php-systzdata.patch

   1 Add support for use of the system timezone database, rather
   2 than embedding a copy.  Discussed upstream but was not desired.
   3
   4 History:
   5 r9: fix another compile error without --with-system-tzdata configured (Michael Heimpold)
   6 r8: fix compile error without --with-system-tzdata configured
   7 r7: improve check for valid timezone id to exclude directories
   8 r6: fix fd leak in r5, fix country code/BC flag use in
   9     timezone_identifiers_list() using system db,
  10     fix use of PECL timezonedb to override system db,
  11 r5: reverts addition of "System/Localtime" fake tzname.
  12     updated for 5.3.0, parses zone.tab to pick up mapping between
  13     timezone name, country code and long/lat coords
  14 r4: added "System/Localtime" tzname which uses /etc/localtime
  15 r3: fix a crash if /usr/share/zoneinfo doesn't exist (Raphael Geissert)
  16 r2: add filesystem trawl to set up name alias index
  17 r1: initial revision
  18
  19 --- a/ext/date/lib/parse_tz.c
  20 +++ b/ext/date/lib/parse_tz.c
  21 @@ -24,6 +24,16 @@
  22
  23  #include "timelib.h"
  24
  25 +#ifdef HAVE_SYSTEM_TZDATA
  26 +#include <sys/mman.h>
  27 +#include <sys/stat.h>
  28 +#include <limits.h>
  29 +#include <fcntl.h>
  30 +#include <unistd.h>
  31 +
  32 +#include "php_scandir.h"
  33 +#endif
  34 +
  35  #include <stdio.h>
  36
  37  #ifdef HAVE_LOCALE_H
  38 @@ -35,7 +45,12 @@
  39  #else
  40  #include <strings.h>
  41  #endif
  42 +
  43 +#ifndef HAVE_SYSTEM_TZDATA
  44  #include "timezonedb.h"
  45 +#endif
  46 +
  47 +#include <ctype.h>
  48
  49  #if (defined(__APPLE__) || defined(__APPLE_CC__)) && (defined(__BIG_ENDIAN__) || defined(__LITTLE_ENDIAN__))
  50  # if defined(__LITTLE_ENDIAN__)
  51 @@ -55,9 +70,14 @@
  52
  53  static void read_preamble(const unsigned char **tzf, timelib_tzinfo *tz)
  54  {
  55 -       /* skip ID */
  56 -       *tzf += 4;
  57 -
  58 +        if (memcmp(tzf, "TZif", 4) == 0) {
  59 +                *tzf += 20;
  60 +                return;
  61 +        }
  62 +
  63 +        /* skip ID */
  64 +        *tzf += 4;
  65 +
  66         /* read BC flag */
  67         tz->bc = (**tzf == '\1');
  68         *tzf += 1;
  69 @@ -260,7 +280,397 @@ void timelib_dump_tzinfo(timelib_tzinfo
  70         }
  71  }
  72
  73 -static int seek_to_tz_position(const unsigned char **tzf, char *timezone, const timelib_tzdb *tzdb)
  74 +#ifdef HAVE_SYSTEM_TZDATA
  75 +
  76 +#ifdef HAVE_SYSTEM_TZDATA_PREFIX
  77 +#define ZONEINFO_PREFIX HAVE_SYSTEM_TZDATA_PREFIX
  78 +#else
  79 +#define ZONEINFO_PREFIX "/usr/share/zoneinfo"
  80 +#endif
  81 +
  82 +/* System timezone database pointer. */
  83 +static const timelib_tzdb *timezonedb_system = NULL;
  84 +
  85 +/* Hash table entry for the cache of the zone.tab mapping table. */
  86 +struct location_info {
  87 +        char code[2];
  88 +        double latitude, longitude;
  89 +        char name[64];
  90 +        char *comment;
  91 +        struct location_info *next;
  92 +};
  93 +
  94 +/* Cache of zone.tab. */
  95 +static struct location_info **system_location_table;
  96 +
  97 +/* Size of the zone.tab hash table; a random-ish prime big enough to
  98 + * prevent too many collisions. */
  99 +#define LOCINFO_HASH_SIZE (1021)
 100 +
 101 +static uint32_t tz_hash(const char *str)
 102 +{
 103 +    const unsigned char *p = (const unsigned char *)str;
 104 +    uint32_t hash = 5381;
 105 +    int c;
 106 +
 107 +    while ((c = *p++) != '\0') {
 108 +        hash = (hash << 5) ^ hash ^ c;
 109 +    }
 110 +
 111 +    return hash % LOCINFO_HASH_SIZE;
 112 +}
 113 +
