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 r10 : make timezone case insensitive
   6 r9: fix another compile error without --with-system-tzdata configured (Michael Heimpold)
   7 r8: fix compile error without --with-system-tzdata configured
   8 r7: improve check for valid timezone id to exclude directories
   9 r6: fix fd leak in r5, fix country code/BC flag use in
  10     timezone_identifiers_list() using system db,
  11     fix use of PECL timezonedb to override system db,
  12 r5: reverts addition of "System/Localtime" fake tzname.
  13     updated for 5.3.0, parses zone.tab to pick up mapping between
  14     timezone name, country code and long/lat coords
  15 r4: added "System/Localtime" tzname which uses /etc/localtime
  16 r3: fix a crash if /usr/share/zoneinfo doesn't exist (Raphael Geissert)
  17 r2: add filesystem trawl to set up name alias index
  18 r1: initial revision
  19
  20 --- php5.orig/ext/date/lib/parse_tz.c
  21 +++ php5/ext/date/lib/parse_tz.c
  22 @@ -24,6 +24,16 @@
  23
  24  #include "timelib.h"
  25
  26 +#ifdef HAVE_SYSTEM_TZDATA
  27 +#include <sys/mman.h>
  28 +#include <sys/stat.h>
  29 +#include <limits.h>
  30 +#include <fcntl.h>
  31 +#include <unistd.h>
  32 +
  33 +#include "php_scandir.h"
  34 +#endif
  35 +
  36  #include <stdio.h>
  37
  38  #ifdef HAVE_LOCALE_H
  39 @@ -35,7 +45,12 @@
  40  #else
  41  #include <strings.h>
  42  #endif
  43 +
  44 +#ifndef HAVE_SYSTEM_TZDATA
  45  #include "timezonedb.h"
  46 +#endif
  47 +
  48 +#include <ctype.h>
  49
  50  #if (defined(__APPLE__) || defined(__APPLE_CC__)) && (defined(__BIG_ENDIAN__) || defined(__LITTLE_ENDIAN__))
  51  # if defined(__LITTLE_ENDIAN__)
  52 @@ -55,9 +70,14 @@
  53
  54  static void read_preamble(const unsigned char **tzf, timelib_tzinfo *tz)
  55  {
  56 -       /* skip ID */
  57 -       *tzf += 4;
  58 -
  59 +        if (memcmp(tzf, "TZif", 4) == 0) {
  60 +                *tzf += 20;
  61 +                return;
  62 +        }
  63 +
  64 +        /* skip ID */
  65 +        *tzf += 4;
  66 +
  67         /* read BC flag */
  68         tz->bc = (**tzf == '\1');
  69         *tzf += 1;
  70 @@ -260,7 +280,405 @@ void timelib_dump_tzinfo(timelib_tzinfo
  71         }
  72  }
  73
  74 -static int seek_to_tz_position(const unsigned char **tzf, char *timezone, const timelib_tzdb *tzdb)
  75 +#ifdef HAVE_SYSTEM_TZDATA
  76 +
  77 +#ifdef HAVE_SYSTEM_TZDATA_PREFIX
  78 +#define ZONEINFO_PREFIX HAVE_SYSTEM_TZDATA_PREFIX
  79 +#else
  80 +#define ZONEINFO_PREFIX "/usr/share/zoneinfo"
  81 +#endif
  82 +
  83 +/* System timezone database pointer. */
  84 +static const timelib_tzdb *timezonedb_system;
  85 +
  86 +/* Hash table entry for the cache of the zone.tab mapping table. */
  87 +struct location_info {
  88 +        char code[2];
  89 +        double latitude, longitude;
  90 +        char name[64];
  91 +        char *comment;
  92 +        struct location_info *next;
  93 +};
  94 +
  95 +/* Cache of zone.tab. */
  96 +static struct location_info **system_location_table;
  97 +
  98 +/* Size of the zone.tab hash table; a random-ish prime big enough to
  99 + * prevent too many collisions. */
 100 +#define LOCINFO_HASH_SIZE (1021)
 101 +
 102 +/* Compute a case insensitive hash of str */
 103 +static uint32_t tz_hash(const char *str)
 104 +{
 105 +    const unsigned char *p = (const unsigned char *)str;
 106 +    uint32_t hash = 5381;
 107 +    int c;
 108 +
 109 +    while ((c = tolower(*p++)) != '\0') {
 110 +        hash = (hash << 5) ^ hash ^ c;
 111 +    }
 112 +
 113 +    return hash % LOCINFO_HASH_SIZE;
 114 +}
 115 +
 116 +/* Parse an ISO-6709 date as used in zone.tab. Returns end of the
 117 + * parsed string on success, or NULL on parse error.  On success,
