsplen, pg_euccn_verifychar, pg_euccn_verifystr, 3}, /* PG_EUC_CN */ {pg_euckr2wchar_with_len, pg_wchar2euc_with_len, pg_euckr_mblen, pg_euckr_dsplen, pg_euckr_verifychar, pg_euckr_verifystr, 3}, /* PG_EUC_KR */ {pg_euctw2wchar_with_len, pg_wchar2euc_with_len, pg_euctw_mblen, pg_euctw_dsplen, pg_euctw_verifychar, pg_euctw_verifystr, 4}, /* PG_EUC_TW */ {pg_eucjp2wchar_with_len, pg_wchar2euc_with_len, pg_eucjp_mblen, pg_eucjp_dsplen, pg_eucjp_verifychar, pg_eucjp_verifystr, 3}, /* PG_EUC_JIS_2004 */ From 9a17efceea935a1d0d9e17e77f6fc5aff228e16e Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Mon, 26 Jan 2026 11:22:32 +1300 Subject: [PATCH 02/11] Fix mb2wchar functions on short input. When converting multibyte to pg_wchar, the UTF-8 implementation would silently ignore an incomplete final character, while the other implementations would cast a single byte to pg_wchar, and then repeat for the remaining byte sequence. While it didn't overrun the buffer, it was surely garbage output. Make all encodings behave like the UTF-8 implementation. A later change for master only will convert this to an error, but we choose not to back-patch that behavior change on the off-chance that someone is relying on the existing UTF-8 behavior. Security: CVE-2026-2006 Backpatch-through: 14 Author: Thomas Munro Reported-by: Noah Misch Reviewed-by: Noah Misch Reviewed-by: Heikki Linnakangas (cherry picked from commit 50863be0b77eeac5c1907ccd46c146eb80524e1a) --- src/common/wchar.c | 52 ++++++++++++++++++++++++++++------------------ 1 file changed, 32 insertions(+), 20 deletions(-) diff --git a/src/common/wchar.c b/src/common/wchar.c index 461cd67a42a8c..d931f2bd5345a 100644 --- a/src/common/wchar.c +++ b/src/common/wchar.c @@ -63,6 +63,9 @@ * subset to the ASCII routines to ensure consistency. */ +/* No error-reporting facility. Ignore incomplete trailing byte sequence. */ +#define MB2CHAR_NEED_AT_LEAST(len, need) if ((len) < (need)) break + /* * SQL/ASCII */ @@ -108,22 +111,24 @@ pg_euc2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) while (len > 0 && *from) { - if (*from == SS2 && len >= 2) /* JIS X 0201 (so called "1 byte - * KANA") */ + if (*from == SS2) /* JIS X 0201 (so called "1 byte KANA") */ { + MB2CHAR_NEED_AT_LEAST(len, 2); from++; *to = (SS2 << 8) | *from++; len -= 2; } - else if (*from == SS3 && len >= 3) /* JIS X 0212 KANJI */ + else if (*from == SS3) /* JIS X 0212 KANJI */ { + MB2CHAR_NEED_AT_LEAST(len, 3); from++; *to = (SS3 << 16) | (*from++ << 8); *to |= *from++; len -= 3; } - else if (IS_HIGHBIT_SET(*from) && len >= 2) /* JIS X 0208 KANJI */ + else if (IS_HIGHBIT_SET(*from)) /* JIS X 0208 KANJI */ { + MB2CHAR_NEED_AT_LEAST(len, 2); *to = *from++ << 8; *to |= *from++; len -= 2; @@ -235,22 +240,25 @@ pg_euccn2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) while (len > 0 && *from) { - if (*from == SS2 && len >= 3) /* code set 2 (unused?) */ + if (*from == SS2) /* code set 2 (unused?) */ { + MB2CHAR_NEED_AT_LEAST(len, 3); from++; *to = (SS2 << 16) | (*from++ << 8); *to |= *from++; len -= 3; } - else if (*from == SS3 && len >= 3) /* code set 3 (unused ?) */ + else if (*from == SS3) /* code set 3 (unused ?) */ { + MB2CHAR_NEED_AT_LEAST(len, 3); from++; *to = (SS3 << 16) | (*from++ << 8); *to |= *from++; len -= 3; } - else if (IS_HIGHBIT_SET(*from) && len >= 2) /* code set 1 */ + else if (IS_HIGHBIT_SET(*from)) /* code set 1 */ { + MB2CHAR_NEED_AT_LEAST(len, 2); *to = *from++ << 8; *to |= *from++; len -= 2; @@ -312,23 +320,26 @@ pg_euctw2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) while (len > 0 && *from) { - if (*from == SS2 && len >= 4) /* code set 2 */ + if (*from == SS2) /* code set 2 */ { + MB2CHAR_NEED_AT_LEAST(len, 4); from++; *to = (((uint32) SS2) << 24) | (*from++ << 16); *to |= *from++ << 8; *to |= *from++; len -= 4; } - else if (*from == SS3 && len >= 3) /* code set 3 (unused?) */ + else if (*from == SS3) /* code set 3 (unused?) */ { + MB2CHAR_NEED_AT_LEAST(len, 3); from++; *to = (SS3 << 16) | (*from++ << 8); *to |= *from++; len -= 3; } - else if (IS_HIGHBIT_SET(*from) && len >= 2) /* code set 2 */ + else if (IS_HIGHBIT_SET(*from)) /* code set 2 */ { + MB2CHAR_NEED_AT_LEAST(len, 2); *to = *from++ << 8; *to |= *from++; len -= 2; @@ -465,8 +476,7 @@ pg_utf2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) } else if ((*from & 0xe0) == 0xc0) { - if (len < 2) - break; /* drop trailing incomplete char */ + MB2CHAR_NEED_AT_LEAST(len, 2); c1 = *from++ & 0x1f; c2 = *from++ & 0x3f; *to = (c1 << 6) | c2; @@ -474,8 +484,7 @@ pg_utf2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) } else if ((*from & 0xf0) == 0xe0) { - if (len < 3) - break; /* drop trailing incomplete char */ + MB2CHAR_NEED_AT_LEAST(len, 3); c1 = *from++ & 0x0f; c2 = *from++ & 0x3f; c3 = *from++ & 0x3f; @@ -484,8 +493,7 @@ pg_utf2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) } else if ((*from & 0xf8) == 0xf0) { - if (len < 4) - break; /* drop trailing incomplete char */ + MB2CHAR_NEED_AT_LEAST(len, 4); c1 = *from++ & 0x07; c2 = *from++ & 0x3f; c3 = *from++ & 0x3f; @@ -748,28 +756,32 @@ pg_mule2wchar_with_len(const unsigned char *from, pg_wchar *to, int len) while (len > 0 && *from) { - if (IS_LC1(*from) && len >= 2) + if (IS_LC1(*from)) { + MB2CHAR_NEED_AT_LEAST(len, 2); *to = *from++ << 16; *to |= *from++; len -= 2; } - else if (IS_LCPRV1(*from) && len >= 3) + else if (IS_LCPRV1(*from)) { + MB2CHAR_NEED_AT_LEAST(len, 3); from++; *to = *from++ << 16; *to |= *from++; len -= 3; } - else if (IS_LC2(*from) && len >= 3) + else if (IS_LC2(*from)) { + MB2CHAR_NEED_AT_LEAST(len, 3); *to = *from++ << 16; *to |= *from++ << 8; *to |= *from++; len -= 3; } - else if (IS_LCPRV2(*from) && len >= 4) + else if (IS_LCPRV2(*from)) { + MB2CHAR_NEED_AT_LEAST(len, 4); from++; *to = *from++ << 16; *to |= *from++ << 8; From 3d47e2e5c5037eea919a85115fee0254dbcdbdc9 Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Wed, 7 Jan 2026 22:14:31 +1300 Subject: [PATCH 03/11] Replace pg_mblen() with bounds-checked versions. A corrupted string could cause code that iterates with pg_mblen() to overrun its buffer. Fix, by converting all callers to one of the following: 1. Callers with a null-terminated string now use pg_mblen_cstr(), which raises an "illegal byte sequence" error if it finds a terminator in the middle of the sequence. 2. Callers with a length or end pointer now use either pg_mblen_with_len() or pg_mblen_range(), for the same effect, depending on which of the two seems more convenient at each site. 3. A small number of cases pre-validate a string, and can use pg_mblen_unbounded(). The traditional pg_mblen() function and COPYCHAR macro still exist for backward compatibility, but are no longer used by core code and are hereby deprecated. The same applies to the t_isXXX() functions. Security: CVE-2026-2006 Backpatch-through: 14 Co-authored-by: Thomas Munro Co-authored-by: Noah Misch Reviewed-by: Heikki Linnakangas Reported-by: Paul Gerste (as part of zeroday.cloud) Reported-by: Moritz Sanft (as part of zeroday.cloud) (cherry picked from commit fd82ddb679d71c976e693f6e922cb5a599755f18) --- contrib/btree_gist/btree_utils_var.c | 21 +++- contrib/dict_xsyn/dict_xsyn.c | 8 +- contrib/hstore/hstore_io.c | 8 +- contrib/ltree/lquery_op.c | 4 +- contrib/ltree/ltree.h | 3 +- contrib/ltree/ltree_io.c | 16 +-- contrib/ltree/ltxtquery_io.c | 4 +- contrib/pageinspect/heapfuncs.c | 2 +- contrib/pg_trgm/trgm.h | 4 +- contrib/pg_trgm/trgm_op.c | 48 +++++--- contrib/pg_trgm/trgm_regexp.c | 21 ++-- contrib/unaccent/unaccent.c | 7 +- src/backend/catalog/pg_proc.c | 2 +- src/backend/tsearch/dict_synonym.c | 8 +- src/backend/tsearch/dict_thesaurus.c | 18 +-- src/backend/tsearch/regis.c | 37 +++--- src/backend/tsearch/spell.c | 123 +++++++++---------- src/backend/tsearch/ts_locale.c | 93 ++++++-------- src/backend/tsearch/ts_utils.c | 4 +- src/backend/tsearch/wparser_def.c | 3 +- src/backend/utils/adt/encode.c | 10 +- src/backend/utils/adt/formatting.c | 22 ++-- src/backend/utils/adt/jsonfuncs.c | 2 +- src/backend/utils/adt/jsonpath_gram.y | 3 +- src/backend/utils/adt/levenshtein.c | 14 ++- src/backend/utils/adt/like.c | 18 +-- src/backend/utils/adt/like_match.c | 3 +- src/backend/utils/adt/oracle_compat.c | 33 +++-- src/backend/utils/adt/regexp.c | 9 +- src/backend/utils/adt/tsquery.c | 25 ++-- src/backend/utils/adt/tsvector.c | 11 +- src/backend/utils/adt/tsvector_op.c | 10 +- src/backend/utils/adt/tsvector_parser.c | 29 ++--- src/backend/utils/adt/varbit.c | 8 +- src/backend/utils/adt/varlena.c | 34 +++-- src/backend/utils/adt/xml.c | 11 +- src/backend/utils/mb/mbutils.c | 150 +++++++++++++++++++++-- src/include/mb/pg_wchar.h | 7 ++ src/include/tsearch/ts_locale.h | 34 ++++- src/include/tsearch/ts_utils.h | 14 +-- src/test/modules/test_regex/test_regex.c | 3 +- 41 files changed, 536 insertions(+), 348 deletions(-) diff --git a/contrib/btree_gist/btree_utils_var.c b/contrib/btree_gist/btree_utils_var.c index 2886c08b85e40..9d93b3c775e56 100644 --- a/contrib/btree_gist/btree_utils_var.c +++ b/contrib/btree_gist/btree_utils_var.c @@ -116,36 +116,47 @@ gbt_var_leaf2node(GBT_VARKEY *leaf, const gbtree_vinfo *tinfo, FmgrInfo *flinfo) /* * returns the common prefix length of a node key + * + * If the underlying type is character data, the prefix length may point in + * the middle of a multibyte character. */ static int32 gbt_var_node_cp_len(const GBT_VARKEY *node, const gbtree_vinfo *tinfo) { GBT_VARKEY_R r = gbt_var_key_readable(node); int32 i = 0; - int32 l = 0; + int32 l_left_to_match = 0; + int32 l_total = 0; int32 t1len = VARSIZE(r.lower) - VARHDRSZ; int32 t2len = VARSIZE(r.upper) - VARHDRSZ; int32 ml = Min(t1len, t2len); char *p1 = VARDATA(r.lower); char *p2 = VARDATA(r.upper); + const char *end1 = p1 + t1len; + const char *end2 = p2 + t2len; if (ml == 0) return 0; while (i < ml) { - if (tinfo->eml > 1 && l == 0) + if (tinfo->eml > 1 && l_left_to_match == 0) { - if ((l = pg_mblen(p1)) != pg_mblen(p2)) + l_total = pg_mblen_range(p1, end1); + if (l_total != pg_mblen_range(p2, end2)) { return i; } + l_left_to_match = l_total; } if (*p1 != *p2) { if (tinfo->eml > 1) { - return (i - l + 1); + int32 l_matched_subset = l_total - l_left_to_match; + + /* end common prefix at final byte of last matching char */ + return i - l_matched_subset; } else { @@ -155,7 +166,7 @@ gbt_var_node_cp_len(const GBT_VARKEY *node, const gbtree_vinfo *tinfo) p1++; p2++; - l--; + l_left_to_match--; i++; } return ml; /* lower == upper */ diff --git a/contrib/dict_xsyn/dict_xsyn.c b/contrib/dict_xsyn/dict_xsyn.c index 584fe44753dbc..a4ced84dd8d9e 100644 --- a/contrib/dict_xsyn/dict_xsyn.c +++ b/contrib/dict_xsyn/dict_xsyn.c @@ -48,15 +48,15 @@ find_word(char *in, char **end) char *start; *end = NULL; - while (*in && t_isspace(in)) - in += pg_mblen(in); + while (*in && t_isspace_cstr(in)) + in += pg_mblen_cstr(in); if (!*in || *in == '#') return NULL; start = in; - while (*in && !t_isspace(in)) - in += pg_mblen(in); + while (*in && !t_isspace_cstr(in)) + in += pg_mblen_cstr(in); *end = in; diff --git a/contrib/hstore/hstore_io.c b/contrib/hstore/hstore_io.c index 03057f085d1db..0b1e0581e84d0 100644 --- a/contrib/hstore/hstore_io.c +++ b/contrib/hstore/hstore_io.c @@ -82,7 +82,7 @@ get_val(HSParser *state, bool ignoreeq, bool *escaped) else if (*(state->ptr) == '=' && !ignoreeq) { elog(ERROR, "Syntax error near \"%.*s\" at position %d", - pg_mblen(state->ptr), state->ptr, + pg_mblen_cstr(state->ptr), state->ptr, (int32) (state->ptr - state->begin)); } else if (*(state->ptr) == '\\') @@ -223,7 +223,7 @@ parse_hstore(HSParser *state) else if (!scanner_isspace((unsigned char) *(state->ptr))) { elog(ERROR, "Syntax error near \"%.*s\" at position %d", - pg_mblen(state->ptr), state->ptr, + pg_mblen_cstr(state->ptr), state->ptr, (int32) (state->ptr - state->begin)); } } @@ -240,7 +240,7 @@ parse_hstore(HSParser *state) else { elog(ERROR, "Syntax error near \"%.*s\" at position %d", - pg_mblen(state->ptr), state->ptr, + pg_mblen_cstr(state->ptr), state->ptr, (int32) (state->ptr - state->begin)); } } @@ -275,7 +275,7 @@ parse_hstore(HSParser *state) else if (!scanner_isspace((unsigned char) *(state->ptr))) { elog(ERROR, "Syntax error near \"%.*s\" at position %d", - pg_mblen(state->ptr), state->ptr, + pg_mblen_cstr(state->ptr), state->ptr, (int32) (state->ptr - state->begin)); } } diff --git a/contrib/ltree/lquery_op.c b/contrib/ltree/lquery_op.c index ef86046fc4bc0..abccbb8927ff0 100644 --- a/contrib/ltree/lquery_op.c +++ b/contrib/ltree/lquery_op.c @@ -26,14 +26,14 @@ getlexeme(char *start, char *end, int *len) char *ptr; int charlen; - while (start < end && (charlen = pg_mblen(start)) == 1 && t_iseq(start, '_')) + while (start < end && (charlen = pg_mblen_range(start, end)) == 1 && t_iseq(start, '_')) start += charlen; ptr = start; if (ptr >= end) return NULL; - while (ptr < end && !((charlen = pg_mblen(ptr)) == 1 && t_iseq(ptr, '_'))) + while (ptr < end && !((charlen = pg_mblen_range(ptr, end)) == 1 && t_iseq(ptr, '_'))) ptr += charlen; *len = ptr - start; diff --git a/contrib/ltree/ltree.h b/contrib/ltree/ltree.h index 564e4fa81b87b..204927d689091 100644 --- a/contrib/ltree/ltree.h +++ b/contrib/ltree/ltree.h @@ -126,7 +126,8 @@ typedef struct #define LQUERY_HASNOT 0x01 -#define ISALNUM(x) ( t_isalpha(x) || t_isdigit(x) || ( pg_mblen(x) == 1 && t_iseq((x), '_') ) ) +/* Caller has already called mblen, so we can use _unbounded variants safely. */ +#define ISALNUM(x) ( t_isalpha_unbounded(x) || t_isdigit_unbounded(x) || ( pg_mblen_unbounded(x) == 1 && t_iseq((x), '_') ) ) /* full text query */ diff --git a/contrib/ltree/ltree_io.c b/contrib/ltree/ltree_io.c index 903115829ef08..385db479540ff 100644 --- a/contrib/ltree/ltree_io.c +++ b/contrib/ltree/ltree_io.c @@ -55,7 +55,7 @@ parse_ltree(const char *buf) ptr = buf; while (*ptr) { - charlen = pg_mblen(ptr); + charlen = pg_mblen_cstr(ptr); if (t_iseq(ptr, '.')) num++; ptr += charlen; @@ -70,7 +70,7 @@ parse_ltree(const char *buf) ptr = buf; while (*ptr) { - charlen = pg_mblen(ptr); + charlen = pg_mblen_cstr(ptr); switch (state) { @@ -286,7 +286,7 @@ parse_lquery(const char *buf) ptr = buf; while (*ptr) { - charlen = pg_mblen(ptr); + charlen = pg_mblen_cstr(ptr); if (t_iseq(ptr, '.')) num++; @@ -306,7 +306,7 @@ parse_lquery(const char *buf) ptr = buf; while (*ptr) { - charlen = pg_mblen(ptr); + charlen = pg_mblen_cstr(ptr); switch (state) { @@ -408,7 +408,7 @@ parse_lquery(const char *buf) case LQPRS_WAITFNUM: if (t_iseq(ptr, ',')) state = LQPRS_WAITSNUM; - else if (t_isdigit(ptr)) + else if (t_isdigit_cstr(ptr)) { int low = atoi(ptr); @@ -426,7 +426,7 @@ parse_lquery(const char *buf) UNCHAR; break; case LQPRS_WAITSNUM: - if (t_isdigit(ptr)) + if (t_isdigit_cstr(ptr)) { int high = atoi(ptr); @@ -457,7 +457,7 @@ parse_lquery(const char *buf) case LQPRS_WAITCLOSE: if (t_iseq(ptr, '}')) state = LQPRS_WAITEND; - else if (!t_isdigit(ptr)) + else if (!t_isdigit_cstr(ptr)) UNCHAR; break; case LQPRS_WAITND: @@ -468,7 +468,7 @@ parse_lquery(const char *buf) } else if (t_iseq(ptr, ',')) state = LQPRS_WAITSNUM; - else if (!t_isdigit(ptr)) + else if (!t_isdigit_cstr(ptr)) UNCHAR; break; case LQPRS_WAITEND: diff --git a/contrib/ltree/ltxtquery_io.c b/contrib/ltree/ltxtquery_io.c index 49b4714ab843e..a4e2e2b33adf2 100644 --- a/contrib/ltree/ltxtquery_io.c +++ b/contrib/ltree/ltxtquery_io.c @@ -59,7 +59,7 @@ gettoken_query(QPRS_STATE *state, int32 *val, int32 *lenval, char **strval, uint for (;;) { - charlen = pg_mblen(state->buf); + charlen = pg_mblen_cstr(state->buf); switch (state->state) { @@ -83,7 +83,7 @@ gettoken_query(QPRS_STATE *state, int32 *val, int32 *lenval, char **strval, uint *lenval = charlen; *flag = 0; } - else if (!t_isspace(state->buf)) + else if (!t_isspace_unbounded(state->buf)) ereport(ERROR, (errcode(ERRCODE_SYNTAX_ERROR), errmsg("operand syntax error"))); diff --git a/contrib/pageinspect/heapfuncs.c b/contrib/pageinspect/heapfuncs.c index 92a17900d61e5..6c3864af9e95e 100644 --- a/contrib/pageinspect/heapfuncs.c +++ b/contrib/pageinspect/heapfuncs.c @@ -101,7 +101,7 @@ text_to_bits(char *str, int len) ereport(ERROR, (errcode(ERRCODE_DATA_CORRUPTED), errmsg("invalid character \"%.