 114 +/* Parse an ISO-6709 date as used in zone.tab. Returns end of the
 115 + * parsed string on success, or NULL on parse error.  On success,
 116 + * writes the parsed number to *result. */
 117 +static char *parse_iso6709(char *p, double *result)
 118 +{
 119 +    double v, sign;
 120 +    char *pend;
 121 +    size_t len;
 122 +
 123 +    if (*p == '+')
 124 +        sign = 1.0;
 125 +    else if (*p == '-')
 126 +        sign = -1.0;
 127 +    else
 128 +        return NULL;
 129 +
 130 +    p++;
 131 +    for (pend = p; *pend >= '0' && *pend <= '9'; pend++)
 132 +        ;;
 133 +
 134 +    /* Annoying encoding used by zone.tab has no decimal point, so use
 135 +     * the length to determine the format:
 136 +     *
 137 +     * 4 = DDMM
 138 +     * 5 = DDDMM
 139 +     * 6 = DDMMSS
 140 +     * 7 = DDDMMSS
 141 +     */
 142 +    len = pend - p;
 143 +    if (len < 4 || len > 7) {
 144 +        return NULL;
 145 +    }
 146 +
 147 +    /* p => [D]DD */
 148 +    v = (p[0] - '0') * 10.0 + (p[1] - '0');
 149 +    p += 2;
 150 +    if (len == 5 || len == 7)
 151 +        v = v * 10.0 + (*p++ - '0');
 152 +    /* p => MM[SS] */
 153 +    v += (10.0 * (p[0] - '0')
 154 +          + p[1] - '0') / 60.0;
 155 +    p += 2;
 156 +    /* p => [SS] */
 157 +    if (len > 5) {
 158 +        v += (10.0 * (p[0] - '0')
 159 +              + p[1] - '0') / 3600.0;
 160 +        p += 2;
 161 +    }
 162 +
 163 +    /* Round to five decimal place, not because it's a good idea,
 164 +     * but, because the builtin data uses rounded data, so, match
 165 +     * that. */
 166 +    *result = round(v * sign * 100000.0) / 100000.0;
 167 +
 168 +    return p;
 169 +}
 170 +
 171 +/* This function parses the zone.tab file to build up the mapping of
 172 + * timezone to country code and geographic location, and returns a
 173 + * hash table.  The hash table is indexed by the function:
 174 + *
 175 + *   tz_hash(timezone-name)
 176 + */
 177 +static struct location_info **create_location_table(void)
 178 +{
 179 +    struct location_info **li, *i;
 180 +    char zone_tab[PATH_MAX];
 181 +    char line[512];
 182 +    FILE *fp;
 183 +
 184 +    strncpy(zone_tab, ZONEINFO_PREFIX "/zone.tab", sizeof zone_tab);
 185 +
 186 +    fp = fopen(zone_tab, "r");
 187 +    if (!fp) {
 188 +        return NULL;
 189 +    }
 190 +
 191 +    li = calloc(LOCINFO_HASH_SIZE, sizeof *li);
 192 +
 193 +    while (fgets(line, sizeof line, fp)) {
 194 +        char *p = line, *code, *name, *comment;
 195 +        uint32_t hash;
 196 +        double latitude, longitude;
 197 +
 198 +        while (isspace(*p))
 199 +            p++;
 200 +
 201 +        if (*p == '#' || *p == '\0' || *p == '\n')
 202 +            continue;
 203 +
 204 +        if (!isalpha(p[0]) || !isalpha(p[1]) || p[2] != '\t')
 205 +            continue;
 206 +
 207 +        /* code => AA */
 208 +        code = p;
 209 +        p[2] = 0;
 210 +        p += 3;
 211 +
 212 +        /* coords => [+-][D]DDMM[SS][+-][D]DDMM[SS] */
 213 +        p = parse_iso6709(p, &latitude);
 214 +        if (!p) {
 215 +            continue;
 216 +        }
 217 +        p = parse_iso6709(p, &longitude);
 218 +        if (!p) {
 219 +            continue;
 220 +        }
 221 +
 222 +        if (!p || *p != '\t') {
 223 +            continue;
 224 +        }
 225 +
 226 +        /* name = string */
 227 +        name = ++p;
 228 +        while (*p != '\t' && *p && *p != '\n')
 229 +            p++;
 230 +
 231 +        *p++ = '\0';
 232 +
 233 +        /* comment = string */
 234 +        comment = p;
 235 +        while (*p != '\t' && *p && *p != '\n')