 118 + * writes the parsed number to *result. */
 119 +static char *parse_iso6709(char *p, double *result)
 120 +{
 121 +    double v, sign;
 122 +    char *pend;
 123 +    size_t len;
 124 +
 125 +    if (*p == '+')
 126 +        sign = 1.0;
 127 +    else if (*p == '-')
 128 +        sign = -1.0;
 129 +    else
 130 +        return NULL;
 131 +
 132 +    p++;
 133 +    for (pend = p; *pend >= '0' && *pend <= '9'; pend++)
 134 +        ;;
 135 +
 136 +    /* Annoying encoding used by zone.tab has no decimal point, so use
 137 +     * the length to determine the format:
 138 +     *
 139 +     * 4 = DDMM
 140 +     * 5 = DDDMM
 141 +     * 6 = DDMMSS
 142 +     * 7 = DDDMMSS
 143 +     */
 144 +    len = pend - p;
 145 +    if (len < 4 || len > 7) {
 146 +        return NULL;
 147 +    }
 148 +
 149 +    /* p => [D]DD */
 150 +    v = (p[0] - '0') * 10.0 + (p[1] - '0');
 151 +    p += 2;
 152 +    if (len == 5 || len == 7)
 153 +        v = v * 10.0 + (*p++ - '0');
 154 +    /* p => MM[SS] */
 155 +    v += (10.0 * (p[0] - '0')
 156 +          + p[1] - '0') / 60.0;
 157 +    p += 2;
 158 +    /* p => [SS] */
 159 +    if (len > 5) {
 160 +        v += (10.0 * (p[0] - '0')
 161 +              + p[1] - '0') / 3600.0;
 162 +        p += 2;
 163 +    }
 164 +
 165 +    /* Round to five decimal place, not because it's a good idea,
 166 +     * but, because the builtin data uses rounded data, so, match
 167 +     * that. */
 168 +    *result = round(v * sign * 100000.0) / 100000.0;
 169 +
 170 +    return p;
 171 +}
 172 +
 173 +/* This function parses the zone.tab file to build up the mapping of
 174 + * timezone to country code and geographic location, and returns a
 175 + * hash table.  The hash table is indexed by the function:
 176 + *
 177 + *   tz_hash(timezone-name)
 178 + */
 179 +static struct location_info **create_location_table(void)
 180 +{
 181 +    struct location_info **li, *i;
 182 +    char zone_tab[PATH_MAX];
 183 +    char line[512];
 184 +    FILE *fp;
 185 +
 186 +    strncpy(zone_tab, ZONEINFO_PREFIX "/zone.tab", sizeof zone_tab);
 187 +
 188 +    fp = fopen(zone_tab, "r");
 189 +    if (!fp) {
 190 +        return NULL;
 191 +    }
 192 +
 193 +    li = calloc(LOCINFO_HASH_SIZE, sizeof *li);
 194 +
 195 +    while (fgets(line, sizeof line, fp)) {
 196 +        char *p = line, *code, *name, *comment;
 197 +        uint32_t hash;
 198 +        double latitude, longitude;
 199 +
 200 +        while (isspace(*p))
 201 +            p++;
 202 +
 203 +        if (*p == '#' || *p == '\0' || *p == '\n')
 204 +            continue;
 205 +
 206 +        if (!isalpha(p[0]) || !isalpha(p[1]) || p[2] != '\t')
 207 +            continue;
 208 +
 209 +        /* code => AA */
 210 +        code = p;
 211 +        p[2] = 0;
 212 +        p += 3;
 213 +
 214 +        /* coords => [+-][D]DDMM[SS][+-][D]DDMM[SS] */
 215 +        p = parse_iso6709(p, &latitude);
 216 +        if (!p) {
 217 +            continue;
 218 +        }
 219 +        p = parse_iso6709(p, &longitude);
 220 +        if (!p) {
 221 +            continue;
 222 +        }
 223 +
 224 +        if (!p || *p != '\t') {
 225 +            continue;
 226 +        }
 227 +
 228 +        /* name = string */
 229 +        name = ++p;
 230 +        while (*p != '\t' && *p && *p != '\n')
 231 +            p++;
 232 +
 233 +        *p++ = '\0';
 234 +
 235 +        /* comment = string */
 236 +        comment = p;
 237 +        while (*p != '\t' && *p && *p != '\n')
 238 +            p++;
 239 +
 240 +        if (*p == '\n' || *p == '\t')
 241 +            *p = '\0';
 242 +
 243 +        hash = tz_hash(name);
 244 +        i = malloc(sizeof *i);
 245 +        memcpy(i->code, code, 2);
