| 1 | /* Copyright (C) 1995-2020 Free Software Foundation, Inc. | 
|---|
| 2 | This file is part of the GNU C Library. | 
|---|
| 3 | Written by Ulrich Drepper <drepper@gnu.org>, 1995. | 
|---|
| 4 |  | 
|---|
| 5 | The GNU C Library is free software; you can redistribute it and/or | 
|---|
| 6 | modify it under the terms of the GNU Lesser General Public | 
|---|
| 7 | License as published by the Free Software Foundation; either | 
|---|
| 8 | version 2.1 of the License, or (at your option) any later version. | 
|---|
| 9 |  | 
|---|
| 10 | The GNU C Library is distributed in the hope that it will be useful, | 
|---|
| 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of | 
|---|
| 12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU | 
|---|
| 13 | Lesser General Public License for more details. | 
|---|
| 14 |  | 
|---|
| 15 | You should have received a copy of the GNU Lesser General Public | 
|---|
| 16 | License along with the GNU C Library; if not, see | 
|---|
| 17 | <https://www.gnu.org/licenses/>.  */ | 
|---|
| 18 |  | 
|---|
| 19 |  | 
|---|
| 20 | #include <assert.h> | 
|---|
| 21 | #include <langinfo.h> | 
|---|
| 22 | #include <locale.h> | 
|---|
| 23 | #include <stddef.h> | 
|---|
| 24 | #include <stdint.h> | 
|---|
| 25 | #include <string.h> | 
|---|
| 26 | #include <sys/param.h> | 
|---|
| 27 | #include <libc-diag.h> | 
|---|
| 28 |  | 
|---|
| 29 | #ifndef STRING_TYPE | 
|---|
| 30 | # define STRING_TYPE char | 
|---|
| 31 | # define USTRING_TYPE unsigned char | 
|---|
| 32 | # define STRCOLL __strcoll_l | 
|---|
| 33 | # define STRCMP strcmp | 
|---|
| 34 | # define WEIGHT_H "../locale/weight.h" | 
|---|
| 35 | # define SUFFIX	MB | 
|---|
| 36 | # define L(arg) arg | 
|---|
| 37 | #endif | 
|---|
| 38 |  | 
|---|
| 39 | #define CONCAT(a,b) CONCAT1(a,b) | 
|---|
| 40 | #define CONCAT1(a,b) a##b | 
|---|
| 41 |  | 
|---|
| 42 | #include "../locale/localeinfo.h" | 
|---|
| 43 | #include WEIGHT_H | 
|---|
| 44 |  | 
|---|
| 45 | /* Track status while looking for sequences in a string.  */ | 
|---|
| 46 | typedef struct | 
|---|
| 47 | { | 
|---|
| 48 | int len;			/* Length of the current sequence.  */ | 
|---|
| 49 | size_t val;			/* Position of the sequence relative to the | 
|---|
| 50 | previous non-ignored sequence.  */ | 
|---|
| 51 | size_t idxmax;		/* Maximum index in sequences.  */ | 
|---|
| 52 | size_t idxcnt;		/* Current count of indices.  */ | 
|---|
| 53 | size_t backw;			/* Current Backward sequence index.  */ | 
|---|
| 54 | size_t backw_stop;		/* Index where the backward sequences stop.  */ | 
|---|
| 55 | const USTRING_TYPE *us;	/* The string.  */ | 
|---|
| 56 | unsigned char rule;		/* Saved rule for the first sequence.  */ | 
|---|
| 57 | int32_t idx;			/* Index to weight of the current sequence.  */ | 
|---|
| 58 | int32_t save_idx;		/* Save looked up index of a forward | 
|---|
| 59 | sequence after the last backward | 
|---|
| 60 | sequence.  */ | 
|---|
| 61 | const USTRING_TYPE *back_us;	/* Beginning of the backward sequence.  */ | 
|---|
| 62 | } coll_seq; | 
|---|
| 63 |  | 
|---|
| 64 | /* Get next sequence.  Traverse the string as required.  */ | 
|---|
| 65 | static __always_inline void | 
|---|
