diff options
| -rw-r--r-- | src/pcre2_compile.c | 18 | ||||
| -rw-r--r-- | src/pcre2_intmodedep.h | 7 |
2 files changed, 16 insertions, 9 deletions
diff --git a/src/pcre2_compile.c b/src/pcre2_compile.c index 8ab10b5..7e44fc8 100644 --- a/src/pcre2_compile.c +++ b/src/pcre2_compile.c @@ -5585,7 +5585,7 @@ for (;; pptr++) they are case partners. This can be optimized to generate a caseless single character match (which also sets first/required code units if relevant). */ - if (meta == META_CLASS && pptr[1] < META_END && pptr[2] < META_END && + if (meta == META_CLASS && pptr[1] < META_END && pptr[2] < META_END && pptr[3] == META_CLASS_END) { uint32_t c = pptr[1]; @@ -5594,10 +5594,18 @@ for (;; pptr++) if (UCD_CASESET(c) == 0) #endif { - uint32_t d = TABLE_GET(c, cb->fcc, c); + uint32_t d; + #ifdef SUPPORT_UNICODE - if (utf && c > 127) d = UCD_OTHERCASE(c); + if (utf && c > 127) d = UCD_OTHERCASE(c); else #endif + { +#if PCRE2_CODE_UNIT_WIDTH != 8 + if (c > 255) d = c; else +#endif + d = TABLE_GET(c, cb->fcc, c); + } + if (c != d && pptr[2] == d) { pptr += 3; /* Move on to class end */ @@ -5607,7 +5615,7 @@ for (;; pptr++) reset_caseful = TRUE; options |= PCRE2_CASELESS; req_caseopt = REQ_CASELESS; - } + } goto CLASS_CASELESS_CHAR; } } @@ -7892,7 +7900,7 @@ for (;; pptr++) zeroreqcuflags = reqcuflags; /* If the character is more than one code unit long, we can set a single - firstcu only if it is not to be matched caselessly. Multiple possible + firstcu only if it is not to be matched caselessly. Multiple possible starting code units may be picked up later in the studying code. */ if (mclength == 1 || req_caseopt == 0) diff --git a/src/pcre2_intmodedep.h b/src/pcre2_intmodedep.h index b03d5c2..15ade47 100644 --- a/src/pcre2_intmodedep.h +++ b/src/pcre2_intmodedep.h @@ -205,19 +205,19 @@ whether its argument, which is assumed to be one code unit, is less than 256. The CHMAX_255 macro does not assume one code unit. The maximum length of a MARK name must fit in one code unit; currently it is set to 255 or 65535. The TABLE_GET macro is used to access elements of tables containing exactly 256 -items. When code points can be greater than 255, a check is needed before -accessing these tables. */ +items. Its argument is a code unit. When code points can be greater than 255, a +check is needed before accessing these tables. */ #if PCRE2_CODE_UNIT_WIDTH == 8 #define MAX_255(c) TRUE #define MAX_MARK ((1u << 8) - 1) +#define TABLE_GET(c, table, default) ((table)[c]) #ifdef SUPPORT_UNICODE #define SUPPORT_WIDE_CHARS #define CHMAX_255(c) ((c) <= 255u) #else #define CHMAX_255(c) TRUE #endif /* SUPPORT_UNICODE */ -#define TABLE_GET(c, table, default) ((table)[c]) #else /* Code units are 16 or 32 bits */ #define CHMAX_255(c) ((c) <= 255u) @@ -228,7 +228,6 @@ accessing these tables. */ #endif - /* ----------------- Character-handling macros ----------------- */ /* There is a proposed future special "UTF-21" mode, in which only the lowest |