 236 +            p++;
 237 +
 238 +        if (*p == '\n' || *p == '\t')
 239 +            *p = '\0';
 240 +
 241 +        hash = tz_hash(name);
 242 +        i = malloc(sizeof *i);
 243 +        memcpy(i->code, code, 2);
 244 +        strncpy(i->name, name, sizeof i->name);
 245 +        i->comment = strdup(comment);
 246 +        i->longitude = longitude;
 247 +        i->latitude = latitude;
 248 +        i->next = li[hash];
 249 +        li[hash] = i;
 250 +        /* printf("%s [%u, %f, %f]\n", name, hash, latitude, longitude); */
 251 +    }
 252 +
 253 +    fclose(fp);
 254 +
 255 +    return li;
 256 +}
 257 +
 258 +/* Return location info from hash table, using given timezone name.
 259 + * Returns NULL if the name could not be found. */
 260 +const struct location_info *find_zone_info(struct location_info **li,
 261 +                                           const char *name)
 262 +{
 263 +    uint32_t hash = tz_hash(name);
 264 +    const struct location_info *l;
 265 +
 266 +    if (!li) {
 267 +        return NULL;
 268 +    }
 269 +
 270 +    for (l = li[hash]; l; l = l->next) {
 271 +        if (strcasecmp(l->name, name) == 0)
 272 +            return l;
 273 +    }
 274 +
 275 +    return NULL;
 276 +}
 277 +
 278 +/* Filter out some non-tzdata files and the posix/right databases, if
 279 + * present. */
 280 +static int index_filter(const struct dirent *ent)
 281 +{
 282 +       return strcmp(ent->d_name, ".") != 0
 283 +               && strcmp(ent->d_name, "..") != 0
 284 +               && strcmp(ent->d_name, "posix") != 0
 285 +               && strcmp(ent->d_name, "posixrules") != 0
 286 +               && strcmp(ent->d_name, "right") != 0
 287 +               && strstr(ent->d_name, ".tab") == NULL;
 288 +}
 289 +
 290 +static int sysdbcmp(const void *first, const void *second)
 291 +{
 292 +        const timelib_tzdb_index_entry *alpha = first, *beta = second;
 293 +
 294 +        return strcmp(alpha->id, beta->id);
 295 +}
 296 +
 297 +
 298 +/* Create the zone identifier index by trawling the filesystem. */
 299 +static void create_zone_index(timelib_tzdb *db)
 300 +{
 301 +       size_t dirstack_size,  dirstack_top;
 302 +       size_t index_size, index_next;
 303 +       timelib_tzdb_index_entry *db_index;
 304 +       char **dirstack;
 305 +
 306 +       /* LIFO stack to hold directory entries to scan; each slot is a
 307 +        * directory name relative to the zoneinfo prefix. */
 308 +       dirstack_size = 32;
 309 +       dirstack = malloc(dirstack_size * sizeof *dirstack);
 310 +       dirstack_top = 1;
 311 +       dirstack[0] = strdup("");
 312 +
 313 +       /* Index array. */
 314 +       index_size = 64;
 315 +       db_index = malloc(index_size * sizeof *db_index);
 316 +       index_next = 0;
 317 +
 318 +       do {
 319 +               struct dirent **ents;
 320 +               char name[PATH_MAX], *top;
 321 +               int count;
 322 +
 323 +               /* Pop the top stack entry, and iterate through its contents. */
 324 +               top = dirstack[--dirstack_top];
 325 +               snprintf(name, sizeof name, ZONEINFO_PREFIX "/%s", top);
 326 +
 327 +               count = php_scandir(name, &ents, index_filter, php_alphasort);
 328 +
 329 +               while (count > 0) {
 330 +                       struct stat st;
 331 +                       const char *leaf = ents[count - 1]->d_name;
 332 +
 333 +                       snprintf(name, sizeof name, ZONEINFO_PREFIX "/%s/%s",
 334 +                                top, leaf);
 335 +
 336 +                       if (strlen(name) && stat(name, &st) == 0) {
