diff options
Diffstat (limited to 'lib/regex_internal.c')
| -rw-r--r-- | lib/regex_internal.c | 150 |
1 files changed, 114 insertions, 36 deletions
diff --git a/lib/regex_internal.c b/lib/regex_internal.c index 55fe298..2129888 100644 --- a/lib/regex_internal.c +++ b/lib/regex_internal.c @@ -1,5 +1,6 @@ /* Extended regular expression matching and search library. - Copyright (C) 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc. + Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 Free Software + Foundation, Inc. This file is part of the GNU C Library. Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. @@ -488,27 +489,34 @@ re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) mbstate_t prev_st; Idx rawbuf_idx; size_t mbclen; - wchar_t wc = 0; + wint_t wc = WEOF; /* Skip the characters which are not necessary to check. */ for (rawbuf_idx = pstr->raw_mbs_idx + pstr->valid_raw_len; rawbuf_idx < new_raw_idx;) { + wchar_t wc2; Idx remain_len; remain_len = pstr->len - rawbuf_idx; prev_st = pstr->cur_state; - mbclen = mbrtowc (&wc, (const char *) pstr->raw_mbs + rawbuf_idx, + mbclen = mbrtowc (&wc2, (const char *) pstr->raw_mbs + rawbuf_idx, remain_len, &pstr->cur_state); if (BE (mbclen == (size_t) -2 || mbclen == (size_t) -1 || mbclen == 0, 0)) { - /* We treat these cases as a singlebyte character. */ + /* We treat these cases as a single byte character. */ + if (mbclen == 0 || remain_len == 0) + wc = L'\0'; + else + wc = *(unsigned char *) (pstr->raw_mbs + rawbuf_idx); mbclen = 1; pstr->cur_state = prev_st; } + else + wc = wc2; /* Then proceed the next character. */ rawbuf_idx += mbclen; } - *last_wc = (wint_t) wc; + *last_wc = wc; return rawbuf_idx; } #endif /* RE_ENABLE_I18N */ @@ -590,34 +598,98 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) if (BE (offset != 0, 1)) { - /* Are the characters which are already checked remain? */ - if (BE (offset < pstr->valid_raw_len, 1) -#ifdef RE_ENABLE_I18N - /* Handling this would enlarge the code too much. - Accept a slowdown in that case. */ - && pstr->offsets_needed == 0 -#endif - ) + /* Should the already checked characters be kept? */ + if (BE (offset < pstr->valid_raw_len, 1)) { /* Yes, move them to the front of the buffer. */ - pstr->tip_context = re_string_context_at (pstr, offset - 1, eflags); #ifdef RE_ENABLE_I18N - if (pstr->mb_cur_max > 1) - memmove (pstr->wcs, pstr->wcs + offset, - (pstr->valid_len - offset) * sizeof (wint_t)); + if (BE (pstr->offsets_needed, 0)) + { + Idx low = 0, high = pstr->valid_len, mid; + do + { + mid = (high + low) / 2; + if (pstr->offsets[mid] > offset) + high = mid; + else if (pstr->offsets[mid] < offset) + low = mid + 1; + else + break; + } + while (low < high); + if (pstr->offsets[mid] < offset) + ++mid; + pstr->tip_context = re_string_context_at (pstr, mid - 1, + eflags); + /* This can be quite complicated, so handle specially + only the common and easy case where the character with + different length representation of lower and upper + case is present at or after offset. */ + if (pstr->valid_len > offset + && mid == offset && pstr->offsets[mid] == offset) + { + memmove (pstr->wcs, pstr->wcs + offset, + (pstr->valid_len - offset) * sizeof (wint_t)); + memmove (pstr->mbs, pstr->mbs + offset, pstr->valid_len - offset); + pstr->valid_len -= offset; + pstr->valid_raw_len -= offset; + for (low = 0; low < pstr->valid_len; low++) + pstr->offsets[low] = pstr->offsets[low + offset] - offset; + } + else + { + /* Otherwise, just find out how long the partial multibyte + character at offset is and fill it with WEOF/255. */ + pstr->len = pstr->raw_len - idx + offset; + pstr->stop = pstr->raw_stop - idx + offset; + pstr->offsets_needed = 0; + while (mid > 0 && pstr->offsets[mid - 1] == offset) + --mid; + while (mid < pstr->valid_len) + if (pstr->wcs[mid] != WEOF) + break; + else + ++mid; + if (mid == pstr->valid_len) + pstr->valid_len = 0; + else + { + pstr->valid_len = pstr->offsets[mid] - offset; + if (pstr->valid_len) + { + for (low = 