| 66 | get_next_seq (coll_seq *seq, int nrules, const unsigned char *rulesets, | 
|---|
| 67 | const USTRING_TYPE *weights, const int32_t *table, | 
|---|
| 68 | const USTRING_TYPE *, const int32_t *indirect, | 
|---|
| 69 | int pass) | 
|---|
| 70 | { | 
|---|
| 71 | size_t val = seq->val = 0; | 
|---|
| 72 | int len = seq->len; | 
|---|
| 73 | size_t backw_stop = seq->backw_stop; | 
|---|
| 74 | size_t backw = seq->backw; | 
|---|
| 75 | size_t idxcnt = seq->idxcnt; | 
|---|
| 76 | size_t idxmax = seq->idxmax; | 
|---|
| 77 | int32_t idx = seq->idx; | 
|---|
| 78 | const USTRING_TYPE *us = seq->us; | 
|---|
| 79 |  | 
|---|
| 80 | while (len == 0) | 
|---|
| 81 | { | 
|---|
| 82 | ++val; | 
|---|
| 83 | if (backw_stop != ~0ul) | 
|---|
| 84 | { | 
|---|
| 85 | /* There is something pushed.  */ | 
|---|
| 86 | if (backw == backw_stop) | 
|---|
| 87 | { | 
|---|
| 88 | /* The last pushed character was handled.  Continue | 
|---|
| 89 | with forward characters.  */ | 
|---|
| 90 | if (idxcnt < idxmax) | 
|---|
| 91 | { | 
|---|
| 92 | idx = seq->save_idx; | 
|---|
| 93 | backw_stop = ~0ul; | 
|---|
| 94 | } | 
|---|
| 95 | else | 
|---|
| 96 | { | 
|---|
| 97 | /* Nothing anymore.  The backward sequence ended with | 
|---|
| 98 | the last sequence in the string.  Note that len is | 
|---|
| 99 | still zero.  */ | 
|---|
| 100 | idx = 0; | 
|---|
| 101 | break; | 
|---|
| 102 | } | 
|---|
| 103 | } | 
|---|
| 104 | else | 
|---|
| 105 | { | 
|---|
| 106 | /* XXX Traverse BACKW sequences from the beginning of | 
|---|
| 107 | BACKW_STOP to get the next sequence.  Is ther a quicker way | 
|---|
| 108 | to do this?  */ | 
|---|
| 109 | size_t i = backw_stop; | 
|---|
| 110 | us = seq->back_us; | 
|---|
| 111 | while (i < backw) | 
|---|
| 112 | { | 
|---|
| 113 | int32_t tmp = findidx (table, indirect, extra, &us, -1); | 
|---|
| 114 | idx = tmp & 0xffffff; | 
|---|
| 115 | i++; | 
|---|
| 116 | } | 
|---|
| 117 | --backw; | 
|---|
| 118 | us = seq->us; | 
|---|
| 119 | } | 
|---|
| 120 | } | 
|---|
| 121 | else | 
|---|
| 122 | { | 
|---|
| 123 | backw_stop = idxmax; | 
|---|
| 124 | int32_t prev_idx = idx; | 
|---|
| 125 |  | 
|---|
| 126 | while (*us != L('\0')) | 
|---|
| 127 | { | 
|---|
| 128 | int32_t tmp = findidx (table, indirect, extra, &us, -1); | 
|---|
| 129 | unsigned char rule = tmp >> 24; | 
|---|
| 130 | prev_idx = idx; | 
|---|
| 131 | idx = tmp & 0xffffff; | 
|---|
| 132 | idxcnt = idxmax++; | 
|---|
| 133 |  | 
|---|
| 134 | /* Save the rule for the first sequence.  */ | 
|---|
| 135 | if (__glibc_unlikely (idxcnt == 0)) | 
|---|
| 136 | seq->rule = rule; | 
|---|
| 137 |  | 
|---|
| 138 | if ((rulesets[rule * nrules + pass] | 
|---|
| 139 | & sort_backward) == 0) | 
|---|
| 140 | /* No more backward characters to push.  */ | 
|---|
| 141 | break; | 
|---|
| 142 | ++idxcnt; | 
|---|
| 143 | } | 
|---|
| 144 |  | 
|---|
| 145 | if (backw_stop >= idxcnt) | 
|---|
| 146 | { | 
|---|
| 147 | /* No sequence at all or just one.  */ | 
|---|
| 148 | if (idxcnt == idxmax || backw_stop > idxcnt) | 
|---|
| 149 | /* Note that len is still zero.  */ | 
|---|
| 150 | break; | 
|---|
| 151 |  | 
|---|
| 152 | backw_stop = ~0ul; | 
|---|
| 153 | } | 
|---|
| 154 | else | 
