php-systzdata.patch

   1
   2 Add support for use of the system timezone database, rather
   3 than embedding a copy.  Discussed upstream but was not desired.
   4
   5 History:
   6 r7: improve check for valid timezone id to exclude directories
   7 r6: fix fd leak in r5, fix country code/BC flag use in
   8     timezone_identifiers_list() using system db,
   9     fix use of PECL timezonedb to override system db,
  10 r5: reverts addition of "System/Localtime" fake tzname.
  11     updated for 5.3.0, parses zone.tab to pick up mapping between
  12     timezone name, country code and long/lat coords
  13 r4: added "System/Localtime" tzname which uses /etc/localtime
  14 r3: fix a crash if /usr/share/zoneinfo doesn't exist (Raphael Geissert)
  15 r2: add filesystem trawl to set up name alias index
  16 r1: initial revision
  17
  18 --- php-5.3.1/ext/date/lib/parse_tz.c.systzdata
  19 +++ php-5.3.1/ext/date/lib/parse_tz.c
  20 @@ -20,6 +20,16 @@
  21
  22  #include "timelib.h"
  23
  24 +#ifdef HAVE_SYSTEM_TZDATA
  25 +#include <sys/mman.h>
  26 +#include <sys/stat.h>
  27 +#include <limits.h>
  28 +#include <fcntl.h>
  29 +#include <unistd.h>
  30 +
  31 +#include "php_scandir.h"
  32 +#endif
  33 +
  34  #include <stdio.h>
  35
  36  #ifdef HAVE_LOCALE_H
  37 @@ -31,7 +41,12 @@
  38  #else
  39  #include <strings.h>
  40  #endif
  41 +
  42 +#ifndef HAVE_SYSTEM_TZDATA
  43  #include "timezonedb.h"
  44 +#endif
  45 +
  46 +#include <ctype.h>
  47
  48  #if (defined(__APPLE__) || defined(__APPLE_CC__)) && (defined(__BIG_ENDIAN__) || defined(__LITTLE_ENDIAN__))
  49  # if defined(__LITTLE_ENDIAN__)
  50 @@ -51,9 +66,14 @@
  51
  52  static void read_preamble(const unsigned char **tzf, timelib_tzinfo *tz)
  53  {
  54 -       /* skip ID */
  55 -       *tzf += 4;
  56 -
  57 +        if (memcmp(tzf, "TZif", 4) == 0) {
  58 +                *tzf += 20;
  59 +                return;
  60 +        }
  61 +
  62 +        /* skip ID */
  63 +        *tzf += 4;
  64 +
  65         /* read BC flag */
  66         tz->bc = (**tzf == '\1');
  67         *tzf += 1;
  68 @@ -253,7 +273,397 @@ void timelib_dump_tzinfo(timelib_tzinfo
  69         }
  70  }
  71
  72 -static int seek_to_tz_position(const unsigned char **tzf, char *timezone, const timelib_tzdb *tzdb)
  73 +#ifdef HAVE_SYSTEM_TZDATA
  74 +
  75 +#ifdef HAVE_SYSTEM_TZDATA_PREFIX
  76 +#define ZONEINFO_PREFIX HAVE_SYSTEM_TZDATA_PREFIX
  77 +#else
  78 +#define ZONEINFO_PREFIX "/usr/share/zoneinfo"
  79 +#endif
  80 +
  81 +/* System timezone database pointer. */
  82 +static const timelib_tzdb *timezonedb_system = NULL;
  83 +
  84 +/* Hash table entry for the cache of the zone.tab mapping table. */
  85 +struct location_info {
  86 +        char code[2];
  87 +        double latitude, longitude;
  88 +        char name[64];
  89 +        char *comment;
  90 +        struct location_info *next;
  91 +};
  92 +
  93 +/* Cache of zone.tab. */
  94 +static struct location_info **system_location_table;
  95 +
  96 +/* Size of the zone.tab hash table; a random-ish prime big enough to
  97 + * prevent too many collisions. */
  98 +#define LOCINFO_HASH_SIZE (1021)
  99 +
 100 +static uint32_t tz_hash(const char *str)
 101 +{
 102 +    const unsigned char *p = (const unsigned char *)str;
 103 +    uint32_t hash = 5381;
 104 +    int c;
 105 +
 106 +    while ((c = *p++) != '\0') {
 107 +        hash = (hash << 5) ^ hash ^ c;
 108 +    }
 109 +
 110 +    return hash % LOCINFO_HASH_SIZE;
 111 +}
 112 +
 113 +/* Parse an ISO-6709 date as used in zone.tab. Returns end of the
 114 + * parsed string on success, or NULL on parse error.  On success,
 115 + * writes the parsed number to *result. */
 116 +static char *parse_iso6709(char *p, double *result)
 117 +{
 118 +    double v, sign;
 119 +    char *pend;
 120 +    size_t len;
 121 +
 122 +    if (*p == '+')
 123 +        sign = 1.0;
 124 +    else if (*p == '-')
 125 +        sign = -1.0;
 126 +    else
 127 +        return NULL;
 128 +
 129 +    p++;
 130 +    for (pend = p; *pend >= '0' && *pend <= '9'; pend++)
 131 +        ;;
 132 +
 133 +    /* Annoying encoding used by zone.tab has no decimal point, so use
 134 +     * the length to determine the format:
 135 +     *
 136 +     * 4 = DDMM
 137 +     * 5 = DDDMM
 138 +     * 6 = DDMMSS
 139 +     * 7 = DDDMMSS
 140 +     */
 141 +    len = pend - p;
 142 +    if (len < 4 || len > 7) {
 143 +        return NULL;
 144 +    }
 145 +
 146 +    /* p => [D]DD */