0; low < pstr->valid_len; ++low) + pstr->wcs[low] = WEOF; + memset (pstr->mbs, 255, pstr->valid_len); + } + } + pstr->valid_raw_len = pstr->valid_len; + } + } + else +#endif + { + pstr->tip_context = re_string_context_at (pstr, offset - 1, + eflags); +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + memmove (pstr->wcs, pstr->wcs + offset, + (pstr->valid_len - offset) * sizeof (wint_t)); #endif /* RE_ENABLE_I18N */ - if (BE (pstr->mbs_allocated, 0)) - memmove (pstr->mbs, pstr->mbs + offset, - pstr->valid_len - offset); - pstr->valid_len -= offset; - pstr->valid_raw_len -= offset; + if (BE (pstr->mbs_allocated, 0)) + memmove (pstr->mbs, pstr->mbs + offset, + pstr->valid_len - offset); + pstr->valid_len -= offset; + pstr->valid_raw_len -= offset; #if DEBUG - assert (pstr->valid_len > 0); + assert (pstr->valid_len > 0); #endif + } } else { /* No, skip all characters until IDX. */ + Idx prev_valid_len = pstr->valid_len; + #ifdef RE_ENABLE_I18N if (BE (pstr->offsets_needed, 0)) { @@ -627,7 +699,6 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) } #endif pstr->valid_len = 0; - pstr->valid_raw_len = 0; #ifdef RE_ENABLE_I18N if (pstr->mb_cur_max > 1) { @@ -636,12 +707,14 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) if (pstr->is_utf8) { - const unsigned char *raw, *p, *q, *end; + const unsigned char *raw, *p, *end; /* Special case UTF-8. Multi-byte chars start with any byte other than 0x80 - 0xbf. */ raw = pstr->raw_mbs + pstr->raw_mbs_idx; end = raw + (offset - pstr->mb_cur_max); + if (end < pstr->raw_mbs) + end = pstr->raw_mbs; p = raw + offset - 1; #ifdef _LIBC /* We know the wchar_t encoding is UCS4, so for the simple @@ -649,7 +722,7 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) if (isascii (*p) && BE (pstr->trans == NULL, 1)) { memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); - pstr->valid_len = 0; + /* pstr->valid_len = 0; */ wc = (wchar_t) *p; } else @@ -663,13 +736,11 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) unsigned char buf[6]; size_t mbclen; - q = p; if (BE (pstr->trans != NULL, 0)) { int i = mlen < 6 ? mlen : 6; while (--i >= 0) buf[i] = pstr->trans[p[i]]; - q = buf; } /* XXX Don't use mbrtowc, we know which conversion to use (UTF-8 -> UCS4). */ @@ -690,25 +761,30 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) if (wc == WEOF) pstr->valid_len = re_string_skip_chars (pstr, idx, &wc) - idx; + if (wc == WEOF) + pstr->tip_context + = re_string_context_at (pstr, prev_valid_len - 1, eflags); + else + pstr->tip_context = ((BE (pstr->word_ops_used != 0, 0) + && IS_WIDE_WORD_CHAR (wc)) + ? CONTEXT_WORD + : ((IS_WIDE_NEWLINE (wc) + && pstr->newline_anchor) + ? CONTEXT_NEWLINE : 0)); if (BE (pstr->valid_len, 0)) { for (wcs_idx = 0; wcs_idx < pstr->valid_len; ++wcs_idx) pstr->wcs[wcs_idx] = WEOF; if (pstr->mbs_allocated) - memset (pstr->mbs, -1, pstr->valid_len); + memset (pstr->mbs, 255, pstr->valid_len); } pstr->valid_raw_len = pstr->valid_len; - pstr->tip_context = ((BE (pstr->word_ops_used != 0, 0) - && IS_WIDE_WORD_CHAR (wc)) - ? CONTEXT_WORD - : ((IS_WIDE_NEWLINE (wc) - && pstr->newline_anchor) - ? CONTEXT_NEWLINE : 0)); } else #endif /* RE_ENABLE_I18N */ { int c = pstr->raw_mbs[pstr->raw_mbs_idx + offset - 1]; + pstr->valid_raw_len = 0; if (pstr->trans) c = pstr->trans[c]; pstr->tip_context = (bitset_contain (pstr->word_char, c) @@ -1329,7 +1405,6 @@ static Idx internal_function re_dfa_add_node (re_dfa_t *dfa, re_token_t token) { - int type = token.type; if (BE (dfa->nodes_len >= dfa->nodes_alloc, 0)) { size_t new_nodes_alloc = dfa->nodes_alloc * 2; @@ -1365,8 +1440,11 @@ re_dfa_add_node (re_dfa_t *dfa, re_token_t token) dfa->nodes[dfa->nodes_len] = token; dfa->nodes[dfa->nodes_len].constraint = 0; #ifdef RE_ENABLE_I18N + { + int type = token.type; dfa->nodes[dfa->nodes_len].accept_mb = (type == OP_PERIOD && dfa->mb_cur_max > 1) || type == COMPLEX_BRACKET; + } #endif dfa->nexts[dfa->nodes_len] = REG_MISSING; re_node_set_init_empty (dfa->edests + dfa->nodes_len); |