|---|
| 155 | { | 
|---|
| 156 | /* We pushed backward sequences.  If the stream ended with the | 
|---|
| 157 | backward sequence, then we process the last sequence we | 
|---|
| 158 | found.  Otherwise we process the sequence before the last | 
|---|
| 159 | one since the last one was a forward sequence.  */ | 
|---|
| 160 | seq->back_us = seq->us; | 
|---|
| 161 | seq->us = us; | 
|---|
| 162 | backw = idxcnt; | 
|---|
| 163 | if (idxmax > idxcnt) | 
|---|
| 164 | { | 
|---|
| 165 | backw--; | 
|---|
| 166 | seq->save_idx = idx; | 
|---|
| 167 | idx = prev_idx; | 
|---|
| 168 | } | 
|---|
| 169 | if (backw > backw_stop) | 
|---|
| 170 | backw--; | 
|---|
| 171 | } | 
|---|
| 172 | } | 
|---|
| 173 |  | 
|---|
| 174 | /* With GCC 5.3 when compiling with -Os the compiler complains | 
|---|
| 175 | that idx, taken from seq->idx (seq1 or seq2 from STRCOLL) may | 
|---|
| 176 | be used uninitialized.  In general this can't possibly be true | 
|---|
| 177 | since seq1.idx and seq2.idx are initialized to zero in the | 
|---|
| 178 | outer function.  Only one case where seq->idx is restored from | 
|---|
| 179 | seq->save_idx might result in an uninitialized idx value, but | 
|---|
| 180 | it is guarded by a sequence of checks against backw_stop which | 
|---|
| 181 | ensures that seq->save_idx was saved to first and contains a | 
|---|
| 182 | valid value.  */ | 
|---|
| 183 | DIAG_PUSH_NEEDS_COMMENT; | 
|---|
| 184 | DIAG_IGNORE_Os_NEEDS_COMMENT (5, "-Wmaybe-uninitialized"); | 
|---|
| 185 | len = weights[idx++]; | 
|---|
| 186 | DIAG_POP_NEEDS_COMMENT; | 
|---|
| 187 | /* Skip over indices of previous levels.  */ | 
|---|
| 188 | for (int i = 0; i < pass; i++) | 
|---|
| 189 | { | 
|---|
| 190 | idx += len; | 
|---|
| 191 | len = weights[idx]; | 
|---|
| 192 | idx++; | 
|---|
| 193 | } | 
|---|
| 194 | } | 
|---|
| 195 |  | 
|---|
| 196 | /* Update the structure.  */ | 
|---|
| 197 | seq->val = val; | 
|---|
| 198 | seq->len = len; | 
|---|
| 199 | seq->backw_stop = backw_stop; | 
|---|
| 200 | seq->backw = backw; | 
|---|
| 201 | seq->idxcnt = idxcnt; | 
|---|
| 202 | seq->idxmax = idxmax; | 
|---|
| 203 | seq->us = us; | 
|---|
| 204 | seq->idx = idx; | 
|---|
| 205 | } | 
|---|
| 206 |  | 
|---|
| 207 | /* Compare two sequences.  */ | 
|---|
| 208 | static __always_inline int | 
|---|
| 209 | do_compare (coll_seq *seq1, coll_seq *seq2, int position, | 
|---|
| 210 | const USTRING_TYPE *weights) | 
|---|
| 211 | { | 
|---|
| 212 | int seq1len = seq1->len; | 
|---|
| 213 | int seq2len = seq2->len; | 
|---|
| 214 | size_t val1 = seq1->val; | 
|---|
| 215 | size_t val2 = seq2->val; | 
|---|
| 216 | int idx1 = seq1->idx; | 
|---|
| 217 | int idx2 = seq2->idx; | 
|---|
| 218 | int result = 0; | 
|---|
| 219 |  | 
|---|
| 220 | /* Test for position if necessary.  */ | 
|---|
| 221 | if (position && val1 != val2) | 
|---|
| 222 | { | 
|---|
| 223 | result = val1 > val2 ? 1 : -1; | 
|---|
| 224 | goto out; | 
|---|
| 225 | } | 
|---|
| 226 |  | 
|---|
| 227 | /* Compare the two sequences.  */ | 
|---|
| 228 | do | 
|---|
| 229 | { | 
|---|
| 230 | if (weights[idx1] != weights[idx2]) | 
|---|
| 231 | { | 
|---|
| 232 | /* The sequences differ.  */ | 
|---|