 337 +                               /* Name, relative to the zoneinfo prefix. */
 338 +                               const char *root = top;
 339 +
 340 +                               if (root[0] == '/') root++;
 341 +
 342 +                               snprintf(name, sizeof name, "%s%s%s", root,
 343 +                                        *root ? "/": "", leaf);
 344 +
 345 +                               if (S_ISDIR(st.st_mode)) {
 346 +                                       if (dirstack_top == dirstack_size) {
 347 +                                               dirstack_size *= 2;
 348 +                                               dirstack = realloc(dirstack,
 349 +                                                                  dirstack_size * sizeof *dirstack);
 350 +                                       }
 351 +                                       dirstack[dirstack_top++] = strdup(name);
 352 +                               }
 353 +                               else {
 354 +                                       if (index_next == index_size) {
 355 +                                               index_size *= 2;
 356 +                                               db_index = realloc(db_index,
 357 +                                                                  index_size * sizeof *db_index);
 358 +                                       }
 359 +
 360 +                                       db_index[index_next++].id = strdup(name);
 361 +                               }
 362 +                       }
 363 +
 364 +                       free(ents[--count]);
 365 +               }
 366 +
 367 +               if (count != -1) free(ents);
 368 +               free(top);
 369 +       } while (dirstack_top);
 370 +
 371 +        qsort(db_index, index_next, sizeof *db_index, sysdbcmp);
 372 +
 373 +       db->index = db_index;
 374 +       db->index_size = index_next;
 375 +
 376 +       free(dirstack);
 377 +}
 378 +
 379 +#define FAKE_HEADER "1234\0??\1??"
 380 +#define FAKE_UTC_POS (7 - 4)
 381 +
 382 +/* Create a fake data segment for database 'sysdb'. */
 383 +static void fake_data_segment(timelib_tzdb *sysdb,
 384 +                              struct location_info **info)
 385 +{
 386 +        size_t n;
 387 +        char *data, *p;
 388 +
 389 +        data = malloc(3 * sysdb->index_size + 7);
 390 +
 391 +        p = mempcpy(data, FAKE_HEADER, sizeof(FAKE_HEADER) - 1);
 392 +
 393 +        for (n = 0; n < sysdb->index_size; n++) {
 394 +                const struct location_info *li;
 395 +                timelib_tzdb_index_entry *ent;
 396 +
 397 +                ent = (timelib_tzdb_index_entry *)&sysdb->index[n];
 398 +
 399 +                /* Lookup the timezone name in the hash table. */
 400 +                if (strcmp(ent->id, "UTC") == 0) {
 401 +                        ent->pos = FAKE_UTC_POS;
 402 +                        continue;
 403 +                }
 404 +
 405 +                li = find_zone_info(info, ent->id);
 406 +                if (li) {
 407 +                        /* If found, append the BC byte and the
 408 +                         * country code; set the position for this
 409 +                         * section of timezone data.  */
 410 +                        ent->pos = (p - data) - 4;
 411 +                        *p++ = '\1';
 412 +                        *p++ = li->code[0];
 413 +                        *p++ = li->code[1];
 414 +                }
 415 +                else {
 416 +                        /* If not found, the timezone data can
 417 +                         * point at the header. */