 246 +        strncpy(i->name, name, sizeof i->name);
 247 +        i->comment = strdup(comment);
 248 +        i->longitude = longitude;
 249 +        i->latitude = latitude;
 250 +        i->next = li[hash];
 251 +        li[hash] = i;
 252 +        /* printf("%s [%u, %f, %f]\n", name, hash, latitude, longitude); */
 253 +    }
 254 +
 255 +    fclose(fp);
 256 +
 257 +    return li;
 258 +}
 259 +
 260 +/* Return location info from hash table, using given timezone name.
 261 + * Returns NULL if the name could not be found. */
 262 +const struct location_info *find_zone_info(struct location_info **li,
 263 +                                           const char *name)
 264 +{
 265 +    uint32_t hash = tz_hash(name);
 266 +    const struct location_info *l;
 267 +
 268 +    if (!li) {
 269 +        return NULL;
 270 +    }
 271 +
 272 +    for (l = li[hash]; l; l = l->next) {
 273 +        if (strcasecmp(l->name, name) == 0)
 274 +            return l;
 275 +    }
 276 +
 277 +    return NULL;
 278 +}
 279 +
 280 +/* Filter out some non-tzdata files and the posix/right databases, if
 281 + * present. */
 282 +static int index_filter(const struct dirent *ent)
 283 +{
 284 +       return strcmp(ent->d_name, ".") != 0
 285 +               && strcmp(ent->d_name, "..") != 0
 286 +               && strcmp(ent->d_name, "posix") != 0
 287 +               && strcmp(ent->d_name, "posixrules") != 0
 288 +               && strcmp(ent->d_name, "right") != 0
 289 +               && strstr(ent->d_name, ".tab") == NULL;
 290 +}
 291 +
 292 +static int sysdbcmp(const void *first, const void *second)
 293 +{
 294 +        const timelib_tzdb_index_entry *alpha = first, *beta = second;
 295 +
 296 +        return strcmp(alpha->id, beta->id);
 297 +}
 298 +
 299 +
 300 +/* Create the zone identifier index by trawling the filesystem. */
 301 +static void create_zone_index(timelib_tzdb *db)
 302 +{
 303 +       size_t dirstack_size,  dirstack_top;
 304 +       size_t index_size, index_next;
 305 +       timelib_tzdb_index_entry *db_index;
 306 +       char **dirstack;
 307 +
 308 +       /* LIFO stack to hold directory entries to scan; each slot is a
 309 +        * directory name relative to the zoneinfo prefix. */
 310 +       dirstack_size = 32;
 311 +       dirstack = malloc(dirstack_size * sizeof *dirstack);
 312 +       dirstack_top = 1;
 313 +       dirstack[0] = strdup("");
 314 +
 315 +       /* Index array. */
 316 +       index_size = 64;
 317 +       db_index = malloc(index_size * sizeof *db_index);
 318 +       index_next = 0;
 319 +
 320 +       do {
 321 +               struct dirent **ents;
 322 +               char name[PATH_MAX], *top;
 323 +               int count;
 324 +
 325 +               /* Pop the top stack entry, and iterate through its contents. */
 326 +               top = dirstack[--dirstack_top];
 327 +               snprintf(name, sizeof name, ZONEINFO_PREFIX "/%s", top);
 328 +
 329 +               count = php_scandir(name, &ents, index_filter, php_alphasort);
 330 +
 331 +               while (count > 0) {
 332 +                       struct stat st;
 333 +                       const char *leaf = ents[count - 1]->d_name;
 334 +
 335 +                       snprintf(name, sizeof name, ZONEINFO_PREFIX "/%s/%s",
 336 +                                top, leaf);
 337 +
 338 +                       if (strlen(name) && stat(name, &st) == 0) {
 339 +                               /* Name, relative to the zoneinfo prefix. */
 340 +                               const char *root = top;
 341 +
 342 +                               if (root[0] == '/') root++;
 343 +
 344 +                               snprintf(name, sizeof name, "%s%s%s", root,
