diff options
Diffstat (limited to 'gl/regex_internal.c')
-rw-r--r-- | gl/regex_internal.c | 394 |
1 files changed, 182 insertions, 212 deletions
diff --git a/gl/regex_internal.c b/gl/regex_internal.c index 899b0ae6..0e6919f3 100644 --- a/gl/regex_internal.c +++ b/gl/regex_internal.c | |||
@@ -1,33 +1,41 @@ | |||
1 | /* Extended regular expression matching and search library. | 1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2013 Free Software Foundation, Inc. | 2 | Copyright (C) 2002-2022 Free Software Foundation, Inc. |
3 | This file is part of the GNU C Library. | 3 | This file is part of the GNU C Library. |
4 | Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. | 4 | Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. |
5 | 5 | ||
6 | The GNU C Library is free software; you can redistribute it and/or | 6 | The GNU C Library is free software; you can redistribute it and/or |
7 | modify it under the terms of the GNU General Public | 7 | modify it under the terms of the GNU Lesser General Public |
8 | License as published by the Free Software Foundation; either | 8 | License as published by the Free Software Foundation; either |
9 | version 3 of the License, or (at your option) any later version. | 9 | version 2.1 of the License, or (at your option) any later version. |
10 | 10 | ||
11 | The GNU C Library is distributed in the hope that it will be useful, | 11 | The GNU C Library is distributed in the hope that it will be useful, |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | 12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | 13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 | General Public License for more details. | 14 | Lesser General Public License for more details. |
15 | 15 | ||
16 | You should have received a copy of the GNU General Public | 16 | You should have received a copy of the GNU Lesser General Public |
17 | License along with the GNU C Library; if not, see | 17 | License along with the GNU C Library; if not, see |
18 | <http://www.gnu.org/licenses/>. */ | 18 | <https://www.gnu.org/licenses/>. */ |
19 | 19 | ||
20 | static void re_string_construct_common (const char *str, Idx len, | 20 | static void re_string_construct_common (const char *str, Idx len, |
21 | re_string_t *pstr, | 21 | re_string_t *pstr, |
22 | RE_TRANSLATE_TYPE trans, bool icase, | 22 | RE_TRANSLATE_TYPE trans, bool icase, |
23 | const re_dfa_t *dfa) internal_function; | 23 | const re_dfa_t *dfa); |
24 | static re_dfastate_t *create_ci_newstate (const re_dfa_t *dfa, | 24 | static re_dfastate_t *create_ci_newstate (const re_dfa_t *dfa, |
25 | const re_node_set *nodes, | 25 | const re_node_set *nodes, |
26 | re_hashval_t hash) internal_function; | 26 | re_hashval_t hash); |
27 | static re_dfastate_t *create_cd_newstate (const re_dfa_t *dfa, | 27 | static re_dfastate_t *create_cd_newstate (const re_dfa_t *dfa, |
28 | const re_node_set *nodes, | 28 | const re_node_set *nodes, |
29 | unsigned int context, | 29 | unsigned int context, |
30 | re_hashval_t hash) internal_function; | 30 | re_hashval_t hash); |
31 | static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, | ||
32 | Idx new_buf_len); | ||
33 | static void build_wcs_buffer (re_string_t *pstr); | ||
34 | static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr); | ||
35 | static void build_upper_buffer (re_string_t *pstr); | ||
36 | static void re_string_translate_buffer (re_string_t *pstr); | ||
37 | static unsigned int re_string_context_at (const re_string_t *input, Idx idx, | ||
38 | int eflags) __attribute__ ((pure)); | ||
31 | 39 | ||
32 | /* Functions for string operation. */ | 40 | /* Functions for string operation. */ |
33 | 41 | ||
@@ -35,7 +43,7 @@ static re_dfastate_t *create_cd_newstate (const re_dfa_t *dfa, | |||
35 | re_string_reconstruct before using the object. */ | 43 | re_string_reconstruct before using the object. */ |
36 | 44 | ||
37 | static reg_errcode_t | 45 | static reg_errcode_t |
38 | internal_function __attribute_warn_unused_result__ | 46 | __attribute_warn_unused_result__ |
39 | re_string_allocate (re_string_t *pstr, const char *str, Idx len, Idx init_len, | 47 | re_string_allocate (re_string_t *pstr, const char *str, Idx len, Idx init_len, |
40 | RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) | 48 | RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) |
41 | { | 49 | { |
@@ -49,7 +57,7 @@ re_string_allocate (re_string_t *pstr, const char *str, Idx len, Idx init_len, | |||
49 | re_string_construct_common (str, len, pstr, trans, icase, dfa); | 57 | re_string_construct_common (str, len, pstr, trans, icase, dfa); |
50 | 58 | ||
51 | ret = re_string_realloc_buffers (pstr, init_buf_len); | 59 | ret = re_string_realloc_buffers (pstr, init_buf_len); |
52 | if (BE (ret != REG_NOERROR, 0)) | 60 | if (__glibc_unlikely (ret != REG_NOERROR)) |
53 | return ret; | 61 | return ret; |
54 | 62 | ||
55 | pstr->word_char = dfa->word_char; | 63 | pstr->word_char = dfa->word_char; |
@@ -63,7 +71,7 @@ re_string_allocate (re_string_t *pstr, const char *str, Idx len, Idx init_len, | |||
63 | /* This function allocate the buffers, and initialize them. */ | 71 | /* This function allocate the buffers, and initialize them. */ |
64 | 72 | ||
65 | static reg_errcode_t | 73 | static reg_errcode_t |
66 | internal_function __attribute_warn_unused_result__ | 74 | __attribute_warn_unused_result__ |
67 | re_string_construct (re_string_t *pstr, const char *str, Idx len, | 75 | re_string_construct (re_string_t *pstr, const char *str, Idx len, |
68 | RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) | 76 | RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) |
69 | { | 77 | { |
@@ -74,41 +82,37 @@ re_string_construct (re_string_t *pstr, const char *str, Idx len, | |||
74 | if (len > 0) | 82 | if (len > 0) |
75 | { | 83 | { |
76 | ret = re_string_realloc_buffers (pstr, len + 1); | 84 | ret = re_string_realloc_buffers (pstr, len + 1); |
77 | if (BE (ret != REG_NOERROR, 0)) | 85 | if (__glibc_unlikely (ret != REG_NOERROR)) |
78 | return ret; | 86 | return ret; |
79 | } | 87 | } |
80 | pstr->mbs = pstr->mbs_allocated ? pstr->mbs : (unsigned char *) str; | 88 | pstr->mbs = pstr->mbs_allocated ? pstr->mbs : (unsigned char *) str; |
81 | 89 | ||
82 | if (icase) | 90 | if (icase) |
83 | { | 91 | { |
84 | #ifdef RE_ENABLE_I18N | ||
85 | if (dfa->mb_cur_max > 1) | 92 | if (dfa->mb_cur_max > 1) |
86 | { | 93 | { |
87 | while (1) | 94 | while (1) |
88 | { | 95 | { |
89 | ret = build_wcs_upper_buffer (pstr); | 96 | ret = build_wcs_upper_buffer (pstr); |
90 | if (BE (ret != REG_NOERROR, 0)) | 97 | if (__glibc_unlikely (ret != REG_NOERROR)) |
91 | return ret; | 98 | return ret; |
92 | if (pstr->valid_raw_len >= len) | 99 | if (pstr->valid_raw_len >= len) |
93 | break; | 100 | break; |
94 | if (pstr->bufs_len > pstr->valid_len + dfa->mb_cur_max) | 101 | if (pstr->bufs_len > pstr->valid_len + dfa->mb_cur_max) |
95 | break; | 102 | break; |
96 | ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2); | 103 | ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2); |
97 | if (BE (ret != REG_NOERROR, 0)) | 104 | if (__glibc_unlikely (ret != REG_NOERROR)) |
98 | return ret; | 105 | return ret; |
99 | } | 106 | } |
100 | } | 107 | } |
101 | else | 108 | else |
102 | #endif /* RE_ENABLE_I18N */ | ||
103 | build_upper_buffer (pstr); | 109 | build_upper_buffer (pstr); |
104 | } | 110 | } |
105 | else | 111 | else |
106 | { | 112 | { |
107 | #ifdef RE_ENABLE_I18N | ||
108 | if (dfa->mb_cur_max > 1) | 113 | if (dfa->mb_cur_max > 1) |
109 | build_wcs_buffer (pstr); | 114 | build_wcs_buffer (pstr); |
110 | else | 115 | else |
111 | #endif /* RE_ENABLE_I18N */ | ||
112 | { | 116 | { |
113 | if (trans != NULL) | 117 | if (trans != NULL) |
114 | re_string_translate_buffer (pstr); | 118 | re_string_translate_buffer (pstr); |
@@ -126,37 +130,36 @@ re_string_construct (re_string_t *pstr, const char *str, Idx len, | |||
126 | /* Helper functions for re_string_allocate, and re_string_construct. */ | 130 | /* Helper functions for re_string_allocate, and re_string_construct. */ |
127 | 131 | ||
128 | static reg_errcode_t | 132 | static reg_errcode_t |
129 | internal_function __attribute_warn_unused_result__ | 133 | __attribute_warn_unused_result__ |
130 | re_string_realloc_buffers (re_string_t *pstr, Idx new_buf_len) | 134 | re_string_realloc_buffers (re_string_t *pstr, Idx new_buf_len) |
131 | { | 135 | { |
132 | #ifdef RE_ENABLE_I18N | ||
133 | if (pstr->mb_cur_max > 1) | 136 | if (pstr->mb_cur_max > 1) |
134 | { | 137 | { |
135 | wint_t *new_wcs; | 138 | wint_t *new_wcs; |
136 | 139 | ||
137 | /* Avoid overflow in realloc. */ | 140 | /* Avoid overflow in realloc. */ |
138 | const size_t max_object_size = MAX (sizeof (wint_t), sizeof (Idx)); | 141 | const size_t max_object_size = MAX (sizeof (wint_t), sizeof (Idx)); |
139 | if (BE (MIN (IDX_MAX, SIZE_MAX / max_object_size) < new_buf_len, 0)) | 142 | if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / max_object_size) |
143 | < new_buf_len)) | ||
140 | return REG_ESPACE; | 144 | return REG_ESPACE; |
141 | 145 | ||
142 | new_wcs = re_realloc (pstr->wcs, wint_t, new_buf_len); | 146 | new_wcs = re_realloc (pstr->wcs, wint_t, new_buf_len); |
143 | if (BE (new_wcs == NULL, 0)) | 147 | if (__glibc_unlikely (new_wcs == NULL)) |
144 | return REG_ESPACE; | 148 | return REG_ESPACE; |
145 | pstr->wcs = new_wcs; | 149 | pstr->wcs = new_wcs; |
