diff --git a/Makefile.am b/Makefile.am index f0aaa7f..f989176 100644 --- a/Makefile.am +++ b/Makefile.am @@ -18,13 +18,15 @@ html: cd doc && make html update-regex: - cd lib && \ - HOST=sources.redhat.com && \ - BASEURL="http://$$HOST/cgi-bin/cvsweb.cgi/~checkout~/libc/posix" && \ - QUERY='cvsroot=glibc&content-type=text/plain' && \ - wget -O regcomp.c "$$BASEURL/regcomp.c?$$QUERY" && \ - wget -O regexec.c "$$BASEURL/regexec.c?$$QUERY" && \ - wget -O regex.c "$$BASEURL/regex.c?$$QUERY" && \ - wget -O regex_.h "$$BASEURL/regex.h?$$QUERY" && \ - wget -O regex_internal.c "$$BASEURL/regex_internal.c?$$QUERY" && \ - wget -O regex_internal.h "$$BASEURL/regex_internal.h?$$QUERY" + HOST=sources.redhat.com && \ + BASEURL="http://$$HOST/cgi-bin/cvsweb.cgi/~checkout~/libc/posix" && \ + QUERY='cvsroot=glibc&content-type=text/plain' && \ + wget -O lib/regcomp.c "$$BASEURL/regcomp.c?$$QUERY" && \ + wget -O lib/regexec.c "$$BASEURL/regexec.c?$$QUERY" && \ + wget -O lib/regex.c "$$BASEURL/regex.c?$$QUERY" && \ + wget -O lib/regex_.h "$$BASEURL/regex.h?$$QUERY" && \ + wget -O lib/regex_internal.c "$$BASEURL/regex_internal.c?$$QUERY" && \ + wget -O lib/regex_internal.h "$$BASEURL/regex_internal.h?$$QUERY" && \ + wget -O testsuite/BOOST.tests "$$BASEURL/BOOST.tests?$$QUERY" && \ + wget -O testsuite/PCRE.tests "$$BASEURL/PCRE.tests?$$QUERY" && \ + wget -O testsuite/SPENCER.tests "$$BASEURL/rxspencer/tests?$$QUERY" diff --git a/lib/regcomp.c b/lib/regcomp.c index 6aafdd7..91958c9 100644 --- a/lib/regcomp.c +++ b/lib/regcomp.c @@ -566,6 +566,23 @@ weak_alias (__regerror, regerror) #endif +#ifdef RE_ENABLE_I18N +/* This static array is used for the map to single-byte characters when + UTF-8 is used. Otherwise we would allocate memory just to initialize + it the same all the time. UTF-8 is the preferred encoding so this is + a worthwhile optimization. */ +static const bitset utf8_sb_map = +{ + /* Set the first 128 bits. */ +# if UINT_MAX == 0xffffffff + 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +# else +# error "Add case for new unsigned int size" +# endif +}; +#endif + + static void free_dfa_content (re_dfa_t *dfa) { @@ -613,7 +630,8 @@ free_dfa_content (re_dfa_t *dfa) } re_free (dfa->state_table); #ifdef RE_ENABLE_I18N - re_free (dfa->sb_char); + if (dfa->sb_char != utf8_sb_map) + re_free (dfa->sb_char); #endif #ifdef DEBUG re_free (dfa->re_str); @@ -824,7 +842,9 @@ init_dfa (dfa, pat_len) int pat_len; { int table_size; + #ifndef _LIBC char *codeset_name; + #endif memset (dfa, '\0', sizeof (re_dfa_t)); @@ -850,9 +870,14 @@ init_dfa (dfa, pat_len) dfa->mb_cur_max = MB_CUR_MAX; #if defined _LIBC codeset_name = _NL_CURRENT (LC_CTYPE, _NL_CTYPE_CODESET_NAME); -#elif defined HAVE_LANGINFO_CODESET - codeset_name = nl_langinfo (CODESET); + if (dfa->mb_cur_max == 6 && strcmp (codeset_name, "UTF-8") == 0) + dfa->is_utf8 = 1; + dfa->map_notascii = (_NL_CURRENT_WORD (LC_CTYPE, _NL_CTYPE_MAP_TO_NONASCII) + != 0); #else +# ifdef HAVE_LANGINFO_CODESET + codeset_name = nl_langinfo (CODESET); +# else codeset_name = getenv ("LC_ALL"); if (codeset_name == NULL || codeset[0] == '\0') codeset_name = getenv ("LC_CTYPE"); @@ -860,19 +885,14 @@ init_dfa (dfa, pat_len) codeset_name = getenv ("LANG"); if (codeset_name == NULL) codeset_name = ""; - else if (strchr (codeset_name, '.') + else if (strchr (codeset_name, '.') != NULL) codeset_name = strchr (codeset_name, '.') + 1; -#endif +# endif - if (dfa->mb_cur_max == 6 - && (strcasecmp (codeset_name, "UTF-8") == 0 - || strcasecmp (codeset_name, "UTF8") == 0)) + if (strcasecmp (codeset_name, "UTF-8") == 0 + || strcasecmp (codeset_name, "UTF8") == 0) dfa->is_utf8 = 1; -#ifdef _LIBC - dfa->map_notascii = (_NL_CURRENT_WORD (LC_CTYPE, _NL_CTYPE_MAP_TO_NONASCII) - != 0); -#else /* We check exhaustively in the loop below if this charset is a superset of ASCII. */ dfa->map_notascii = 0; @@ -881,28 +901,35 @@ init_dfa (dfa, pat_len) #ifdef RE_ENABLE_I18N