 345 +                                        *root ? "/": "", leaf);
 346 +
 347 +                               if (S_ISDIR(st.st_mode)) {
 348 +                                       if (dirstack_top == dirstack_size) {
 349 +                                               dirstack_size *= 2;
 350 +                                               dirstack = realloc(dirstack,
 351 +                                                                  dirstack_size * sizeof *dirstack);
 352 +                                       }
 353 +                                       dirstack[dirstack_top++] = strdup(name);
 354 +                               }
 355 +                               else {
 356 +                                       if (index_next == index_size) {
 357 +                                               index_size *= 2;
 358 +                                               db_index = realloc(db_index,
 359 +                                                                  index_size * sizeof *db_index);
 360 +                                       }
 361 +
 362 +                                       db_index[index_next++].id = strdup(name);
 363 +                               }
 364 +                       }
 365 +
 366 +                       free(ents[--count]);
 367 +               }
 368 +
 369 +               if (count != -1) free(ents);
 370 +               free(top);
 371 +       } while (dirstack_top);
 372 +
 373 +        qsort(db_index, index_next, sizeof *db_index, sysdbcmp);
 374 +
 375 +       db->index = db_index;
 376 +       db->index_size = index_next;
 377 +
 378 +       free(dirstack);
 379 +}
 380 +
 381 +#define FAKE_HEADER "1234\0??\1??"
 382 +#define FAKE_UTC_POS (7 - 4)
 383 +
 384 +/* Create a fake data segment for database 'sysdb'. */
 385 +static void fake_data_segment(timelib_tzdb *sysdb,
 386 +                              struct location_info **info)
 387 +{
 388 +        size_t n;
 389 +        char *data, *p;
 390 +
 391 +        data = malloc(3 * sysdb->index_size + 7);
 392 +
 393 +        p = mempcpy(data, FAKE_HEADER, sizeof(FAKE_HEADER) - 1);
 394 +
 395 +        for (n = 0; n < sysdb->index_size; n++) {
 396 +                const struct location_info *li;
 397 +                timelib_tzdb_index_entry *ent;
 398 +
 399 +                ent = (timelib_tzdb_index_entry *)&sysdb->index[n];
 400 +
 401 +                /* Lookup the timezone name in the hash table. */
 402 +                if (strcmp(ent->id, "UTC") == 0) {
 403 +                        ent->pos = FAKE_UTC_POS;
 404 +                        continue;
 405 +                }
 406 +
 407 +                li = find_zone_info(info, ent->id);
 408 +                if (li) {
 409 +                        /* If found, append the BC byte and the
 410 +                         * country code; set the position for this
 411 +                         * section of timezone data.  */
 412 +                        ent->pos = (p - data) - 4;
 413 +                        *p++ = '\1';
 414 +                        *p++ = li->code[0];
 415 +                        *p++ = li->code[1];
 416 +                }
 417 +                else {
 418 +                        /* If not found, the timezone data can
 419 +                         * point at the header. */
 420 +                        ent->pos = 0;
 421 +                }
 422 +        }
 423 +
 424 +        sysdb->data = (unsigned char *)data;
 425 +}
 426 +
 427 +/* Returns true if the passed-in stat structure describes a
 428 + * probably-valid timezone file. */
 429 +static int is_valid_tzfile(const struct stat *st)
 430 +{
 431 +       return S_ISREG(st->st_mode) && st->st_size > 20;
 432 +}
 433 +