146 | if (pstr->offsets != NULL) | 150 | if (pstr->offsets != NULL) |
147 | { | 151 | { |
148 | Idx *new_offsets = re_realloc (pstr->offsets, Idx, new_buf_len); | 152 | Idx *new_offsets = re_realloc (pstr->offsets, Idx, new_buf_len); |
149 | if (BE (new_offsets == NULL, 0)) | 153 | if (__glibc_unlikely (new_offsets == NULL)) |
150 | return REG_ESPACE; | 154 | return REG_ESPACE; |
151 | pstr->offsets = new_offsets; | 155 | pstr->offsets = new_offsets; |
152 | } | 156 | } |
153 | } | 157 | } |
154 | #endif /* RE_ENABLE_I18N */ | ||
155 | if (pstr->mbs_allocated) | 158 | if (pstr->mbs_allocated) |
156 | { | 159 | { |
157 | unsigned char *new_mbs = re_realloc (pstr->mbs, unsigned char, | 160 | unsigned char *new_mbs = re_realloc (pstr->mbs, unsigned char, |
158 | new_buf_len); | 161 | new_buf_len); |
159 | if (BE (new_mbs == NULL, 0)) | 162 | if (__glibc_unlikely (new_mbs == NULL)) |
160 | return REG_ESPACE; | 163 | return REG_ESPACE; |
161 | pstr->mbs = new_mbs; | 164 | pstr->mbs = new_mbs; |
162 | } | 165 | } |
@@ -166,7 +169,6 @@ re_string_realloc_buffers (re_string_t *pstr, Idx new_buf_len) | |||
166 | 169 | ||
167 | 170 | ||
168 | static void | 171 | static void |
169 | internal_function | ||
170 | re_string_construct_common (const char *str, Idx len, re_string_t *pstr, | 172 | re_string_construct_common (const char *str, Idx len, re_string_t *pstr, |
171 | RE_TRANSLATE_TYPE trans, bool icase, | 173 | RE_TRANSLATE_TYPE trans, bool icase, |
172 | const re_dfa_t *dfa) | 174 | const re_dfa_t *dfa) |
@@ -184,7 +186,6 @@ re_string_construct_common (const char *str, Idx len, re_string_t *pstr, | |||
184 | pstr->raw_stop = pstr->stop; | 186 | pstr->raw_stop = pstr->stop; |
185 | } | 187 | } |
186 | 188 | ||
187 | #ifdef RE_ENABLE_I18N | ||
188 | 189 | ||
189 | /* Build wide character buffer PSTR->WCS. | 190 | /* Build wide character buffer PSTR->WCS. |
190 | If the byte sequence of the string are: | 191 | If the byte sequence of the string are: |
@@ -198,12 +199,11 @@ re_string_construct_common (const char *str, Idx len, re_string_t *pstr, | |||
198 | built and starts from PSTR->VALID_LEN. */ | 199 | built and starts from PSTR->VALID_LEN. */ |
199 | 200 | ||
200 | static void | 201 | static void |
201 | internal_function | ||
202 | build_wcs_buffer (re_string_t *pstr) | 202 | build_wcs_buffer (re_string_t *pstr) |
203 | { | 203 | { |
204 | #ifdef _LIBC | 204 | #ifdef _LIBC |
205 | unsigned char buf[MB_LEN_MAX]; | 205 | unsigned char buf[MB_LEN_MAX]; |
206 | assert (MB_LEN_MAX >= pstr->mb_cur_max); | 206 | DEBUG_ASSERT (MB_LEN_MAX >= pstr->mb_cur_max); |
207 | #else | 207 | #else |
208 | unsigned char buf[64]; | 208 | unsigned char buf[64]; |
209 | #endif | 209 | #endif |
@@ -222,7 +222,7 @@ build_wcs_buffer (re_string_t *pstr) | |||
222 | remain_len = end_idx - byte_idx; | 222 | remain_len = end_idx - byte_idx; |
223 | prev_st = pstr->cur_state; | 223 | prev_st = pstr->cur_state; |
224 | /* Apply the translation if we need. */ | 224 | /* Apply the translation if we need. */ |
225 | if (BE (pstr->trans != NULL, 0)) | 225 | if (__glibc_unlikely (pstr->trans != NULL)) |
226 | { | 226 | { |
227 | int i, ch; | 227 | int i, ch; |
228 | 228 | ||
@@ -236,17 +236,18 @@ build_wcs_buffer (re_string_t *pstr) | |||
236 | else | 236 | else |
237 | p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + byte_idx; | 237 | p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + byte_idx; |
238 | mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); | 238 | mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); |
239 | if (BE (mbclen == (size_t) -1 || mbclen == 0 | 239 | if (__glibc_unlikely (mbclen == (size_t) -1 || mbclen == 0 |
240 | || (mbclen == (size_t) -2 && pstr->bufs_len >= pstr->len), 0)) | 240 | || (mbclen == (size_t) -2 |
241 | && pstr->bufs_len >= pstr->len))) | ||
241 | { | 242 | { |
242 | /* We treat these cases as a singlebyte character. */ | 243 | /* We treat these cases as a singlebyte character. */ |
243 | mbclen = 1; | 244 | mbclen = 1; |
244 | wc = (wchar_t) pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; | 245 | wc = (wchar_t) pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; |
245 | if (BE (pstr->trans != NULL, 0)) | 246 | if (__glibc_unlikely (pstr->trans != NULL)) |
246 | wc = pstr->trans[wc]; | 247 | wc = pstr->trans[wc]; |
247 | pstr->cur_state = prev_st; | 248 | pstr->cur_state = prev_st; |
248 | } | 249 | } |
249 | else if (BE (mbclen == (size_t) -2, 0)) | 250 | else if (__glibc_unlikely (mbclen == (size_t) -2)) |
250 | { | 251 | { |
251 | /* The buffer doesn't have enough space, finish to build. */ | 252 | /* The buffer doesn't have enough space, finish to build. */ |
252 | pstr->cur_state = prev_st; | 253 | pstr->cur_state = prev_st; |
@@ -267,7 +268,7 @@ build_wcs_buffer (re_string_t *pstr) | |||
267 | but for REG_ICASE. */ | 268 | but for REG_ICASE. */ |
268 | 269 | ||
269 | static reg_errcode_t | 270 | static reg_errcode_t |
270 | internal_function __attribute_warn_unused_result__ | 271 | __attribute_warn_unused_result__ |
271 | build_wcs_upper_buffer (re_string_t *pstr) | 272 | build_wcs_upper_buffer (re_string_t *pstr) |
272 | { | 273 | { |
273 | mbstate_t prev_st; | 274 | mbstate_t prev_st; |
@@ -275,7 +276,7 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
275 | size_t mbclen; | 276 | size_t mbclen; |
276 | #ifdef _LIBC | 277 | #ifdef _LIBC |
277 | char buf[MB_LEN_MAX]; | 278 | char buf[MB_LEN_MAX]; |
278 | assert (MB_LEN_MAX >= pstr->mb_cur_max); | 279 | DEBUG_ASSERT (pstr->mb_cur_max <= MB_LEN_MAX); |
279 | #else | 280 | #else |
280 | char buf[64]; | 281 | char buf[64]; |
281 | #endif | 282 | #endif |
@@ -290,18 +291,20 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
290 | while (byte_idx < end_idx) | 291 | while (byte_idx < end_idx) |
291 | { | 292 | { |
292 | wchar_t wc; | 293 | wchar_t wc; |
294 | unsigned char ch = pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; | ||
293 | 295 | ||
294 | if (isascii (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]) | 296 | if (isascii (ch) && mbsinit (&pstr->cur_state)) |
295 | && mbsinit (&pstr->cur_state)) | ||
296 | { | 297 | { |
297 | /* In case of a singlebyte character. */ | ||
298 | pstr->mbs[byte_idx] | ||
299 | = toupper (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]); | ||
300 | /* The next step uses the assumption that wchar_t is encoded | 298 | /* The next step uses the assumption that wchar_t is encoded |
301 | ASCII-safe: all ASCII values can be converted like this. */ | 299 | ASCII-safe: all ASCII values can be converted like this. */ |
302 | pstr->wcs[byte_idx] = (wchar_t) pstr->mbs[byte_idx]; | 300 | wchar_t wcu = __towupper (ch); |
303 | ++byte_idx; | 301 | if (isascii (wcu)) |
304 | continue; | 302 | { |
303 | pstr->mbs[byte_idx] = wcu; | ||
304 | pstr->wcs[byte_idx] = wcu; | ||
305 | byte_idx++; | ||
306 | continue; | ||
307 | } | ||
305 | } | 308 | } |
306 | 309 | ||
307 | remain_len = end_idx - byte_idx; | 310 | remain_len = end_idx - byte_idx; |
@@ -309,16 +312,15 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
309 | mbclen = __mbrtowc (&wc, | 312 | mbclen = __mbrtowc (&wc, |
310 | ((const char *) pstr->raw_mbs + pstr->raw_mbs_idx | 313 | ((const char *) pstr->raw_mbs + pstr->raw_mbs_idx |
311 | + byte_idx), remain_len, &pstr->cur_state); | 314 | + byte_idx), remain_len, &pstr->cur_state); |
312 | if (BE (mbclen < (size_t) -2, 1)) | 315 | if (__glibc_likely (0 < mbclen && mbclen < (size_t) -2)) |
313 | { | 316 | { |
314 | wchar_t wcu = wc; | 317 | wchar_t wcu = __towupper (wc); |
315 | if (iswlower (wc)) | 318 | if (wcu != wc) |
316 | { | 319 | { |
317 | size_t mbcdlen; | 320 | size_t mbcdlen; |
318 | 321 | ||
319 | wcu = towupper (wc); | 322 | mbcdlen = __wcrtomb (buf, wcu, &prev_st); |
320 | mbcdlen = wcrtomb (buf, wcu, &prev_st); | 323 | if (__glibc_likely (mbclen == mbcdlen)) |
321 | if (BE (mbclen == mbcdlen, 1)) | ||
322 | memcpy (pstr->mbs + byte_idx, buf, mbclen); | 324 | memcpy (pstr->mbs + byte_idx, buf, mbclen); |
323 | else | 325 | else |
324 | { | 326 | { |
@@ -339,11 +341,10 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
339 | { | 341 | { |
340 | /* It is an invalid character, an incomplete character | 342 | /* It is an invalid character, an incomplete character |
341 | at the end of the string, or '\0'. Just use the byte. */ | 343 | at the end of the string, or '\0'. Just use the byte. */ |
342 | int ch = pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; | ||
343 | pstr->mbs[byte_idx] = ch; | 344 | pstr->mbs[byte_idx] = ch; |
344 | /* And also cast it to wide char. */ | 345 | /* And also cast it to wide char. */ |
345 | pstr->wcs[byte_idx++] = (wchar_t) ch; | 346 | pstr->wcs[byte_idx++] = (wchar_t) ch; |
346 | if (BE (mbclen == (size_t) -1, 0)) | 347 | if (__glibc_unlikely (mbclen == (size_t) -1)) |
347 | pstr->cur_state = prev_st; | 348 | pstr->cur_state = prev_st; |
348 | } | 349 | } |
349 | else | 350 | else |
@@ -365,7 +366,7 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
365 | offsets_needed: | 366 | offsets_needed: |
366 | remain_len = end_idx - byte_idx; | 367 | remain_len = end_idx - byte_idx; |
367 | prev_st = pstr->cur_state; | 368 | prev_st = pstr->cur_state; |
368 | if (BE (pstr->trans != NULL, 0)) | 369 | if (__glibc_unlikely (pstr->trans != NULL)) |
369 | { | 370 | { |
370 | int i, ch; | 371 | int i, ch; |
371 | 372 | ||