 418 +                        ent->pos = 0;
 419 +                }
 420 +        }
 421 +
 422 +        sysdb->data = (unsigned char *)data;
 423 +}
 424 +
 425 +/* Returns true if the passed-in stat structure describes a
 426 + * probably-valid timezone file. */
 427 +static int is_valid_tzfile(const struct stat *st)
 428 +{
 429 +       return S_ISREG(st->st_mode) && st->st_size > 20;
 430 +}
 431 +
 432 +/* Return the mmap()ed tzfile if found, else NULL.  On success, the
 433 + * length of the mapped data is placed in *length. */
 434 +static char *map_tzfile(const char *timezone, size_t *length)
 435 +{
 436 +       char fname[PATH_MAX];
 437 +       struct stat st;
 438 +       char *p;
 439 +       int fd;
 440 +
 441 +       if (timezone[0] == '\0' || strstr(timezone, "..") != NULL) {
 442 +               return NULL;
 443 +       }
 444 +
 445 +       snprintf(fname, sizeof fname, ZONEINFO_PREFIX "/%s", timezone);
 446 +
 447 +       fd = open(fname, O_RDONLY);
 448 +       if (fd == -1) {
 449 +               return NULL;
 450 +       } else if (fstat(fd, &st) != 0 || !is_valid_tzfile(&st)) {
 451 +               close(fd);
 452 +               return NULL;
 453 +       }
 454 +
 455 +       *length = st.st_size;
 456 +       p = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0);
 457 +       close(fd);
 458 +
 459 +       return p != MAP_FAILED ? p : NULL;
 460 +}
 461 +
 462 +#endif
 463 +
 464 +static int inmem_seek_to_tz_position(const unsigned char **tzf, char *timezone, const timelib_tzdb *tzdb)
 465  {
 466         int left = 0, right = tzdb->index_size - 1;
 467  #ifdef HAVE_SETLOCALE
 468 @@ -299,36 +709,128 @@ static int seek_to_tz_position(const uns
 469         return 0;
 470  }
 471
 472 +static int seek_to_tz_position(const unsigned char **tzf, char *timezone,
 473 +                              char **map, size_t *maplen,
 474 +                              const timelib_tzdb *tzdb)
 475 +{
 476 +#ifdef HAVE_SYSTEM_TZDATA
 477 +       if (tzdb == timezonedb_system) {
 478 +               char *orig;
 479 +
 480 +               orig = map_tzfile(timezone, maplen);
 481 +               if (orig == NULL) {
 482 +                       return 0;
 483 +               }
 484 +
 485 +               (*tzf) = (unsigned char *)orig ;
 486 +               *map = orig;
 487 +
 488 +                return 1;
 489 +       }
 490 +       else
 491 +#endif
 492 +       {
 493 +               return inmem_seek_to_tz_position(tzf, timezone, tzdb);
 494 +       }
 495 +}
 496 +
 497  const timelib_tzdb *timelib_builtin_db(void)
 498  {
 499 +#ifdef HAVE_SYSTEM_TZDATA
 500 +       if (timezonedb_system == NULL) {
 501 +               timelib_tzdb *tmp = malloc(sizeof *tmp);
 502 +
 503 +               tmp->version = "0.system";
 504 +               tmp->data = NULL;
 505 +               create_zone_index(tmp);
 506 +               system_location_table = create_location_table();
 507 +                fake_data_segment(tmp, system_location_table);
 508 +               timezonedb_system = tmp;
 509 +       }
 510 +
 511 +
 512 +       return timezonedb_system;
 513 +#else
 514         return &timezonedb_builtin;
 515 +#endif
 516  }
 517
 518  const timelib_tzdb_index_entry *timelib_timezone_builtin_identifiers_list(int *count)
 519  {
 520 +#ifdef HAVE_SYSTEM_TZDATA
 521 +       *count = timezonedb_system->index_size;
 522 +       return timezonedb_system->index;
 523 +#else
 524         *count = sizeof(timezonedb_idx_builtin) / sizeof(*timezonedb_idx_builtin);
 525         return timezonedb_idx_builtin;
 526 +#endif
 527  }
 528
 529  int timelib_timezone_id_is_valid(char *timezone, const timelib_tzdb *tzdb)