 434 +/* Return the mmap()ed tzfile if found, else NULL.  On success, the
 435 + * length of the mapped data is placed in *length. */
 436 +static char *map_tzfile(const char *timezone, size_t *length)
 437 +{
 438 +       char fname[PATH_MAX];
 439 +       struct stat st;
 440 +       char *p;
 441 +       int fd;
 442 +
 443 +       if (timezone[0] == '\0' || strstr(timezone, "..") != NULL) {
 444 +               return NULL;
 445 +       }
 446 +
 447 +    if (system_location_table) {
 448 +        const struct location_info *li;
 449 +        if ((li = find_zone_info(system_location_table, timezone)) != NULL) {
 450 +            /* Use the stored name to avoid case issue */
 451 +            timezone = li->name;
 452 +        }
 453 +    }
 454 +       snprintf(fname, sizeof fname, ZONEINFO_PREFIX "/%s", timezone);
 455 +
 456 +       fd = open(fname, O_RDONLY);
 457 +       if (fd == -1) {
 458 +               return NULL;
 459 +       } else if (fstat(fd, &st) != 0 || !is_valid_tzfile(&st)) {
 460 +               close(fd);
 461 +               return NULL;
 462 +       }
 463 +
 464 +       *length = st.st_size;
 465 +       p = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0);
 466 +       close(fd);
 467 +
 468 +       return p != MAP_FAILED ? p : NULL;
 469 +}
 470 +
 471 +#endif
 472 +
 473 +static int inmem_seek_to_tz_position(const unsigned char **tzf, char *timezone, const timelib_tzdb *tzdb)
 474  {
 475         int left = 0, right = tzdb->index_size - 1;
 476  #ifdef HAVE_SETLOCALE
 477 @@ -299,36 +717,135 @@ static int seek_to_tz_position(const uns
 478         return 0;
 479  }
 480
 481 +static int seek_to_tz_position(const unsigned char **tzf, char *timezone,
 482 +                              char **map, size_t *maplen,
 483 +                              const timelib_tzdb *tzdb)
 484 +{
 485 +#ifdef HAVE_SYSTEM_TZDATA
 486 +       if (tzdb == timezonedb_system) {
 487 +               char *orig;
 488 +
 489 +               orig = map_tzfile(timezone, maplen);
 490 +               if (orig == NULL) {
 491 +                       return 0;
 492 +               }
 493 +
 494 +               (*tzf) = (unsigned char *)orig ;
 495 +               *map = orig;
 496 +
 497 +                return 1;
 498 +       }
 499 +       else
 500 +#endif
 501 +       {
 502 +               return inmem_seek_to_tz_position(tzf, timezone, tzdb);
 503 +       }
 504 +}
 505 +
 506  const timelib_tzdb *timelib_builtin_db(void)
 507  {
 508 +#ifdef HAVE_SYSTEM_TZDATA
 509 +       if (timezonedb_system == NULL) {
 510 +               timelib_tzdb *tmp = malloc(sizeof *tmp);
 511 +
 512 +               tmp->version = "0.system";
 513 +               tmp->data = NULL;
 514 +               create_zone_index(tmp);
 515 +               system_location_table = create_location_table();
 516 +                fake_data_segment(tmp, system_location_table);
 517 +               timezonedb_system = tmp;
 518 +       }
 519 +
 520 +
 521 +       return timezonedb_system;
 522 +#else
 523         return &timezonedb_builtin;
 524 +#endif
 525  }
 526
 527  const timelib_tzdb_index_entry *timelib_timezone_builtin_identifiers_list(int *count)
 528  {
 529 +#ifdef HAVE_SYSTEM_TZDATA
 530 +       *count = timezonedb_system->index_size;
 531 +       return timezonedb_system->index;
 532 +#else
 533         *count = sizeof(timezonedb_idx_builtin) / sizeof(*timezonedb_idx_builtin);
 534         return timezonedb_idx_builtin;
 535 +#endif
 536  }
 537
 538  int timelib_timezone_id_is_valid(char *timezone, const timelib_tzdb *tzdb)
 539  {
 540         const unsigned char *tzf;
 541 -       return (seek_to_tz_position(&tzf, timezone, tzdb));
 542 +
 543 +#ifdef HAVE_SYSTEM_TZDATA