@@ -379,16 +380,15 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
379 | else | 380 | else |
380 | p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + src_idx; | 381 | p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + src_idx; |
381 | mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); | 382 | mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); |
382 | if (BE (mbclen < (size_t) -2, 1)) | 383 | if (__glibc_likely (0 < mbclen && mbclen < (size_t) -2)) |
383 | { | 384 | { |
384 | wchar_t wcu = wc; | 385 | wchar_t wcu = __towupper (wc); |
385 | if (iswlower (wc)) | 386 | if (wcu != wc) |
386 | { | 387 | { |
387 | size_t mbcdlen; | 388 | size_t mbcdlen; |
388 | 389 | ||
389 | wcu = towupper (wc); | 390 | mbcdlen = __wcrtomb ((char *) buf, wcu, &prev_st); |
390 | mbcdlen = wcrtomb ((char *) buf, wcu, &prev_st); | 391 | if (__glibc_likely (mbclen == mbcdlen)) |
391 | if (BE (mbclen == mbcdlen, 1)) | ||
392 | memcpy (pstr->mbs + byte_idx, buf, mbclen); | 392 | memcpy (pstr->mbs + byte_idx, buf, mbclen); |
393 | else if (mbcdlen != (size_t) -1) | 393 | else if (mbcdlen != (size_t) -1) |
394 | { | 394 | { |
@@ -438,7 +438,7 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
438 | else | 438 | else |
439 | memcpy (pstr->mbs + byte_idx, p, mbclen); | 439 | memcpy (pstr->mbs + byte_idx, p, mbclen); |
440 | 440 | ||
441 | if (BE (pstr->offsets_needed != 0, 0)) | 441 | if (__glibc_unlikely (pstr->offsets_needed != 0)) |
442 | { | 442 | { |
443 | size_t i; | 443 | size_t i; |
444 | for (i = 0; i < mbclen; ++i) | 444 | for (i = 0; i < mbclen; ++i) |
@@ -457,17 +457,17 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
457 | /* It is an invalid character or '\0'. Just use the byte. */ | 457 | /* It is an invalid character or '\0'. Just use the byte. */ |
458 | int ch = pstr->raw_mbs[pstr->raw_mbs_idx + src_idx]; | 458 | int ch = pstr->raw_mbs[pstr->raw_mbs_idx + src_idx]; |
459 | 459 | ||
460 | if (BE (pstr->trans != NULL, 0)) | 460 | if (__glibc_unlikely (pstr->trans != NULL)) |
461 | ch = pstr->trans [ch]; | 461 | ch = pstr->trans [ch]; |
462 | pstr->mbs[byte_idx] = ch; | 462 | pstr->mbs[byte_idx] = ch; |
463 | 463 | ||
464 | if (BE (pstr->offsets_needed != 0, 0)) | 464 | if (__glibc_unlikely (pstr->offsets_needed != 0)) |
465 | pstr->offsets[byte_idx] = src_idx; | 465 | pstr->offsets[byte_idx] = src_idx; |
466 | ++src_idx; | 466 | ++src_idx; |
467 | 467 | ||
468 | /* And also cast it to wide char. */ | 468 | /* And also cast it to wide char. */ |
469 | pstr->wcs[byte_idx++] = (wchar_t) ch; | 469 | pstr->wcs[byte_idx++] = (wchar_t) ch; |
470 | if (BE (mbclen == (size_t) -1, 0)) | 470 | if (__glibc_unlikely (mbclen == (size_t) -1)) |
471 | pstr->cur_state = prev_st; | 471 | pstr->cur_state = prev_st; |
472 | } | 472 | } |
473 | else | 473 | else |
@@ -486,7 +486,6 @@ build_wcs_upper_buffer (re_string_t *pstr) | |||
486 | Return the index. */ | 486 | Return the index. */ |
487 | 487 | ||
488 | static Idx | 488 | static Idx |
489 | internal_function | ||
490 | re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) | 489 | re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) |
491 | { | 490 | { |
492 | mbstate_t prev_st; | 491 | mbstate_t prev_st; |
@@ -503,7 +502,8 @@ re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) | |||
503 | prev_st = pstr->cur_state; | 502 | prev_st = pstr->cur_state; |
504 | mbclen = __mbrtowc (&wc2, (const char *) pstr->raw_mbs + rawbuf_idx, | 503 | mbclen = __mbrtowc (&wc2, (const char *) pstr->raw_mbs + rawbuf_idx, |
505 | remain_len, &pstr->cur_state); | 504 | remain_len, &pstr->cur_state); |
506 | if (BE (mbclen == (size_t) -2 || mbclen == (size_t) -1 || mbclen == 0, 0)) | 505 | if (__glibc_unlikely (mbclen == (size_t) -2 || mbclen == (size_t) -1 |
506 | || mbclen == 0)) | ||
507 | { | 507 | { |
508 | /* We treat these cases as a single byte character. */ | 508 | /* We treat these cases as a single byte character. */ |
509 | if (mbclen == 0 || remain_len == 0) | 509 | if (mbclen == 0 || remain_len == 0) |
@@ -521,13 +521,11 @@ re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) | |||
521 | *last_wc = wc; | 521 | *last_wc = wc; |
522 | return rawbuf_idx; | 522 | return rawbuf_idx; |
523 | } | 523 | } |
524 | #endif /* RE_ENABLE_I18N */ | ||
525 | 524 | ||
526 | /* Build the buffer PSTR->MBS, and apply the translation if we need. | 525 | /* Build the buffer PSTR->MBS, and apply the translation if we need. |
527 | This function is used in case of REG_ICASE. */ | 526 | This function is used in case of REG_ICASE. */ |
528 | 527 | ||
529 | static void | 528 | static void |
530 | internal_function | ||
531 | build_upper_buffer (re_string_t *pstr) | 529 | build_upper_buffer (re_string_t *pstr) |
532 | { | 530 | { |
533 | Idx char_idx, end_idx; | 531 | Idx char_idx, end_idx; |
@@ -536,12 +534,9 @@ build_upper_buffer (re_string_t *pstr) | |||
536 | for (char_idx = pstr->valid_len; char_idx < end_idx; ++char_idx) | 534 | for (char_idx = pstr->valid_len; char_idx < end_idx; ++char_idx) |
537 | { | 535 | { |
538 | int ch = pstr->raw_mbs[pstr->raw_mbs_idx + char_idx]; | 536 | int ch = pstr->raw_mbs[pstr->raw_mbs_idx + char_idx]; |
539 | if (BE (pstr->trans != NULL, 0)) | 537 | if (__glibc_unlikely (pstr->trans != NULL)) |
540 | ch = pstr->trans[ch]; | 538 | ch = pstr->trans[ch]; |
541 | if (islower (ch)) | 539 | pstr->mbs[char_idx] = toupper (ch); |
542 | pstr->mbs[char_idx] = toupper (ch); | ||
543 | else | ||
544 | pstr->mbs[char_idx] = ch; | ||
545 | } | 540 | } |
546 | pstr->valid_len = char_idx; | 541 | pstr->valid_len = char_idx; |
547 | pstr->valid_raw_len = char_idx; | 542 | pstr->valid_raw_len = char_idx; |
@@ -550,7 +545,6 @@ build_upper_buffer (re_string_t *pstr) | |||
550 | /* Apply TRANS to the buffer in PSTR. */ | 545 | /* Apply TRANS to the buffer in PSTR. */ |
551 | 546 | ||
552 | static void | 547 | static void |
553 | internal_function | ||
554 | re_string_translate_buffer (re_string_t *pstr) | 548 | re_string_translate_buffer (re_string_t *pstr) |
555 | { | 549 | { |
556 | Idx buf_idx, end_idx; | 550 | Idx buf_idx, end_idx; |
@@ -571,20 +565,18 @@ re_string_translate_buffer (re_string_t *pstr) | |||
571 | convert to upper case in case of REG_ICASE, apply translation. */ | 565 | convert to upper case in case of REG_ICASE, apply translation. */ |
572 | 566 | ||
573 | static reg_errcode_t | 567 | static reg_errcode_t |
574 | internal_function __attribute_warn_unused_result__ | 568 | __attribute_warn_unused_result__ |
575 | re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | 569 | re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) |
576 | { | 570 | { |
577 | Idx offset; | 571 | Idx offset; |
578 | 572 | ||
579 | if (BE (pstr->raw_mbs_idx <= idx, 0)) | 573 | if (__glibc_unlikely (pstr->raw_mbs_idx <= idx)) |
580 | offset = idx - pstr->raw_mbs_idx; | 574 | offset = idx - pstr->raw_mbs_idx; |
581 | else | 575 | else |
582 | { | 576 | { |
583 | /* Reset buffer. */ | 577 | /* Reset buffer. */ |
584 | #ifdef RE_ENABLE_I18N | ||
585 | if (pstr->mb_cur_max > 1) | 578 | if (pstr->mb_cur_max > 1) |
586 | memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); | 579 | memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); |
587 | #endif /* RE_ENABLE_I18N */ | ||
588 | pstr->len = pstr->raw_len; | 580 | pstr->len = pstr->raw_len; |
589 | pstr->stop = pstr->raw_stop; | 581 | pstr->stop = pstr->raw_stop; |
590 | pstr->valid_len = 0; | 582 | pstr->valid_len = 0; |
@@ -598,14 +590,13 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
598 | offset = idx; | 590 | offset = idx; |
599 | } | 591 | } |
600 | 592 | ||
601 | if (BE (offset != 0, 1)) | 593 | if (__glibc_likely (offset != 0)) |
602 | { | 594 | { |
603 | /* Should the already checked characters be kept? */ | 595 | /* Should the already checked characters be kept? */ |
604 | if (BE (offset < pstr->valid_raw_len, 1)) | 596 | if (__glibc_likely (offset < pstr->valid_raw_len)) |
605 | { | 597 | { |
606 | /* Yes, move them to the front of the buffer. */ | 598 | /* Yes, move them to the front of the buffer. */ |
607 | #ifdef RE_ENABLE_I18N | 599 | if (__glibc_unlikely (pstr->offsets_needed)) |
608 | if (BE (pstr->offsets_needed, 0)) | ||
609 | { | 600 | { |
610 | Idx low = 0, high = pstr->valid_len, mid; | 601 | Idx low = 0, high = pstr->valid_len, mid; |
611 | do | 602 | do |
@@ -668,40 +659,32 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
668 | } | 659 | } |
669 | } | 660 | } |
670 | else | 661 | else |
671 | #endif | ||
672 | { | 662 | { |
673 | pstr->tip_context = re_string_context_at (pstr, offset - 1, | 663 | pstr->tip_context = re_string_context_at (pstr, offset - 1, |
674 | eflags); | 664 | eflags); |
675 | #ifdef RE_ENABLE_I18N | ||
676 | if (pstr->mb_cur_max > 1) | 665 | if (pstr->mb_cur_max > 1) |
677 | memmove (pstr->wcs, pstr->wcs + offset, | 666 | memmove (pstr->wcs, pstr->wcs + offset, |
678 | (pstr->valid_len - offset) * sizeof (wint_t)); | 667 | (pstr->valid_len - offset) * sizeof (wint_t)); |
679 | #endif /* RE_ENABLE_I18N */ | 668 | if (__glibc_unlikely (pstr->mbs_allocated)) |
680 | if (BE (pstr->mbs_allocated, 0)) | ||
681 | memmove (pstr->mbs, pstr->mbs + offset, | 669 | memmove (pstr->mbs, pstr->mbs + offset, |
682 | pstr->valid_len - offset); | 670 | pstr->valid_len - offset); |
683 | pstr->valid_len -= offset; | 671 | pstr->valid_len -= offset; |