*s\" in t_bits string", - pg_mblen(str + off), str + off))); + pg_mblen_cstr(str + off), str + off))); if (off % 8 == 7) bits[off / 8] = byte; diff --git a/contrib/pg_trgm/trgm.h b/contrib/pg_trgm/trgm.h index 405a1d95528da..06d3994e69277 100644 --- a/contrib/pg_trgm/trgm.h +++ b/contrib/pg_trgm/trgm.h @@ -52,10 +52,10 @@ typedef char trgm[3]; } while(0) #ifdef KEEPONLYALNUM -#define ISWORDCHR(c) (t_isalpha(c) || t_isdigit(c)) +#define ISWORDCHR(c, len) (t_isalpha_with_len(c, len) || t_isdigit_with_len(c, len)) #define ISPRINTABLECHAR(a) ( isascii( *(unsigned char*)(a) ) && (isalnum( *(unsigned char*)(a) ) || *(unsigned char*)(a)==' ') ) #else -#define ISWORDCHR(c) (!t_isspace(c)) +#define ISWORDCHR(c, len) (!t_isspace_with_len(c, len)) #define ISPRINTABLECHAR(a) ( isascii( *(unsigned char*)(a) ) && isprint( *(unsigned char*)(a) ) ) #endif #define ISPRINTABLETRGM(t) ( ISPRINTABLECHAR( ((char*)(t)) ) && ISPRINTABLECHAR( ((char*)(t))+1 ) && ISPRINTABLECHAR( ((char*)(t))+2 ) ) diff --git a/contrib/pg_trgm/trgm_op.c b/contrib/pg_trgm/trgm_op.c index e9b7981619fb6..fe7f1ca4412d1 100644 --- a/contrib/pg_trgm/trgm_op.c +++ b/contrib/pg_trgm/trgm_op.c @@ -173,18 +173,29 @@ static char * find_word(char *str, int lenstr, char **endword, int *charlen) { char *beginword = str; + const char *endstr = str + lenstr; - while (beginword - str < lenstr && !ISWORDCHR(beginword)) - beginword += pg_mblen(beginword); + while (beginword < endstr) + { + int clen = pg_mblen_range(beginword, endstr); - if (beginword - str >= lenstr) + if (ISWORDCHR(beginword, clen)) + break; + beginword += clen; + } + + if (beginword >= endstr) return NULL; *endword = beginword; *charlen = 0; - while (*endword - str < lenstr && ISWORDCHR(*endword)) + while (*endword < endstr) { - *endword += pg_mblen(*endword); + int clen = pg_mblen_range(*endword, endstr); + + if (!ISWORDCHR(*endword, clen)) + break; + *endword += clen; (*charlen)++; } @@ -232,9 +243,9 @@ make_trigrams(trgm *tptr, char *str, int bytelen, int charlen) if (bytelen > charlen) { /* Find multibyte character boundaries and apply compact_trigram */ - int lenfirst = pg_mblen(str), - lenmiddle = pg_mblen(str + lenfirst), - lenlast = pg_mblen(str + lenfirst + lenmiddle); + int lenfirst = pg_mblen_unbounded(str), + lenmiddle = pg_mblen_unbounded(str + lenfirst), + lenlast = pg_mblen_unbounded(str + lenfirst + lenmiddle); while ((ptr - str) + lenfirst + lenmiddle + lenlast <= bytelen) { @@ -245,7 +256,7 @@ make_trigrams(trgm *tptr, char *str, int bytelen, int charlen) lenfirst = lenmiddle; lenmiddle = lenlast; - lenlast = pg_mblen(ptr + lenfirst + lenmiddle); + lenlast = pg_mblen_unbounded(ptr + lenfirst + lenmiddle); } } else @@ -725,6 +736,7 @@ get_wildcard_part(const char *str, int lenstr, { const char *beginword = str; const char *endword; + const char *endstr = str + lenstr; char *s = buf; bool in_leading_wildcard_meta = false; bool in_trailing_wildcard_meta = false; @@ -737,11 +749,13 @@ get_wildcard_part(const char *str, int lenstr, * from this loop to the next one, since we may exit at a word character * that is in_escape. */ - while (beginword - str < lenstr) + while (beginword < endstr) { + clen = pg_mblen_range(beginword, endstr); + if (in_escape) { - if (ISWORDCHR(beginword)) + if (ISWORDCHR(beginword, clen)) break; in_escape = false; in_leading_wildcard_meta = false; @@ -752,12 +766,12 @@ get_wildcard_part(const char *str, int lenstr, in_escape = true; else if (ISWILDCARDCHAR(beginword)) in_leading_wildcard_meta = true; - else if (ISWORDCHR(beginword)) + else if (ISWORDCHR(beginword, clen)) break; else in_leading_wildcard_meta = false; } - beginword += pg_mblen(beginword); + beginword += clen; } /* @@ -790,12 +804,12 @@ get_wildcard_part(const char *str, int lenstr, * string boundary. Strip escapes during copy. */ endword = beginword; - while (endword - str < lenstr) + while (endword < endstr) { - clen = pg_mblen(endword); + clen = pg_mblen_range(endword, endstr); if (in_escape) { - if (ISWORDCHR(endword)) + if (ISWORDCHR(endword, clen)) { memcpy(s, endword, clen); (*charlen)++; @@ -823,7 +837,7 @@ get_wildcard_part(const char *str, int lenstr, in_trailing_wildcard_meta = true; break; } - else if (ISWORDCHR(endword)) + else if (ISWORDCHR(endword, clen)) { memcpy(s, endword, clen); (*charlen)++; diff --git a/contrib/pg_trgm/trgm_regexp.c b/contrib/pg_trgm/trgm_regexp.c index 3fc8a9ec6f0d6..f675bd0643743 100644 --- a/contrib/pg_trgm/trgm_regexp.c +++ b/contrib/pg_trgm/trgm_regexp.c @@ -480,7 +480,7 @@ static TRGM *createTrgmNFAInternal(regex_t *regex, TrgmPackedGraph **graph, static void RE_compile(regex_t *regex, text *text_re, int cflags, Oid collation); static void getColorInfo(regex_t *regex, TrgmNFA *trgmNFA); -static bool convertPgWchar(pg_wchar c, trgm_mb_char *result); +static int convertPgWchar(pg_wchar c, trgm_mb_char *result); static void transformGraph(TrgmNFA *trgmNFA); static void processState(TrgmNFA *trgmNFA, TrgmState *state); static void addKey(TrgmNFA *trgmNFA, TrgmState *state, TrgmStateKey *key); @@ -818,10 +818,11 @@ getColorInfo(regex_t *regex, TrgmNFA *trgmNFA) for (j = 0; j < charsCount; j++) { trgm_mb_char c; + int clen = convertPgWchar(chars[j], &c); - if (!convertPgWchar(chars[j], &c)) + if (!clen) continue; /* ok to ignore it altogether */ - if (ISWORDCHR(c.bytes)) + if (ISWORDCHR(c.bytes, clen)) colorInfo->wordChars[colorInfo->wordCharsCount++] = c; else colorInfo->containsNonWord = true; @@ -833,13 +834,15 @@ getColorInfo(regex_t *regex, TrgmNFA *trgmNFA) /* * Convert pg_wchar to multibyte format. - * Returns false if the character should be ignored completely. + * Returns 0 if the character should be ignored completely, else returns its + * byte length. */ -static bool +static int convertPgWchar(pg_wchar c, trgm_mb_char *result) { /* "s" has enough space for a multibyte character and a trailing NUL */ char s[MAX_MULTIBYTE_CHAR_LEN + 1]; + int clen; /* * We can ignore the NUL character, since it can never appear in a PG text @@ -847,11 +850,11 @@ convertPgWchar(pg_wchar c, trgm_mb_char *result) * reconstructing trigrams. */ if (c == 0) - return false; + return 0; /* Do the conversion, making sure the result is NUL-terminated */ memset(s, 0, sizeof(s)); - pg_wchar2mb_with_len(&c, s, 1); + clen = pg_wchar2mb_with_len(&c, s, 1); /* * In IGNORECASE mode, we can ignore uppercase characters. We assume that @@ -873,7 +876,7 @@ convertPgWchar(pg_wchar c, trgm_mb_char *result) if (strcmp(lowerCased, s) != 0) { pfree(lowerCased); - return false; + return 0; } pfree(lowerCased); } @@ -881,7 +884,7 @@ convertPgWchar(pg_wchar c, trgm_mb_char *result) /* Fill result with exactly MAX_MULTIBYTE_CHAR_LEN bytes */ memcpy(result->bytes, s, MAX_MULTIBYTE_CHAR_LEN); - return true; + return clen; } diff --git a/contrib/unaccent/unaccent.c b/contrib/unaccent/unaccent.c index 77ecd76528289..de20ac1ae8b47 100644 --- a/contrib/unaccent/unaccent.c +++ b/contrib/unaccent/unaccent.c @@ -149,9 +149,9 @@ initTrie(const char *filename) state = 0; for (ptr = line; *ptr; ptr += ptrlen) { - ptrlen = pg_mblen(ptr); + ptrlen = pg_mblen_cstr(ptr); /* ignore whitespace, but end src or trg */ - if (t_isspace(ptr)) + if (t_isspace_cstr(ptr)) { if (state == 1) state = 2; @@ -315,6 +315,7 @@ unaccent_lexize(PG_FUNCTION_ARGS) char *srcchar = (char *) PG_GETARG_POINTER(1); int32 len = PG_GETARG_INT32(2); char *srcstart = srcchar; + const char *srcend = srcstart + len; TSLexeme *res; StringInfoData buf; @@ -342,7 +343,7 @@ unaccent_lexize(PG_FUNCTION_ARGS) } else { - matchlen = pg_mblen(srcchar); + matchlen = pg_mblen_range(srcchar, srcend); if (buf.data != NULL) appendBinaryStringInfo(&buf, srcchar, matchlen); } diff --git a/src/backend/catalog/pg_proc.c b/src/backend/catalog/pg_proc.c index 5fd8a385a570b..1acd28f7b8b84 100644 --- a/src/backend/catalog/pg_proc.c +++ b/src/backend/catalog/pg_proc.c @@ -1175,7 +1175,7 @@ match_prosrc_to_literal(const char *prosrc, const char *literal, if (cursorpos > 0) newcp++; } - chlen = pg_mblen(prosrc); + chlen = pg_mblen_cstr(prosrc); if (strncmp(prosrc, literal, chlen) != 0) goto fail; prosrc += chlen; diff --git a/src/backend/tsearch/dict_synonym.c b/src/backend/tsearch/dict_synonym.c index 65e34e91c97b5..b7a8109143b21 100644 --- a/src/backend/tsearch/dict_synonym.c +++ b/src/backend/tsearch/dict_synonym.c @@ -47,8 +47,8 @@ findwrd(char *in, char **end, uint16 *flags) char *lastchar; /* Skip leading spaces */ - while (*in && t_isspace(in)) - in += pg_mblen(in); + while (*in && t_isspace_cstr(in)) + in += pg_mblen_cstr(in); /* Return NULL on empty lines */ if (*in == '\0') @@ -60,10 +60,10 @@ findwrd(char *in, char **end, uint16 *flags) lastchar = start = in; /* Find end of word */ - while (*in && !t_isspace(in)) + while (*in && !t_isspace_cstr(in)) { lastchar = in; - in += pg_mblen(in); + in += pg_mblen_cstr(in); } if (in - lastchar == 1 && t_iseq(lastchar, '*') && flags) diff --git a/src/backend/tsearch/dict_thesaurus.c b/src/backend/tsearch/dict_thesaurus.c index b8c08bcf7ba50..cc6b3829c9032 100644 --- a/src/backend/tsearch/dict_thesaurus.c +++ b/src/backend/tsearch/dict_thesaurus.c @@ -190,8 +190,8 @@ thesaurusRead(const char *filename, DictThesaurus *d) ptr = line; /* is it a comment? */ - while (*ptr && t_isspace(ptr)) - ptr += pg_mblen(ptr); + while (*ptr && t_isspace_cstr(ptr)) + ptr += pg_mblen_cstr(ptr); if (t_iseq(ptr, '#') || *ptr == '\0' || t_iseq(ptr, '\n') || t_iseq(ptr, '\r')) @@ -212,7 +212,7 @@ thesaurusRead(const char *filename, DictThesaurus *d) errmsg("unexpected delimiter"))); state = TR_WAITSUBS; } - else if (!t_isspace(ptr)) + else if (!t_isspace_cstr(ptr)) { beginwrd = ptr; state = TR_INLEX; @@ -225,7 +225,7 @@ thesaurusRead(const char *filename, DictThesaurus *d) newLexeme(d, beginwrd, ptr, idsubst, posinsubst++); state = TR_WAITSUBS; } - else if (t_isspace(ptr)) + else if (t_isspace_cstr(ptr)) { newLexeme(d, beginwrd, ptr, idsubst, posinsubst++); state = TR_WAITLEX; @@ -237,15 +237,15 @@ thesaurusRead(const char *filename, DictThesaurus *d) { useasis = true; state = TR_INSUBS; - beginwrd = ptr + pg_mblen(ptr); + beginwrd = ptr + pg_mblen_cstr(ptr); } else if (t_iseq(ptr, '\\')) { useasis = false; state = TR_INSUBS; - beginwrd = ptr + pg_mblen(ptr); + beginwrd = ptr + pg_mblen_cstr(ptr); } - else if (!t_isspace(ptr)) + else if (!t_isspace_cstr(ptr)) { useasis = false; beginwrd = ptr; @@ -254,7 +254,7 @@ thesaurusRead(const char *filename, DictThesaurus *d) } else if (state == TR_INSUBS) { - if (t_isspace(ptr)) + if (t_isspace_cstr(ptr)) { if (ptr == beginwrd) ereport(ERROR, @@ -267,7 +267,7 @@ thesaurusRead(const char *filename, DictThesaurus *d) else elog(ERROR, "unrecognized thesaurus state: %d", state); - ptr += pg_mblen(ptr); + ptr += pg_mblen_cstr(ptr); } if (state == TR_INSUBS) diff --git a/src/backend/tsearch/regis.c b/src/backend/tsearch/regis.c index 43cab72f472ec..1db03ab0b8260 100644 --- a/src/backend/tsearch/regis.c +++ b/src/backend/tsearch/regis.c @@ -37,7 +37,7 @@ RS_isRegis(const char *str) { if (state == RS_IN_WAIT) { - if (t_isalpha(c)) + if (t_isalpha_cstr(c)) /* okay */ ; else if (t_iseq(c, '[')) state = RS_IN_ONEOF; @@ -48,14 +48,14 @@ RS_isRegis(const char *str) { if (t_iseq(c, '^')) state = RS_IN_NONEOF; - else if (t_isalpha(c)) + else if (t_isalpha_cstr(c)) state = RS_IN_ONEOF_IN; else return false; } else if (state == RS_IN_ONEOF_IN || state == RS_IN_NONEOF) { - if (t_isalpha(c)) + if (t_isalpha_cstr(c)) /* okay */ ; else if (t_iseq(c, ']')) state = RS_IN_WAIT; @@ -64,7 +64,7 @@ RS_isRegis(const char *str) } else elog(ERROR, "internal error in RS_isRegis: state %d", state); - c += pg_mblen(c); + c += pg_mblen_cstr(c); } return (state == RS_IN_WAIT); @@ -96,15 +96,14 @@ RS_compile(Regis *r, bool issuffix, const char *str) { if (state == RS_IN_WAIT) { - if (t_isalpha(c)) + if (t_isalpha_cstr(c)) { if (ptr) ptr = newRegisNode(ptr, len); else ptr = r->node = newRegisNode(NULL, len); - COPYCHAR(ptr->data, c); ptr->type = RSF_ONEOF; - ptr->len = pg_mblen(c); + ptr->len = ts_copychar_cstr(ptr->data, c); } else if (t_iseq(c, '[')) { @@ -125,10 +124,9 @@ RS_compile(Regis *r, bool issuffix, const char *str) ptr->type = RSF_NONEOF; state = RS_IN_NONEOF; } - else if (t_isalpha(c)) + else if (t_isalpha_cstr(c)) { - COPYCHAR(ptr->data, c); - ptr->len = pg_mblen(c); + ptr->len = ts_copychar_cstr(ptr->data, c); state = RS_IN_ONEOF_IN; } else /* shouldn't get here */ @@ -136,11 +134,8 @@ RS_compile(Regis *r, bool issuffix, const char *str) } else if (state == RS_IN_ONEOF_IN || state == RS_IN_NONEOF) { - if (t_isalpha(c)) - { - COPYCHAR(ptr->data + ptr->len, c); - ptr->len += pg_mblen(c); - } + if (t_isalpha_cstr(c)) + ptr->len += ts_copychar_cstr(ptr->data + ptr->len, c); else if (t_iseq(c, ']')) state = RS_IN_WAIT; else /* shouldn't get here */ @@ -148,7 +143,7 @@ RS_compile(Regis *r, bool issuffix, const char *str) } else elog(ERROR, "internal error in RS_compile: state %d", state); - c += pg_mblen(c); + c += pg_mblen_cstr(c); } if (state != RS_IN_WAIT) /* shouldn't get here */ @@ -187,10 +182,10 @@ mb_strchr(char *str, char *c) char *ptr = str; bool res = false; - clen = pg_mblen(c); + clen = pg_mblen_cstr(c); while (*ptr && !res) { - plen = pg_mblen(ptr); + plen = pg_mblen_cstr(ptr); if (plen == clen) { i = plen; @@ -219,7 +214,7 @@ RS_execute(Regis *r, char *str) while (*c) { len++; - c += pg_mblen(c); + c += pg_mblen_cstr(c); } if (len < r->nchar) @@ -230,7 +225,7 @@ RS_execute(Regis *r, char *str) { len -= r->nchar; while (len-- > 0) - c += pg_mblen(c); + c += pg_mblen_cstr(c); } @@ -250,7 +245,7 @@ RS_execute(Regis *r, char *str) elog(ERROR, "unrecognized regis node type: %d", ptr->type); } ptr = ptr->next; - c += pg_mblen(c); + c += pg_mblen_cstr(c); } return true; diff --git a/src/backend/tsearch/spell.c b/src/backend/tsearch/spell.c index 03a17f293dbf3..98fdf4890f297 100644 --- a/src/backend/tsearch/spell.c +++ b/src/backend/tsearch/spell.c @@ -232,7 +232,7 @@ findchar(char *str, int c) { if (t_iseq(str, c)) return str; - str += pg_mblen(str); + str += pg_mblen_cstr(str); } return NULL; @@ -245,7 +245,7 @@ findchar2(char *str, int c1, int c2) { if (t_iseq(str, c1) || t_iseq(str, c2)) return str; - str += pg_mblen(str); + str += pg_mblen_cstr(str); } return NULL; @@ -352,6 +352,7 @@ getNextFlagFromString(IspellDict *Conf, char **sflagset, char *sflag) char *next, *sbuf = *sflagset; int maxstep; + int clen; bool stop = false; bool met_comma = false; @@ -363,11 +364,11 @@ getNextFlagFromString(IspellDict *Conf, char **sflagset, char *sflag) { case FM_LONG: case FM_CHAR: - COPYCHAR(sflag, *sflagset); - sflag += pg_mblen(*sflagset); + clen = ts_copychar_cstr(sflag, *sflagset); + sflag += clen; /* Go to start of the next flag */ - *sflagset += pg_mblen(*sflagset); + *sflagset += clen; /* Check if we get all characters of flag */ maxstep--; @@ -390,7 +391,7 @@ getNextFlagFromString(IspellDict *Conf, char **sflagset, char *sflag) *sflagset = next; while (**sflagset) { - if (t_isdigit(*sflagset)) + if (t_isdigit_cstr(*sflagset)) { if (!met_comma) ereport(ERROR, @@ -408,7 +409,7 @@ getNextFlagFromString(IspellDict *Conf, char **sflagset, char *sflag) *sflagset))); met_comma = true; } - else if (!t_isspace(*sflagset)) + else if (!t_isspace_cstr(*sflagset)) { ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), @@ -416,7 +417,7 @@ getNextFlagFromString(IspellDict *Conf, char **sflagset, char *sflag) *sflagset))); } - *sflagset += pg_mblen(*sflagset); + *sflagset += pg_mblen_cstr(*sflagset); } stop = true; break; @@ -542,7 +543,7 @@ NIImportDictionary(IspellDict *Conf, const char *filename) while (*s) { /* we allow only single encoded flags for faster works */ - if (pg_mblen(s) == 1 && t_isprint(s) && !t_isspace(s)) + if (pg_mblen_cstr(s) == 1 && t_isprint_unbounded(s) && !t_isspace_unbounded(s)) s++; else { @@ -558,12 +559,12 @@ NIImportDictionary(IspellDict *Conf, const char *filename) s = line; while (*s) { - if (t_isspace(s)) + if (t_isspace_cstr(s)) { *s = '\0'; break; } - s += pg_mblen(s); + s += pg_mblen_cstr(s); } pstr = lowerstr_ctx(Conf, line); @@ -815,17 +816,17 @@ get_nextfield(char **str, char *next) while (**str) { + int clen = pg_mblen_cstr(*str); + if (state == PAE_WAIT_MASK) { if (t_iseq(*str, '#')) return false; - else if (!t_isspace(*str)) + else if (!t_isspace_cstr(*str)) { - int clen = pg_mblen(*str); - if (clen < avail) { - COPYCHAR(next, *str); + ts_copychar_with_len(next, *str, clen); next += clen; avail -= clen; } @@ -834,24 +835,22 @@ get_nextfield(char **str, char *next) } else /* state == PAE_INMASK */ { - if (t_isspace(*str)) + if (t_isspace_cstr(*str)) { *next = '\0'; return true; } else { - int clen = pg_mblen(*str); - if (clen < avail) { - COPYCHAR(next, *str); + ts_copychar_with_len(next, *str, clen); next += clen; avail -= clen; } } } - *str += pg_mblen(*str); + *str += clen; } *next = '\0'; @@ -941,14 +940,15 @@ parse_affentry(char *str, char *mask, char *find, char *repl) while (*str) { + int clen = pg_mblen_cstr(str); + if (state == PAE_WAIT_MASK) { if (t_iseq(str, '#')) return false; - else if (!t_isspace(str)) + else if (!t_isspace_cstr(str)) { - COPYCHAR(pmask, str); - pmask += pg_mblen(str); + pmask += ts_copychar_with_len(pmask, str, clen); state = PAE_INMASK; } } @@ -959,10 +959,9 @@ parse_affentry(char *str, char *mask, char *find, char *repl) *pmask = '\0'; state = PAE_WAIT_FIND; } - else if (!t_isspace(str)) + else if (!t_isspace_cstr(str)) { - COPYCHAR(pmask, str); - pmask += pg_mblen(str); + pmask += ts_copychar_with_len(pmask, str, clen); } } else if (state == PAE_WAIT_FIND) @@ -971,13 +970,12 @@ parse_affentry(char *str, char *mask, char *find, char *repl) { state = PAE_INFIND; } - else if (t_isalpha(str) || t_iseq(str, '\'') /* english 's */ ) + else if (t_isalpha_cstr(str) || t_iseq(str, '\'') /* english 's */ ) { - COPYCHAR(prepl, str); - prepl += pg_mblen(str); + prepl += ts_copychar_with_len(prepl, str, clen); state = PAE_INREPL; } - else if (!t_isspace(str)) + else if (!t_isspace_cstr(str)) ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("syntax error"))); @@ -989,12 +987,11 @@ parse_affentry(char *str, char *mask, char *find, char *repl) *pfind = '\0'; state = PAE_WAIT_REPL; } - else if (t_isalpha(str)) + else if (t_isalpha_cstr(str)) { - COPYCHAR(pfind, str); - pfind += pg_mblen(str); + pfind += ts_copychar_with_len(pfind, str, clen); } - else if (!t_isspace(str)) + else if (!t_isspace_cstr(str)) ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("syntax error"))); @@ -1005,13 +1002,12 @@ parse_affentry(char *str, char *mask, char *find, char *repl) { break; /* void repl */ } - else if (t_isalpha(str)) + else if (t_isalpha_cstr(str)) { - COPYCHAR(prepl, str); - prepl += pg_mblen(str); + prepl += ts_copychar_with_len(prepl, str, clen); state = PAE_INREPL; } - else if (!t_isspace(str)) + else if (!t_isspace_cstr(str)) ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("syntax error"))); @@ -1023,12 +1019,11 @@ parse_affentry(char *str, char *mask, char *find, char *repl) *prepl = '\0'; break; } - else if (t_isalpha(str)) + else if (t_isalpha_cstr(str)) { - COPYCHAR(prepl, str); - prepl += pg_mblen(str); + prepl += ts_copychar_with_len(prepl, str, clen); } - else if (!t_isspace(str)) + else if (!t_isspace_cstr(str)) ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("syntax error"))); @@ -1036,7 +1031,7 @@ parse_affentry(char *str, char *mask, char *find, char *repl) else elog(ERROR, "unrecognized state in parse_affentry: %d", state); - str += pg_mblen(str); + str += clen; } *pmask = *pfind = *prepl = '\0'; @@ -1088,10 +1083,9 @@ addCompoundAffixFlagValue(IspellDict *Conf, char *s, uint32 val) CompoundAffixFlag *newValue; char sbuf[BUFSIZ]; char *sflag; - int clen; - while (*s && t_isspace(s)) - s += pg_mblen(s); + while (*s && t_isspace_cstr(s)) + s += pg_mblen_cstr(s); if (!