if (dfa->mb_cur_max > 1) { - int i, j, ch; - - dfa->sb_char = (re_bitset_ptr_t) calloc (sizeof (bitset), 1); - if (BE (dfa->sb_char == NULL, 0)) - return REG_ESPACE; if (dfa->is_utf8) - memset (dfa->sb_char, 255, sizeof (unsigned int) * BITSET_UINTS / 2); + dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map; else - for (i = 0, ch = 0; i < BITSET_UINTS; ++i) - for (j = 0; j < UINT_BITS; ++j, ++ch) - { - wchar_t wch = __btowc (ch); - if (wch != WEOF) - dfa->sb_char[i] |= 1 << j; -#ifndef _LIBC - if (ch <= 127 && wch != (wchar_t) ch) - dfa->map_notascii = 1; -#endif - } - } -#endif + { + int i, j, ch; + dfa->sb_char = (re_bitset_ptr_t) calloc (sizeof (bitset), 1); + if (BE (dfa->sb_char == NULL, 0)) + return REG_ESPACE; + + /* Clear all bits by, then set those corresponding to single + byte chars. */ + bitset_empty (dfa->sb_char); + + for (i = 0, ch = 0; i < BITSET_UINTS; ++i) + for (j = 0; j < UINT_BITS; ++j, ++ch) + { + wchar_t wch = __btowc (ch); + if (wch != WEOF) + dfa->sb_char[i] |= 1 << j; +# ifndef _LIBC + if (isascii (ch) && wch != (wchar_t) ch) + dfa->map_notascii = 1; +# endif + } + } + } + #endif + if (BE (dfa->nodes == NULL || dfa->state_table == NULL || dfa->subexps == NULL, 0)) return REG_ESPACE; @@ -1575,7 +1602,9 @@ calc_eclosure_iter (new_set, dfa, node, root) ? dfa->nodes[node].opr.ctx_type : 0); /* If the current node has constraints, duplicate all nodes. Since they must inherit the constraints. */ - if (constraint && !dfa->nodes[dfa->edests[node].elems[0]].duplicated) + if (constraint + && dfa->edests[node].nelem + && !dfa->nodes[dfa->edests[node].elems[0]].duplicated) { int org_node, cur_node; org_node = cur_node = node; diff --git a/lib/regex_internal.c b/lib/regex_internal.c index b12b76f..d6f44b6 100644 --- a/lib/regex_internal.c +++ b/lib/regex_internal.c @@ -293,8 +293,8 @@ build_wcs_upper_buffer (pstr) byte_idx = pstr->valid_len; end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; - /* The following optimization assumes that the wchar_t encoding is - always ISO 10646. */ + /* The following optimization assumes that ASCII characters can be + mapped to wide characters with a simple cast. */ if (! pstr->map_notascii && pstr->trans == NULL && !pstr->offsets_needed) { while (byte_idx < end_idx) @@ -308,8 +308,7 @@ build_wcs_upper_buffer (pstr) pstr->mbs[byte_idx] = toupper (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]); /* The next step uses the assumption that wchar_t is encoded - with ISO 10646: all ASCII values can be converted like - this. */ + ASCII-safe: all ASCII values can be converted like this. */ pstr->wcs[byte_idx] = (wchar_t) pstr->mbs[byte_idx]; ++byte_idx; continue; @@ -371,7 +370,7 @@ build_wcs_upper_buffer (pstr) { wchar_t wc; const char *p; -offsets_needed: + offsets_needed: remain_len = end_idx - byte_idx; prev_st = pstr->cur_state; if (BE (pstr->trans != NULL, 0)) @@ -682,6 +681,7 @@ re_string_reconstruct (pstr, idx, eflags) break; } } + if (wc == WEOF) pstr->valid_len = re_string_skip_chars (pstr, idx, &wc) - idx; if (BE (pstr->valid_len, 0)) diff --git a/lib/regex_internal.h b/lib/regex_internal.h index e5eda5d..904f133 100644 --- a/lib/regex_internal.h +++ b/lib/regex_internal.h @@ -547,7 +547,9 @@ struct re_backref_cache_entry int str_idx; int subexp_from; int subexp_to; - char more; + /* We need only one byte from the following field. If other small + fields are added the type could be changed to 'char'. */ + int more; }; typedef struct diff --git a/lib/regexec.c b/lib/regexec.c index cc09592..2579434 100644 --- a/lib/regexec.c +++ b/lib/regexec.c @@ -54,7 +54,7 @@ static unsigned re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, int nregs, int regs_allocated) internal_function; static inline re_dfastate_t *acquire_init_state_context (reg_errcode_t *err, const re_match_context_t *mctx, int idx) - internal_function; + __attribute ((always_inline)) internal_function; static reg_errcode_t