684 | pstr->valid_raw_len -= offset; | 672 | pstr->valid_raw_len -= offset; |
685 | #if DEBUG | 673 | DEBUG_ASSERT (pstr->valid_len > 0); |
686 | assert (pstr->valid_len > 0); | ||
687 | #endif | ||
688 | } | 674 | } |
689 | } | 675 | } |
690 | else | 676 | else |
691 | { | 677 | { |
692 | #ifdef RE_ENABLE_I18N | ||
693 | /* No, skip all characters until IDX. */ | 678 | /* No, skip all characters until IDX. */ |
694 | Idx prev_valid_len = pstr->valid_len; | 679 | Idx prev_valid_len = pstr->valid_len; |
695 | 680 | ||
696 | if (BE (pstr->offsets_needed, 0)) | 681 | if (__glibc_unlikely (pstr->offsets_needed)) |
697 | { | 682 | { |
698 | pstr->len = pstr->raw_len - idx + offset; | 683 | pstr->len = pstr->raw_len - idx + offset; |
699 | pstr->stop = pstr->raw_stop - idx + offset; | 684 | pstr->stop = pstr->raw_stop - idx + offset; |
700 | pstr->offsets_needed = 0; | 685 | pstr->offsets_needed = 0; |
701 | } | 686 | } |
702 | #endif | ||
703 | pstr->valid_len = 0; | 687 | pstr->valid_len = 0; |
704 | #ifdef RE_ENABLE_I18N | ||
705 | if (pstr->mb_cur_max > 1) | 688 | if (pstr->mb_cur_max > 1) |
706 | { | 689 | { |
707 | Idx wcs_idx; | 690 | Idx wcs_idx; |
@@ -721,7 +704,7 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
721 | #ifdef _LIBC | 704 | #ifdef _LIBC |
722 | /* We know the wchar_t encoding is UCS4, so for the simple | 705 | /* We know the wchar_t encoding is UCS4, so for the simple |
723 | case, ASCII characters, skip the conversion step. */ | 706 | case, ASCII characters, skip the conversion step. */ |
724 | if (isascii (*p) && BE (pstr->trans == NULL, 1)) | 707 | if (isascii (*p) && __glibc_likely (pstr->trans == NULL)) |
725 | { | 708 | { |
726 | memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); | 709 | memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); |
727 | /* pstr->valid_len = 0; */ | 710 | /* pstr->valid_len = 0; */ |
@@ -739,7 +722,7 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
739 | size_t mbclen; | 722 | size_t mbclen; |
740 | 723 | ||
741 | const unsigned char *pp = p; | 724 | const unsigned char *pp = p; |
742 | if (BE (pstr->trans != NULL, 0)) | 725 | if (__glibc_unlikely (pstr->trans != NULL)) |
743 | { | 726 | { |
744 | int i = mlen < 6 ? mlen : 6; | 727 | int i = mlen < 6 ? mlen : 6; |
745 | while (--i >= 0) | 728 | while (--i >= 0) |
@@ -769,13 +752,13 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
769 | pstr->tip_context | 752 | pstr->tip_context |
770 | = re_string_context_at (pstr, prev_valid_len - 1, eflags); | 753 | = re_string_context_at (pstr, prev_valid_len - 1, eflags); |
771 | else | 754 | else |
772 | pstr->tip_context = ((BE (pstr->word_ops_used != 0, 0) | 755 | pstr->tip_context = ((__glibc_unlikely (pstr->word_ops_used != 0) |
773 | && IS_WIDE_WORD_CHAR (wc)) | 756 | && IS_WIDE_WORD_CHAR (wc)) |
774 | ? CONTEXT_WORD | 757 | ? CONTEXT_WORD |
775 | : ((IS_WIDE_NEWLINE (wc) | 758 | : ((IS_WIDE_NEWLINE (wc) |
776 | && pstr->newline_anchor) | 759 | && pstr->newline_anchor) |
777 | ? CONTEXT_NEWLINE : 0)); | 760 | ? CONTEXT_NEWLINE : 0)); |
778 | if (BE (pstr->valid_len, 0)) | 761 | if (__glibc_unlikely (pstr->valid_len)) |
779 | { | 762 | { |
780 | for (wcs_idx = 0; wcs_idx < pstr->valid_len; ++wcs_idx) | 763 | for (wcs_idx = 0; wcs_idx < pstr->valid_len; ++wcs_idx) |
781 | pstr->wcs[wcs_idx] = WEOF; | 764 | pstr->wcs[wcs_idx] = WEOF; |
@@ -785,7 +768,6 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
785 | pstr->valid_raw_len = pstr->valid_len; | 768 | pstr->valid_raw_len = pstr->valid_len; |
786 | } | 769 | } |
787 | else | 770 | else |
788 | #endif /* RE_ENABLE_I18N */ | ||
789 | { | 771 | { |
790 | int c = pstr->raw_mbs[pstr->raw_mbs_idx + offset - 1]; | 772 | int c = pstr->raw_mbs[pstr->raw_mbs_idx + offset - 1]; |
791 | pstr->valid_raw_len = 0; | 773 | pstr->valid_raw_len = 0; |
@@ -797,7 +779,7 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
797 | ? CONTEXT_NEWLINE : 0)); | 779 | ? CONTEXT_NEWLINE : 0)); |
798 | } | 780 | } |
799 | } | 781 | } |
800 | if (!BE (pstr->mbs_allocated, 0)) | 782 | if (!__glibc_unlikely (pstr->mbs_allocated)) |
801 | pstr->mbs += offset; | 783 | pstr->mbs += offset; |
802 | } | 784 | } |
803 | pstr->raw_mbs_idx = idx; | 785 | pstr->raw_mbs_idx = idx; |
@@ -805,21 +787,19 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
805 | pstr->stop -= offset; | 787 | pstr->stop -= offset; |
806 | 788 | ||
807 | /* Then build the buffers. */ | 789 | /* Then build the buffers. */ |
808 | #ifdef RE_ENABLE_I18N | ||
809 | if (pstr->mb_cur_max > 1) | 790 | if (pstr->mb_cur_max > 1) |
810 | { | 791 | { |
811 | if (pstr->icase) | 792 | if (pstr->icase) |
812 | { | 793 | { |
813 | reg_errcode_t ret = build_wcs_upper_buffer (pstr); | 794 | reg_errcode_t ret = build_wcs_upper_buffer (pstr); |
814 | if (BE (ret != REG_NOERROR, 0)) | 795 | if (__glibc_unlikely (ret != REG_NOERROR)) |
815 | return ret; | 796 | return ret; |
816 | } | 797 | } |
817 | else | 798 | else |
818 | build_wcs_buffer (pstr); | 799 | build_wcs_buffer (pstr); |
819 | } | 800 | } |
820 | else | 801 | else |
821 | #endif /* RE_ENABLE_I18N */ | 802 | if (__glibc_unlikely (pstr->mbs_allocated)) |
822 | if (BE (pstr->mbs_allocated, 0)) | ||
823 | { | 803 | { |
824 | if (pstr->icase) | 804 | if (pstr->icase) |
825 | build_upper_buffer (pstr); | 805 | build_upper_buffer (pstr); |
@@ -834,50 +814,42 @@ re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) | |||
834 | } | 814 | } |
835 | 815 | ||
836 | static unsigned char | 816 | static unsigned char |
837 | internal_function __attribute__ ((pure)) | 817 | __attribute__ ((pure)) |
838 | re_string_peek_byte_case (const re_string_t *pstr, Idx idx) | 818 | re_string_peek_byte_case (const re_string_t *pstr, Idx idx) |
839 | { | 819 | { |
840 | int ch; | 820 | int ch; |
841 | Idx off; | 821 | Idx off; |
842 | 822 | ||
843 | /* Handle the common (easiest) cases first. */ | 823 | /* Handle the common (easiest) cases first. */ |
844 | if (BE (!pstr->mbs_allocated, 1)) | 824 | if (__glibc_likely (!pstr->mbs_allocated)) |
845 | return re_string_peek_byte (pstr, idx); | 825 | return re_string_peek_byte (pstr, idx); |
846 | 826 | ||
847 | #ifdef RE_ENABLE_I18N | ||
848 | if (pstr->mb_cur_max > 1 | 827 | if (pstr->mb_cur_max > 1 |
849 | && ! re_string_is_single_byte_char (pstr, pstr->cur_idx + idx)) | 828 | && ! re_string_is_single_byte_char (pstr, pstr->cur_idx + idx)) |
850 | return re_string_peek_byte (pstr, idx); | 829 | return re_string_peek_byte (pstr, idx); |
851 | #endif | ||
852 | 830 | ||
853 | off = pstr->cur_idx + idx; | 831 | off = pstr->cur_idx + idx; |
854 | #ifdef RE_ENABLE_I18N | ||
855 | if (pstr->offsets_needed) | 832 | if (pstr->offsets_needed) |
856 | off = pstr->offsets[off]; | 833 | off = pstr->offsets[off]; |
857 | #endif | ||
858 | 834 | ||
859 | ch = pstr->raw_mbs[pstr->raw_mbs_idx + off]; | 835 | ch = pstr->raw_mbs[pstr->raw_mbs_idx + off]; |
860 | 836 | ||
861 | #ifdef RE_ENABLE_I18N | ||
862 | /* Ensure that e.g. for tr_TR.UTF-8 BACKSLASH DOTLESS SMALL LETTER I | 837 | /* Ensure that e.g. for tr_TR.UTF-8 BACKSLASH DOTLESS SMALL LETTER I |
863 | this function returns CAPITAL LETTER I instead of first byte of | 838 | this function returns CAPITAL LETTER I instead of first byte of |
864 | DOTLESS SMALL LETTER I. The latter would confuse the parser, | 839 | DOTLESS SMALL LETTER I. The latter would confuse the parser, |
865 | since peek_byte_case doesn't advance cur_idx in any way. */ | 840 | since peek_byte_case doesn't advance cur_idx in any way. */ |
866 | if (pstr->offsets_needed && !isascii (ch)) | 841 | if (pstr->offsets_needed && !isascii (ch)) |
867 | return re_string_peek_byte (pstr, idx); | 842 | return re_string_peek_byte (pstr, idx); |
868 | #endif | ||
869 | 843 | ||
870 | return ch; | 844 | return ch; |
871 | } | 845 | } |
872 | 846 | ||
873 | static unsigned char | 847 | static unsigned char |
874 | internal_function | ||
875 | re_string_fetch_byte_case (re_string_t *pstr) | 848 | re_string_fetch_byte_case (re_string_t *pstr) |
876 | { | 849 | { |
877 | if (BE (!pstr->mbs_allocated, 1)) | 850 | if (__glibc_likely (!pstr->mbs_allocated)) |
878 | return re_string_fetch_byte (pstr); | 851 | return re_string_fetch_byte (pstr); |
879 | 852 | ||
880 | #ifdef RE_ENABLE_I18N | ||
881 | if (pstr->offsets_needed) | 853 | if (pstr->offsets_needed) |
882 | { | 854 | { |
883 | Idx off; | 855 | Idx off; |
@@ -903,19 +875,15 @@ re_string_fetch_byte_case (re_string_t *pstr) | |||
903 | re_string_char_size_at (pstr, pstr->cur_idx)); | 875 | re_string_char_size_at (pstr, pstr->cur_idx)); |
904 | return ch; | 876 | return ch; |
905 | } | 877 | } |
906 | #endif | ||
907 | 878 | ||
908 | return pstr->raw_mbs[pstr->raw_mbs_idx + pstr->cur_idx++]; | 879 | return pstr->raw_mbs[pstr->raw_mbs_idx + pstr->cur_idx++]; |
909 | } | 880 | } |
910 | 881 | ||
911 | static void | 882 | static void |
912 | internal_function | ||
913 | re_string_destruct (re_string_t *pstr) | 883 | re_string_destruct (re_string_t *pstr) |
914 | { | 884 | { |
915 | #ifdef RE_ENABLE_I18N | ||
916 | re_free (pstr->wcs); | 885 | re_free (pstr->wcs); |
917 | re_free (pstr->offsets); | 886 | re_free (pstr->offsets); |
918 | #endif /* RE_ENABLE_I18N */ | ||
919 | if (pstr->mbs_allocated) | 887 | if (pstr->mbs_allocated) |
920 | re_free (pstr->mbs); | 888 | re_free (pstr->mbs); |
921 | } | 889 | } |
@@ -923,40 +891,35 @@ re_string_destruct (re_string_t *pstr) | |||