*s) ereport(ERROR, @@ -1100,10 +1094,10 @@ addCompoundAffixFlagValue(IspellDict *Conf, char *s, uint32 val) /* Get flag without \n */ sflag = sbuf; - while (*s && !t_isspace(s) && *s != '\n') + while (*s && !t_isspace_cstr(s) && *s != '\n') { - clen = pg_mblen(s); - COPYCHAR(sflag, s); + int clen = ts_copychar_cstr(sflag, s); + sflag += clen; s += clen; } @@ -1245,7 +1239,7 @@ NIImportOOAffixes(IspellDict *Conf, const char *filename) while ((recoded = tsearch_readline(&trst)) != NULL) { - if (*recoded == '\0' || t_isspace(recoded) || t_iseq(recoded, '#')) + if (*recoded == '\0' || t_isspace_cstr(recoded) || t_iseq(recoded, '#')) { pfree(recoded); continue; @@ -1282,8 +1276,8 @@ NIImportOOAffixes(IspellDict *Conf, const char *filename) { char *s = recoded + strlen("FLAG"); - while (*s && t_isspace(s)) - s += pg_mblen(s); + while (*s && t_isspace_cstr(s)) + s += pg_mblen_cstr(s); if (*s) { @@ -1318,7 +1312,7 @@ NIImportOOAffixes(IspellDict *Conf, const char *filename) { int fields_read; - if (*recoded == '\0' || t_isspace(recoded) || t_iseq(recoded, '#')) + if (*recoded == '\0' || t_isspace_cstr(recoded) || t_iseq(recoded, '#')) goto nextline; fields_read = parse_ooaffentry(recoded, type, sflag, find, repl, mask); @@ -1481,12 +1475,12 @@ NIImportAffixes(IspellDict *Conf, const char *filename) s = findchar2(recoded, 'l', 'L'); if (s) { - while (*s && !t_isspace(s)) - s += pg_mblen(s); - while (*s && t_isspace(s)) - s += pg_mblen(s); + while (*s && !t_isspace_cstr(s)) + s += pg_mblen_cstr(s); + while (*s && t_isspace_cstr(s)) + s += pg_mblen_cstr(s); - if (*s && pg_mblen(s) == 1) + if (*s && pg_mblen_cstr(s) == 1) { addCompoundAffixFlagValue(Conf, s, FF_COMPOUNDFLAG); Conf->usecompound = true; @@ -1514,8 +1508,8 @@ NIImportAffixes(IspellDict *Conf, const char *filename) s = recoded + 4; /* we need non-lowercased string */ flagflags = 0; - while (*s && t_isspace(s)) - s += pg_mblen(s); + while (*s && t_isspace_cstr(s)) + s += pg_mblen_cstr(s); if (*s == '*') { @@ -1536,14 +1530,13 @@ NIImportAffixes(IspellDict *Conf, const char *filename) * be followed by EOL, whitespace, or ':'. Otherwise this is a * new-format flag command. */ - if (*s && pg_mblen(s) == 1) + if (*s && pg_mblen_cstr(s) == 1) { - COPYCHAR(flag, s); + flag[0] = *s++; flag[1] = '\0'; - s++; if (*s == '\0' || *s == '#' || *s == '\n' || *s == ':' || - t_isspace(s)) + t_isspace_cstr(s)) { oldformat = true; goto nextline; @@ -1766,7 +1759,7 @@ NISortDictionary(IspellDict *Conf) (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("invalid affix alias \"%s\"", Conf->Spell[i]->p.flag))); - if (*end != '\0' && !t_isdigit(end) && !t_isspace(end)) + if (*end != '\0' && !t_isdigit_cstr(end) && !t_isspace_cstr(end)) ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("invalid affix alias \"%s\"", diff --git a/src/backend/tsearch/ts_locale.c b/src/backend/tsearch/ts_locale.c index 3a475a0f5fcb7..96e35e04c51f8 100644 --- a/src/backend/tsearch/ts_locale.c +++ b/src/backend/tsearch/ts_locale.c @@ -33,66 +33,43 @@ static void tsearch_readline_callback(void *arg); */ #define WC_BUF_LEN 3 -int -t_isdigit(const char *ptr) -{ - int clen = pg_mblen(ptr); - wchar_t character[WC_BUF_LEN]; - pg_locale_t mylocale = 0; /* TODO */ - - if (clen == 1 || database_ctype_is_c) - return isdigit(TOUCHAR(ptr)); - - char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale); - - return iswdigit((wint_t) character[0]); -} - -int -t_isspace(const char *ptr) -{ - int clen = pg_mblen(ptr); - wchar_t character[WC_BUF_LEN]; - pg_locale_t mylocale = 0; /* TODO */ - - if (clen == 1 || database_ctype_is_c) - return isspace(TOUCHAR(ptr)); - - char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale); - - return iswspace((wint_t) character[0]); -} - -int -t_isalpha(const char *ptr) -{ - int clen = pg_mblen(ptr); - wchar_t character[WC_BUF_LEN]; - pg_locale_t mylocale = 0; /* TODO */ - - if (clen == 1 || database_ctype_is_c) - return isalpha(TOUCHAR(ptr)); - - char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale); - - return iswalpha((wint_t) character[0]); -} - -int -t_isprint(const char *ptr) -{ - int clen = pg_mblen(ptr); - wchar_t character[WC_BUF_LEN]; - pg_locale_t mylocale = 0; /* TODO */ - - if (clen == 1 || database_ctype_is_c) - return isprint(TOUCHAR(ptr)); - - char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale); - - return iswprint((wint_t) character[0]); +#define GENERATE_T_ISCLASS_DEF(character_class) \ +/* mblen shall be that of the first character */ \ +int \ +t_is##character_class##_with_len(const char *ptr, int mblen) \ +{ \ + int clen = pg_mblen_with_len(ptr, mblen); \ + wchar_t character[WC_BUF_LEN]; \ + pg_locale_t mylocale = 0; /* TODO */ \ + if (clen == 1 || database_ctype_is_c) \ + return is##character_class(TOUCHAR(ptr)); \ + char2wchar(character, WC_BUF_LEN, ptr, clen, mylocale); \ + return isw##character_class((wint_t) character[0]); \ +} \ +\ +/* ptr shall point to a NUL-terminated string */ \ +int \ +t_is##character_class##_cstr(const char *ptr) \ +{ \ + return t_is##character_class##_with_len(ptr, pg_mblen_cstr(ptr)); \ +} \ +/* ptr shall point to a string with pre-validated encoding */ \ +int \ +t_is##character_class##_unbounded(const char *ptr) \ +{ \ + return t_is##character_class##_with_len(ptr, pg_mblen_unbounded(ptr)); \ +} \ +/* historical name for _unbounded */ \ +int \ +t_is##character_class(const char *ptr) \ +{ \ + return t_is##character_class##_unbounded(ptr); \ } +GENERATE_T_ISCLASS_DEF(alpha) +GENERATE_T_ISCLASS_DEF(digit) +GENERATE_T_ISCLASS_DEF(print) +GENERATE_T_ISCLASS_DEF(space) /* * Set up to read a file using tsearch_readline(). This facility is diff --git a/src/backend/tsearch/ts_utils.c b/src/backend/tsearch/ts_utils.c index 7743bdfd592c1..9685e54eb2b6b 100644 --- a/src/backend/tsearch/ts_utils.c +++ b/src/backend/tsearch/ts_utils.c @@ -88,8 +88,8 @@ readstoplist(const char *fname, StopList *s, char *(*wordop) (const char *)) char *pbuf = line; /* Trim trailing space */ - while (*pbuf && !t_isspace(pbuf)) - pbuf += pg_mblen(pbuf); + while (*pbuf && !t_isspace_cstr(pbuf)) + pbuf += pg_mblen_cstr(pbuf); *pbuf = '\0'; /* Skip empty lines */ diff --git a/src/backend/tsearch/wparser_def.c b/src/backend/tsearch/wparser_def.c index 57df42237c92b..9ffa5cf24067a 100644 --- a/src/backend/tsearch/wparser_def.c +++ b/src/backend/tsearch/wparser_def.c @@ -1727,7 +1727,8 @@ TParserGet(TParser *prs) prs->state->charlen = 0; else prs->state->charlen = (prs->charmaxlen == 1) ? prs->charmaxlen : - pg_mblen(prs->str + prs->state->posbyte); + pg_mblen_range(prs->str + prs->state->posbyte, + prs->str + prs->lenstr); Assert(prs->state->posbyte + prs->state->charlen <= prs->lenstr); Assert(prs->state->state >= TPS_Base && prs->state->state < TPS_Null); diff --git a/src/backend/utils/adt/encode.c b/src/backend/utils/adt/encode.c index feb3e830e4fd8..10a7ad89a9e84 100644 --- a/src/backend/utils/adt/encode.c +++ b/src/backend/utils/adt/encode.c @@ -172,7 +172,7 @@ hex_encode(const char *src, size_t len, char *dst) } static inline char -get_hex(const char *cp) +get_hex(const char *cp, const char *end) { unsigned char c = (unsigned char) *cp; int res = -1; @@ -184,7 +184,7 @@ get_hex(const char *cp) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("invalid hexadecimal digit: \"%.*s\"", - pg_mblen(cp), cp))); + pg_mblen_range(cp, end), cp))); return (char) res; } @@ -208,14 +208,14 @@ hex_decode(const char *src, size_t len, char *dst) s++; continue; } - v1 = get_hex(s) << 4; + v1 = get_hex(s, srcend) << 4; s++; if (s >= srcend) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("invalid hexadecimal data: odd number of digits"))); - v2 = get_hex(s); + v2 = get_hex(s, srcend); s++; *p++ = v1 | v2; } @@ -344,7 +344,7 @@ pg_base64_decode(const char *src, size_t len, char *dst) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("invalid symbol \"%.*s\" found while decoding base64 sequence", - pg_mblen(s - 1), s - 1))); + pg_mblen_range(s - 1, srcend), s - 1))); } /* add it to buffer */ buf = (buf << 6) + b; diff --git a/src/backend/utils/adt/formatting.c b/src/backend/utils/adt/formatting.c index 1d6dee9b6d60e..0fb41cc1282f0 100644 --- a/src/backend/utils/adt/formatting.c +++ b/src/backend/utils/adt/formatting.c @@ -1427,7 +1427,7 @@ parse_format(FormatNode *node, const char *str, const KeyWord *kw, ereport(ERROR, (errcode(ERRCODE_INVALID_DATETIME_FORMAT), errmsg("invalid datetime format separator: \"%s\"", - pnstrdup(str, pg_mblen(str))))); + pnstrdup(str, pg_mblen_cstr(str))))); if (*str == ' ') n->type = NODE_TYPE_SPACE; @@ -1457,7 +1457,7 @@ parse_format(FormatNode *node, const char *str, const KeyWord *kw, /* backslash quotes the next character, if any */ if (*str == '\\' && *(str + 1)) str++; - chlen = pg_mblen(str); + chlen = pg_mblen_cstr(str); n->type = NODE_TYPE_CHAR; memcpy(n->character, str, chlen); n->character[chlen] = '\0'; @@ -1475,7 +1475,7 @@ parse_format(FormatNode *node, const char *str, const KeyWord *kw, */ if (*str == '\\' && *(str + 1) == '"') str++; - chlen = pg_mblen(str); + chlen = pg_mblen_cstr(str); if ((flags & DCH_FLAG) && is_separator_char(str)) n->type = NODE_TYPE_SEPARATOR; @@ -2180,8 +2180,8 @@ asc_toupper_z(const char *buff) do { \ if (S_THth(_suf)) \ { \ - if (*(ptr)) (ptr) += pg_mblen(ptr); \ - if (*(ptr)) (ptr) += pg_mblen(ptr); \ + if (*(ptr)) (ptr) += pg_mblen_cstr(ptr); \ + if (*(ptr)) (ptr) += pg_mblen_cstr(ptr); \ } \ } while (0) @@ -3396,7 +3396,7 @@ DCH_from_char(FormatNode *node, const char *in, TmFromChar *out, * insist that the consumed character match the format's * character. */ - s += pg_mblen(s); + s += pg_mblen_cstr(s); } continue; } @@ -3418,11 +3418,11 @@ DCH_from_char(FormatNode *node, const char *in, TmFromChar *out, if (extra_skip > 0) extra_skip--; else - s += pg_mblen(s); + s += pg_mblen_cstr(s); } else { - int chlen = pg_mblen(s); + int chlen = pg_mblen_cstr(s); /* * Standard mode requires strict match of format characters. @@ -5610,13 +5610,15 @@ NUM_numpart_to_char(NUMProc *Np, int id) static void NUM_eat_non_data_chars(NUMProc *Np, int n, int input_len) { + const char *end = Np->inout + input_len; + while (n-- > 0) { if (OVERLOAD_TEST) break; /* end of input */ if (strchr("0123456789.,+-", *Np->inout_p) != NULL) break; /* it's a data character */ - Np->inout_p += pg_mblen(Np->inout_p); + Np->inout_p += pg_mblen_range(Np->inout_p, end); } } @@ -6073,7 +6075,7 @@ NUM_processor(FormatNode *node, NUMDesc *Num, char *inout, } else { - Np->inout_p += pg_mblen(Np->inout_p); + Np->inout_p += pg_mblen_range(Np->inout_p, Np->inout + input_len); } continue; } diff --git a/src/backend/utils/adt/jsonfuncs.c b/src/backend/utils/adt/jsonfuncs.c index adfae6cb3b08b..49d91bd013abf 100644 --- a/src/backend/utils/adt/jsonfuncs.c +++ b/src/backend/utils/adt/jsonfuncs.c @@ -663,7 +663,7 @@ report_json_context(JsonLexContext *lex) { /* Advance to next multibyte character */ if (IS_HIGHBIT_SET(*context_start)) - context_start += pg_mblen(context_start); + context_start += pg_mblen_range(context_start, context_end); else context_start++; } diff --git a/src/backend/utils/adt/jsonpath_gram.y b/src/backend/utils/adt/jsonpath_gram.y index 91e4308d6559f..b79165fc608ea 100644 --- a/src/backend/utils/adt/jsonpath_gram.y +++ b/src/backend/utils/adt/jsonpath_gram.y @@ -528,7 +528,8 @@ makeItemLikeRegex(JsonPathParseItem *expr, JsonPathString *pattern, (errcode(ERRCODE_SYNTAX_ERROR), errmsg("invalid input syntax for type %s", "jsonpath"), errdetail("Unrecognized flag character \"%.*s\" in LIKE_REGEX predicate.", - pg_mblen(flags->val + i), flags->val + i))); + pg_mblen_range(flags->val + i, flags->val + flags->len), + flags->val + i))); break; } } diff --git a/src/backend/utils/adt/levenshtein.c b/src/backend/utils/adt/levenshtein.c index 3026cc2431117..c6445cdcbf7a1 100644 --- a/src/backend/utils/adt/levenshtein.c +++ b/src/backend/utils/adt/levenshtein.c @@ -84,6 +84,8 @@ varstr_levenshtein(const char *source, int slen, int i, j; const char *y; + const char *send = source + slen; + const char *tend = target + tlen; /* * For varstr_levenshtein_less_equal, we have real variables called @@ -184,10 +186,10 @@ varstr_levenshtein(const char *source, int slen, #endif /* - * In order to avoid calling pg_mblen() repeatedly on each character in s, - * we cache all the lengths before starting the main loop -- but if all - * the characters in both strings are single byte, then we skip this and - * use a fast-path in the main loop. If only one string contains + * In order to avoid calling pg_mblen_range() repeatedly on each character + * in s, we cache all the lengths before starting the main loop -- but if + * all the characters in both strings are single byte, then we skip this + * and use a fast-path in the main loop. If only one string contains * multi-byte characters, we still build the array, so that the fast-path * needn't deal with the case where the array hasn't been initialized. */ @@ -199,7 +201,7 @@ varstr_levenshtein(const char *source, int slen, s_char_len = (int *) palloc((m + 1) * sizeof(int)); for (i = 0; i < m; ++i) { - s_char_len[i] = pg_mblen(cp); + s_char_len[i] = pg_mblen_range(cp, send); cp += s_char_len[i]; } s_char_len[i] = 0; @@ -225,7 +227,7 @@ varstr_levenshtein(const char *source, int slen, { int *temp; const char *x = source; - int y_char_len = n != tlen + 1 ? pg_mblen(y) : 1; + int y_char_len = n != tlen + 1 ? pg_mblen_range(y, tend) : 1; #ifdef LEVENSHTEIN_LESS_EQUAL diff --git a/src/backend/utils/adt/like.c b/src/backend/utils/adt/like.c index e02fc3725ad8c..b37bc0ceb55fa 100644 --- a/src/backend/utils/adt/like.c +++ b/src/backend/utils/adt/like.c @@ -54,20 +54,20 @@ static int Generic_Text_IC_like(text *str, text *pat, Oid collation); *-------------------- */ static inline int -wchareq(const char *p1, const char *p2) +wchareq(const char *p1, int p1len, const char *p2, int p2len) { - int p1_len; + int p1clen; /* Optimization: quickly compare the first byte. */ if (*p1 != *p2) return 0; - p1_len = pg_mblen(p1); - if (pg_mblen(p2) != p1_len) + p1clen = pg_mblen_with_len(p1, p1len); + if (pg_mblen_with_len(p2, p2len) != p1clen) return 0; /* They are the same length */ - while (p1_len--) + while (p1clen--) { if (*p1++ != *p2++) return 0; @@ -106,11 +106,11 @@ SB_lower_char(unsigned char c, pg_locale_t locale, bool locale_is_c) #define NextByte(p, plen) ((p)++, (plen)--) /* Set up to compile like_match.c for multibyte characters */ -#define CHAREQ(p1, p2) wchareq((p1), (p2)) +#define CHAREQ(p1, p1len, p2, p2len) wchareq((p1), (p1len), (p2), (p2len)) #define NextChar(p, plen) \ - do { int __l = pg_mblen(p); (p) +=__l; (plen) -=__l; } while (0) + do { int __l = pg_mblen_with_len((p), (plen)); (p) +=__l; (plen) -=__l; } while (0) #define CopyAdvChar(dst, src, srclen) \ - do { int __l = pg_mblen(src); \ + do { int __l = pg_mblen_with_len((src), (srclen)); \ (srclen) -= __l; \ while (__l-- > 0) \ *(dst)++ = *(src)++; \ @@ -122,7 +122,7 @@ SB_lower_char(unsigned char c, pg_locale_t locale, bool locale_is_c) #include "like_match.c" /* Set up to compile like_match.c for single-byte characters */ -#define CHAREQ(p1, p2) (*(p1) == *(p2)) +#define CHAREQ(p1, p1len, p2, p2len) (*(p1) == *(p2)) #define NextChar(p, plen) NextByte((p), (plen)) #define CopyAdvChar(dst, src, srclen) (*(dst)++ = *(src)++, (srclen)--) diff --git a/src/backend/utils/adt/like_match.c b/src/backend/utils/adt/like_match.c index e876560991332..e2e8c28ccfba7 100644 --- a/src/backend/utils/adt/like_match.c +++ b/src/backend/utils/adt/like_match.c @@ -294,6 +294,7 @@ do_like_escape(text *pat, text *esc) errhint("Escape string must be empty or one character."))); e = VARDATA_ANY(esc); + elen = VARSIZE_ANY_EXHDR(esc); /* * If specified escape is '\', just copy the pattern as-is. @@ -312,7 +313,7 @@ do_like_escape(text *pat, text *esc) afterescape = false; while (plen > 0) { - if (CHAREQ(p, e) && !afterescape) + if (CHAREQ(p, plen, e, elen) && !afterescape) { *r++ = '\\'; NextChar(p, plen); diff --git a/src/backend/utils/adt/oracle_compat.c b/src/backend/utils/adt/oracle_compat.c index 6a5ce1cc10241..102500192c089 100644 --- a/src/backend/utils/adt/oracle_compat.c +++ b/src/backend/utils/adt/oracle_compat.c @@ -152,8 +152,8 @@ lpad(PG_FUNCTION_ARGS) char *ptr1, *ptr2, *ptr2start, - *ptr2end, *ptr_ret; + const char *ptr2end; int m, s1len, s2len; @@ -198,7 +198,7 @@ lpad(PG_FUNCTION_ARGS) while (m--) { - int mlen = pg_mblen(ptr2); + int mlen = pg_mblen_range(ptr2, ptr2end); memcpy(ptr_ret, ptr2, mlen); ptr_ret += mlen; @@ -211,7 +211,7 @@ lpad(PG_FUNCTION_ARGS) while (s1len--) { - int mlen = pg_mblen(ptr1); + int mlen = pg_mblen_unbounded(ptr1); memcpy(ptr_ret, ptr1, mlen); ptr_ret += mlen; @@ -250,8 +250,8 @@ rpad(PG_FUNCTION_ARGS) char *ptr1, *ptr2, *ptr2start, - *ptr2end, *ptr_ret; + const char *ptr2end; int m, s1len, s2len; @@ -291,11 +291,12 @@ rpad(PG_FUNCTION_ARGS) m = len - s1len; ptr1 = VARDATA_ANY(string1); + ptr_ret = VARDATA(ret); while (s1len--) { - int mlen = pg_mblen(ptr1); + int mlen = pg_mblen_unbounded(ptr1); memcpy(ptr_ret, ptr1, mlen); ptr_ret += mlen; @@ -307,7 +308,7 @@ rpad(PG_FUNCTION_ARGS) while (m--) { - int mlen = pg_mblen(ptr2); + int mlen = pg_mblen_range(ptr2, ptr2end); memcpy(ptr_ret, ptr2, mlen); ptr_ret += mlen; @@ -392,6 +393,7 @@ dotrim(const char *string, int stringlen, */ const char **stringchars; const char **setchars; + const char *setend; int *stringmblen; int *setmblen; int stringnchars; @@ -399,6 +401,7 @@ dotrim(const char *string, int stringlen, int resultndx; int resultnchars; const char *p; + const char *pend; int len; int mblen; const char *str_pos; @@ -409,10 +412,11 @@ dotrim(const char *string, int stringlen, stringnchars = 0; p = string; len = stringlen; + pend = p + len; while (len > 0) { stringchars[stringnchars] = p; - stringmblen[stringnchars] = mblen = pg_mblen(p); + stringmblen[stringnchars] = mblen = pg_mblen_range(p, pend); stringnchars++; p += mblen; len -= mblen; @@ -423,10 +427,11 @@ dotrim(const char *string, int stringlen, setnchars = 0; p = set; len = setlen; + setend = set + setlen; while (len > 0) { setchars[setnchars] = p; - setmblen[setnchars] = mblen = pg_mblen(p); + setmblen[setnchars] = mblen = pg_mblen_range(p, setend); setnchars++; p += mblen; len -= mblen; @@ -804,6 +809,8 @@ translate(PG_FUNCTION_ARGS) *to_end; char *source, *target; + const char *source_end; + const char *from_end; int m, fromlen, tolen, @@ -818,9 +825,11 @@ translate(PG_FUNCTION_ARGS) if (m <= 0) PG_RETURN_TEXT_P(string); source = VARDATA_ANY(string); + source_end = source + m; fromlen = VARSIZE_ANY_EXHDR(from); from_ptr = VARDATA_ANY(from); + from_end = from_ptr + fromlen; tolen = VARSIZE_ANY_EXHDR(to); to_ptr = VARDATA_ANY(to); to_end = to_ptr + tolen; @@ -844,12 +853,12 @@ translate(PG_FUNCTION_ARGS) while (m > 0) { - source_len = pg_mblen(source); + source_len = pg_mblen_range(source, source_end); from_index = 0; for (i = 0; i < fromlen; i += len) { - len = pg_mblen(&from_ptr[i]); + len = pg_mblen_range(&from_ptr[i], from_end); if (len == source_len && memcmp(source, &from_ptr[i], len) == 0) break; @@ -865,11 +874,11 @@ translate(PG_FUNCTION_ARGS) { if (p >= to_end) break; - p += pg_mblen(p); + p += pg_mblen_range(p, to_end); } if (p < to_end) { - len = pg_mblen(p); + len = pg_mblen_range(p, to_end); memcpy(target, p, len); target += len; retlen += len; diff --git a/src/backend/utils/adt/regexp.c b/src/backend/utils/adt/regexp.c index dcf6681ef24c3..f65bc5cdf6d57 100644 --- a/src/backend/utils/adt/regexp.c +++ b/src/backend/utils/adt/regexp.c @@ -429,7 +429,7 @@ parse_re_flags(pg_re_flags *flags, text *opts) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("invalid regular expression option: \"%.