 530  {
 531         const unsigned char *tzf;
 532 -       return (seek_to_tz_position(&tzf, timezone, tzdb));
 533 +
 534 +#ifdef HAVE_SYSTEM_TZDATA
 535 +        if (tzdb == timezonedb_system) {
 536 +            char fname[PATH_MAX];
 537 +            struct stat st;
 538 +
 539 +            if (timezone[0] == '\0' || strstr(timezone, "..") != NULL) {
 540 +               return 0;
 541 +            }
 542 +
 543 +            snprintf(fname, sizeof fname, ZONEINFO_PREFIX "/%s", timezone);
 544 +
 545 +            return stat(fname, &st) == 0 && is_valid_tzfile(&st);
 546 +        }
 547 +#endif
 548 +
 549 +       return (inmem_seek_to_tz_position(&tzf, timezone, tzdb));
 550  }
 551
 552  timelib_tzinfo *timelib_parse_tzfile(char *timezone, const timelib_tzdb *tzdb)
 553  {
 554         const unsigned char *tzf;
 555 +       char *memmap = NULL;
 556 +       size_t maplen;
 557         timelib_tzinfo *tmp;
 558
 559 -       if (seek_to_tz_position(&tzf, timezone, tzdb)) {
 560 +       if (seek_to_tz_position(&tzf, timezone, &memmap, &maplen, tzdb)) {
 561                 tmp = timelib_tzinfo_ctor(timezone);
 562
 563                 read_preamble(&tzf, tmp);
 564                 read_header(&tzf, tmp);
 565                 read_transistions(&tzf, tmp);
 566                 read_types(&tzf, tmp);
 567 -               read_location(&tzf, tmp);
 568 +
 569 +#ifdef HAVE_SYSTEM_TZDATA
 570 +               if (memmap) {
 571 +                       const struct location_info *li;
 572 +
 573 +                       /* TZif-style - grok the location info from the system database,
 574 +                        * if possible. */
 575 +
 576 +                       if ((li = find_zone_info(system_location_table, timezone)) != NULL) {
 577 +                               tmp->location.comments = strdup(li->comment);
 578 +                                strncpy(tmp->location.country_code, li->code, 2);
 579 +                               tmp->location.longitude = li->longitude;
 580 +                               tmp->location.latitude = li->latitude;
 581 +                               tmp->bc = 1;
 582 +                       }
 583 +                       else {
 584 +                               strcpy(tmp->location.country_code, "??");
 585 +                               tmp->bc = 0;
 586 +                               tmp->location.comments = strdup("");
 587 +                       }
 588 +
 589 +                       /* Now done with the mmap segment - discard it. */
 590 +                       munmap(memmap, maplen);
 591 +               } else
 592 +#endif
 593 +               {
 594 +                       /* PHP-style - use the embedded info. */
 595 +                       read_location(&tzf, tmp);
 596 +               }
 597         } else {
 598                 tmp = NULL;
 599         }
 600 --- a/ext/date/lib/timelib.m4
 601 +++ b/ext/date/lib/timelib.m4
 602 @@ -78,3 +78,17 @@ stdlib.h
 603
 604  dnl Check for strtoll, atoll
 605  AC_CHECK_FUNCS(strtoll atoll strftime)
 606 +
 607 +PHP_ARG_WITH(system-tzdata, for use of system timezone data,
 608 +[  --with-system-tzdata[=DIR]      to specify use of system timezone data],
 609 +no, no)
 610 +
 611 +if test "$PHP_SYSTEM_TZDATA" != "no"; then
 612 +   AC_DEFINE(HAVE_SYSTEM_TZDATA, 1, [Define if system timezone data is used])
 613 +
 614 +   if test "$PHP_SYSTEM_TZDATA" != "yes"; then
 615 +      AC_DEFINE_UNQUOTED(HAVE_SYSTEM_TZDATA_PREFIX, "$PHP_SYSTEM_TZDATA",
 616 +                         [Define for location of system timezone data])
 617 +   fi
 618 +fi
 619 +