923 | /* Return the context at IDX in INPUT. */ | 891 | /* Return the context at IDX in INPUT. */ |
924 | 892 | ||
925 | static unsigned int | 893 | static unsigned int |
926 | internal_function | ||
927 | re_string_context_at (const re_string_t *input, Idx idx, int eflags) | 894 | re_string_context_at (const re_string_t *input, Idx idx, int eflags) |
928 | { | 895 | { |
929 | int c; | 896 | int c; |
930 | if (BE (! REG_VALID_INDEX (idx), 0)) | 897 | if (__glibc_unlikely (idx < 0)) |
931 | /* In this case, we use the value stored in input->tip_context, | 898 | /* In this case, we use the value stored in input->tip_context, |
932 | since we can't know the character in input->mbs[-1] here. */ | 899 | since we can't know the character in input->mbs[-1] here. */ |
933 | return input->tip_context; | 900 | return input->tip_context; |
934 | if (BE (idx == input->len, 0)) | 901 | if (__glibc_unlikely (idx == input->len)) |
935 | return ((eflags & REG_NOTEOL) ? CONTEXT_ENDBUF | 902 | return ((eflags & REG_NOTEOL) ? CONTEXT_ENDBUF |
936 | : CONTEXT_NEWLINE | CONTEXT_ENDBUF); | 903 | : CONTEXT_NEWLINE | CONTEXT_ENDBUF); |
937 | #ifdef RE_ENABLE_I18N | ||
938 | if (input->mb_cur_max > 1) | 904 | if (input->mb_cur_max > 1) |
939 | { | 905 | { |
940 | wint_t wc; | 906 | wint_t wc; |
941 | Idx wc_idx = idx; | 907 | Idx wc_idx = idx; |
942 | while(input->wcs[wc_idx] == WEOF) | 908 | while(input->wcs[wc_idx] == WEOF) |
943 | { | 909 | { |
944 | #ifdef DEBUG | 910 | DEBUG_ASSERT (wc_idx >= 0); |
945 | /* It must not happen. */ | ||
946 | assert (REG_VALID_INDEX (wc_idx)); | ||
947 | #endif | ||
948 | --wc_idx; | 911 | --wc_idx; |
949 | if (! REG_VALID_INDEX (wc_idx)) | 912 | if (wc_idx < 0) |
950 | return input->tip_context; | 913 | return input->tip_context; |
951 | } | 914 | } |
952 | wc = input->wcs[wc_idx]; | 915 | wc = input->wcs[wc_idx]; |
953 | if (BE (input->word_ops_used != 0, 0) && IS_WIDE_WORD_CHAR (wc)) | 916 | if (__glibc_unlikely (input->word_ops_used != 0) |
917 | && IS_WIDE_WORD_CHAR (wc)) | ||
954 | return CONTEXT_WORD; | 918 | return CONTEXT_WORD; |
955 | return (IS_WIDE_NEWLINE (wc) && input->newline_anchor | 919 | return (IS_WIDE_NEWLINE (wc) && input->newline_anchor |
956 | ? CONTEXT_NEWLINE : 0); | 920 | ? CONTEXT_NEWLINE : 0); |
957 | } | 921 | } |
958 | else | 922 | else |
959 | #endif | ||
960 | { | 923 | { |
961 | c = re_string_byte_at (input, idx); | 924 | c = re_string_byte_at (input, idx); |
962 | if (bitset_contain (input->word_char, c)) | 925 | if (bitset_contain (input->word_char, c)) |
@@ -968,25 +931,26 @@ re_string_context_at (const re_string_t *input, Idx idx, int eflags) | |||
968 | /* Functions for set operation. */ | 931 | /* Functions for set operation. */ |
969 | 932 | ||
970 | static reg_errcode_t | 933 | static reg_errcode_t |
971 | internal_function __attribute_warn_unused_result__ | 934 | __attribute_warn_unused_result__ |
972 | re_node_set_alloc (re_node_set *set, Idx size) | 935 | re_node_set_alloc (re_node_set *set, Idx size) |
973 | { | 936 | { |
974 | set->alloc = size; | 937 | set->alloc = size; |
975 | set->nelem = 0; | 938 | set->nelem = 0; |
976 | set->elems = re_malloc (Idx, size); | 939 | set->elems = re_malloc (Idx, size); |
977 | if (BE (set->elems == NULL, 0) && (MALLOC_0_IS_NONNULL || size != 0)) | 940 | if (__glibc_unlikely (set->elems == NULL) |
941 | && (MALLOC_0_IS_NONNULL || size != 0)) | ||
978 | return REG_ESPACE; | 942 | return REG_ESPACE; |
979 | return REG_NOERROR; | 943 | return REG_NOERROR; |
980 | } | 944 | } |
981 | 945 | ||
982 | static reg_errcode_t | 946 | static reg_errcode_t |
983 | internal_function __attribute_warn_unused_result__ | 947 | __attribute_warn_unused_result__ |
984 | re_node_set_init_1 (re_node_set *set, Idx elem) | 948 | re_node_set_init_1 (re_node_set *set, Idx elem) |
985 | { | 949 | { |
986 | set->alloc = 1; | 950 | set->alloc = 1; |
987 | set->nelem = 1; | 951 | set->nelem = 1; |
988 | set->elems = re_malloc (Idx, 1); | 952 | set->elems = re_malloc (Idx, 1); |
989 | if (BE (set->elems == NULL, 0)) | 953 | if (__glibc_unlikely (set->elems == NULL)) |
990 | { | 954 | { |
991 | set->alloc = set->nelem = 0; | 955 | set->alloc = set->nelem = 0; |
992 | return REG_ESPACE; | 956 | return REG_ESPACE; |
@@ -996,12 +960,12 @@ re_node_set_init_1 (re_node_set *set, Idx elem) | |||
996 | } | 960 | } |
997 | 961 | ||
998 | static reg_errcode_t | 962 | static reg_errcode_t |
999 | internal_function __attribute_warn_unused_result__ | 963 | __attribute_warn_unused_result__ |
1000 | re_node_set_init_2 (re_node_set *set, Idx elem1, Idx elem2) | 964 | re_node_set_init_2 (re_node_set *set, Idx elem1, Idx elem2) |
1001 | { | 965 | { |
1002 | set->alloc = 2; | 966 | set->alloc = 2; |
1003 | set->elems = re_malloc (Idx, 2); | 967 | set->elems = re_malloc (Idx, 2); |
1004 | if (BE (set->elems == NULL, 0)) | 968 | if (__glibc_unlikely (set->elems == NULL)) |
1005 | return REG_ESPACE; | 969 | return REG_ESPACE; |
1006 | if (elem1 == elem2) | 970 | if (elem1 == elem2) |
1007 | { | 971 | { |
@@ -1026,7 +990,7 @@ re_node_set_init_2 (re_node_set *set, Idx elem1, Idx elem2) | |||
1026 | } | 990 | } |
1027 | 991 | ||
1028 | static reg_errcode_t | 992 | static reg_errcode_t |
1029 | internal_function __attribute_warn_unused_result__ | 993 | __attribute_warn_unused_result__ |
1030 | re_node_set_init_copy (re_node_set *dest, const re_node_set *src) | 994 | re_node_set_init_copy (re_node_set *dest, const re_node_set *src) |
1031 | { | 995 | { |
1032 | dest->nelem = src->nelem; | 996 | dest->nelem = src->nelem; |
@@ -1034,7 +998,7 @@ re_node_set_init_copy (re_node_set *dest, const re_node_set *src) | |||
1034 | { | 998 | { |
1035 | dest->alloc = dest->nelem; | 999 | dest->alloc = dest->nelem; |
1036 | dest->elems = re_malloc (Idx, dest->alloc); | 1000 | dest->elems = re_malloc (Idx, dest->alloc); |
1037 | if (BE (dest->elems == NULL, 0)) | 1001 | if (__glibc_unlikely (dest->elems == NULL)) |
1038 | { | 1002 | { |
1039 | dest->alloc = dest->nelem = 0; | 1003 | dest->alloc = dest->nelem = 0; |
1040 | return REG_ESPACE; | 1004 | return REG_ESPACE; |
@@ -1051,7 +1015,7 @@ re_node_set_init_copy (re_node_set *dest, const re_node_set *src) | |||
1051 | Note: We assume dest->elems is NULL, when dest->alloc is 0. */ | 1015 | Note: We assume dest->elems is NULL, when dest->alloc is 0. */ |
1052 | 1016 | ||
1053 | static reg_errcode_t | 1017 | static reg_errcode_t |
1054 | internal_function __attribute_warn_unused_result__ | 1018 | __attribute_warn_unused_result__ |
1055 | re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, | 1019 | re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, |
1056 | const re_node_set *src2) | 1020 | const re_node_set *src2) |
1057 | { | 1021 | { |
@@ -1065,7 +1029,7 @@ re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, | |||
1065 | { | 1029 | { |
1066 | Idx new_alloc = src1->nelem + src2->nelem + dest->alloc; | 1030 | Idx new_alloc = src1->nelem + src2->nelem + dest->alloc; |
1067 | Idx *new_elems = re_realloc (dest->elems, Idx, new_alloc); | 1031 | Idx *new_elems = re_realloc (dest->elems, Idx, new_alloc); |
1068 | if (BE (new_elems == NULL, 0)) | 1032 | if (__glibc_unlikely (new_elems == NULL)) |
1069 | return REG_ESPACE; | 1033 | return REG_ESPACE; |
1070 | dest->elems = new_elems; | 1034 | dest->elems = new_elems; |
1071 | dest->alloc = new_alloc; | 1035 | dest->alloc = new_alloc; |
@@ -1082,25 +1046,25 @@ re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, | |||
1082 | if (src1->elems[i1] == src2->elems[i2]) | 1046 | if (src1->elems[i1] == src2->elems[i2]) |
1083 | { | 1047 | { |
1084 | /* Try to find the item in DEST. Maybe we could binary search? */ | 1048 | /* Try to find the item in DEST. Maybe we could binary search? */ |
1085 | while (REG_VALID_INDEX (id) && dest->elems[id] > src1->elems[i1]) | 1049 | while (id >= 0 && dest->elems[id] > src1->elems[i1]) |
1086 | --id; | 1050 | --id; |
1087 | 1051 | ||
1088 | if (! REG_VALID_INDEX (id) || dest->elems[id] != src1->elems[i1]) | 1052 | if (id < 0 || dest->elems[id] != src1->elems[i1]) |
1089 | dest->elems[--sbase] = src1->elems[i1]; | 1053 | dest->elems[--sbase] = src1->elems[i1]; |
1090 | 1054 | ||
1091 | if (! REG_VALID_INDEX (--i1) || ! REG_VALID_INDEX (--i2)) | 1055 | if (--i1 < 0 || --i2 < 0) |
1092 | break; | 1056 | break; |
1093 | } | 1057 | } |
1094 | 1058 | ||
1095 | /* Lower the highest of the two items. */ | 1059 | /* Lower the highest of the two items. */ |
1096 | else if (src1->elems[i1] < src2->elems[i2]) | 1060 | else if (src1->elems[i1] < src2->elems[i2]) |
1097 | { | 1061 | { |
1098 | if (! REG_VALID_INDEX (--i2)) | 1062 | if (--i2 < 0) |
1099 | break; | 1063 | break; |
1100 | } | 1064 | } |
1101 | else | 1065 | else |
1102 | { | 1066 | { |
1103 | if (! REG_VALID_INDEX (--i1)) | 1067 | if (--i1 < 0) |
1104 | break; | 1068 | break; |
1105 | } | 1069 | } |
1106 | } | 1070 | } |
@@ -1113,7 +1077,7 @@ re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, | |||
1113 | DEST elements are already in place; this is more or | 1077 | DEST elements are already in place; this is more or |
1114 | less the same loop that is in re_node_set_merge. */ | 1078 | less the same loop that is in re_node_set_merge. */ |
1115 | dest->nelem += delta; | 1079 | dest->nelem += delta; |
1116 | if (delta > 0 && REG_VALID_INDEX (id)) | 1080 | if (delta > 0 && id >= 0) |
1117 | for (;;) | 1081 | for (;;) |