*s\"", - pg_mblen(opt_p + i), opt_p + i))); + pg_mblen_range(opt_p + i, opt_p + opt_len), opt_p + i))); break; } } @@ -659,12 +659,13 @@ textregexreplace(PG_FUNCTION_ARGS) if (VARSIZE_ANY_EXHDR(opt) > 0) { char *opt_p = VARDATA_ANY(opt); + const char *end_p = opt_p + VARSIZE_ANY_EXHDR(opt); if (*opt_p >= '0' && *opt_p <= '9') ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("invalid regular expression option: \"%.*s\"", - pg_mblen(opt_p), opt_p), + pg_mblen_range(opt_p, end_p), opt_p), errhint("If you meant to use regexp_replace() with a start parameter, cast the fourth argument to integer explicitly."))); } @@ -758,12 +759,14 @@ similar_escape_internal(text *pat_text, text *esc_text) *r; int plen, elen; + const char *pend; bool afterescape = false; bool incharclass = false; int nquotes = 0; p = VARDATA_ANY(pat_text); plen = VARSIZE_ANY_EXHDR(pat_text); + pend = p + plen; if (esc_text == NULL) { /* No ESCAPE clause provided; default to backslash as escape */ @@ -852,7 +855,7 @@ similar_escape_internal(text *pat_text, text *esc_text) if (elen > 1) { - int mblen = pg_mblen(p); + int mblen = pg_mblen_range(p, pend); if (mblen > 1) { diff --git a/src/backend/utils/adt/tsquery.c b/src/backend/utils/adt/tsquery.c index 2bb3d0fc3f399..5289dd90a96ce 100644 --- a/src/backend/utils/adt/tsquery.c +++ b/src/backend/utils/adt/tsquery.c @@ -109,7 +109,7 @@ get_modifiers(char *buf, int16 *weight, bool *prefix) return buf; buf++; - while (*buf && pg_mblen(buf) == 1) + while (*buf && pg_mblen_cstr(buf) == 1) { switch (*buf) { @@ -186,7 +186,7 @@ parse_phrase_operator(TSQueryParserState pstate, int16 *distance) continue; } - if (!t_isdigit(ptr)) + if (!t_isdigit_cstr(ptr)) return false; errno = 0; @@ -248,12 +248,12 @@ parse_or_operator(TSQueryParserState pstate) return false; /* it shouldn't be a part of any word */ - if (t_iseq(ptr, '-') || t_iseq(ptr, '_') || t_isalpha(ptr) || t_isdigit(ptr)) + if (t_iseq(ptr, '-') || t_iseq(ptr, '_') || t_isalpha_cstr(ptr) || t_isdigit_cstr(ptr)) return false; for (;;) { - ptr += pg_mblen(ptr); + ptr += pg_mblen_cstr(ptr); if (*ptr == '\0') /* got end of string without operand */ return false; @@ -263,7 +263,7 @@ parse_or_operator(TSQueryParserState pstate) * So we still treat OR literal as operation with possibly incorrect * operand and will not search it as lexeme */ - if (!t_isspace(ptr)) + if (!t_isspace_cstr(ptr)) break; } @@ -306,7 +306,7 @@ gettoken_query_standard(TSQueryParserState state, int8 *operator, errmsg("syntax error in tsquery: \"%s\"", state->buffer))); } - else if (!t_isspace(state->buf)) + else if (!t_isspace_cstr(state->buf)) { /* * We rely on the tsvector parser to parse the value for @@ -364,14 +364,14 @@ gettoken_query_standard(TSQueryParserState state, int8 *operator, { return (state->count) ? PT_ERR : PT_END; } - else if (!t_isspace(state->buf)) + else if (!t_isspace_cstr(state->buf)) { return PT_ERR; } break; } - state->buf += pg_mblen(state->buf); + state->buf += pg_mblen_cstr(state->buf); } } @@ -425,7 +425,7 @@ gettoken_query_websearch(TSQueryParserState state, int8 *operator, state->state = WAITOPERAND; continue; } - else if (!t_isspace(state->buf)) + else if (!t_isspace_cstr(state->buf)) { /* * We rely on the tsvector parser to parse the value for @@ -468,7 +468,7 @@ gettoken_query_websearch(TSQueryParserState state, int8 *operator, state->buf++; continue; } - else if (!t_isspace(state->buf)) + else if (!t_isspace_cstr(state->buf)) { /* insert implicit AND between operands */ state->state = WAITOPERAND; @@ -478,7 +478,7 @@ gettoken_query_websearch(TSQueryParserState state, int8 *operator, break; } - state->buf += pg_mblen(state->buf); + state->buf += pg_mblen_cstr(state->buf); } } @@ -961,9 +961,8 @@ infix(INFIX *in, int parentPriority, bool rightPhraseOp) *(in->cur) = '\\'; in->cur++; } - COPYCHAR(in->cur, op); - clen = pg_mblen(op); + clen = ts_copychar_cstr(in->cur, op); op += clen; in->cur += clen; } diff --git a/src/backend/utils/adt/tsvector.c b/src/backend/utils/adt/tsvector.c index cb36893b96070..9038bf84173ec 100644 --- a/src/backend/utils/adt/tsvector.c +++ b/src/backend/utils/adt/tsvector.c @@ -316,9 +316,9 @@ tsvectorout(PG_FUNCTION_ARGS) lenbuf = 0, pp; WordEntry *ptr = ARRPTR(out); - char *curbegin, - *curin, + char *curin, *curout; + const char *curend; lenbuf = out->size * 2 /* '' */ + out->size - 1 /* space */ + 2 /* \0 */ ; for (i = 0; i < out->size; i++) @@ -331,13 +331,14 @@ tsvectorout(PG_FUNCTION_ARGS) curout = outbuf = (char *) palloc(lenbuf); for (i = 0; i < out->size; i++) { - curbegin = curin = STRPTR(out) + ptr->pos; + curin = STRPTR(out) + ptr->pos; + curend = curin + ptr->len; if (i != 0) *curout++ = ' '; *curout++ = '\''; - while (curin - curbegin < ptr->len) + while (curin < curend) { - int len = pg_mblen(curin); + int len = pg_mblen_range(curin, curend); if (t_iseq(curin, '\'')) *curout++ = '\''; diff --git a/src/backend/utils/adt/tsvector_op.c b/src/backend/utils/adt/tsvector_op.c index d1d880cfa27b8..0a96fba95cb32 100644 --- a/src/backend/utils/adt/tsvector_op.c +++ b/src/backend/utils/adt/tsvector_op.c @@ -2439,11 +2439,15 @@ ts_stat_sql(MemoryContext persistentContext, text *txt, text *ws) if (ws) { char *buf; + const char *end; buf = VARDATA_ANY(ws); - while (buf - VARDATA_ANY(ws) < VARSIZE_ANY_EXHDR(ws)) + end = buf + VARSIZE_ANY_EXHDR(ws); + while (buf < end) { - if (pg_mblen(buf) == 1) + int len = pg_mblen_range(buf, end); + + if (len == 1) { switch (*buf) { @@ -2467,7 +2471,7 @@ ts_stat_sql(MemoryContext persistentContext, text *txt, text *ws) stat->weight |= 0; } } - buf += pg_mblen(buf); + buf += len; } } diff --git a/src/backend/utils/adt/tsvector_parser.c b/src/backend/utils/adt/tsvector_parser.c index e2460d393ab11..f7152029b0e87 100644 --- a/src/backend/utils/adt/tsvector_parser.c +++ b/src/backend/utils/adt/tsvector_parser.c @@ -185,10 +185,9 @@ gettoken_tsvector(TSVectorParseState state, else if ((state->oprisdelim && ISOPERATOR(state->prsbuf)) || (state->is_web && t_iseq(state->prsbuf, '"'))) PRSSYNTAXERROR; - else if (!t_isspace(state->prsbuf)) + else if (!t_isspace_cstr(state->prsbuf)) { - COPYCHAR(curpos, state->prsbuf); - curpos += pg_mblen(state->prsbuf); + curpos += ts_copychar_cstr(curpos, state->prsbuf); statecode = WAITENDWORD; } } @@ -202,8 +201,7 @@ gettoken_tsvector(TSVectorParseState state, else { RESIZEPRSBUF; - COPYCHAR(curpos, state->prsbuf); - curpos += pg_mblen(state->prsbuf); + curpos += ts_copychar_cstr(curpos, state->prsbuf); Assert(oldstate != 0); statecode = oldstate; } @@ -215,7 +213,7 @@ gettoken_tsvector(TSVectorParseState state, statecode = WAITNEXTCHAR; oldstate = WAITENDWORD; } - else if (t_isspace(state->prsbuf) || *(state->prsbuf) == '\0' || + else if (t_isspace_cstr(state->prsbuf) || *(state->prsbuf) == '\0' || (state->oprisdelim && ISOPERATOR(state->prsbuf)) || (state->is_web && t_iseq(state->prsbuf, '"'))) { @@ -238,8 +236,7 @@ gettoken_tsvector(TSVectorParseState state, else { RESIZEPRSBUF; - COPYCHAR(curpos, state->prsbuf); - curpos += pg_mblen(state->prsbuf); + curpos += ts_copychar_cstr(curpos, state->prsbuf); } } else if (statecode == WAITENDCMPLX) @@ -258,8 +255,7 @@ gettoken_tsvector(TSVectorParseState state, else { RESIZEPRSBUF; - COPYCHAR(curpos, state->prsbuf); - curpos += pg_mblen(state->prsbuf); + curpos += ts_copychar_cstr(curpos, state->prsbuf); } } else if (statecode == WAITCHARCMPLX) @@ -267,8 +263,7 @@ gettoken_tsvector(TSVectorParseState state, if (!state->is_web && t_iseq(state->prsbuf, '\'')) { RESIZEPRSBUF; - COPYCHAR(curpos, state->prsbuf); - curpos += pg_mblen(state->prsbuf); + curpos += ts_copychar_cstr(curpos, state->prsbuf); statecode = WAITENDCMPLX; } else @@ -279,7 +274,7 @@ gettoken_tsvector(TSVectorParseState state, PRSSYNTAXERROR; if (state->oprisdelim) { - /* state->prsbuf+=pg_mblen(state->prsbuf); */ + /* state->prsbuf+=pg_mblen_cstr(state->prsbuf); */ RETURN_TOKEN; } else @@ -296,7 +291,7 @@ gettoken_tsvector(TSVectorParseState state, } else if (statecode == INPOSINFO) { - if (t_isdigit(state->prsbuf)) + if (t_isdigit_cstr(state->prsbuf)) { if (posalen == 0) { @@ -351,10 +346,10 @@ gettoken_tsvector(TSVectorParseState state, PRSSYNTAXERROR; WEP_SETWEIGHT(pos[npos - 1], 0); } - else if (t_isspace(state->prsbuf) || + else if (t_isspace_cstr(state->prsbuf) || *(state->prsbuf) == '\0') RETURN_TOKEN; - else if (!t_isdigit(state->prsbuf)) + else if (!t_isdigit_cstr(state->prsbuf)) PRSSYNTAXERROR; } else /* internal error */ @@ -362,6 +357,6 @@ gettoken_tsvector(TSVectorParseState state, statecode); /* get next char */ - state->prsbuf += pg_mblen(state->prsbuf); + state->prsbuf += pg_mblen_cstr(state->prsbuf); } } diff --git a/src/backend/utils/adt/varbit.c b/src/backend/utils/adt/varbit.c index 73e41e0808fb5..532b6dcd097b7 100644 --- a/src/backend/utils/adt/varbit.c +++ b/src/backend/utils/adt/varbit.c @@ -232,7 +232,7 @@ bit_in(PG_FUNCTION_ARGS) ereport(ERROR, (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), errmsg("\"%.*s\" is not a valid binary digit", - pg_mblen(sp), sp))); + pg_mblen_cstr(sp), sp))); x >>= 1; if (x == 0) @@ -257,7 +257,7 @@ bit_in(PG_FUNCTION_ARGS) ereport(ERROR, (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), errmsg("\"%.*s\" is not a valid hexadecimal digit", - pg_mblen(sp), sp))); + pg_mblen_cstr(sp), sp))); if (bc) { @@ -533,7 +533,7 @@ varbit_in(PG_FUNCTION_ARGS) ereport(ERROR, (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), errmsg("\"%.*s\" is not a valid binary digit", - pg_mblen(sp), sp))); + pg_mblen_cstr(sp), sp))); x >>= 1; if (x == 0) @@ -558,7 +558,7 @@ varbit_in(PG_FUNCTION_ARGS) ereport(ERROR, (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), errmsg("\"%.*s\" is not a valid hexadecimal digit", - pg_mblen(sp), sp))); + pg_mblen_cstr(sp), sp))); if (bc) { diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c index 7df2edc529236..a6d138a4058bf 100644 --- a/src/backend/utils/adt/varlena.c +++ b/src/backend/utils/adt/varlena.c @@ -778,8 +778,11 @@ text_catenate(text *t1, text *t2) * charlen_to_bytelen() * Compute the number of bytes occupied by n characters starting at *p * - * It is caller's responsibility that there actually are n characters; - * the string need not be null-terminated. + * The caller shall ensure there are n complete characters. Callers achieve + * this by deriving "n" from regmatch_t findings from searching a wchar array. + * pg_mb2wchar_with_len() skips any trailing incomplete character, so regex + * matches will end no later than the last complete character. (The string + * need not be null-terminated.) */ static int charlen_to_bytelen(const char *p, int n) @@ -794,7 +797,7 @@ charlen_to_bytelen(const char *p, int n) const char *s; for (s = p; n > 0; n--) - s += pg_mblen(s); + s += pg_mblen_unbounded(s); /* caller verified encoding */ return s - p; } @@ -927,6 +930,7 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) int32 slice_start; int32 slice_size; int32 slice_strlen; + int32 slice_len; text *slice; int32 E1; int32 i; @@ -996,7 +1000,8 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) slice = (text *) DatumGetPointer(str); /* see if we got back an empty string */ - if (VARSIZE_ANY_EXHDR(slice) == 0) + slice_len = VARSIZE_ANY_EXHDR(slice); + if (slice_len == 0) { if (slice != (text *) DatumGetPointer(str)) pfree(slice); @@ -1005,7 +1010,7 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) /* Now we can get the actual length of the slice in MB characters */ slice_strlen = pg_mbstrlen_with_len(VARDATA_ANY(slice), - VARSIZE_ANY_EXHDR(slice)); + slice_len); /* * Check that the start position wasn't > slice_strlen. If so, SQL99 @@ -1032,7 +1037,7 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) */ p = VARDATA_ANY(slice); for (i = 0; i < S1 - 1; i++) - p += pg_mblen(p); + p += pg_mblen_unbounded(p); /* hang onto a pointer to our start position */ s = p; @@ -1042,7 +1047,7 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) * length. */ for (i = S1; i < E1; i++) - p += pg_mblen(p); + p += pg_mblen_unbounded(p); ret = (text *) palloc(VARHDRSZ + (p - s)); SET_VARSIZE(ret, VARHDRSZ + (p - s)); @@ -1340,6 +1345,8 @@ text_position_next(TextPositionState *state) */ if (state->is_multibyte_char_in_char) { + const char *haystack_end = state->str1 + state->len1; + /* Walk one character at a time, until we reach the match. */ /* the search should never move backwards. */ @@ -1348,7 +1355,7 @@ text_position_next(TextPositionState *state) while (state->refpoint < matchptr) { /* step to next character. */ - state->refpoint += pg_mblen(state->refpoint); + state->refpoint += pg_mblen_range(state->refpoint, haystack_end); state->refpos++; /* @@ -4940,6 +4947,8 @@ split_text(FunctionCallInfo fcinfo, SplitTextOutputData *tstate) } else { + const char *end_ptr; + /* * When fldsep is NULL, each character in the input string becomes a * separate element in the result set. The separator is effectively @@ -4948,10 +4957,11 @@ split_text(FunctionCallInfo fcinfo, SplitTextOutputData *tstate) inputstring_len = VARSIZE_ANY_EXHDR(inputstring); start_ptr = VARDATA_ANY(inputstring); + end_ptr = start_ptr + inputstring_len; while (inputstring_len > 0) { - int chunk_len = pg_mblen(start_ptr); + int chunk_len = pg_mblen_range(start_ptr, end_ptr); CHECK_FOR_INTERRUPTS(); @@ -5630,7 +5640,7 @@ text_reverse(PG_FUNCTION_ARGS) { int sz; - sz = pg_mblen(p); + sz = pg_mblen_range(p, endp); dst -= sz; memcpy(dst, p, sz); p += sz; @@ -5791,7 +5801,7 @@ text_format(PG_FUNCTION_ARGS) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("unrecognized format() type specifier \"%.*s\"", - pg_mblen(cp), cp), + pg_mblen_range(cp, end_ptr), cp), errhint("For a single \"%%\" use \"%%%%\"."))); /* If indirect width was specified, get its value */ @@ -5912,7 +5922,7 @@ text_format(PG_FUNCTION_ARGS) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("unrecognized format() type specifier \"%.*s\"", - pg_mblen(cp), cp), + pg_mblen_range(cp, end_ptr), cp), errhint("For a single \"%%\" use \"%%%%\"."))); break; } diff --git a/src/backend/utils/adt/xml.c b/src/backend/utils/adt/xml.c index 85f726079be90..c80bfce41441b 100644 --- a/src/backend/utils/adt/xml.c +++ b/src/backend/utils/adt/xml.c @@ -2042,8 +2042,7 @@ sqlchar_to_unicode(const char *s) char *utf8string; pg_wchar ret[2]; /* need space for trailing zero */ - /* note we're not assuming s is null-terminated */ - utf8string = pg_server_to_any(s, pg_mblen(s), PG_UTF8); + utf8string = pg_server_to_any(s, pg_mblen_cstr(s), PG_UTF8); pg_encoding_mb2wchar_with_len(PG_UTF8, utf8string, ret, pg_encoding_mblen(PG_UTF8, utf8string)); @@ -2096,7 +2095,7 @@ map_sql_identifier_to_xml_name(const char *ident, bool fully_escaped, initStringInfo(&buf); - for (p = ident; *p; p += pg_mblen(p)) + for (p = ident; *p; p += pg_mblen_cstr(p)) { if (*p == ':' && (p == ident || fully_escaped)) appendStringInfoString(&buf, "_x003A_"); @@ -2121,7 +2120,7 @@ map_sql_identifier_to_xml_name(const char *ident, bool fully_escaped, : !is_valid_xml_namechar(u)) appendStringInfo(&buf, "_x%04X_", (unsigned int) u); else - appendBinaryStringInfo(&buf, p, pg_mblen(p)); + appendBinaryStringInfo(&buf, p, pg_mblen_cstr(p)); } } @@ -2144,7 +2143,7 @@ map_xml_name_to_sql_identifier(const char *name) initStringInfo(&buf); - for (p = name; *p; p += pg_mblen(p)) + for (p = name; *p; p += pg_mblen_cstr(p)) { if (*p == '_' && *(p + 1) == 'x' && isxdigit((unsigned char) *(p + 2)) @@ -2162,7 +2161,7 @@ map_xml_name_to_sql_identifier(const char *name) p += 6; } else - appendBinaryStringInfo(&buf, p, pg_mblen(p)); + appendBinaryStringInfo(&buf, p, pg_mblen_cstr(p)); } return buf.data; diff --git a/src/backend/utils/mb/mbutils.c b/src/backend/utils/mb/mbutils.c index fc279b3125b14..c0f96cad60c90 100644 --- a/src/backend/utils/mb/mbutils.c +++ b/src/backend/utils/mb/mbutils.c @@ -38,6 +38,7 @@ #include "catalog/namespace.h" #include "mb/pg_wchar.h" #include "utils/builtins.h" +#include "utils/memdebug.h" #include "utils/memutils.h" #include "utils/syscache.h" @@ -96,6 +97,13 @@ static char *perform_default_encoding_conversion(const char *src, int len, bool is_client_to_server); static int cliplen(const char *str, int len, int limit); +pg_attribute_noreturn() +static void report_invalid_encoding_int(int encoding, const char *mbstr, + int mblen, int len); + +pg_attribute_noreturn() +static void report_invalid_encoding_db(const char *mbstr, int mblen, int len); + /* * Prepare for a future call to SetClientEncoding. Success should mean @@ -961,11 +969,126 @@ pg_encoding_wchar2mb_with_len(int encoding, return pg_wchar_table[encoding].wchar2mb_with_len(from, (unsigned char *) to, len); } -/* returns the byte length of a multibyte character */ +/* + * Returns the byte length of a multibyte character sequence in a + * null-terminated string. Raises an illegal byte sequence error if the + * sequence would hit a null terminator. + * + * The caller is expected to have checked for a terminator at *mbstr == 0 + * before calling, but some callers want 1 in that case, so this function + * continues that tradition. + * + * This must only be used for strings that have a null-terminator to enable + * bounds detection. + */ +int +pg_mblen_cstr(const char *mbstr) +{ + int length = pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); + + /* + * The .mblen functions return 1 when given a pointer to a terminator. + * Some callers depend on that, so we tolerate it for now. Well-behaved + * callers check the leading byte for a terminator *before* calling. + */ + for (int i = 1; i < length; ++i) + if (unlikely(mbstr[i] == 0)) + report_invalid_encoding_db(mbstr, length, i); + + /* + * String should be NUL-terminated, but checking that would make typical + * callers O(N^2), tripling Valgrind check-world time. Unless + * VALGRIND_EXPENSIVE, check 1 byte after each actual character. (If we + * found a character, not a terminator, the next byte must be a terminator + * or the start of the next character.) If the caller iterates the whole + * string, the last call will diagnose a missing terminator. + */ + if (mbstr[0] != '\0') + { +#ifdef VALGRIND_EXPENSIVE + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, strlen(mbstr)); +#else + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr + length, 1); +#endif + } + + return length; +} + +/* + * Returns the byte length of a multibyte character sequence bounded by a range + * [mbstr, end) of at least one byte in size. Raises an illegal byte sequence + * error if the sequence would exceed the range. + */ +int +pg_mblen_range(const char *mbstr, const char *end) +{ + int length = pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); + + Assert(end > mbstr); +#ifdef VALGRIND_EXPENSIVE + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, end - mbstr); +#else + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, length); +#endif + + if (unlikely(mbstr + length > end)) + report_invalid_encoding_db(mbstr, length, end - mbstr); + + return length; +} + +/* + * Returns the byte length of a multibyte character sequence bounded by a range + * extending for 'limit' bytes, which must be at least one. Raises an illegal + * byte sequence error if the sequence would exceed the range. + */ +int +pg_mblen_with_len(const char *mbstr, int limit) +{ + int length = pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); + + Assert(limit >= 1); +#ifdef VALGRIND_EXPENSIVE + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, limit); +#else + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, length); +#endif + + if (unlikely(length > limit)) + report_invalid_encoding_db(mbstr, length, limit); + + return length; +} + + +/* + * Returns the length of a multibyte character sequence, without any + * validation of bounds. + * + * PLEASE NOTE: This function can only be used safely if the caller has + * already verified the input string, since otherwise there is a risk of + * overrunning the buffer if the string is invalid. A prior call to a + * pg_mbstrlen* function suffices. + */ +int +pg_mblen_unbounded(const char *mbstr) +{ + int length = pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); + + VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, length); + + return length; +} + +/* + * Historical name for pg_mblen_unbounded(). Should not be used and will be + * removed in a later version. + */ int pg_mblen(const char *mbstr) { - return pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); + return pg_mblen_unbounded(mbstr); } /* returns the display length of a multibyte character */ @@ -987,14 +1110,14 @@ pg_mbstrlen(const char *mbstr) while (*mbstr) { - mbstr += pg_mblen(mbstr); + mbstr += pg_mblen_cstr(mbstr); len++; } return len; } /* returns the length (counted in wchars) of a multibyte string - * (not necessarily NULL terminated) + * (stops at the first of "limit" or a NUL) */ int pg_mbstrlen_with_len(const char *mbstr, int limit) @@ -1007,7 +1130,7 @@ pg_mbstrlen_with_len(const char *mbstr, int limit) while (limit > 0 && *mbstr) { - int l = pg_mblen(mbstr); + int l = pg_mblen_with_len(mbstr, limit); limit -= l; mbstr += l; @@ -1077,7 +1200,7 @@ pg_mbcharcliplen(const char *mbstr, int len, int limit) while (len > 0 && *mbstr) { - l = pg_mblen(mbstr); + l = pg_mblen_with_len(mbstr, len); nch++; if (nch > limit) break; @@ -1647,12 +1770,19 @@ void report_invalid_encoding(int encoding, const char *mbstr, int len) { int l = pg_encoding_mblen_or_incomplete(encoding, mbstr, len); + + report_invalid_encoding_int(encoding, mbstr, l, len); +} + +static void +report_invalid_encoding_int(int encoding, const char *mbstr, int mblen, int len) +{ char buf[8 * 5 + 1]; char *p = buf; int j, jlimit; - jlimit = Min(l, len); + jlimit = Min(mblen, len); jlimit = Min(jlimit, 8); /* prevent buffer overrun */ for (j = 0; j < jlimit; j++) @@ -1669,6 +1799,12 @@ report_invalid_encoding(int encoding, const char *mbstr, int len) buf))); } +static void +report_invalid_encoding_db(const char *mbstr, int mblen, int len) +{ + report_invalid_encoding_int(GetDatabaseEncoding(), mbstr, mblen, len); +} + /* * report_untranslatable_char: complain about untranslatable character * diff --git a/src/include/mb/pg_wchar.h b/src/include/mb/pg_wchar.h index 45ef9cbba70c9..0412f3886e587 100644 --- a/src/include/mb/pg_wchar.h +++ b/src/include/mb/pg_wchar.h @@ -608,7 +608,14 @@ extern int pg_char_and_wchar_strcmp(const char *s1, const pg_wchar *s2); extern int pg_wchar_strncmp(const pg_wchar *s1, const pg_wchar *s2, size_t n); extern int pg_char_and_wchar_strncmp(const char *s1, const pg_wchar *s2, size_t n); extern size_t pg_wchar_strlen(const pg_wchar *wstr); +extern int pg_mblen_cstr(const char *mbstr); +extern int pg_mblen_range(const char *mbstr, const char *end); +extern int pg_mblen_with_len(const char *mbstr, int limit); +extern int pg_mblen_unbounded(const char *mbstr); + +/* deprecated */ extern int pg_mblen(const char *mbstr); + extern int pg_dsplen(const char *mbstr); extern int pg_mbstrlen(const char *mbstr); extern int pg_mbstrlen_with_len(const char *mbstr, int len); diff --git a/src/include/tsearch/ts_locale.h b/src/include/tsearch/ts_locale.h index 7d7c4e16c621c..1a1e713a89207 100644 --- a/src/include/tsearch/ts_locale.h +++ b/src/include/tsearch/ts_locale.h @@ -45,12 +45,36 @@ typedef struct /* The second argument of t_iseq() must be a plain ASCII character */ #define t_iseq(x,c) (TOUCHAR(x) == (unsigned char) (c)) -#define COPYCHAR(d,s) memcpy(d, s, pg_mblen(s)) +/* Copy multibyte character of known byte length, return byte length. */ +static inline int +ts_copychar_with_len(void *dest, const void *src, int length) +{ + memcpy(dest, src, length); + return length; +} + +/* Copy multibyte character from null-terminated string, return byte length. */ +static inline int +ts_copychar_cstr(void *dest, const void *src) +{ + return ts_copychar_with_len(dest, src, pg_mblen_cstr((const char *) src)); +} + +/* Historical macro for the above. */ +#define COPYCHAR ts_copychar_cstr + +#define GENERATE_T_ISCLASS_DECL(character_class) \ +extern int t_is##character_class##_with_len(const char *ptr, int len); \ +extern int t_is##character_class##_cstr(const char *ptr); \ +extern int t_is##character_class##_unbounded(const char *ptr); \ +\ +/* deprecated */ \ +extern int t_is##character_class(const char *ptr); -extern int t_isdigit(const char *ptr); -extern int t_isspace(const char *ptr); -extern int t_isalpha(const char *ptr); -extern int t_isprint(const char *ptr); +GENERATE_T_ISCLASS_DECL(alpha); +GENERATE_T_ISCLASS_DECL(digit); +GENERATE_T_ISCLASS_DECL(print); +GENERATE_T_ISCLASS_DECL(space); extern char *lowerstr(const char *str); extern char *lowerstr_with_len(const char *str, int len); diff --git a/src/include/tsearch/ts_utils.h b/src/include/tsearch/ts_utils.h index c36c711dae0d1..94ca47588ff9e 100644 --- a/src/include/tsearch/ts_utils.h +++ b/src/include/tsearch/ts_utils.h @@ -38,14 +38,12 @@ extern bool gettoken_tsvector(TSVectorParseState state, extern void close_tsvector_parser(TSVectorParseState state); /* phrase operator begins with '<' */ -#define ISOPERATOR(x) \ - ( pg_mblen(x) == 1 && ( *(x) == '!' || \ - *(x) == '&' || \ - *(x) == '|' || \ - *(x) == '(' || \ - *(x) == ')' || \ - *(x) == '<' \ - ) ) +#define ISOPERATOR(x) (*(x) == '!' || \ + *(x) == '&' || \ + *(x) == '|' || \ + *(x) == '(' || \ + *(x) == ')' || \ + *(x) == '<') /* parse_tsquery */ diff --git a/src/test/modules/test_regex/test_regex.c b/src/test/modules/test_regex/test_regex.c index e23a0bd0d7f5d..f7bf50b760916 100644 --- a/src/test/modules/test_regex/test_regex.c +++ b/src/test/modules/test_regex/test_regex.c @@ -424,7 +424,8 @@ parse_test_flags(test_re_flags *flags, text *opts) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("invalid regular expression test option: \"%.*s\"", - pg_mblen(opt_p + i), opt_p + i))); + pg_mblen_range(opt_p + i, opt_p + opt_len), + opt_p + i))); break; } } From 7408df83bcbfa84eed23a65c6672d31757eabfbc Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Mon, 12 Jan 2026 10:20:06 +1300 Subject: [PATCH 04/11] Code coverage for most pg_mblen* calls. A security patch changed them today, so close the coverage gap now. Test that buffer overrun is avoided when pg_mblen*() requires more than the number of bytes remaining. This does not cover the calls in dict_thesaurus.c or in dict_synonym.c. That code is straightforward. To change that code's input, one must have access to modify installed OS files, so low-privilege users are not a threat. Testing this would likewise require changing installed share/postgresql/tsearch_data, which was enough of an obstacle to not bother. Security: CVE-2026-2006 Backpatch-through: 14 Co-authored-by: Thomas Munro Co-authored-by: Noah Misch Reviewed-by: Heikki Linnakangas (cherry picked from commit 757bf8145e243b4ad1a76460264f6f4df7e0fb1f) Conflicts resolved: - src/test/regress/parallel_schedule - parallel test group line "test: geometry horology ... database": - Upstream appends its two new tests, "encoding euc_kr", to the end of this group line. - YB f885e22323 ([#26335] YSQL: Import SetStats functions from upstream PG) appended "stats_import" to the end of the same line, so the two single-line edits overlap. - Keep both: the line now ends in "... database stats_import encoding euc_kr". - src/backend/utils/adt/arrayfuncs.c - construct_array_builtin(): - Upstream backpatches construct_array_builtin() to REL_15 for the new test code, with a case list omitting FLOAT8OID and XIDOID. - YB f885e22323 already imported construct_array_builtin() with FLOAT8OID and XIDOID cases, so git pairs the two function bodies and conflicts on those two extra cases. - Keep YB's superset: all upstream cases plus FLOAT8OID and XIDOID. --- contrib/pg_trgm/Makefile | 2 +- contrib/pg_trgm/data/trgm_utf8.data | 50 +++ contrib/pg_trgm/expected/pg_utf8_trgm.out | 8 + contrib/pg_trgm/expected/pg_utf8_trgm_1.out | 3 + contrib/pg_trgm/sql/pg_utf8_trgm.sql | 9 + src/backend/utils/adt/arrayfuncs.c | 75 ++++ src/include/utils/array.h | 3 + src/test/regress/expected/encoding.out | 401 ++++++++++++++++++++ src/test/regress/expected/encoding_1.out | 4 + src/test/regress/expected/euc_kr.out | 16 + src/test/regress/expected/euc_kr_1.out | 6 + src/test/regress/parallel_schedule | 2 +- src/test/regress/regress.c | 139 +++++++ src/test/regress/sql/encoding.sql | 228 +++++++++++ src/test/regress/sql/euc_kr.sql | 12 + 15 files changed, 956 insertions(+), 2 deletions(-) create mode 100644 contrib/pg_trgm/data/trgm_utf8.data create mode 100644 contrib/pg_trgm/expected/pg_utf8_trgm.out create mode 100644 contrib/pg_trgm/expected/pg_utf8_trgm_1.out create mode 100644 contrib/pg_trgm/sql/pg_utf8_trgm.sql create mode 100644 src/test/regress/expected/encoding.out create mode 100644 src/test/regress/expected/encoding_1.out create mode 100644 src/test/regress/expected/euc_kr.out create mode 100644 src/test/regress/expected/euc_kr_1.out create mode 100644 src/test/regress/sql/encoding.sql create mode 100644 src/test/regress/sql/euc_kr.sql diff --git a/contrib/pg_trgm/Makefile b/contrib/pg_trgm/Makefile index 1fbdc9ec1ef43..c1756993ec7ba 100644 --- a/contrib/pg_trgm/Makefile +++ b/contrib/pg_trgm/Makefile @@ -14,7 +14,7 @@ DATA = pg_trgm--1.5--1.6.sql pg_trgm--1.4--1.5.sql pg_trgm--1.3--1.4.sql \ pg_trgm--1.0--1.1.sql PGFILEDESC = "pg_trgm - trigram matching" -REGRESS = pg_trgm pg_word_trgm pg_strict_word_trgm +REGRESS = pg_trgm pg_utf8_trgm pg_word_trgm pg_strict_word_trgm ifdef USE_PGXS PG_CONFIG = pg_config diff --git a/contrib/pg_trgm/data/trgm_utf8.data b/contrib/pg_trgm/data/trgm_utf8.data new file mode 100644 index 0000000000000..713856e76a625 --- /dev/null +++ b/contrib/pg_trgm/data/trgm_utf8.data @@ -0,0 +1,50 @@ +Mathematics +数学 +गणित +Matemáticas +رياضيات +Mathématiques +গণিত +Matemática +Математика +ریاضی +Matematika +Mathematik +数学 +Mathematics +गणित +గణితం +Matematik +கணிதம் +數學 +Toán học +Matematika +数学 +수학 +ریاضی +Lissafi +Hisabati +Matematika +Matematica +ریاضی +ಗಣಿತ +ગણિત +คณิตศาสตร์ +ሂሳብ +गणित +ਗਣਿਤ +數學 +数学 +Iṣiro +數學 +သင်္ချာ +Herrega +رياضي +गणित +Математика +Matematyka +ഗണിതം +Matematika +رياضي +Matematika +Matematică diff --git a/contrib/pg_trgm/expected/pg_utf8_trgm.out b/contrib/pg_trgm/expected/pg_utf8_trgm.out new file mode 100644 index 0000000000000..0768e7d6a8320 --- /dev/null +++ b/contrib/pg_trgm/expected/pg_utf8_trgm.out @@ -0,0 +1,8 @@ +SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset +\if :skip_test +\quit +\endif +-- Index 50 translations of the word "Mathematics" +CREATE TEMP TABLE mb (s text); +\copy mb from 'data/trgm_utf8.data' +CREATE INDEX ON mb USING gist(s gist_trgm_ops); diff --git a/contrib/pg_trgm/expected/pg_utf8_trgm_1.out b/contrib/pg_trgm/expected/pg_utf8_trgm_1.out new file mode 100644 index 0000000000000..8505c4fa55262 --- /dev/null +++ b/contrib/pg_trgm/expected/pg_utf8_trgm_1.out @@ -0,0 +1,3 @@ +SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset +\if :skip_test +\quit diff --git a/contrib/pg_trgm/sql/pg_utf8_trgm.sql b/contrib/pg_trgm/sql/pg_utf8_trgm.sql new file mode 100644 index 0000000000000..0dd962ced8310 --- /dev/null +++ b/contrib/pg_trgm/sql/pg_utf8_trgm.sql @@ -0,0 +1,9 @@ +SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset +\if :skip_test +\quit +\endif + +-- Index 50 translations of the word "Mathematics" +CREATE TEMP TABLE mb (s text); +\copy mb from 'data/trgm_utf8.data' +CREATE INDEX ON mb USING gist(s gist_trgm_ops); diff --git a/src/backend/utils/adt/arrayfuncs.c b/src/backend/utils/adt/arrayfuncs.c index 648ab6b5fbca9..ec5049162f661 100644 --- a/src/backend/utils/adt/arrayfuncs.c +++ b/src/backend/utils/adt/arrayfuncs.c @@ -3700,6 +3700,81 @@ deconstruct_array(ArrayType *array, } } +/* + * Like deconstruct_array(), where elmtype must be a built-in type, and + * elmlen/elmbyval/elmalign is looked up from hardcoded data. This is often + * useful when manipulating arrays from/for system catalogs. + */ +void +deconstruct_array_builtin(ArrayType *array, + Oid elmtype, + Datum **elemsp, bool **nullsp, int *nelemsp) +{ + int elmlen; + bool elmbyval; + char elmalign; + + switch (elmtype) + { + case CHAROID: + elmlen = 1; + elmbyval = true; + elmalign = TYPALIGN_CHAR; + break; + + case CSTRINGOID: + elmlen = -2; + elmbyval = false; + elmalign = TYPALIGN_CHAR; + break; + + case FLOAT8OID: + elmlen = sizeof(float8); + elmbyval = FLOAT8PASSBYVAL; + elmalign = TYPALIGN_DOUBLE; + break; + + case INT2OID: + elmlen = sizeof(int16); + elmbyval = true; + elmalign = TYPALIGN_SHORT; + break; + + case INT4OID: + elmlen = sizeof(int32); + elmbyval = true; + elmalign = TYPALIGN_INT; + break; + + case OIDOID: + elmlen = sizeof(Oid); + elmbyval = true; + elmalign = TYPALIGN_INT; + break; + + case TEXTOID: + elmlen = -1; + elmbyval = false; + elmalign = TYPALIGN_INT; + break; + + case TIDOID: + elmlen = sizeof(ItemPointerData); + elmbyval = false; + elmalign = TYPALIGN_SHORT; + break; + + default: + elog(ERROR, "type %u not supported by deconstruct_array_builtin()", elmtype); + /* keep compiler quiet */ + elmlen = 0; + elmbyval = false; + elmalign = 0; + } + + deconstruct_array(array, elmtype, elmlen, elmbyval, elmalign, elemsp, nullsp, nelemsp); +} + /* * array_contains_nulls --- detect whether an array has any null elements * diff --git a/src/include/utils/array.h b/src/include/utils/array.h index 51295c6a1050d..6ef28dfcf3b9d 100644 --- a/src/include/utils/array.h +++ b/src/include/utils/array.h @@ -409,6 +409,9 @@ extern void deconstruct_array(ArrayType *array, Oid elmtype, int elmlen, bool elmbyval, char elmalign, Datum **elemsp, bool **nullsp, int *nelemsp); +extern void deconstruct_array_builtin(ArrayType *array, + Oid elmtype, + Datum **elemsp, bool **nullsp, int *nelemsp); extern bool array_contains_nulls(ArrayType *array); extern ArrayBuildState *initArrayResult(Oid element_type, diff --git a/src/test/regress/expected/encoding.out b/src/test/regress/expected/encoding.out new file mode 100644 index 0000000000000..ea1f38cff41d8 --- /dev/null +++ b/src/test/regress/expected/encoding.out @@ -0,0 +1,401 @@ +/* skip test if not UTF8 server encoding */ +SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset +\if :skip_test +\quit +\endif +\getenv libdir PG_LIBDIR +\getenv dlsuffix PG_DLSUFFIX +\set regresslib :libdir '/regress' :dlsuffix +CREATE FUNCTION test_bytea_to_text(bytea) RETURNS text + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_text_to_bytea(text) RETURNS bytea + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_mblen_func(text, text, text, int) RETURNS int + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_text_to_wchars(text, text) RETURNS int[] + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_wchars_to_text(text, int[]) RETURNS text + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_valid_server_encoding(text) RETURNS boolean + AS :'regresslib' LANGUAGE C STRICT; +CREATE TABLE regress_encoding(good text, truncated text, with_nul text, truncated_with_nul text); +INSERT INTO regress_encoding +VALUES ('café', + 'caf' || test_bytea_to_text('\xc3'), + 'café' || test_bytea_to_text('\x00') || 'dcba', + 'caf' || test_bytea_to_text('\xc300') || 'dcba'); +SELECT good, truncated, with_nul FROM regress_encoding; + good | truncated | with_nul +------+-----------+---------- + café | caf | café +(1 row) + +SELECT length(good) FROM regress_encoding; + length +-------- + 4 +(1 row) + +SELECT substring(good, 3, 1) FROM regress_encoding; + substring +----------- + f +(1 row) + +SELECT substring(good, 4, 1) FROM regress_encoding; + substring +----------- + é +(1 row) + +SELECT regexp_replace(good, '^caf(.)$', '\1') FROM regress_encoding; + regexp_replace +---------------- + é +(1 row) + +SELECT reverse(good) FROM regress_encoding; + reverse +--------- + éfac +(1 row) + +-- invalid short mb character = error +SELECT length(truncated) FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 +SELECT substring(truncated, 1, 1) FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 +SELECT reverse(truncated) FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 +-- invalid short mb character = silently dropped +SELECT regexp_replace(truncated, '^caf(.)$', '\1') FROM regress_encoding; + regexp_replace +---------------- + caf +(1 row) + +-- PostgreSQL doesn't allow strings to contain NUL. If a corrupted string +-- contains NUL at a character boundary position, some functions treat it as a +-- character while others treat it as a terminator, as implementation details. +-- NUL = terminator +SELECT length(with_nul) FROM regress_encoding; + length +-------- + 4 +(1 row) + +SELECT substring(with_nul, 3, 1) FROM regress_encoding; + substring +----------- + f +(1 row) + +SELECT substring(with_nul, 4, 1) FROM regress_encoding; + substring +----------- + é +(1 row) + +SELECT substring(with_nul, 5, 1) FROM regress_encoding; + substring +----------- + +(1 row) + +SELECT convert_to(substring(with_nul, 5, 1), 'UTF8') FROM regress_encoding; + convert_to +------------ + \x +(1 row) + +SELECT regexp_replace(with_nul, '^caf(.)