 544 +        if (tzdb == timezonedb_system) {
 545 +            char fname[PATH_MAX];
 546 +            struct stat st;
 547 +
 548 +            if (timezone[0] == '\0' || strstr(timezone, "..") != NULL) {
 549 +                       return 0;
 550 +            }
 551 +
 552 +            if (system_location_table) {
 553 +                if (find_zone_info(system_location_table, timezone) != NULL) {
 554 +                    /* found in cache */
 555 +                    return 1;
 556 +                }
 557 +            }
 558 +
 559 +            snprintf(fname, sizeof fname, ZONEINFO_PREFIX "/%s", timezone);
 560 +
 561 +            return stat(fname, &st) == 0 && is_valid_tzfile(&st);
 562 +        }
 563 +#endif
 564 +
 565 +       return (inmem_seek_to_tz_position(&tzf, timezone, tzdb));
 566  }
 567
 568  timelib_tzinfo *timelib_parse_tzfile(char *timezone, const timelib_tzdb *tzdb)
 569  {
 570         const unsigned char *tzf;
 571 +       char *memmap = NULL;
 572 +       size_t maplen;
 573         timelib_tzinfo *tmp;
 574
 575 -       if (seek_to_tz_position(&tzf, timezone, tzdb)) {
 576 +       if (seek_to_tz_position(&tzf, timezone, &memmap, &maplen, tzdb)) {
 577                 tmp = timelib_tzinfo_ctor(timezone);
 578
 579                 read_preamble(&tzf, tmp);
 580                 read_header(&tzf, tmp);
 581                 read_transistions(&tzf, tmp);
 582                 read_types(&tzf, tmp);
 583 -               read_location(&tzf, tmp);
 584 +
 585 +#ifdef HAVE_SYSTEM_TZDATA
 586 +               if (memmap) {
 587 +                       const struct location_info *li;
 588 +
 589 +                       /* TZif-style - grok the location info from the system database,
 590 +                        * if possible. */
 591 +
 592 +                       if ((li = find_zone_info(system_location_table, timezone)) != NULL) {
 593 +                               tmp->location.comments = strdup(li->comment);
 594 +                                strncpy(tmp->location.country_code, li->code, 2);
 595 +                               tmp->location.longitude = li->longitude;
 596 +                               tmp->location.latitude = li->latitude;
 597 +                               tmp->bc = 1;
 598 +                       }
 599 +                       else {
 600 +                               strcpy(tmp->location.country_code, "??");
 601 +                               tmp->bc = 0;
 602 +                               tmp->location.comments = strdup("");
 603 +                       }
 604 +
 605 +                       /* Now done with the mmap segment - discard it. */
 606 +                       munmap(memmap, maplen);
 607 +               } else
 608 +#endif
 609 +               {
 610 +                       /* PHP-style - use the embedded info. */
 611 +                       read_location(&tzf, tmp);
 612 +               }
 613         } else {
 614                 tmp = NULL;
 615         }
 616 --- php5.orig/ext/date/lib/timelib.m4
 617 +++ php5/ext/date/lib/timelib.m4
 618 @@ -78,3 +78,17 @@ stdlib.h
 619
 620  dnl Check for strtoll, atoll
 621  AC_CHECK_FUNCS(strtoll atoll strftime)
 622 +
 623 +PHP_ARG_WITH(system-tzdata, for use of system timezone data,
 624 +[  --with-system-tzdata[=DIR]      to specify use of system timezone data],
 625 +no, no)
 626 +
 627 +if test "$PHP_SYSTEM_TZDATA" != "no"; then
 628 +   AC_DEFINE(HAVE_SYSTEM_TZDATA, 1, [Define if system timezone data is used])
 629 +
 630 +   if test "$PHP_SYSTEM_TZDATA" != "yes"; then
 631 +      AC_DEFINE_UNQUOTED(HAVE_SYSTEM_TZDATA_PREFIX, "$PHP_SYSTEM_TZDATA",
 632 +                         [Define for location of system timezone data])
 633 +   fi
 634 +fi
 635 +