1118 | { | 1082 | { |
1119 | if (dest->elems[is] > dest->elems[id]) | 1083 | if (dest->elems[is] > dest->elems[id]) |
@@ -1127,7 +1091,7 @@ re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, | |||
1127 | { | 1091 | { |
1128 | /* Slide from the bottom. */ | 1092 | /* Slide from the bottom. */ |
1129 | dest->elems[id + delta] = dest->elems[id]; | 1093 | dest->elems[id + delta] = dest->elems[id]; |
1130 | if (! REG_VALID_INDEX (--id)) | 1094 | if (--id < 0) |
1131 | break; | 1095 | break; |
1132 | } | 1096 | } |
1133 | } | 1097 | } |
@@ -1142,7 +1106,7 @@ re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, | |||
1142 | DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ | 1106 | DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ |
1143 | 1107 | ||
1144 | static reg_errcode_t | 1108 | static reg_errcode_t |
1145 | internal_function __attribute_warn_unused_result__ | 1109 | __attribute_warn_unused_result__ |
1146 | re_node_set_init_union (re_node_set *dest, const re_node_set *src1, | 1110 | re_node_set_init_union (re_node_set *dest, const re_node_set *src1, |
1147 | const re_node_set *src2) | 1111 | const re_node_set *src2) |
1148 | { | 1112 | { |
@@ -1151,7 +1115,7 @@ re_node_set_init_union (re_node_set *dest, const re_node_set *src1, | |||
1151 | { | 1115 | { |
1152 | dest->alloc = src1->nelem + src2->nelem; | 1116 | dest->alloc = src1->nelem + src2->nelem; |
1153 | dest->elems = re_malloc (Idx, dest->alloc); | 1117 | dest->elems = re_malloc (Idx, dest->alloc); |
1154 | if (BE (dest->elems == NULL, 0)) | 1118 | if (__glibc_unlikely (dest->elems == NULL)) |
1155 | return REG_ESPACE; | 1119 | return REG_ESPACE; |
1156 | } | 1120 | } |
1157 | else | 1121 | else |
@@ -1195,7 +1159,7 @@ re_node_set_init_union (re_node_set *dest, const re_node_set *src1, | |||
1195 | DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ | 1159 | DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ |
1196 | 1160 | ||
1197 | static reg_errcode_t | 1161 | static reg_errcode_t |
1198 | internal_function __attribute_warn_unused_result__ | 1162 | __attribute_warn_unused_result__ |
1199 | re_node_set_merge (re_node_set *dest, const re_node_set *src) | 1163 | re_node_set_merge (re_node_set *dest, const re_node_set *src) |
1200 | { | 1164 | { |
1201 | Idx is, id, sbase, delta; | 1165 | Idx is, id, sbase, delta; |
@@ -1205,14 +1169,18 @@ re_node_set_merge (re_node_set *dest, const re_node_set *src) | |||
1205 | { | 1169 | { |
1206 | Idx new_alloc = 2 * (src->nelem + dest->alloc); | 1170 | Idx new_alloc = 2 * (src->nelem + dest->alloc); |
1207 | Idx *new_buffer = re_realloc (dest->elems, Idx, new_alloc); | 1171 | Idx *new_buffer = re_realloc (dest->elems, Idx, new_alloc); |
1208 | if (BE (new_buffer == NULL, 0)) | 1172 | if (__glibc_unlikely (new_buffer == NULL)) |
1209 | return REG_ESPACE; | 1173 | return REG_ESPACE; |
1210 | dest->elems = new_buffer; | 1174 | dest->elems = new_buffer; |
1211 | dest->alloc = new_alloc; | 1175 | dest->alloc = new_alloc; |
1212 | } | 1176 | } |
1213 | 1177 | ||
1214 | if (BE (dest->nelem == 0, 0)) | 1178 | if (__glibc_unlikely (dest->nelem == 0)) |
1215 | { | 1179 | { |
1180 | /* Although we already guaranteed above that dest->alloc != 0 and | ||
1181 | therefore dest->elems != NULL, add a debug assertion to pacify | ||
1182 | GCC 11.2.1's -fanalyzer. */ | ||
1183 | DEBUG_ASSERT (dest->elems); | ||
1216 | dest->nelem = src->nelem; | 1184 | dest->nelem = src->nelem; |
1217 | memcpy (dest->elems, src->elems, src->nelem * sizeof (Idx)); | 1185 | memcpy (dest->elems, src->elems, src->nelem * sizeof (Idx)); |
1218 | return REG_NOERROR; | 1186 | return REG_NOERROR; |
@@ -1221,8 +1189,7 @@ re_node_set_merge (re_node_set *dest, const re_node_set *src) | |||
1221 | /* Copy into the top of DEST the items of SRC that are not | 1189 | /* Copy into the top of DEST the items of SRC that are not |
1222 | found in DEST. Maybe we could binary search in DEST? */ | 1190 | found in DEST. Maybe we could binary search in DEST? */ |
1223 | for (sbase = dest->nelem + 2 * src->nelem, | 1191 | for (sbase = dest->nelem + 2 * src->nelem, |
1224 | is = src->nelem - 1, id = dest->nelem - 1; | 1192 | is = src->nelem - 1, id = dest->nelem - 1; is >= 0 && id >= 0; ) |
1225 | REG_VALID_INDEX (is) && REG_VALID_INDEX (id); ) | ||
1226 | { | 1193 | { |
1227 | if (dest->elems[id] == src->elems[is]) | 1194 | if (dest->elems[id] == src->elems[is]) |
1228 | is--, id--; | 1195 | is--, id--; |
@@ -1232,7 +1199,7 @@ re_node_set_merge (re_node_set *dest, const re_node_set *src) | |||
1232 | --id; | 1199 | --id; |
1233 | } | 1200 | } |
1234 | 1201 | ||
1235 | if (REG_VALID_INDEX (is)) | 1202 | if (is >= 0) |
1236 | { | 1203 | { |
1237 | /* If DEST is exhausted, the remaining items of SRC must be unique. */ | 1204 | /* If DEST is exhausted, the remaining items of SRC must be unique. */ |
1238 | sbase -= is + 1; | 1205 | sbase -= is + 1; |
@@ -1261,7 +1228,7 @@ re_node_set_merge (re_node_set *dest, const re_node_set *src) | |||
1261 | { | 1228 | { |
1262 | /* Slide from the bottom. */ | 1229 | /* Slide from the bottom. */ |
1263 | dest->elems[id + delta] = dest->elems[id]; | 1230 | dest->elems[id + delta] = dest->elems[id]; |
1264 | if (! REG_VALID_INDEX (--id)) | 1231 | if (--id < 0) |
1265 | { | 1232 | { |
1266 | /* Copy remaining SRC elements. */ | 1233 | /* Copy remaining SRC elements. */ |
1267 | memcpy (dest->elems, dest->elems + sbase, | 1234 | memcpy (dest->elems, dest->elems + sbase, |
@@ -1279,17 +1246,20 @@ re_node_set_merge (re_node_set *dest, const re_node_set *src) | |||
1279 | Return true if successful. */ | 1246 | Return true if successful. */ |
1280 | 1247 | ||
1281 | static bool | 1248 | static bool |
1282 | internal_function __attribute_warn_unused_result__ | 1249 | __attribute_warn_unused_result__ |
1283 | re_node_set_insert (re_node_set *set, Idx elem) | 1250 | re_node_set_insert (re_node_set *set, Idx elem) |
1284 | { | 1251 | { |
1285 | Idx idx; | 1252 | Idx idx; |
1286 | /* In case the set is empty. */ | 1253 | /* In case the set is empty. */ |
1287 | if (set->alloc == 0) | 1254 | if (set->alloc == 0) |
1288 | return BE (re_node_set_init_1 (set, elem) == REG_NOERROR, 1); | 1255 | return __glibc_likely (re_node_set_init_1 (set, elem) == REG_NOERROR); |
1289 | 1256 | ||
1290 | if (BE (set->nelem, 0) == 0) | 1257 | if (__glibc_unlikely (set->nelem) == 0) |
1291 | { | 1258 | { |
1292 | /* We already guaranteed above that set->alloc != 0. */ | 1259 | /* Although we already guaranteed above that set->alloc != 0 and |
1260 | therefore set->elems != NULL, add a debug assertion to pacify | ||
1261 | GCC 11.2 -fanalyzer. */ | ||
1262 | DEBUG_ASSERT (set->elems); | ||
1293 | set->elems[0] = elem; | 1263 | set->elems[0] = elem; |
1294 | ++set->nelem; | 1264 | ++set->nelem; |
1295 | return true; | 1265 | return true; |
@@ -1301,7 +1271,7 @@ re_node_set_insert (re_node_set *set, Idx elem) | |||
1301 | Idx *new_elems; | 1271 | Idx *new_elems; |
1302 | set->alloc = set->alloc * 2; | 1272 | set->alloc = set->alloc * 2; |
1303 | new_elems = re_realloc (set->elems, Idx, set->alloc); | 1273 | new_elems = re_realloc (set->elems, Idx, set->alloc); |
1304 | if (BE (new_elems == NULL, 0)) | 1274 | if (__glibc_unlikely (new_elems == NULL)) |
1305 | return false; | 1275 | return false; |
1306 | set->elems = new_elems; | 1276 | set->elems = new_elems; |
1307 | } | 1277 | } |
@@ -1310,7 +1280,6 @@ re_node_set_insert (re_node_set *set, Idx elem) | |||
1310 | first element separately to skip a check in the inner loop. */ | 1280 | first element separately to skip a check in the inner loop. */ |
1311 | if (elem < set->elems[0]) | 1281 | if (elem < set->elems[0]) |
1312 | { | 1282 | { |
1313 | idx = 0; | ||
1314 | for (idx = set->nelem; idx > 0; idx--) | 1283 | for (idx = set->nelem; idx > 0; idx--) |
1315 | set->elems[idx] = set->elems[idx - 1]; | 1284 | set->elems[idx] = set->elems[idx - 1]; |
1316 | } | 1285 | } |
@@ -1318,6 +1287,7 @@ re_node_set_insert (re_node_set *set, Idx elem) | |||
1318 | { | 1287 | { |
1319 | for (idx = set->nelem; set->elems[idx - 1] > elem; idx--) | 1288 | for (idx = set->nelem; set->elems[idx - 1] > elem; idx--) |
1320 | set->elems[idx] = set->elems[idx - 1]; | 1289 | set->elems[idx] = set->elems[idx - 1]; |
1290 | DEBUG_ASSERT (set->elems[idx - 1] < elem); | ||
1321 | } | 1291 | } |
1322 | 1292 | ||
1323 | /* Insert the new element. */ | 1293 | /* Insert the new element. */ |
@@ -1331,7 +1301,7 @@ re_node_set_insert (re_node_set *set, Idx elem) | |||
1331 | Return true if successful. */ | 1301 | Return true if successful. */ |
1332 | 1302 | ||
1333 | static bool | 1303 | static bool |
1334 | internal_function __attribute_warn_unused_result__ | 1304 | __attribute_warn_unused_result__ |
1335 | re_node_set_insert_last (re_node_set *set, Idx elem) | 1305 | re_node_set_insert_last (re_node_set *set, Idx elem) |
1336 | { | 1306 | { |
1337 | /* Realloc if we need. */ | 1307 | /* Realloc if we need. */ |
@@ -1340,7 +1310,7 @@ re_node_set_insert_last (re_node_set *set, Idx elem) | |||
1340 | Idx *new_elems; | 1310 | Idx *new_elems; |
1341 | set->alloc = (set->alloc + 1) * 2; | 1311 | set->alloc = (set->alloc + 1) * 2; |
1342 | new_elems = re_realloc (set->elems, Idx, set->alloc); | 1312 | new_elems = re_realloc (set->elems, Idx, set->alloc); |
1343 | if (BE (new_elems == NULL, 0)) | 1313 | if (__glibc_unlikely (new_elems == NULL)) |