$', '\1') FROM regress_encoding; + regexp_replace +---------------- + é +(1 row) + +-- NUL = character +SELECT with_nul, reverse(with_nul), reverse(reverse(with_nul)) FROM regress_encoding; + with_nul | reverse | reverse +----------+---------+--------- + café | abcd | café +(1 row) + +-- If a corrupted string contains NUL in the tail bytes of a multibyte +-- character (invalid in all encodings), it is considered part of the +-- character for length purposes. An error will only be raised in code paths +-- that convert or verify encodings. +SELECT length(truncated_with_nul) FROM regress_encoding; + length +-------- + 8 +(1 row) + +SELECT substring(truncated_with_nul, 3, 1) FROM regress_encoding; + substring +----------- + f +(1 row) + +SELECT substring(truncated_with_nul, 4, 1) FROM regress_encoding; + substring +----------- + +(1 row) + +SELECT convert_to(substring(truncated_with_nul, 4, 1), 'UTF8') FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 0x00 +SELECT substring(truncated_with_nul, 5, 1) FROM regress_encoding; + substring +----------- + d +(1 row) + +SELECT regexp_replace(truncated_with_nul, '^caf(.)dcba$', '\1') = test_bytea_to_text('\xc300') FROM regress_encoding; + ?column? +---------- + t +(1 row) + +SELECT reverse(truncated_with_nul) FROM regress_encoding; + reverse +--------- + abcd +(1 row) + +-- unbounded: sequence would overrun the string! +SELECT test_mblen_func('pg_mblen_unbounded', 'UTF8', truncated, 3) +FROM regress_encoding; + test_mblen_func +----------------- + 2 +(1 row) + +-- condition detected when using the length/range variants +SELECT test_mblen_func('pg_mblen_with_len', 'UTF8', truncated, 3) +FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 +SELECT test_mblen_func('pg_mblen_range', 'UTF8', truncated, 3) +FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 +-- unbounded: sequence would overrun the string, if the terminator were really +-- the end of it +SELECT test_mblen_func('pg_mblen_unbounded', 'UTF8', truncated_with_nul, 3) +FROM regress_encoding; + test_mblen_func +----------------- + 2 +(1 row) + +SELECT test_mblen_func('pg_encoding_mblen', 'GB18030', truncated_with_nul, 3) +FROM regress_encoding; + test_mblen_func +----------------- + 2 +(1 row) + +-- condition detected when using the cstr variants +SELECT test_mblen_func('pg_mblen_cstr', 'UTF8', truncated_with_nul, 3) +FROM regress_encoding; +ERROR: invalid byte sequence for encoding "UTF8": 0xc3 +DROP TABLE regress_encoding; +-- mb<->wchar conversions +CREATE FUNCTION test_encoding(encoding text, description text, input bytea) +RETURNS VOID LANGUAGE plpgsql AS +$$ +DECLARE + prefix text; + len int; + wchars int[]; + round_trip bytea; + result text; +BEGIN + prefix := rpad(encoding || ' ' || description || ':', 28); + + -- XXX could also test validation, length functions and include client + -- only encodings with these test cases + + IF test_valid_server_encoding(encoding) THEN + wchars := test_text_to_wchars(encoding, test_bytea_to_text(input)); + round_trip = test_text_to_bytea(test_wchars_to_text(encoding, wchars)); + if input = round_trip then + result := 'OK'; + elsif length(input) > length(round_trip) and round_trip = substr(input, 1, length(round_trip)) then + result := 'truncated'; + else + result := 'failed'; + end if; + RAISE NOTICE '% % -> % -> % = %', prefix, input, wchars, round_trip, result; + END IF; +END; +$$; +-- No validation is done on the encoding itself, just the length to avoid +-- overruns, so some of the byte sequences below are bogus. They cover +-- all code branches, server encodings only for now. +CREATE TABLE encoding_tests (encoding text, description text, input bytea); +INSERT INTO encoding_tests VALUES + -- LATIN1, other single-byte encodings + ('LATIN1', 'ASCII', 'a'), + ('LATIN1', 'extended', '\xe9'), + -- EUC_JP, EUC_JIS_2004, EUR_KR (for the purposes of wchar conversion): + -- 2 8e (CS2, not used by EUR_KR but arbitrarily considered to have EUC_JP length) + -- 3 8f (CS3, not used by EUR_KR but arbitrarily considered to have EUC_JP length) + -- 2 80..ff (CS1) + ('EUC_JP', 'ASCII', 'a'), + ('EUC_JP', 'CS1, short', '\x80'), + ('EUC_JP', 'CS1', '\x8002'), + ('EUC_JP', 'CS2, short', '\x8e'), + ('EUC_JP', 'CS2', '\x8e02'), + ('EUC_JP', 'CS3, short', '\x8f'), + ('EUC_JP', 'CS3, short', '\x8f02'), + ('EUC_JP', 'CS3', '\x8f0203'), + -- EUC_CN + -- 3 8e (CS2, not used but arbitrarily considered to have length 3) + -- 3 8f (CS3, not used but arbitrarily considered to have length 3) + -- 2 80..ff (CS1) + ('EUC_CN', 'ASCII', 'a'), + ('EUC_CN', 'CS1, short', '\x80'), + ('EUC_CN', 'CS1', '\x8002'), + ('EUC_CN', 'CS2, short', '\x8e'), + ('EUC_CN', 'CS2, short', '\x8e02'), + ('EUC_CN', 'CS2', '\x8e0203'), + ('EUC_CN', 'CS3, short', '\x8f'), + ('EUC_CN', 'CS3, short', '\x8f02'), + ('EUC_CN', 'CS3', '\x8f0203'), + -- EUC_TW: + -- 4 8e (CS2) + -- 3 8f (CS3, not used but arbitrarily considered to have length 3) + -- 2 80..ff (CS1) + ('EUC_TW', 'ASCII', 'a'), + ('EUC_TW', 'CS1, short', '\x80'), + ('EUC_TW', 'CS1', '\x8002'), + ('EUC_TW', 'CS2, short', '\x8e'), + ('EUC_TW', 'CS2, short', '\x8e02'), + ('EUC_TW', 'CS2, short', '\x8e0203'), + ('EUC_TW', 'CS2', '\x8e020304'), + ('EUC_TW', 'CS3, short', '\x8f'), + ('EUC_TW', 'CS3, short', '\x8f02'), + ('EUC_TW', 'CS3', '\x8f0203'), + -- UTF8 + -- 2 c0..df + -- 3 e0..ef + -- 4 f0..f7 (but maximum real codepoint U+10ffff has f4) + -- 5 f8..fb (not supported) + -- 6 fc..fd (not supported) + ('UTF8', 'ASCII', 'a'), + ('UTF8', '2 byte, short', '\xdf'), + ('UTF8', '2 byte', '\xdf82'), + ('UTF8', '3 byte, short', '\xef'), + ('UTF8', '3 byte, short', '\xef82'), + ('UTF8', '3 byte', '\xef8283'), + ('UTF8', '4 byte, short', '\xf7'), + ('UTF8', '4 byte, short', '\xf782'), + ('UTF8', '4 byte, short', '\xf78283'), + ('UTF8', '4 byte', '\xf7828384'), + ('UTF8', '5 byte, unsupported', '\xfb'), + ('UTF8', '5 byte, unsupported', '\xfb82'), + ('UTF8', '5 byte, unsupported', '\xfb8283'), + ('UTF8', '5 byte, unsupported', '\xfb828384'), + ('UTF8', '5 byte, unsupported', '\xfb82838485'), + ('UTF8', '6 byte, unsupported', '\xfd'), + ('UTF8', '6 byte, unsupported', '\xfd82'), + ('UTF8', '6 byte, unsupported', '\xfd8283'), + ('UTF8', '6 byte, unsupported', '\xfd828384'), + ('UTF8', '6 byte, unsupported', '\xfd82838485'), + ('UTF8', '6 byte, unsupported', '\xfd8283848586'), + -- MULE_INTERNAL + -- 2 81..8d LC1 + -- 3 90..99 LC2 + ('MULE_INTERNAL', 'ASCII', 'a'), + ('MULE_INTERNAL', 'LC1, short', '\x81'), + ('MULE_INTERNAL', 'LC1', '\x8182'), + ('MULE_INTERNAL', 'LC2, short', '\x90'), + ('MULE_INTERNAL', 'LC2, short', '\x9082'), + ('MULE_INTERNAL', 'LC2', '\x908283'); +SELECT COUNT(test_encoding(encoding, description, input)) > 0 +FROM encoding_tests; +NOTICE: LATIN1 ASCII: \x61 -> {97} -> \x61 = OK +NOTICE: LATIN1 extended: \xe9 -> {233} -> \xe9 = OK +NOTICE: EUC_JP ASCII: \x61 -> {97} -> \x61 = OK +NOTICE: EUC_JP CS1, short: \x80 -> {} -> \x = truncated +NOTICE: EUC_JP CS1: \x8002 -> {32770} -> \x8002 = OK +NOTICE: EUC_JP CS2, short: \x8e -> {} -> \x = truncated +NOTICE: EUC_JP CS2: \x8e02 -> {36354} -> \x8e02 = OK +NOTICE: EUC_JP CS3, short: \x8f -> {} -> \x = truncated +NOTICE: EUC_JP CS3, short: \x8f02 -> {} -> \x = truncated +NOTICE: EUC_JP CS3: \x8f0203 -> {9372163} -> \x8f0203 = OK +NOTICE: EUC_CN ASCII: \x61 -> {97} -> \x61 = OK +NOTICE: EUC_CN CS1, short: \x80 -> {} -> \x = truncated +NOTICE: EUC_CN CS1: \x8002 -> {32770} -> \x8002 = OK +NOTICE: EUC_CN CS2, short: \x8e -> {} -> \x = truncated +NOTICE: EUC_CN CS2, short: \x8e02 -> {} -> \x = truncated +NOTICE: EUC_CN CS2: \x8e0203 -> {9306627} -> \x8e0203 = OK +NOTICE: EUC_CN CS3, short: \x8f -> {} -> \x = truncated +NOTICE: EUC_CN CS3, short: \x8f02 -> {} -> \x = truncated +NOTICE: EUC_CN CS3: \x8f0203 -> {9372163} -> \x8f0203 = OK +NOTICE: EUC_TW ASCII: \x61 -> {97} -> \x61 = OK +NOTICE: EUC_TW CS1, short: \x80 -> {} -> \x = truncated +NOTICE: EUC_TW CS1: \x8002 -> {32770} -> \x8002 = OK +NOTICE: EUC_TW CS2, short: \x8e -> {} -> \x = truncated +NOTICE: EUC_TW CS2, short: \x8e02 -> {} -> \x = truncated +NOTICE: EUC_TW CS2, short: \x8e0203 -> {} -> \x = truncated +NOTICE: EUC_TW CS2: \x8e020304 -> {-1912470780} -> \x8e020304 = OK +NOTICE: EUC_TW CS3, short: \x8f -> {} -> \x = truncated +NOTICE: EUC_TW CS3, short: \x8f02 -> {} -> \x = truncated +NOTICE: EUC_TW CS3: \x8f0203 -> {9372163} -> \x8f0203 = OK +NOTICE: UTF8 ASCII: \x61 -> {97} -> \x61 = OK +NOTICE: UTF8 2 byte, short: \xdf -> {} -> \x = truncated +NOTICE: UTF8 2 byte: \xdf82 -> {1986} -> \xdf82 = OK +NOTICE: UTF8 3 byte, short: \xef -> {} -> \x = truncated +NOTICE: UTF8 3 byte, short: \xef82 -> {} -> \x = truncated +NOTICE: UTF8 3 byte: \xef8283 -> {61571} -> \xef8283 = OK +NOTICE: UTF8 4 byte, short: \xf7 -> {} -> \x = truncated +NOTICE: UTF8 4 byte, short: \xf782 -> {} -> \x = truncated +NOTICE: UTF8 4 byte, short: \xf78283 -> {} -> \x = truncated +NOTICE: UTF8 4 byte: \xf7828384 -> {1843396} -> \xf7828384 = OK +NOTICE: UTF8 5 byte, unsupported: \xfb -> {251} -> \xc3bb = failed +NOTICE: UTF8 5 byte, unsupported: \xfb82 -> {251,130} -> \xc3bbc282 = failed +NOTICE: UTF8 5 byte, unsupported: \xfb8283 -> {251,130,131} -> \xc3bbc282c283 = failed +NOTICE: UTF8 5 byte, unsupported: \xfb828384 -> {251,130,131,132} -> \xc3bbc282c283c284 = failed +NOTICE: UTF8 5 byte, unsupported: \xfb82838485 -> {251,130,131,132,133} -> \xc3bbc282c283c284c285 = failed +NOTICE: UTF8 6 byte, unsupported: \xfd -> {253} -> \xc3bd = failed +NOTICE: UTF8 6 byte, unsupported: \xfd82 -> {253,130} -> \xc3bdc282 = failed +NOTICE: UTF8 6 byte, unsupported: \xfd8283 -> {253,130,131} -> \xc3bdc282c283 = failed +NOTICE: UTF8 6 byte, unsupported: \xfd828384 -> {253,130,131,132} -> \xc3bdc282c283c284 = failed +NOTICE: UTF8 6 byte, unsupported: \xfd82838485 -> {253,130,131,132,133} -> \xc3bdc282c283c284c285 = failed +NOTICE: UTF8 6 byte, unsupported: \xfd8283848586 -> {253,130,131,132,133,134} -> \xc3bdc282c283c284c285c286 = failed +NOTICE: MULE_INTERNAL ASCII: \x61 -> {97} -> \x61 = OK +NOTICE: MULE_INTERNAL LC1, short: \x81 -> {} -> \x = truncated +NOTICE: MULE_INTERNAL LC1: \x8182 -> {8454274} -> \x8182 = OK +NOTICE: MULE_INTERNAL LC2, short: \x90 -> {} -> \x = truncated +NOTICE: MULE_INTERNAL LC2, short: \x9082 -> {} -> \x = truncated +NOTICE: MULE_INTERNAL LC2: \x908283 -> {9470595} -> \x908283 = OK + ?column? +---------- + t +(1 row) + +DROP TABLE encoding_tests; +DROP FUNCTION test_encoding; +DROP FUNCTION test_text_to_wchars; +DROP FUNCTION test_mblen_func; +DROP FUNCTION test_bytea_to_text; +DROP FUNCTION test_text_to_bytea; +-- substring slow path: multi-byte escape char vs. multi-byte pattern char. +SELECT SUBSTRING('a' SIMILAR U&'\00AC' ESCAPE U&'\00A7'); + substring +----------- + +(1 row) + +-- Levenshtein distance metric: exercise character length cache. +SELECT U&"real\00A7_name" FROM (select 1) AS x(real_name); +ERROR: column "real§_name" does not exist +LINE 1: SELECT U&"real\00A7_name" FROM (select 1) AS x(real_name); + ^ +HINT: Perhaps you meant to reference the column "x.real_name". +-- JSON errcontext: truncate long data. +SELECT repeat(U&'\00A7', 30)::json; +ERROR: invalid input syntax for type json +DETAIL: Token "§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§" is invalid. +CONTEXT: JSON data, line 1: ...§§§§§§§§§§§§§§§§§§§§§§§§ diff --git a/src/test/regress/expected/encoding_1.out b/src/test/regress/expected/encoding_1.out new file mode 100644 index 0000000000000..a5b02090901b3 --- /dev/null +++ b/src/test/regress/expected/encoding_1.out @@ -0,0 +1,4 @@ +/* skip test if not UTF8 server encoding */ +SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset +\if :skip_test +\quit diff --git a/src/test/regress/expected/euc_kr.out b/src/test/regress/expected/euc_kr.out new file mode 100644 index 0000000000000..7a61c89a43ad9 --- /dev/null +++ b/src/test/regress/expected/euc_kr.out @@ -0,0 +1,16 @@ +-- This test is about EUC_KR encoding, chosen as perhaps the most prevalent +-- non-UTF8, multibyte encoding as of 2026-01. Since UTF8 can represent all +-- of EUC_KR, also run the test in UTF8. +SELECT getdatabaseencoding() NOT IN ('EUC_KR', 'UTF8') AS skip_test \gset +\if :skip_test +\quit +\endif +-- Exercise is_multibyte_char_in_char (non-UTF8) slow path. +SELECT POSITION( + convert_from('\xbcf6c7d0', 'EUC_KR') IN + convert_from('\xb0fac7d02c20bcf6c7d02c20b1e2bcfa2c20bbee', 'EUC_KR')); + position +---------- + 5 +(1 row) + diff --git a/src/test/regress/expected/euc_kr_1.out b/src/test/regress/expected/euc_kr_1.out new file mode 100644 index 0000000000000..faaac5d635524 --- /dev/null +++ b/src/test/regress/expected/euc_kr_1.out @@ -0,0 +1,6 @@ +-- This test is about EUC_KR encoding, chosen as perhaps the most prevalent +-- non-UTF8, multibyte encoding as of 2026-01. Since UTF8 can represent all +-- of EUC_KR, also run the test in UTF8. +SELECT getdatabaseencoding() NOT IN ('EUC_KR', 'UTF8') AS skip_test \gset +\if :skip_test +\quit diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule index 27c7863ee821d..def499bb43c3b 100644 --- a/src/test/regress/parallel_schedule +++ b/src/test/regress/parallel_schedule @@ -33,7 +33,7 @@ test: strings numerology point lseg line box path polygon circle date time timet # geometry depends on point, lseg, line, box, path, polygon, circle # horology depends on date, time, timetz, timestamp, timestamptz, interval # ---------- -test: geometry horology tstypes regex type_sanity opr_sanity misc_sanity comments expressions unicode xid mvcc database stats_import +test: geometry horology tstypes regex type_sanity opr_sanity misc_sanity comments expressions unicode xid mvcc database stats_import encoding euc_kr # ---------- # Load huge amounts of data diff --git a/src/test/regress/regress.c b/src/test/regress/regress.c index 5c0ecbace42a2..60378f643446f 100644 --- a/src/test/regress/regress.c +++ b/src/test/regress/regress.c @@ -1280,6 +1280,145 @@ test_enc_conversion(PG_FUNCTION_ARGS) PG_RETURN_DATUM(HeapTupleGetDatum(tuple)); } +/* Convert bytea to text without validation for corruption tests from SQL. */ +PG_FUNCTION_INFO_V1(test_bytea_to_text); +Datum +test_bytea_to_text(PG_FUNCTION_ARGS) +{ + PG_RETURN_TEXT_P(PG_GETARG_BYTEA_PP(0)); +} + +/* And the reverse. */ +PG_FUNCTION_INFO_V1(test_text_to_bytea); +Datum +test_text_to_bytea(PG_FUNCTION_ARGS) +{ + PG_RETURN_BYTEA_P(PG_GETARG_TEXT_PP(0)); +} + +/* Corruption tests in C. */ +PG_FUNCTION_INFO_V1(test_mblen_func); +Datum +test_mblen_func(PG_FUNCTION_ARGS) +{ + const char *func = text_to_cstring(PG_GETARG_BYTEA_PP(0)); + const char *encoding = text_to_cstring(PG_GETARG_BYTEA_PP(1)); + text *string = PG_GETARG_BYTEA_PP(2); + int offset = PG_GETARG_INT32(3); + const char *data = VARDATA_ANY(string); + size_t size = VARSIZE_ANY_EXHDR(string); + int result = 0; + + if (strcmp(func, "pg_mblen_unbounded") == 0) + result = pg_mblen_unbounded(data + offset); + else if (strcmp(func, "pg_mblen_cstr") == 0) + result = pg_mblen_cstr(data + offset); + else if (strcmp(func, "pg_mblen_with_len") == 0) + result = pg_mblen_with_len(data + offset, size - offset); + else if (strcmp(func, "pg_mblen_range") == 0) + result = pg_mblen_range(data + offset, data + size); + else if (strcmp(func, "pg_encoding_mblen") == 0) + result = pg_encoding_mblen(pg_char_to_encoding(encoding), data + offset); + else + elog(ERROR, "unknown function"); + + PG_RETURN_INT32(result); +} + +PG_FUNCTION_INFO_V1(test_text_to_wchars); +Datum +test_text_to_wchars(PG_FUNCTION_ARGS) +{ + const char *encoding_name = text_to_cstring(PG_GETARG_BYTEA_PP(0)); + text *string = PG_GETARG_TEXT_PP(1); + const char *data = VARDATA_ANY(string); + size_t size = VARSIZE_ANY_EXHDR(string); + pg_wchar *wchars = palloc(sizeof(pg_wchar) * (size + 1)); + Datum *datums; + int wlen; + int encoding; + + encoding = pg_char_to_encoding(encoding_name); + if (encoding < 0) + elog(ERROR, "unknown encoding name: %s", encoding_name); + + if (size > 0) + { + datums = palloc(sizeof(Datum) * size); + wlen = pg_encoding_mb2wchar_with_len(encoding, + data, + wchars, + size); + Assert(wlen >= 0); + Assert(wlen <= size); + Assert(wchars[wlen] == 0); + + for (int i = 0; i < wlen; ++i) + datums[i] = UInt32GetDatum(wchars[i]); + } + else + { + datums = NULL; + wlen = 0; + } + + PG_RETURN_ARRAYTYPE_P(construct_array_builtin(datums, wlen, INT4OID)); +} + +PG_FUNCTION_INFO_V1(test_wchars_to_text); +Datum +test_wchars_to_text(PG_FUNCTION_ARGS) +{ + const char *encoding_name = text_to_cstring(PG_GETARG_BYTEA_PP(0)); + ArrayType *array = PG_GETARG_ARRAYTYPE_P(1); + Datum *datums; + bool *nulls; + char *mb; + text *result; + int wlen; + int bytes; + int encoding; + + encoding = pg_char_to_encoding(encoding_name); + if (encoding < 0) + elog(ERROR, "unknown encoding name: %s", encoding_name); + + deconstruct_array_builtin(array, INT4OID, &datums, &nulls, &wlen); + + if (wlen > 0) + { + pg_wchar *wchars = palloc(sizeof(pg_wchar) * wlen); + + for (int i = 0; i < wlen; ++i) + { + if (nulls[i]) + elog(ERROR, "unexpected NULL in array"); + wchars[i] = DatumGetInt32(datums[i]); + } + + mb = palloc(pg_encoding_max_length(encoding) * wlen + 1); + bytes = pg_encoding_wchar2mb_with_len(encoding, wchars, mb, wlen); + } + else + { + mb = ""; + bytes = 0; + } + + result = palloc(bytes + VARHDRSZ); + SET_VARSIZE(result, bytes + VARHDRSZ); + memcpy(VARDATA(result), mb, bytes); + + PG_RETURN_TEXT_P(result); +} + +PG_FUNCTION_INFO_V1(test_valid_server_encoding); +Datum +test_valid_server_encoding(PG_FUNCTION_ARGS) +{ + return pg_valid_server_encoding(text_to_cstring(PG_GETARG_TEXT_PP(0))); +} + /* Provide SQL access to IsBinaryCoercible() */ PG_FUNCTION_INFO_V1(binary_coercible); Datum diff --git a/src/test/regress/sql/encoding.sql b/src/test/regress/sql/encoding.sql new file mode 100644 index 0000000000000..b9543c0cb3262 --- /dev/null +++ b/src/test/regress/sql/encoding.sql @@ -0,0 +1,228 @@ +/* skip test if not UTF8 server encoding */ +SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset +\if :skip_test +\quit +\endif + +\getenv libdir PG_LIBDIR +\getenv dlsuffix PG_DLSUFFIX + +\set regresslib :libdir '/regress' :dlsuffix + +CREATE FUNCTION test_bytea_to_text(bytea) RETURNS text + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_text_to_bytea(text) RETURNS bytea + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_mblen_func(text, text, text, int) RETURNS int + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_text_to_wchars(text, text) RETURNS int[] + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_wchars_to_text(text, int[]) RETURNS text + AS :'regresslib' LANGUAGE C STRICT; +CREATE FUNCTION test_valid_server_encoding(text) RETURNS boolean + AS :'regresslib' LANGUAGE C STRICT; + + +CREATE TABLE regress_encoding(good text, truncated text, with_nul text, truncated_with_nul text); +INSERT INTO regress_encoding +VALUES ('café', + 'caf' || test_bytea_to_text('\xc3'), + 'café' || test_bytea_to_text('\x00') || 'dcba', + 'caf' || test_bytea_to_text('\xc300') || 'dcba'); + +SELECT good, truncated, with_nul FROM regress_encoding; + +SELECT length(good) FROM regress_encoding; +SELECT substring(good, 3, 1) FROM regress_encoding; +SELECT substring(good, 4, 1) FROM regress_encoding; +SELECT regexp_replace(good, '^caf(.)