| 233 | result = weights[idx1] - weights[idx2]; | 
|---|
| 234 | goto out; | 
|---|
| 235 | } | 
|---|
| 236 |  | 
|---|
| 237 | /* Increment the offsets.  */ | 
|---|
| 238 | ++idx1; | 
|---|
| 239 | ++idx2; | 
|---|
| 240 |  | 
|---|
| 241 | --seq1len; | 
|---|
| 242 | --seq2len; | 
|---|
| 243 | } | 
|---|
| 244 | while (seq1len > 0 && seq2len > 0); | 
|---|
| 245 |  | 
|---|
| 246 | if (position && seq1len != seq2len) | 
|---|
| 247 | result = seq1len - seq2len; | 
|---|
| 248 |  | 
|---|
| 249 | out: | 
|---|
| 250 | seq1->len = seq1len; | 
|---|
| 251 | seq2->len = seq2len; | 
|---|
| 252 | seq1->idx = idx1; | 
|---|
| 253 | seq2->idx = idx2; | 
|---|
| 254 | return result; | 
|---|
| 255 | } | 
|---|
| 256 |  | 
|---|
| 257 | int | 
|---|
| 258 | STRCOLL (const STRING_TYPE *s1, const STRING_TYPE *s2, locale_t l) | 
|---|
| 259 | { | 
|---|
| 260 | struct __locale_data *current = l->__locales[LC_COLLATE]; | 
|---|
| 261 | uint_fast32_t nrules = current->values[_NL_ITEM_INDEX (_NL_COLLATE_NRULES)].word; | 
|---|
| 262 | /* We don't assign the following values right away since it might be | 
|---|
| 263 | unnecessary in case there are no rules.  */ | 
|---|
| 264 | const unsigned char *rulesets; | 
|---|
| 265 | const int32_t *table; | 
|---|
| 266 | const USTRING_TYPE *weights; | 
|---|
| 267 | const USTRING_TYPE *; | 
|---|
| 268 | const int32_t *indirect; | 
|---|
| 269 |  | 
|---|
| 270 | if (nrules == 0) | 
|---|
| 271 | return STRCMP (s1, s2); | 
|---|
| 272 |  | 
|---|
| 273 | /* Catch empty strings.  */ | 
|---|
| 274 | if (__glibc_unlikely (*s1 == '\0') || __glibc_unlikely (*s2 == '\0')) | 
|---|
| 275 | return (*s1 != '\0') - (*s2 != '\0'); | 
|---|
| 276 |  | 
|---|
| 277 | rulesets = (const unsigned char *) | 
|---|
| 278 | current->values[_NL_ITEM_INDEX (_NL_COLLATE_RULESETS)].string; | 
|---|
| 279 | table = (const int32_t *) | 
|---|
| 280 | current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_TABLE,SUFFIX))].string; | 
|---|
| 281 | weights = (const USTRING_TYPE *) | 
|---|
| 282 | current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_WEIGHT,SUFFIX))].string; | 
|---|
| 283 | extra = (const USTRING_TYPE *) | 
|---|
| 284 | current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_EXTRA,SUFFIX))].string; | 
|---|
| 285 | indirect = (const int32_t *) | 
|---|
| 286 | current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_INDIRECT,SUFFIX))].string; | 
|---|
| 287 |  | 
|---|
| 288 | assert (((uintptr_t) table) % __alignof__ (table[0]) == 0); | 
|---|
| 289 | assert (((uintptr_t) weights) % __alignof__ (weights[0]) == 0); | 
|---|
| 290 | assert (((uintptr_t) extra) % __alignof__ (extra[0]) == 0); | 
|---|
| 291 | assert (((uintptr_t) indirect) % __alignof__ (indirect[0]) == 0); | 
|---|
| 292 |  | 
|---|
| 293 | int result = 0, rule = 0; | 
|---|
| 294 |  | 
|---|
| 295 | /* With GCC 7 when compiling with -Os the compiler warns that | 
|---|
| 296 | seq1.back_us and seq2.back_us might be used uninitialized. | 
|---|
| 297 | Sometimes this warning appears at locations in locale/weightwc.h | 
|---|
| 298 | where the actual use is, but on architectures other than x86_64, | 
|---|
| 299 | x86 and s390x, a warning appears at the definitions of seq1 and | 