1344 | return false; | 1314 | return false; |
1345 | set->elems = new_elems; | 1315 | set->elems = new_elems; |
1346 | } | 1316 | } |
@@ -1354,13 +1324,13 @@ re_node_set_insert_last (re_node_set *set, Idx elem) | |||
1354 | Return true if SET1 and SET2 are equivalent. */ | 1324 | Return true if SET1 and SET2 are equivalent. */ |
1355 | 1325 | ||
1356 | static bool | 1326 | static bool |
1357 | internal_function __attribute__ ((pure)) | 1327 | __attribute__ ((pure)) |
1358 | re_node_set_compare (const re_node_set *set1, const re_node_set *set2) | 1328 | re_node_set_compare (const re_node_set *set1, const re_node_set *set2) |
1359 | { | 1329 | { |
1360 | Idx i; | 1330 | Idx i; |
1361 | if (set1 == NULL || set2 == NULL || set1->nelem != set2->nelem) | 1331 | if (set1 == NULL || set2 == NULL || set1->nelem != set2->nelem) |
1362 | return false; | 1332 | return false; |
1363 | for (i = set1->nelem ; REG_VALID_INDEX (--i) ; ) | 1333 | for (i = set1->nelem ; --i >= 0 ; ) |
1364 | if (set1->elems[i] != set2->elems[i]) | 1334 | if (set1->elems[i] != set2->elems[i]) |
1365 | return false; | 1335 | return false; |
1366 | return true; | 1336 | return true; |
@@ -1369,11 +1339,11 @@ re_node_set_compare (const re_node_set *set1, const re_node_set *set2) | |||
1369 | /* Return (idx + 1) if SET contains the element ELEM, return 0 otherwise. */ | 1339 | /* Return (idx + 1) if SET contains the element ELEM, return 0 otherwise. */ |
1370 | 1340 | ||
1371 | static Idx | 1341 | static Idx |
1372 | internal_function __attribute__ ((pure)) | 1342 | __attribute__ ((pure)) |
1373 | re_node_set_contains (const re_node_set *set, Idx elem) | 1343 | re_node_set_contains (const re_node_set *set, Idx elem) |
1374 | { | 1344 | { |
1375 | __re_size_t idx, right, mid; | 1345 | __re_size_t idx, right, mid; |
1376 | if (! REG_VALID_NONZERO_INDEX (set->nelem)) | 1346 | if (set->nelem <= 0) |
1377 | return 0; | 1347 | return 0; |
1378 | 1348 | ||
1379 | /* Binary search the element. */ | 1349 | /* Binary search the element. */ |
@@ -1391,7 +1361,6 @@ re_node_set_contains (const re_node_set *set, Idx elem) | |||
1391 | } | 1361 | } |
1392 | 1362 | ||
1393 | static void | 1363 | static void |
1394 | internal_function | ||
1395 | re_node_set_remove_at (re_node_set *set, Idx idx) | 1364 | re_node_set_remove_at (re_node_set *set, Idx idx) |
1396 | { | 1365 | { |
1397 | if (idx < 0 || idx >= set->nelem) | 1366 | if (idx < 0 || idx >= set->nelem) |
@@ -1403,13 +1372,12 @@ re_node_set_remove_at (re_node_set *set, Idx idx) | |||
1403 | 1372 | ||
1404 | 1373 | ||
1405 | /* Add the token TOKEN to dfa->nodes, and return the index of the token. | 1374 | /* Add the token TOKEN to dfa->nodes, and return the index of the token. |
1406 | Or return REG_MISSING if an error occurred. */ | 1375 | Or return -1 if an error occurred. */ |
1407 | 1376 | ||
1408 | static Idx | 1377 | static Idx |
1409 | internal_function | ||
1410 | re_dfa_add_node (re_dfa_t *dfa, re_token_t token) | 1378 | re_dfa_add_node (re_dfa_t *dfa, re_token_t token) |
1411 | { | 1379 | { |
1412 | if (BE (dfa->nodes_len >= dfa->nodes_alloc, 0)) | 1380 | if (__glibc_unlikely (dfa->nodes_len >= dfa->nodes_alloc)) |
1413 | { | 1381 | { |
1414 | size_t new_nodes_alloc = dfa->nodes_alloc * 2; | 1382 | size_t new_nodes_alloc = dfa->nodes_alloc * 2; |
1415 | Idx *new_nexts, *new_indices; | 1383 | Idx *new_nexts, *new_indices; |
@@ -1420,41 +1388,43 @@ re_dfa_add_node (re_dfa_t *dfa, re_token_t token) | |||
1420 | const size_t max_object_size = MAX (sizeof (re_token_t), | 1388 | const size_t max_object_size = MAX (sizeof (re_token_t), |
1421 | MAX (sizeof (re_node_set), | 1389 | MAX (sizeof (re_node_set), |
1422 | sizeof (Idx))); | 1390 | sizeof (Idx))); |
1423 | if (BE (MIN (IDX_MAX, SIZE_MAX / max_object_size) < new_nodes_alloc, 0)) | 1391 | if (__glibc_unlikely (MIN (IDX_MAX, SIZE_MAX / max_object_size) |
1424 | return REG_MISSING; | 1392 | < new_nodes_alloc)) |
1393 | return -1; | ||
1425 | 1394 | ||
1426 | new_nodes = re_realloc (dfa->nodes, re_token_t, new_nodes_alloc); | 1395 | new_nodes = re_realloc (dfa->nodes, re_token_t, new_nodes_alloc); |
1427 | if (BE (new_nodes == NULL, 0)) | 1396 | if (__glibc_unlikely (new_nodes == NULL)) |
1428 | return REG_MISSING; | 1397 | return -1; |
1429 | dfa->nodes = new_nodes; | 1398 | dfa->nodes = new_nodes; |
1399 | dfa->nodes_alloc = new_nodes_alloc; | ||
1430 | new_nexts = re_realloc (dfa->nexts, Idx, new_nodes_alloc); | 1400 | new_nexts = re_realloc (dfa->nexts, Idx, new_nodes_alloc); |
1401 | if (new_nexts != NULL) | ||
1402 | dfa->nexts = new_nexts; | ||
1431 | new_indices = re_realloc (dfa->org_indices, Idx, new_nodes_alloc); | 1403 | new_indices = re_realloc (dfa->org_indices, Idx, new_nodes_alloc); |
1404 | if (new_indices != NULL) | ||
1405 | dfa->org_indices = new_indices; | ||
1432 | new_edests = re_realloc (dfa->edests, re_node_set, new_nodes_alloc); | 1406 | new_edests = re_realloc (dfa->edests, re_node_set, new_nodes_alloc); |
1407 | if (new_edests != NULL) | ||
1408 | dfa->edests = new_edests; | ||
1433 | new_eclosures = re_realloc (dfa->eclosures, re_node_set, new_nodes_alloc); | 1409 | new_eclosures = re_realloc (dfa->eclosures, re_node_set, new_nodes_alloc); |
1434 | if (BE (new_nexts == NULL || new_indices == NULL | 1410 | if (new_eclosures != NULL) |
1435 | || new_edests == NULL || new_eclosures == NULL, 0)) | 1411 | dfa->eclosures = new_eclosures; |
1436 | return REG_MISSING; | 1412 | if (__glibc_unlikely (new_nexts == NULL || new_indices == NULL |
1437 | dfa->nexts = new_nexts; | 1413 | || new_edests == NULL || new_eclosures == NULL)) |
1438 | dfa->org_indices = new_indices; | 1414 | return -1; |
1439 | dfa->edests = new_edests; | ||
1440 | dfa->eclosures = new_eclosures; | ||
1441 | dfa->nodes_alloc = new_nodes_alloc; | ||
1442 | } | 1415 | } |
1443 | dfa->nodes[dfa->nodes_len] = token; | 1416 | dfa->nodes[dfa->nodes_len] = token; |
1444 | dfa->nodes[dfa->nodes_len].constraint = 0; | 1417 | dfa->nodes[dfa->nodes_len].constraint = 0; |
1445 | #ifdef RE_ENABLE_I18N | ||
1446 | dfa->nodes[dfa->nodes_len].accept_mb = | 1418 | dfa->nodes[dfa->nodes_len].accept_mb = |
1447 | ((token.type == OP_PERIOD && dfa->mb_cur_max > 1) | 1419 | ((token.type == OP_PERIOD && dfa->mb_cur_max > 1) |
1448 | || token.type == COMPLEX_BRACKET); | 1420 | || token.type == COMPLEX_BRACKET); |
1449 | #endif | 1421 | dfa->nexts[dfa->nodes_len] = -1; |
1450 | dfa->nexts[dfa->nodes_len] = REG_MISSING; | ||
1451 | re_node_set_init_empty (dfa->edests + dfa->nodes_len); | 1422 | re_node_set_init_empty (dfa->edests + dfa->nodes_len); |
1452 | re_node_set_init_empty (dfa->eclosures + dfa->nodes_len); | 1423 | re_node_set_init_empty (dfa->eclosures + dfa->nodes_len); |
1453 | return dfa->nodes_len++; | 1424 | return dfa->nodes_len++; |
1454 | } | 1425 | } |
1455 | 1426 | ||
1456 | static re_hashval_t | 1427 | static re_hashval_t |
1457 | internal_function | ||
1458 | calc_state_hash (const re_node_set *nodes, unsigned int context) | 1428 | calc_state_hash (const re_node_set *nodes, unsigned int context) |
1459 | { | 1429 | { |
1460 | re_hashval_t hash = nodes->nelem + context; | 1430 | re_hashval_t hash = nodes->nelem + context; |
@@ -1474,7 +1444,7 @@ calc_state_hash (const re_node_set *nodes, unsigned int context) | |||
1474 | optimization. */ | 1444 | optimization. */ |
1475 | 1445 | ||
1476 | static re_dfastate_t * | 1446 | static re_dfastate_t * |
1477 | internal_function __attribute_warn_unused_result__ | 1447 | __attribute_warn_unused_result__ |
1478 | re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, | 1448 | re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, |
1479 | const re_node_set *nodes) | 1449 | const re_node_set *nodes) |
1480 | { | 1450 | { |
@@ -1482,11 +1452,11 @@ re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, | |||
1482 | re_dfastate_t *new_state; | 1452 | re_dfastate_t *new_state; |
1483 | struct re_state_table_entry *spot; | 1453 | struct re_state_table_entry *spot; |
1484 | Idx i; | 1454 | Idx i; |
1485 | #ifdef lint | 1455 | #if defined GCC_LINT || defined lint |
1486 | /* Suppress bogus uninitialized-variable warnings. */ | 1456 | /* Suppress bogus uninitialized-variable warnings. */ |
1487 | *err = REG_NOERROR; | 1457 | *err = REG_NOERROR; |
1488 | #endif | 1458 | #endif |
1489 | if (BE (nodes->nelem == 0, 0)) | 1459 | if (__glibc_unlikely (nodes->nelem == 0)) |
1490 | { | 1460 | { |
1491 | *err = REG_NOERROR; | 1461 | *err = REG_NOERROR; |
1492 | return NULL; | 1462 | return NULL; |
@@ -1505,7 +1475,7 @@ re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, | |||
1505 | 1475 | ||
1506 | /* There are no appropriate state in the dfa, create the new one. */ | 1476 | /* There are no appropriate state in the dfa, create the new one. */ |
1507 | new_state = create_ci_newstate (dfa, nodes, hash); | 1477 | new_state = create_ci_newstate (dfa, nodes, hash); |
1508 | if (BE (new_state == NULL, 0)) | 1478 | if (__glibc_unlikely (new_state == NULL)) |
1509 | *err = REG_ESPACE; | 1479 | *err = REG_ESPACE; |
1510 | 1480 | ||
1511 | return new_state; | 1481 | return new_state; |
@@ -1522,7 +1492,7 @@ re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, | |||
1522 | optimization. */ | 1492 | optimization. */ |
1523 | 1493 | ||
1524 | static re_dfastate_t * | 1494 | static re_dfastate_t * |
1525 | internal_function __attribute_warn_unused_result__ | 1495 | __attribute_warn_unused_result__ |