$', '\1') FROM regress_encoding; +SELECT reverse(good) FROM regress_encoding; + +-- invalid short mb character = error +SELECT length(truncated) FROM regress_encoding; +SELECT substring(truncated, 1, 1) FROM regress_encoding; +SELECT reverse(truncated) FROM regress_encoding; +-- invalid short mb character = silently dropped +SELECT regexp_replace(truncated, '^caf(.)$', '\1') FROM regress_encoding; + +-- PostgreSQL doesn't allow strings to contain NUL. If a corrupted string +-- contains NUL at a character boundary position, some functions treat it as a +-- character while others treat it as a terminator, as implementation details. + +-- NUL = terminator +SELECT length(with_nul) FROM regress_encoding; +SELECT substring(with_nul, 3, 1) FROM regress_encoding; +SELECT substring(with_nul, 4, 1) FROM regress_encoding; +SELECT substring(with_nul, 5, 1) FROM regress_encoding; +SELECT convert_to(substring(with_nul, 5, 1), 'UTF8') FROM regress_encoding; +SELECT regexp_replace(with_nul, '^caf(.)$', '\1') FROM regress_encoding; +-- NUL = character +SELECT with_nul, reverse(with_nul), reverse(reverse(with_nul)) FROM regress_encoding; + +-- If a corrupted string contains NUL in the tail bytes of a multibyte +-- character (invalid in all encodings), it is considered part of the +-- character for length purposes. An error will only be raised in code paths +-- that convert or verify encodings. + +SELECT length(truncated_with_nul) FROM regress_encoding; +SELECT substring(truncated_with_nul, 3, 1) FROM regress_encoding; +SELECT substring(truncated_with_nul, 4, 1) FROM regress_encoding; +SELECT convert_to(substring(truncated_with_nul, 4, 1), 'UTF8') FROM regress_encoding; +SELECT substring(truncated_with_nul, 5, 1) FROM regress_encoding; +SELECT regexp_replace(truncated_with_nul, '^caf(.)dcba$', '\1') = test_bytea_to_text('\xc300') FROM regress_encoding; +SELECT reverse(truncated_with_nul) FROM regress_encoding; + +-- unbounded: sequence would overrun the string! +SELECT test_mblen_func('pg_mblen_unbounded', 'UTF8', truncated, 3) +FROM regress_encoding; + +-- condition detected when using the length/range variants +SELECT test_mblen_func('pg_mblen_with_len', 'UTF8', truncated, 3) +FROM regress_encoding; +SELECT test_mblen_func('pg_mblen_range', 'UTF8', truncated, 3) +FROM regress_encoding; + +-- unbounded: sequence would overrun the string, if the terminator were really +-- the end of it +SELECT test_mblen_func('pg_mblen_unbounded', 'UTF8', truncated_with_nul, 3) +FROM regress_encoding; +SELECT test_mblen_func('pg_encoding_mblen', 'GB18030', truncated_with_nul, 3) +FROM regress_encoding; + +-- condition detected when using the cstr variants +SELECT test_mblen_func('pg_mblen_cstr', 'UTF8', truncated_with_nul, 3) +FROM regress_encoding; + +DROP TABLE regress_encoding; + +-- mb<->wchar conversions +CREATE FUNCTION test_encoding(encoding text, description text, input bytea) +RETURNS VOID LANGUAGE plpgsql AS +$$ +DECLARE + prefix text; + len int; + wchars int[]; + round_trip bytea; + result text; +BEGIN + prefix := rpad(encoding || ' ' || description || ':', 28); + + -- XXX could also test validation, length functions and include client + -- only encodings with these test cases + + IF test_valid_server_encoding(encoding) THEN + wchars := test_text_to_wchars(encoding, test_bytea_to_text(input)); + round_trip = test_text_to_bytea(test_wchars_to_text(encoding, wchars)); + if input = round_trip then + result := 'OK'; + elsif length(input) > length(round_trip) and round_trip = substr(input, 1, length(round_trip)) then + result := 'truncated'; + else + result := 'failed'; + end if; + RAISE NOTICE '% % -> % -> % = %', prefix, input, wchars, round_trip, result; + END IF; +END; +$$; +-- No validation is done on the encoding itself, just the length to avoid +-- overruns, so some of the byte sequences below are bogus. They cover +-- all code branches, server encodings only for now. +CREATE TABLE encoding_tests (encoding text, description text, input bytea); +INSERT INTO encoding_tests VALUES + -- LATIN1, other single-byte encodings + ('LATIN1', 'ASCII', 'a'), + ('LATIN1', 'extended', '\xe9'), + -- EUC_JP, EUC_JIS_2004, EUR_KR (for the purposes of wchar conversion): + -- 2 8e (CS2, not used by EUR_KR but arbitrarily considered to have EUC_JP length) + -- 3 8f (CS3, not used by EUR_KR but arbitrarily considered to have EUC_JP length) + -- 2 80..ff (CS1) + ('EUC_JP', 'ASCII', 'a'), + ('EUC_JP', 'CS1, short', '\x80'), + ('EUC_JP', 'CS1', '\x8002'), + ('EUC_JP', 'CS2, short', '\x8e'), + ('EUC_JP', 'CS2', '\x8e02'), + ('EUC_JP', 'CS3, short', '\x8f'), + ('EUC_JP', 'CS3, short', '\x8f02'), + ('EUC_JP', 'CS3', '\x8f0203'), + -- EUC_CN + -- 3 8e (CS2, not used but arbitrarily considered to have length 3) + -- 3 8f (CS3, not used but arbitrarily considered to have length 3) + -- 2 80..ff (CS1) + ('EUC_CN', 'ASCII', 'a'), + ('EUC_CN', 'CS1, short', '\x80'), + ('EUC_CN', 'CS1', '\x8002'), + ('EUC_CN', 'CS2, short', '\x8e'), + ('EUC_CN', 'CS2, short', '\x8e02'), + ('EUC_CN', 'CS2', '\x8e0203'), + ('EUC_CN', 'CS3, short', '\x8f'), + ('EUC_CN', 'CS3, short', '\x8f02'), + ('EUC_CN', 'CS3', '\x8f0203'), + -- EUC_TW: + -- 4 8e (CS2) + -- 3 8f (CS3, not used but arbitrarily considered to have length 3) + -- 2 80..ff (CS1) + ('EUC_TW', 'ASCII', 'a'), + ('EUC_TW', 'CS1, short', '\x80'), + ('EUC_TW', 'CS1', '\x8002'), + ('EUC_TW', 'CS2, short', '\x8e'), + ('EUC_TW', 'CS2, short', '\x8e02'), + ('EUC_TW', 'CS2, short', '\x8e0203'), + ('EUC_TW', 'CS2', '\x8e020304'), + ('EUC_TW', 'CS3, short', '\x8f'), + ('EUC_TW', 'CS3, short', '\x8f02'), + ('EUC_TW', 'CS3', '\x8f0203'), + -- UTF8 + -- 2 c0..df + -- 3 e0..ef + -- 4 f0..f7 (but maximum real codepoint U+10ffff has f4) + -- 5 f8..fb (not supported) + -- 6 fc..fd (not supported) + ('UTF8', 'ASCII', 'a'), + ('UTF8', '2 byte, short', '\xdf'), + ('UTF8', '2 byte', '\xdf82'), + ('UTF8', '3 byte, short', '\xef'), + ('UTF8', '3 byte, short', '\xef82'), + ('UTF8', '3 byte', '\xef8283'), + ('UTF8', '4 byte, short', '\xf7'), + ('UTF8', '4 byte, short', '\xf782'), + ('UTF8', '4 byte, short', '\xf78283'), + ('UTF8', '4 byte', '\xf7828384'), + ('UTF8', '5 byte, unsupported', '\xfb'), + ('UTF8', '5 byte, unsupported', '\xfb82'), + ('UTF8', '5 byte, unsupported', '\xfb8283'), + ('UTF8', '5 byte, unsupported', '\xfb828384'), + ('UTF8', '5 byte, unsupported', '\xfb82838485'), + ('UTF8', '6 byte, unsupported', '\xfd'), + ('UTF8', '6 byte, unsupported', '\xfd82'), + ('UTF8', '6 byte, unsupported', '\xfd8283'), + ('UTF8', '6 byte, unsupported', '\xfd828384'), + ('UTF8', '6 byte, unsupported', '\xfd82838485'), + ('UTF8', '6 byte, unsupported', '\xfd8283848586'), + -- MULE_INTERNAL + -- 2 81..8d LC1 + -- 3 90..99 LC2 + ('MULE_INTERNAL', 'ASCII', 'a'), + ('MULE_INTERNAL', 'LC1, short', '\x81'), + ('MULE_INTERNAL', 'LC1', '\x8182'), + ('MULE_INTERNAL', 'LC2, short', '\x90'), + ('MULE_INTERNAL', 'LC2, short', '\x9082'), + ('MULE_INTERNAL', 'LC2', '\x908283'); + +SELECT COUNT(test_encoding(encoding, description, input)) > 0 +FROM encoding_tests; + +DROP TABLE encoding_tests; +DROP FUNCTION test_encoding; +DROP FUNCTION test_text_to_wchars; +DROP FUNCTION test_mblen_func; +DROP FUNCTION test_bytea_to_text; +DROP FUNCTION test_text_to_bytea; + + +-- substring slow path: multi-byte escape char vs. multi-byte pattern char. +SELECT SUBSTRING('a' SIMILAR U&'\00AC' ESCAPE U&'\00A7'); +-- Levenshtein distance metric: exercise character length cache. +SELECT U&"real\00A7_name" FROM (select 1) AS x(real_name); +-- JSON errcontext: truncate long data. +SELECT repeat(U&'\00A7', 30)::json; diff --git a/src/test/regress/sql/euc_kr.sql b/src/test/regress/sql/euc_kr.sql new file mode 100644 index 0000000000000..1851b2a8c1405 --- /dev/null +++ b/src/test/regress/sql/euc_kr.sql @@ -0,0 +1,12 @@ +-- This test is about EUC_KR encoding, chosen as perhaps the most prevalent +-- non-UTF8, multibyte encoding as of 2026-01. Since UTF8 can represent all +-- of EUC_KR, also run the test in UTF8. +SELECT getdatabaseencoding() NOT IN ('EUC_KR', 'UTF8') AS skip_test \gset +\if :skip_test +\quit +\endif + +-- Exercise is_multibyte_char_in_char (non-UTF8) slow path. +SELECT POSITION( + convert_from('\xbcf6c7d0', 'EUC_KR') IN + convert_from('\xb0fac7d02c20bcf6c7d02c20b1e2bcfa2c20bbee', 'EUC_KR')); From 67d63666149b1d1556b4d0cb259dcbde44b93a3b Mon Sep 17 00:00:00 2001 From: Noah Misch Date: Mon, 9 Feb 2026 06:14:47 -0800 Subject: [PATCH 05/11] Require PGP-decrypted text to pass encoding validation. pgp_sym_decrypt() and pgp_pub_decrypt() will raise such errors, while bytea variants will not. The existing "dat3" test decrypted to non-UTF8 text, so switch that query to bytea. The long-term intent is for type "text" to always be valid in the database encoding. pgcrypto has long been known as a source of exceptions to that intent, but a report about exploiting invalid values of type "text" brought this module to the forefront. This particular exception is straightforward to fix, with reasonable effect on user queries. Back-patch to v14 (all supported versions). Reported-by: Paul Gerste (as part of zeroday.cloud) Reported-by: Moritz Sanft (as part of zeroday.cloud) Author: shihao zhong Reviewed-by: cary huang Discussion: https://postgr.es/m/CAGRkXqRZyo0gLxPJqUsDqtWYBbgM14betsHiLRPj9mo2=z9VvA@mail.gmail.com Backpatch-through: 14 Security: CVE-2026-2006 (cherry picked from commit 8f8b1ffac063afb22e3a07bdc6ddcb048116cfbb) --- contrib/pgcrypto/expected/pgp-decrypt.out | 23 ++++++++++++++++++++- contrib/pgcrypto/expected/pgp-decrypt_1.out | 23 ++++++++++++++++++++- contrib/pgcrypto/pgp-pgsql.c | 2 ++ contrib/pgcrypto/sql/pgp-decrypt.sql | 22 +++++++++++++++++++- 4 files changed, 67 insertions(+), 3 deletions(-) diff --git a/contrib/pgcrypto/expected/pgp-decrypt.out b/contrib/pgcrypto/expected/pgp-decrypt.out index eb049ba9d4443..1db89e8c00a56 100644 --- a/contrib/pgcrypto/expected/pgp-decrypt.out +++ b/contrib/pgcrypto/expected/pgp-decrypt.out @@ -315,7 +315,7 @@ SaV9L04ky1qECNDx3XjnoKLC+H7IOQ== \xda39a3ee5e6b4b0d3255bfef95601890afd80709 (1 row) -select digest(pgp_sym_decrypt(dearmor(' +select digest(pgp_sym_decrypt_bytea(dearmor(' -----BEGIN PGP MESSAGE----- Comment: dat3.aes.sha1.mdc.s2k3.z0 @@ -387,6 +387,27 @@ ERROR: Wrong key or corrupt data select pgp_sym_decrypt(pgp_sym_encrypt_bytea('P', 'key'), 'key', 'debug=1'); NOTICE: dbg: parse_literal_data: data type=b ERROR: Not text data +-- NUL byte in text decrypt. Ciphertext source: +-- printf 'a\x00\xc' | gpg --homedir /nonexistent --textmode \ +-- --personal-cipher-preferences aes --no-emit-version --batch \ +-- --symmetric --passphrase key --armor +do $$ +begin + perform pgp_sym_decrypt(dearmor(' +-----BEGIN PGP MESSAGE----- + +jA0EBwMCLd9OvySmZNZg0jgBe7vGTmnje5HGXI+zsIQ99WPZu4Zs/P6pQcZ+HZ4n +SZQHOfE8tagjB6Rqow82QpSBiOfWn4qjhQ== +=c2cz +-----END PGP MESSAGE----- +'), 'key', 'debug=1'); +exception when others then + raise '%', + regexp_replace(sqlerrm, 'encoding "[^"]*"', 'encoding [REDACTED]'); +end +$$; +ERROR: invalid byte sequence for encoding [REDACTED]: 0x00 +CONTEXT: PL/pgSQL function inline_code_block line 12 at RAISE -- Decryption with a certain incorrect key yields an apparent BZip2-compressed -- plaintext. Ciphertext source: iterative pgp_sym_encrypt('secret', 'key') -- until the random prefix gave rise to that property. diff --git a/contrib/pgcrypto/expected/pgp-decrypt_1.out b/contrib/pgcrypto/expected/pgp-decrypt_1.out index 80a4c48613d00..d214e0bc0e08f 100644 --- a/contrib/pgcrypto/expected/pgp-decrypt_1.out +++ b/contrib/pgcrypto/expected/pgp-decrypt_1.out @@ -311,7 +311,7 @@ SaV9L04ky1qECNDx3XjnoKLC+H7IOQ== \xda39a3ee5e6b4b0d3255bfef95601890afd80709 (1 row) -select digest(pgp_sym_decrypt(dearmor(' +select digest(pgp_sym_decrypt_bytea(dearmor(' -----BEGIN PGP MESSAGE----- Comment: dat3.aes.sha1.mdc.s2k3.z0 @@ -383,6 +383,27 @@ ERROR: Wrong key or corrupt data select pgp_sym_decrypt(pgp_sym_encrypt_bytea('P', 'key'), 'key', 'debug=1'); NOTICE: dbg: parse_literal_data: data type=b ERROR: Not text data +-- NUL byte in text decrypt. Ciphertext source: +-- printf 'a\x00\xc' | gpg --homedir /nonexistent --textmode \ +-- --personal-cipher-preferences aes --no-emit-version --batch \ +-- --symmetric --passphrase key --armor +do $$ +begin + perform pgp_sym_decrypt(dearmor(' +-----BEGIN PGP MESSAGE----- + +jA0EBwMCLd9OvySmZNZg0jgBe7vGTmnje5HGXI+zsIQ99WPZu4Zs/P6pQcZ+HZ4n +SZQHOfE8tagjB6Rqow82QpSBiOfWn4qjhQ== +=c2cz +-----END PGP MESSAGE----- +'), 'key', 'debug=1'); +exception when others then + raise '%', + regexp_replace(sqlerrm, 'encoding "[^"]*"', 'encoding [REDACTED]'); +end +$$; +ERROR: invalid byte sequence for encoding [REDACTED]: 0x00 +CONTEXT: PL/pgSQL function inline_code_block line 12 at RAISE -- Decryption with a certain incorrect key yields an apparent BZip2-compressed -- plaintext. Ciphertext source: iterative pgp_sym_encrypt('secret', 'key') -- until the random prefix gave rise to that property. diff --git a/contrib/pgcrypto/pgp-pgsql.c b/contrib/pgcrypto/pgp-pgsql.c index 0536bfb8921c9..cf315b126b748 100644 --- a/contrib/pgcrypto/pgp-pgsql.c +++ b/contrib/pgcrypto/pgp-pgsql.c @@ -631,6 +631,7 @@ pgp_sym_decrypt_text(PG_FUNCTION_ARGS) arg = PG_GETARG_BYTEA_PP(2); res = decrypt_internal(0, 1, data, key, NULL, arg); + pg_verifymbstr(VARDATA_ANY(res), VARSIZE_ANY_EXHDR(res), false); PG_FREE_IF_COPY(data, 0); PG_FREE_IF_COPY(key, 1); @@ -732,6 +733,7 @@ pgp_pub_decrypt_text(PG_FUNCTION_ARGS) arg = PG_GETARG_BYTEA_PP(3); res = decrypt_internal(1, 1, data, key, psw, arg); + pg_verifymbstr(VARDATA_ANY(res), VARSIZE_ANY_EXHDR(res), false); PG_FREE_IF_COPY(data, 0); PG_FREE_IF_COPY(key, 1); diff --git a/contrib/pgcrypto/sql/pgp-decrypt.sql b/contrib/pgcrypto/sql/pgp-decrypt.sql index 49a0267bbcbcc..2fe498f2f02e4 100644 --- a/contrib/pgcrypto/sql/pgp-decrypt.sql +++ b/contrib/pgcrypto/sql/pgp-decrypt.sql @@ -228,7 +228,7 @@ SaV9L04ky1qECNDx3XjnoKLC+H7IOQ== -----END PGP MESSAGE----- '), '0123456789abcdefghij'), 'sha1'); -select digest(pgp_sym_decrypt(dearmor(' +select digest(pgp_sym_decrypt_bytea(dearmor(' -----BEGIN PGP MESSAGE----- Comment: dat3.aes.sha1.mdc.s2k3.z0 @@ -282,6 +282,26 @@ VsxxqLSPzNLAeIspJk5G -- Routine text/binary mismatch. select pgp_sym_decrypt(pgp_sym_encrypt_bytea('P', 'key'), 'key', 'debug=1'); +-- NUL byte in text decrypt. Ciphertext source: +-- printf 'a\x00\xc' | gpg --homedir /nonexistent --textmode \ +-- --personal-cipher-preferences aes --no-emit-version --batch \ +-- --symmetric --passphrase key --armor +do $$ +begin + perform pgp_sym_decrypt(dearmor(' +-----BEGIN PGP MESSAGE----- + +jA0EBwMCLd9OvySmZNZg0jgBe7vGTmnje5HGXI+zsIQ99WPZu4Zs/P6pQcZ+HZ4n +SZQHOfE8tagjB6Rqow82QpSBiOfWn4qjhQ== +=c2cz +-----END PGP MESSAGE----- +'), 'key', 'debug=1'); +exception when others then + raise '%', + regexp_replace(sqlerrm, 'encoding "[^"]*"', 'encoding [REDACTED]'); +end +$$; + -- Decryption with a certain incorrect key yields an apparent BZip2-compressed -- plaintext. Ciphertext source: iterative pgp_sym_encrypt('secret', 'key') -- until the random prefix gave rise to that property. From 44ddde1e705d7e25cf231ff1f366cba57e240e2a Mon Sep 17 00:00:00 2001 From: Noah Misch Date: Mon, 9 Feb 2026 09:08:10 -0800 Subject: [PATCH 06/11] Fix test "NUL byte in text decrypt" for --without-zlib builds. Backpatch-through: 14 Security: CVE-2026-2006 (cherry picked from commit 6f741bcb6aa4d78b2e3c96b6f3830b371c8cdf47) --- contrib/pgcrypto/expected/pgp-decrypt.out | 9 +++++---- contrib/pgcrypto/expected/pgp-decrypt_1.out | 9 +++++---- contrib/pgcrypto/sql/pgp-decrypt.sql | 9 +++++---- 3 files changed, 15 insertions(+), 12 deletions(-) diff --git a/contrib/pgcrypto/expected/pgp-decrypt.out b/contrib/pgcrypto/expected/pgp-decrypt.out index 1db89e8c00a56..8ce6466f2e9a8 100644 --- a/contrib/pgcrypto/expected/pgp-decrypt.out +++ b/contrib/pgcrypto/expected/pgp-decrypt.out @@ -388,7 +388,8 @@ select pgp_sym_decrypt(pgp_sym_encrypt_bytea('P', 'key'), 'key', 'debug=1'); NOTICE: dbg: parse_literal_data: data type=b ERROR: Not text data -- NUL byte in text decrypt. Ciphertext source: --- printf 'a\x00\xc' | gpg --homedir /nonexistent --textmode \ +-- printf 'a\x00\xc' | gpg --homedir /nonexistent \ +-- --personal-compress-preferences uncompressed --textmode \ -- --personal-cipher-preferences aes --no-emit-version --batch \ -- --symmetric --passphrase key --armor do $$ @@ -396,9 +397,9 @@ begin perform pgp_sym_decrypt(dearmor(' -----BEGIN PGP MESSAGE----- -jA0EBwMCLd9OvySmZNZg0jgBe7vGTmnje5HGXI+zsIQ99WPZu4Zs/P6pQcZ+HZ4n -SZQHOfE8tagjB6Rqow82QpSBiOfWn4qjhQ== -=c2cz +jA0EBwMCXLc8pozB10Fg0jQBVUID59TLvWutJp0j6eh9ZgjqIRzdYaIymFB8y4XH +vu0YlJP5D5BX7yqZ+Pry7TlDmiFO +=rV7z -----END PGP MESSAGE----- '), 'key', 'debug=1'); exception when others then diff --git a/contrib/pgcrypto/expected/pgp-decrypt_1.out b/contrib/pgcrypto/expected/pgp-decrypt_1.out index d214e0bc0e08f..ee57ad43cb75f 100644 --- a/contrib/pgcrypto/expected/pgp-decrypt_1.out +++ b/contrib/pgcrypto/expected/pgp-decrypt_1.out @@ -384,7 +384,8 @@ select pgp_sym_decrypt(pgp_sym_encrypt_bytea('P', 'key'), 'key', 'debug=1'); NOTICE: dbg: parse_literal_data: data type=b ERROR: Not text data -- NUL byte in text decrypt. Ciphertext source: --- printf 'a\x00\xc' | gpg --homedir /nonexistent --textmode \ +-- printf 'a\x00\xc' | gpg --homedir /nonexistent \ +-- --personal-compress-preferences uncompressed --textmode \ -- --personal-cipher-preferences aes --no-emit-version --batch \ -- --symmetric --passphrase key --armor do $$ @@ -392,9 +393,9 @@ begin perform pgp_sym_decrypt(dearmor(' -----BEGIN PGP MESSAGE----- -jA0EBwMCLd9OvySmZNZg0jgBe7vGTmnje5HGXI+zsIQ99WPZu4Zs/P6pQcZ+HZ4n -SZQHOfE8tagjB6Rqow82QpSBiOfWn4qjhQ== -=c2cz +jA0EBwMCXLc8pozB10Fg0jQBVUID59TLvWutJp0j6eh9ZgjqIRzdYaIymFB8y4XH +vu0YlJP5D5BX7yqZ+Pry7TlDmiFO +=rV7z -----END PGP MESSAGE----- '), 'key', 'debug=1'); exception when others then diff --git a/contrib/pgcrypto/sql/pgp-decrypt.sql b/contrib/pgcrypto/sql/pgp-decrypt.sql index 2fe498f2f02e4..b499bf757b0d4 100644 --- a/contrib/pgcrypto/sql/pgp-decrypt.sql +++ b/contrib/pgcrypto/sql/pgp-decrypt.sql @@ -283,7 +283,8 @@ VsxxqLSPzNLAeIspJk5G select pgp_sym_decrypt(pgp_sym_encrypt_bytea('P', 'key'), 'key', 'debug=1'); -- NUL byte in text decrypt. Ciphertext source: --- printf 'a\x00\xc' | gpg --homedir /nonexistent --textmode \ +-- printf 'a\x00\xc' | gpg --homedir /nonexistent \ +-- --personal-compress-preferences uncompressed --textmode \ -- --personal-cipher-preferences aes --no-emit-version --batch \ -- --symmetric --passphrase key --armor do $$ @@ -291,9 +292,9 @@ begin perform pgp_sym_decrypt(dearmor(' -----BEGIN PGP MESSAGE----- -jA0EBwMCLd9OvySmZNZg0jgBe7vGTmnje5HGXI+zsIQ99WPZu4Zs/P6pQcZ+HZ4n -SZQHOfE8tagjB6Rqow82QpSBiOfWn4qjhQ== -=c2cz +jA0EBwMCXLc8pozB10Fg0jQBVUID59TLvWutJp0j6eh9ZgjqIRzdYaIymFB8y4XH +vu0YlJP5D5BX7yqZ+Pry7TlDmiFO +=rV7z -----END PGP MESSAGE----- '), 'key', 'debug=1'); exception when others then From 215fd186b7c54f71a20761b05bf0cae585c029e4 Mon Sep 17 00:00:00 2001 From: Noah Misch Date: Sat, 14 Feb 2026 12:16:16 -0800 Subject: [PATCH 07/11] pg_mblen_range, pg_mblen_with_len: Valgrind after encoding ereport. The prior order caused spurious Valgrind errors. They're spurious because the ereport(ERROR) non-local exit discards the pointer in question. pg_mblen_cstr() ordered the checks correctly, but these other two did not. Back-patch to v14, like commit 1e7fe06c10c0a8da9dd6261a6be8d405dc17c728. Reviewed-by: Thomas Munro Discussion: https://postgr.es/m/20260214053821.fa.noahmisch@microsoft.com Backpatch-through: 14 (cherry picked from commit 5b305ebcc9b56b053415997625c6425aa7b878b2) --- src/backend/utils/mb/mbutils.