|---|
| 300 | seq2.  This uninitialized use is impossible for the same reason | 
|---|
| 301 | as described in comments in locale/weightwc.h.  */ | 
|---|
| 302 | DIAG_PUSH_NEEDS_COMMENT; | 
|---|
| 303 | DIAG_IGNORE_Os_NEEDS_COMMENT (7, "-Wmaybe-uninitialized"); | 
|---|
| 304 | coll_seq seq1, seq2; | 
|---|
| 305 | DIAG_POP_NEEDS_COMMENT; | 
|---|
| 306 | seq1.len = 0; | 
|---|
| 307 | seq1.idxmax = 0; | 
|---|
| 308 | seq1.rule = 0; | 
|---|
| 309 | seq2.len = 0; | 
|---|
| 310 | seq2.idxmax = 0; | 
|---|
| 311 |  | 
|---|
| 312 | for (int pass = 0; pass < nrules; ++pass) | 
|---|
| 313 | { | 
|---|
| 314 | seq1.idxcnt = 0; | 
|---|
| 315 | seq1.idx = 0; | 
|---|
| 316 | seq2.idx = 0; | 
|---|
| 317 | seq1.backw_stop = ~0ul; | 
|---|
| 318 | seq1.backw = ~0ul; | 
|---|
| 319 | seq2.idxcnt = 0; | 
|---|
| 320 | seq2.backw_stop = ~0ul; | 
|---|
| 321 | seq2.backw = ~0ul; | 
|---|
| 322 |  | 
|---|
| 323 | /* We need the elements of the strings as unsigned values since they | 
|---|
| 324 | are used as indices.  */ | 
|---|
| 325 | seq1.us = (const USTRING_TYPE *) s1; | 
|---|
| 326 | seq2.us = (const USTRING_TYPE *) s2; | 
|---|
| 327 |  | 
|---|
| 328 | /* We assume that if a rule has defined `position' in one section | 
|---|
| 329 | this is true for all of them.  Please note that the localedef programs | 
|---|
| 330 | makes sure that `position' is not used at the first level.  */ | 
|---|
| 331 |  | 
|---|
| 332 | int position = rulesets[rule * nrules + pass] & sort_position; | 
|---|
| 333 |  | 
|---|
| 334 | while (1) | 
|---|
| 335 | { | 
|---|
| 336 | get_next_seq (&seq1, nrules, rulesets, weights, table, | 
|---|
| 337 | extra, indirect, pass); | 
|---|
| 338 | get_next_seq (&seq2, nrules, rulesets, weights, table, | 
|---|
| 339 | extra, indirect, pass); | 
|---|
| 340 | /* See whether any or both strings are empty.  */ | 
|---|
| 341 | if (seq1.len == 0 || seq2.len == 0) | 
|---|
| 342 | { | 
|---|
| 343 | if (seq1.len == seq2.len) | 
|---|
| 344 | { | 
|---|
| 345 | /* Both strings ended and are equal at this level.  Do a | 
|---|
| 346 | byte-level comparison to ensure that we don't waste time | 
|---|
| 347 | going through multiple passes for totally equal strings | 
|---|
| 348 | before proceeding to subsequent passes.  */ | 
|---|
| 349 | if (pass == 0 && STRCMP (s1, s2) == 0) | 
|---|
| 350 | return result; | 
|---|
| 351 | else | 
|---|
| 352 | break; | 
|---|
| 353 | } | 
|---|
| 354 |  | 
|---|
| 355 | /* This means one string is shorter than the other.  Find out | 
|---|
| 356 | which one and return an appropriate value.  */ | 
|---|
| 357 | return seq1.len == 0 ? -1 : 1; | 
|---|
| 358 | } | 
|---|
| 359 |  | 
|---|
| 360 | result = do_compare (&seq1, &seq2, position, weights); | 
|---|
| 361 | if (result != 0) | 
|---|
| 362 | return result; | 
|---|
| 363 | } | 
|---|
| 364 |  | 
|---|
| 365 | rule = seq1.rule; | 
|---|
| 366 | } | 
|---|
| 367 |  | 
|---|
| 368 | return result; | 
|---|
| 369 | } | 
|---|
| 370 | libc_hidden_def (STRCOLL) | 
|---|
| 371 |  | 
|---|
| 372 | #ifndef WIDE_CHAR_VERSION | 
|---|
| 373 | weak_alias (__strcoll_l, strcoll_l) | 
|---|
| 374 | #endif | 
|---|
| 375 |  | 
|---|