prune_impossible_nodes (re_match_context_t *mctx) internal_function; static int check_matching (re_match_context_t *mctx, int fl_longest_match, @@ -121,8 +121,8 @@ static reg_errcode_t check_subexp_limits (re_dfa_t *dfa, struct re_backref_cache_entry *bkref_ents, int str_idx) internal_function; static reg_errcode_t sift_states_bkref (re_match_context_t *mctx, - re_sift_context_t *sctx, int str_idx, - const re_node_set *candidates) internal_function; + re_sift_context_t *sctx, + int str_idx, const re_node_set *candidates) internal_function; static reg_errcode_t clean_state_log_if_needed (re_match_context_t *mctx, int next_state_log_idx) internal_function; static reg_errcode_t merge_state_array (re_dfa_t *dfa, re_dfastate_t **dst, @@ -1525,7 +1525,6 @@ sift_states_backward (mctx, sctx) re_match_context_t *mctx; re_sift_context_t *sctx; { - re_dfa_t *const dfa = mctx->dfa; reg_errcode_t err; int null_cnt = 0; int str_idx = sctx->last_str_idx; @@ -1547,7 +1546,6 @@ sift_states_backward (mctx, sctx) /* Then check each states in the state_log. */ while (str_idx > 0) { - int ret; /* Update counters. */ null_cnt = (sctx->sifted_states[str_idx] == NULL) ? null_cnt + 1 : 0; if (null_cnt > mctx->max_mb_elem_len) @@ -1590,7 +1588,7 @@ build_sifted_states (mctx, sctx, str_idx, cur_dest) { re_dfa_t *const dfa = mctx->dfa; re_node_set *cur_src = &mctx->state_log[str_idx]->nodes; - int i, ret, err; + int i; /* Then build the next sifted state. We build the next sifted state on `cur_dest', and update @@ -1603,6 +1601,7 @@ build_sifted_states (mctx, sctx, str_idx, cur_dest) int prev_node = cur_src->elems[i]; int naccepted = 0; re_token_type_t type = dfa->nodes[prev_node].type; + int ret; if (IS_EPSILON_NODE (type)) continue; @@ -1721,7 +1720,7 @@ update_cur_sifted_state (mctx, sctx, str_idx, dest_nodes) err = add_epsilon_src_nodes (dfa, dest_nodes, candidates); if (BE (err != REG_NOERROR, 0)) return err; - + /* Then, check the limitations in the current sift_context. */ if (sctx->limits.nelem) { @@ -1939,7 +1938,6 @@ check_dst_limits_calc_pos (mctx, limit, subexp_idx, from_node, str_idx, bkref_id re_match_context_t *mctx; int limit, subexp_idx, from_node, str_idx, bkref_idx; { - re_dfa_t *const dfa = mctx->dfa; struct re_backref_cache_entry *lim = mctx->bkref_ents + limit; int boundaries; diff --git a/testsuite/SPENCER.tests b/testsuite/SPENCER.tests index f125523..30fff15 100644 --- a/testsuite/SPENCER.tests +++ b/testsuite/SPENCER.tests @@ -1,24 +1,3 @@ -# Copyright 1992, 1993, 1994, 1997 Henry Spencer. All rights reserved. -# This software is not subject to any license of the American Telephone -# and Telegraph Company or of the Regents of the University of California. -# -# Permission is granted to anyone to use this software for any purpose on -# any computer system, and to alter it and redistribute it, subject -# to the following restrictions: -# -# 1. The author is not responsible for the consequences of use of this -# software, no matter how awful, even if they arise from flaws in it. -# -# 2. The origin of this software must not be misrepresented, either by -# explicit claim or by omission. Since few users ever read sources, -# credits must appear in the documentation. -# -# 3. Altered versions must be plainly marked as such, and must not be -# misrepresented as being the original software. Since few users -# ever read sources, credits must appear in the documentation. -# -# 4. This notice may not be removed or altered. -# # regular expression test set # Lines are at least three fields, separated by one or more tabs. "" stands # for an empty field. First field is an RE. Second field is flags. If @@ -526,3 +505,8 @@ Char \([a-z0-9_]*\)\[.* b Char xyz[k Char xyz[k xyz a?b - ab ab -\{0,1\}[0-9]*$ b -5 -5 a*a*a*a*a*a*a* & aaaaaa aaaaaa +(\b){0} - x @x - +\(\b\)\{0,0\} b abc @abc - +a(\b){0}c - ac ac - +a(.*)b(\0){0}c - abc abc @bc,- +a(.*)b(\0){0}c - axbc axbc x,-