1526 | re_acquire_state_context (reg_errcode_t *err, const re_dfa_t *dfa, | 1496 | re_acquire_state_context (reg_errcode_t *err, const re_dfa_t *dfa, |
1527 | const re_node_set *nodes, unsigned int context) | 1497 | const re_node_set *nodes, unsigned int context) |
1528 | { | 1498 | { |
@@ -1530,7 +1500,7 @@ re_acquire_state_context (reg_errcode_t *err, const re_dfa_t *dfa, | |||
1530 | re_dfastate_t *new_state; | 1500 | re_dfastate_t *new_state; |
1531 | struct re_state_table_entry *spot; | 1501 | struct re_state_table_entry *spot; |
1532 | Idx i; | 1502 | Idx i; |
1533 | #ifdef lint | 1503 | #if defined GCC_LINT || defined lint |
1534 | /* Suppress bogus uninitialized-variable warnings. */ | 1504 | /* Suppress bogus uninitialized-variable warnings. */ |
1535 | *err = REG_NOERROR; | 1505 | *err = REG_NOERROR; |
1536 | #endif | 1506 | #endif |
@@ -1552,7 +1522,7 @@ re_acquire_state_context (reg_errcode_t *err, const re_dfa_t *dfa, | |||
1552 | } | 1522 | } |
1553 | /* There are no appropriate state in 'dfa', create the new one. */ | 1523 | /* There are no appropriate state in 'dfa', create the new one. */ |
1554 | new_state = create_cd_newstate (dfa, nodes, context, hash); | 1524 | new_state = create_cd_newstate (dfa, nodes, context, hash); |
1555 | if (BE (new_state == NULL, 0)) | 1525 | if (__glibc_unlikely (new_state == NULL)) |
1556 | *err = REG_ESPACE; | 1526 | *err = REG_ESPACE; |
1557 | 1527 | ||
1558 | return new_state; | 1528 | return new_state; |
@@ -1573,7 +1543,7 @@ register_state (const re_dfa_t *dfa, re_dfastate_t *newstate, | |||
1573 | 1543 | ||
1574 | newstate->hash = hash; | 1544 | newstate->hash = hash; |
1575 | err = re_node_set_alloc (&newstate->non_eps_nodes, newstate->nodes.nelem); | 1545 | err = re_node_set_alloc (&newstate->non_eps_nodes, newstate->nodes.nelem); |
1576 | if (BE (err != REG_NOERROR, 0)) | 1546 | if (__glibc_unlikely (err != REG_NOERROR)) |
1577 | return REG_ESPACE; | 1547 | return REG_ESPACE; |
1578 | for (i = 0; i < newstate->nodes.nelem; i++) | 1548 | for (i = 0; i < newstate->nodes.nelem; i++) |
1579 | { | 1549 | { |
@@ -1584,12 +1554,12 @@ register_state (const re_dfa_t *dfa, re_dfastate_t *newstate, | |||
1584 | } | 1554 | } |
1585 | 1555 | ||
1586 | spot = dfa->state_table + (hash & dfa->state_hash_mask); | 1556 | spot = dfa->state_table + (hash & dfa->state_hash_mask); |
1587 | if (BE (spot->alloc <= spot->num, 0)) | 1557 | if (__glibc_unlikely (spot->alloc <= spot->num)) |
1588 | { | 1558 | { |
1589 | Idx new_alloc = 2 * spot->num + 2; | 1559 | Idx new_alloc = 2 * spot->num + 2; |
1590 | re_dfastate_t **new_array = re_realloc (spot->array, re_dfastate_t *, | 1560 | re_dfastate_t **new_array = re_realloc (spot->array, re_dfastate_t *, |
1591 | new_alloc); | 1561 | new_alloc); |
1592 | if (BE (new_array == NULL, 0)) | 1562 | if (__glibc_unlikely (new_array == NULL)) |
1593 | return REG_ESPACE; | 1563 | return REG_ESPACE; |
1594 | spot->array = new_array; | 1564 | spot->array = new_array; |
1595 | spot->alloc = new_alloc; | 1565 | spot->alloc = new_alloc; |
@@ -1618,7 +1588,7 @@ free_state (re_dfastate_t *state) | |||
1618 | Return the new state if succeeded, otherwise return NULL. */ | 1588 | Return the new state if succeeded, otherwise return NULL. */ |
1619 | 1589 | ||
1620 | static re_dfastate_t * | 1590 | static re_dfastate_t * |
1621 | internal_function __attribute_warn_unused_result__ | 1591 | __attribute_warn_unused_result__ |
1622 | create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | 1592 | create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, |
1623 | re_hashval_t hash) | 1593 | re_hashval_t hash) |
1624 | { | 1594 | { |
@@ -1627,10 +1597,10 @@ create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1627 | re_dfastate_t *newstate; | 1597 | re_dfastate_t *newstate; |
1628 | 1598 | ||
1629 | newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); | 1599 | newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); |
1630 | if (BE (newstate == NULL, 0)) | 1600 | if (__glibc_unlikely (newstate == NULL)) |
1631 | return NULL; | 1601 | return NULL; |
1632 | err = re_node_set_init_copy (&newstate->nodes, nodes); | 1602 | err = re_node_set_init_copy (&newstate->nodes, nodes); |
1633 | if (BE (err != REG_NOERROR, 0)) | 1603 | if (__glibc_unlikely (err != REG_NOERROR)) |
1634 | { | 1604 | { |
1635 | re_free (newstate); | 1605 | re_free (newstate); |
1636 | return NULL; | 1606 | return NULL; |
@@ -1643,9 +1613,7 @@ create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1643 | re_token_type_t type = node->type; | 1613 | re_token_type_t type = node->type; |
1644 | if (type == CHARACTER && !node->constraint) | 1614 | if (type == CHARACTER && !node->constraint) |
1645 | continue; | 1615 | continue; |
1646 | #ifdef RE_ENABLE_I18N | ||
1647 | newstate->accept_mb |= node->accept_mb; | 1616 | newstate->accept_mb |= node->accept_mb; |
1648 | #endif /* RE_ENABLE_I18N */ | ||
1649 | 1617 | ||
1650 | /* If the state has the halt node, the state is a halt state. */ | 1618 | /* If the state has the halt node, the state is a halt state. */ |
1651 | if (type == END_OF_RE) | 1619 | if (type == END_OF_RE) |
@@ -1656,7 +1624,7 @@ create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1656 | newstate->has_constraint = 1; | 1624 | newstate->has_constraint = 1; |
1657 | } | 1625 | } |
1658 | err = register_state (dfa, newstate, hash); | 1626 | err = register_state (dfa, newstate, hash); |
1659 | if (BE (err != REG_NOERROR, 0)) | 1627 | if (__glibc_unlikely (err != REG_NOERROR)) |
1660 | { | 1628 | { |
1661 | free_state (newstate); | 1629 | free_state (newstate); |
1662 | newstate = NULL; | 1630 | newstate = NULL; |
@@ -1668,7 +1636,7 @@ create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1668 | Return the new state if succeeded, otherwise return NULL. */ | 1636 | Return the new state if succeeded, otherwise return NULL. */ |
1669 | 1637 | ||
1670 | static re_dfastate_t * | 1638 | static re_dfastate_t * |
1671 | internal_function __attribute_warn_unused_result__ | 1639 | __attribute_warn_unused_result__ |
1672 | create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | 1640 | create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, |
1673 | unsigned int context, re_hashval_t hash) | 1641 | unsigned int context, re_hashval_t hash) |
1674 | { | 1642 | { |
@@ -1677,10 +1645,10 @@ create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1677 | re_dfastate_t *newstate; | 1645 | re_dfastate_t *newstate; |
1678 | 1646 | ||
1679 | newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); | 1647 | newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); |
1680 | if (BE (newstate == NULL, 0)) | 1648 | if (__glibc_unlikely (newstate == NULL)) |
1681 | return NULL; | 1649 | return NULL; |
1682 | err = re_node_set_init_copy (&newstate->nodes, nodes); | 1650 | err = re_node_set_init_copy (&newstate->nodes, nodes); |
1683 | if (BE (err != REG_NOERROR, 0)) | 1651 | if (__glibc_unlikely (err != REG_NOERROR)) |
1684 | { | 1652 | { |
1685 | re_free (newstate); | 1653 | re_free (newstate); |
1686 | return NULL; | 1654 | return NULL; |
@@ -1697,9 +1665,7 @@ create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1697 | 1665 | ||
1698 | if (type == CHARACTER && !constraint) | 1666 | if (type == CHARACTER && !constraint) |
1699 | continue; | 1667 | continue; |
1700 | #ifdef RE_ENABLE_I18N | ||
1701 | newstate->accept_mb |= node->accept_mb; | 1668 | newstate->accept_mb |= node->accept_mb; |
1702 | #endif /* RE_ENABLE_I18N */ | ||
1703 | 1669 | ||
1704 | /* If the state has the halt node, the state is a halt state. */ | 1670 | /* If the state has the halt node, the state is a halt state. */ |
1705 | if (type == END_OF_RE) | 1671 | if (type == END_OF_RE) |
@@ -1711,15 +1677,19 @@ create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1711 | { | 1677 | { |
1712 | if (newstate->entrance_nodes == &newstate->nodes) | 1678 | if (newstate->entrance_nodes == &newstate->nodes) |
1713 | { | 1679 | { |
1714 | newstate->entrance_nodes = re_malloc (re_node_set, 1); | 1680 | re_node_set *entrance_nodes = re_malloc (re_node_set, 1); |
1715 | if (BE (newstate->entrance_nodes == NULL, 0)) | 1681 | if (__glibc_unlikely (entrance_nodes == NULL)) |
1716 | { | 1682 | { |
1717 | free_state (newstate); | 1683 | free_state (newstate); |
1718 | return NULL; | 1684 | return NULL; |
1719 | } | 1685 | } |
1686 | newstate->entrance_nodes = entrance_nodes; | ||
1720 | if (re_node_set_init_copy (newstate->entrance_nodes, nodes) | 1687 | if (re_node_set_init_copy (newstate->entrance_nodes, nodes) |
1721 | != REG_NOERROR) | 1688 | != REG_NOERROR) |
1722 | return NULL; | 1689 | { |
1690 | free_state (newstate); | ||
1691 | return NULL; | ||
1692 | } | ||
1723 | nctx_nodes = 0; | 1693 | nctx_nodes = 0; |
1724 | newstate->has_constraint = 1; | 1694 | newstate->has_constraint = 1; |
1725 | } | 1695 | } |
@@ -1732,7 +1702,7 @@ create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, | |||
1732 | } | 1702 | } |
1733 | } | 1703 | } |
1734 | err = register_state (dfa, newstate, hash); | 1704 | err = register_state (dfa, newstate, hash); |
1735 | if (BE (err != REG_NOERROR, 0)) | 1705 | if (__glibc_unlikely (err != REG_NOERROR)) |
1736 | { | 1706 | { |
1737 | free_state (newstate); | 1707 | free_state (newstate); |
1738 | newstate = NULL; | 1708 | newstate = NULL; |