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/src/backend/utils/mb/mbutils.c b/src/backend/utils/mb/mbutils.c index c0f96cad60c90..33ffeda40cf30 100644 --- a/src/backend/utils/mb/mbutils.c +++ b/src/backend/utils/mb/mbutils.c @@ -1026,15 +1026,16 @@ pg_mblen_range(const char *mbstr, const char *end) int length = pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); Assert(end > mbstr); + + if (unlikely(mbstr + length > end)) + report_invalid_encoding_db(mbstr, length, end - mbstr); + #ifdef VALGRIND_EXPENSIVE VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, end - mbstr); #else VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, length); #endif - if (unlikely(mbstr + length > end)) - report_invalid_encoding_db(mbstr, length, end - mbstr); - return length; } @@ -1049,15 +1050,16 @@ pg_mblen_with_len(const char *mbstr, int limit) int length = pg_wchar_table[DatabaseEncoding->encoding].mblen((const unsigned char *) mbstr); Assert(limit >= 1); + + if (unlikely(length > limit)) + report_invalid_encoding_db(mbstr, length, limit); + #ifdef VALGRIND_EXPENSIVE VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, limit); #else VALGRIND_CHECK_MEM_IS_DEFINED(mbstr, length); #endif - if (unlikely(length > limit)) - report_invalid_encoding_db(mbstr, length, limit); - return length; } From 8424c5087a393cbdf9c2eca756e49eead67a4090 Mon Sep 17 00:00:00 2001 From: Noah Misch Date: Sat, 14 Feb 2026 12:16:16 -0800 Subject: [PATCH 08/11] Fix SUBSTRING() for toasted multibyte characters. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 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 Reviewed-by: Thomas Munro Bug: #19406 Discussion: https://postgr.es/m/19406-9867fddddd724fca@postgresql.org Backpatch-through: 14 (cherry picked from commit a20eb248c51ad6741bbbb3b3bac103d5788bd4f5) --- src/backend/utils/adt/varlena.c | 62 +++++++++++++++++++++----- src/test/regress/expected/encoding.out | 44 +++++++++++++++++- src/test/regress/sql/encoding.sql | 19 +++++++- 3 files changed, 111 insertions(+), 14 deletions(-) diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c index a6d138a4058bf..19d80ce4c8ab3 100644 --- a/src/backend/utils/adt/varlena.c +++ b/src/backend/utils/adt/varlena.c @@ -134,6 +134,7 @@ static text *text_substring(Datum str, 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); @@ -870,7 +871,7 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) 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, but we still must fetch from the @@ -967,11 +968,11 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) 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(""); /* @@ -981,11 +982,11 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) 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; } @@ -1008,9 +1009,17 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) 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 @@ -1065,6 +1074,35 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) 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 diff --git a/src/test/regress/expected/encoding.out b/src/test/regress/expected/encoding.out index ea1f38cff41d8..cac1cb7478266 100644 --- a/src/test/regress/expected/encoding.out +++ b/src/test/regress/expected/encoding.out @@ -63,7 +63,13 @@ SELECT reverse(good) FROM regress_encoding; -- 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 @@ -375,7 +381,43 @@ NOTICE: MULE_INTERNAL LC2: \x908283 -> {9470595} -> \x908283 = OK 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; diff --git a/src/test/regress/sql/encoding.sql b/src/test/regress/sql/encoding.sql index b9543c0cb3262..782f90f0d6292 100644 --- a/src/test/regress/sql/encoding.sql +++ b/src/test/regress/sql/encoding.sql @@ -40,7 +40,8 @@ SELECT reverse(good) FROM regress_encoding; -- 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; @@ -212,7 +213,23 @@ INSERT INTO encoding_tests VALUES 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; From 986be126e2428e106c7896e8afb1c29e4043fbc9 Mon Sep 17 00:00:00 2001 From: Noah Misch Date: Mon, 16 Feb 2026 18:04:58 -0800 Subject: [PATCH 09/11] Suppress new "may be used uninitialized" warning. Various buildfarm members, having compilers like gcc 8.5 and 6.3, fail to deduce that text_substring() variable "E" is initialized if slice_size!=-1. This suppression approach quiets gcc 8.5; I did not reproduce the warning elsewhere. Back-patch to v14, like commit 9f4fd119b2cbb9a41ec0c19a8d6ec9b59b92c125. Reported-by: Tom Lane Discussion: https://postgr.es/m/1157953.1771266105@sss.pgh.pa.us Backpatch-through: 14 (cherry picked from commit ec86152e02090d086caf170c77b3f559ce27e830) --- src/backend/utils/adt/varlena.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c index 19d80ce4c8ab3..8788765130b77 100644 --- a/src/backend/utils/adt/varlena.c +++ b/src/backend/utils/adt/varlena.c @@ -948,14 +948,14 @@ text_substring(Datum str, int32 start, int32 length, bool length_not_specified) if (length_not_specified) /* special case - get length to end of * string */ - slice_size = L1 = -1; + E = slice_size = L1 = -1; else if (length < 0) { /* SQL99 says to throw an error for E < S, i.e., negative length */ ereport(ERROR, (errcode(ERRCODE_SUBSTRING_ERROR), errmsg("negative substring length not allowed"))); - slice_size = L1 = -1; /* silence stupider compilers */ + E = slice_size = L1 = -1; /* silence stupider compilers */ } else if (pg_add_s32_overflow(S, length, &E)) { From dad4e36258c5f0780b465c7661f373675c8658fc Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Tue, 17 Feb 2026 13:53:32 +1300 Subject: [PATCH 10/11] Fix test_valid_server_encoding helper function. Commit c67bef3f325 introduced this test helper function for use by src/test/regress/sql/encoding.sql, but its logic was incorrect. It confused an encoding ID for a boolean so it gave the wrong results for some inputs, and also forgot the usual return macro. The mistake didn't affect values actually used in the test, so there is no change in behavior. Also drop it and another missed function at the end of the test, for consistency. Backpatch-through: 14 Author: Zsolt Parragi (cherry picked from commit e3bfa4f589a8a7127824f48a5892166ae7ac0e24) --- src/test/regress/expected/encoding.out | 2 ++ src/test/regress/regress.c | 2 +- src/test/regress/sql/encoding.sql | 2 ++ 3 files changed, 5 insertions(+), 1 deletion(-) diff --git a/src/test/regress/expected/encoding.out b/src/test/regress/expected/encoding.out index cac1cb7478266..b3655527b0ab9 100644 --- a/src/test/regress/expected/encoding.out +++ b/src/test/regress/expected/encoding.out @@ -419,7 +419,9 @@ 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_wchars_to_text; DROP FUNCTION test_text_to_wchars; +DROP FUNCTION test_valid_server_encoding; DROP FUNCTION test_mblen_func; DROP FUNCTION test_bytea_to_text; DROP FUNCTION test_text_to_bytea; diff --git a/src/test/regress/regress.c b/src/test/regress/regress.c index 60378f643446f..7e77d3e9670a3 100644 --- a/src/test/regress/regress.c +++ b/src/test/regress/regress.c @@ -1416,7 +1416,7 @@ PG_FUNCTION_INFO_V1(test_valid_server_encoding); Datum test_valid_server_encoding(PG_FUNCTION_ARGS) { - return pg_valid_server_encoding(text_to_cstring(PG_GETARG_TEXT_PP(0))); + PG_RETURN_BOOL(pg_valid_server_encoding(text_to_cstring(PG_GETARG_TEXT_PP(0))) >= 0); } /* Provide SQL access to IsBinaryCoercible() */ diff --git a/src/test/regress/sql/encoding.sql b/src/test/regress/sql/encoding.sql index 782f90f0d6292..d591818c3ebde 100644 --- a/src/test/regress/sql/encoding.sql +++ b/src/test/regress/sql/encoding.sql @@ -231,7 +231,9 @@ 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_wchars_to_text; DROP FUNCTION test_text_to_wchars; +DROP FUNCTION test_valid_server_encoding; DROP FUNCTION test_mblen_func; DROP FUNCTION test_bytea_to_text; DROP FUNCTION test_text_to_bytea; From d987a73f925dfc48af644661b54d0f9813599834 Mon Sep 17 00:00:00 2001 From: Noah Misch Date: Wed, 25 Feb 2026 18:13:22 -0800 Subject: [PATCH 11/11] EUC_CN, EUC_JP, EUC_KR, EUC_TW: Skip U+00A0 tests instead of failing. Settings that ran the new test euc_kr.sql to completion would fail these older src/pl tests. Use alternative expected outputs, for which psql \gset and \if have reduced the maintenance burden. This fixes "LANG=ko_KR.euckr LC_MESSAGES=C make check-world". (LC_MESSAGES=C fixes IO::Pty usage in tests 010_tab_completion and 001_password.) That file is new in commit c67bef3f3252a3a38bf347f9f119944176a796ce. Back-patch to v14, like that commit. Discussion: https://postgr.es/m/20260217184758.da.noahmisch@microsoft.com Backpatch-through: 14 (cherry picked from commit 44d29f5c6c0febc8ef3c88da3eab099b89546296) --- src/pl/plperl/GNUmakefile | 2 +- src/pl/plperl/expected/plperl_elog.out | 13 ------------- src/pl/plperl/expected/plperl_elog_1.out | 13 ------------- src/pl/plperl/expected/plperl_unicode.out | 18 ++++++++++++++++++ src/pl/plperl/expected/plperl_unicode_1.out | 10 ++++++++++ src/pl/plperl/sql/plperl_elog.sql | 15 --------------- src/pl/plperl/sql/plperl_unicode.sql | 19 +++++++++++++++++++ src/pl/plpython/expected/plpython_unicode.out | 13 +++++++++---- .../plpython/expected/plpython_unicode_1.out | 12 ++++++++++++ src/pl/plpython/sql/plpython_unicode.sql | 13 +++++++++---- src/pl/tcl/expected/pltcl_unicode.out | 13 +++++++++---- src/pl/tcl/expected/pltcl_unicode_1.out | 12 ++++++++++++ src/pl/tcl/sql/pltcl_unicode.sql | 13 +++++++++---- 13 files changed, 108 insertions(+), 58 deletions(-) create mode 100644 src/pl/plperl/expected/plperl_unicode.out create mode 100644 src/pl/plperl/expected/plperl_unicode_1.out create mode 100644 src/pl/plperl/sql/plperl_unicode.sql create mode 100644 src/pl/plpython/expected/plpython_unicode_1.out create mode 100644 src/pl/tcl/expected/pltcl_unicode_1.out diff --git a/src/pl/plperl/GNUmakefile b/src/pl/plperl/GNUmakefile index 01588d016a063..f07d0e317708f 100644 --- a/src/pl/plperl/GNUmakefile +++ b/src/pl/plperl/GNUmakefile @@ -57,7 +57,7 @@ SHLIB_LINK = $(perl_embed_ldflags) REGRESS_OPTS = --dbname=$(PL_TESTDB) --dlpath=$(top_builddir)/src/test/regress REGRESS = plperl_setup plperl plperl_lc plperl_trigger plperl_shared \ - plperl_elog plperl_util plperl_init plperlu plperl_array \ + plperl_elog plperl_unicode plperl_util plperl_init plperlu plperl_array \ plperl_call plperl_transaction plperl_env # if Perl can support two interpreters in one backend, # test plperl-and-plperlu cases diff --git a/src/pl/plperl/expected/plperl_elog.out b/src/pl/plperl/expected/plperl_elog.out index a6d35cb79c4f9..3f9449a9659b4 100644 --- a/src/pl/plperl/expected/plperl_elog.out +++ b/src/pl/plperl/expected/plperl_elog.out @@ -97,16 +97,3 @@ NOTICE: caught die 2 (1 row) --- Test non-ASCII error messages --- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. -SET client_encoding TO UTF8; -create or replace function error_with_nbsp() returns void language plperl as $$ - elog(ERROR, "this message contains a no-break space"); -$$; -select error_with_nbsp(); -ERROR: this message contains a no-break space at line 2. -CONTEXT: PL/Perl function "error_with_nbsp" diff --git a/src/pl/plperl/expected/plperl_elog_1.out b/src/pl/plperl/expected/plperl_elog_1.out index 85aa460ec4c2c..34d5d5836da22 100644 --- a/src/pl/plperl/expected/plperl_elog_1.out +++ b/src/pl/plperl/expected/plperl_elog_1.out @@ -97,16 +97,3 @@ NOTICE: caught die 2 (1 row) --- Test non-ASCII error messages --- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. -SET client_encoding TO UTF8; -create or replace function error_with_nbsp() returns void language plperl as $$ - elog(ERROR, "this message contains a no-break space"); -$$; -select error_with_nbsp(); -ERROR: this message contains a no-break space at line 2. -CONTEXT: PL/Perl function "error_with_nbsp" diff --git a/src/pl/plperl/expected/plperl_unicode.out b/src/pl/plperl/expected/plperl_unicode.out new file mode 100644 index 0000000000000..3c48f2e961139 --- /dev/null +++ b/src/pl/plperl/expected/plperl_unicode.out @@ -0,0 +1,18 @@ +-- Test non-ASCII error messages +-- +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit +\endif +SET client_encoding TO UTF8; +create or replace function error_with_nbsp() returns void language plperl as $$ + elog(ERROR, "this message contains a no-break space"); +$$; +select error_with_nbsp(); +ERROR: this message contains a no-break space at line 2. +CONTEXT: PL/Perl function "error_with_nbsp" diff --git a/src/pl/plperl/expected/plperl_unicode_1.out b/src/pl/plperl/expected/plperl_unicode_1.out new file mode 100644 index 0000000000000..761de04b1ee57 --- /dev/null +++ b/src/pl/plperl/expected/plperl_unicode_1.out @@ -0,0 +1,10 @@ +-- Test non-ASCII error messages +-- +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit diff --git a/src/pl/plperl/sql/plperl_elog.sql b/src/pl/plperl/sql/plperl_elog.sql index 9ea1350069b86..032fd8b8ba74a 100644 --- a/src/pl/plperl/sql/plperl_elog.sql +++ b/src/pl/plperl/sql/plperl_elog.sql @@ -76,18 +76,3 @@ return $a + $b; $$; select indirect_die_caller(); - --- Test non-ASCII error messages --- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. - -SET client_encoding TO UTF8; - -create or replace function error_with_nbsp() returns void language plperl as $$ - elog(ERROR, "this message contains a no-break space"); -$$; - -select error_with_nbsp(); diff --git a/src/pl/plperl/sql/plperl_unicode.sql b/src/pl/plperl/sql/plperl_unicode.sql new file mode 100644 index 0000000000000..7e1ad745cdd51 --- /dev/null +++ b/src/pl/plperl/sql/plperl_unicode.sql @@ -0,0 +1,19 @@ +-- Test non-ASCII error messages +-- +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit +\endif + +SET client_encoding TO UTF8; + +create or replace function error_with_nbsp() returns void language plperl as $$ + elog(ERROR, "this message contains a no-break space"); +$$; + +select error_with_nbsp(); diff --git a/src/pl/plpython/expected/plpython_unicode.out b/src/pl/plpython/expected/plpython_unicode.out index fd54b0b88e8f8..bd8d9c561c956 100644 --- a/src/pl/plpython/expected/plpython_unicode.out +++ b/src/pl/plpython/expected/plpython_unicode.out @@ -1,11 +1,16 @@ -- -- Unicode handling -- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. -- +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit +\endif SET client_encoding TO UTF8; CREATE TABLE unicode_test ( testvalue text NOT NULL diff --git a/src/pl/plpython/expected/plpython_unicode_1.out b/src/pl/plpython/expected/plpython_unicode_1.out new file mode 100644 index 0000000000000..f8b21fd7eb7ab --- /dev/null +++ b/src/pl/plpython/expected/plpython_unicode_1.out @@ -0,0 +1,12 @@ +-- +-- Unicode handling +-- +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. +-- +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit diff --git a/src/pl/plpython/sql/plpython_unicode.sql b/src/pl/plpython/sql/plpython_unicode.sql index 14f7b4e0053db..f45844b906e47 100644 --- a/src/pl/plpython/sql/plpython_unicode.sql +++ b/src/pl/plpython/sql/plpython_unicode.sql @@ -1,11 +1,16 @@ -- -- Unicode handling -- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. -- +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit +\endif SET client_encoding TO UTF8; diff --git a/src/pl/tcl/expected/pltcl_unicode.out b/src/pl/tcl/expected/pltcl_unicode.out index eea7d70664f47..d33afd7548f0f 100644 --- a/src/pl/tcl/expected/pltcl_unicode.out +++ b/src/pl/tcl/expected/pltcl_unicode.out @@ -1,11 +1,16 @@ -- -- Unicode handling -- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. -- +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit +\endif SET client_encoding TO UTF8; CREATE TABLE unicode_test ( testvalue text NOT NULL diff --git a/src/pl/tcl/expected/pltcl_unicode_1.out b/src/pl/tcl/expected/pltcl_unicode_1.out new file mode 100644 index 0000000000000..f8b21fd7eb7ab --- /dev/null +++ b/src/pl/tcl/expected/pltcl_unicode_1.out @@ -0,0 +1,12 @@ +-- +-- Unicode handling +-- +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. +-- +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit diff --git a/src/pl/tcl/sql/pltcl_unicode.sql b/src/pl/tcl/sql/pltcl_unicode.sql index f0006046127d4..a09e4998b2bf6 100644 --- a/src/pl/tcl/sql/pltcl_unicode.sql +++ b/src/pl/tcl/sql/pltcl_unicode.sql @@ -1,11 +1,16 @@ -- -- Unicode handling -- --- Note: this test case is known to fail if the database encoding is --- EUC_CN, EUC_JP, EUC_KR, or EUC_TW, for lack of any equivalent to --- U+00A0 (no-break space) in those encodings. However, testing with --- plain ASCII data would be rather useless, so we must live with that. +-- This test case would fail if the database encoding is EUC_CN, EUC_JP, +-- EUC_KR, or EUC_TW, for lack of any equivalent to U+00A0 (no-break space) in +-- those encodings. However, testing with plain ASCII data would be rather +-- useless, so we must live with that. -- +SELECT getdatabaseencoding() IN ('EUC_CN', 'EUC_JP', 'EUC_KR', 'EUC_TW') + AS skip_test \gset +\if :skip_test +\quit +\endif SET client_encoding TO UTF8;