 147 +    v = (p[0] - '0') * 10.0 + (p[1] - '0');
 148 +    p += 2;
 149 +    if (len == 5 || len == 7)
 150 +        v = v * 10.0 + (*p++ - '0');
 151 +    /* p => MM[SS] */
 152 +    v += (10.0 * (p[0] - '0')
 153 +          + p[1] - '0') / 60.0;
 154 +    p += 2;
 155 +    /* p => [SS] */
 156 +    if (len > 5) {
 157 +        v += (10.0 * (p[0] - '0')
 158 +              + p[1] - '0') / 3600.0;
 159 +        p += 2;
 160 +    }
 161 +
 162 +    /* Round to five decimal place, not because it's a good idea,
 163 +     * but, because the builtin data uses rounded data, so, match
 164 +     * that. */
 165 +    *result = round(v * sign * 100000.0) / 100000.0;
 166 +
 167 +    return p;
 168 +}
 169 +
 170 +/* This function parses the zone.tab file to build up the mapping of
 171 + * timezone to country code and geographic location, and returns a
 172 + * hash table.  The hash table is indexed by the function:
 173 + *
 174 + *   tz_hash(timezone-name)
 175 + */
 176 +static struct location_info **create_location_table(void)
 177 +{
 178 +    struct location_info **li, *i;
 179 +    char zone_tab[PATH_MAX];
 180 +    char line[512];
 181 +    FILE *fp;
 182 +
 183 +    strncpy(zone_tab, ZONEINFO_PREFIX "/zone.tab", sizeof zone_tab);
 184 +
 185 +    fp = fopen(zone_tab, "r");
 186 +    if (!fp) {
 187 +        return NULL;
 188 +    }
 189 +
 190 +    li = calloc(LOCINFO_HASH_SIZE, sizeof *li);
 191 +
 192 +    while (fgets(line, sizeof line, fp)) {
 193 +        char *p = line, *code, *name, *comment;
 194 +        uint32_t hash;
 195 +        double latitude, longitude;
 196 +
 197 +        while (isspace(*p))
 198 +            p++;
 199 +
 200 +        if (*p == '#' || *p == '\0' || *p == '\n')
 201 +            continue;
 202 +
 203 +        if (!isalpha(p[0]) || !isalpha(p[1]) || p[2] != '\t')
 204 +            continue;
 205 +
 206 +        /* code => AA */
 207 +        code = p;
 208 +        p[2] = 0;
 209 +        p += 3;
 210 +
 211 +        /* coords => [+-][D]DDMM[SS][+-][D]DDMM[SS] */
 212 +        p = parse_iso6709(p, &latitude);
 213 +        if (!p) {
 214 +            continue;
 215 +        }
 216 +        p = parse_iso6709(p, &longitude);
 217 +        if (!p) {
 218 +            continue;
 219 +        }
 220 +
 221 +        if (!p || *p != '\t') {
 222 +            continue;
 223 +        }
 224 +
 225 +        /* name = string */
 226 +        name = ++p;
 227 +        while (*p != '\t' && *p && *p != '\n')
 228 +            p++;
 229 +
 230 +        *p++ = '\0';
 231 +
 232 +        /* comment = string */
 233 +        comment = p;
 234 +        while (*p != '\t' && *p && *p != '\n')
 235 +            p++;
 236 +
 237 +        if (*p == '\n' || *p == '\t')
 238 +            *p = '\0';
 239 +
 240 +        hash = tz_hash(name);
 241 +        i = malloc(sizeof *i);
 242 +        memcpy(i->code, code, 2);
 243 +        strncpy(i->name, name, sizeof i->name);
 244 +        i->comment = strdup(comment);
 245 +        i->longitude = longitude;
 246 +        i->latitude = latitude;
 247 +        i->next = li[hash];
 248 +        li[hash] = i;
 249 +        /* printf("%s [%u, %f, %f]\n", name, hash, latitude, longitude); */
 250 +    }
 251 +
 252 +    fclose(fp);
 253 +
 254 +    return li;
 255 +}
 256 +
 257 +/* Return location info from hash table, using given timezone name.
 258 + * Returns NULL if the name could not be found. */
 259 +const struct location_info *find_zone_info(struct location_info **li,
 260 +                                           const char *name)
 261 +{
 262 +    uint32_t hash = tz_hash(name);
 263 +    const struct location_info *l;
 264 +
 265 +    if (!li) {
 266 +        return NULL;
 267 +    }
 268 +
 269 +    for (l = li[hash]; l; l = l->next) {
 270 +        if (strcasecmp(l->name, name) == 0)
 271 +            return l;
 272 +    }
 273 +
 274 +    return NULL;
 275 +}
 276 +
 277 +/* Filter out some non-tzdata files and the posix/right databases, if
 278 + * present. */
 279 +static int index_filter(const struct dirent *ent)
 280 +{
 281 +       return strcmp(ent->d_name, ".") != 0
 282 +               && strcmp(ent->d_name, "..") != 0
 283 +               && strcmp(ent->d_name, "posix") != 0
 284 +               && strcmp(ent->d_name, "posixrules") != 0
 285 +               && strcmp(ent->d_name, "right") != 0
 286 +               && strstr(ent->d_name, ".tab") == NULL;
 287 +}
 288 +
 289 +static int sysdbcmp(const void *first, const void *second)
 290 +{
 291 +        const timelib_tzdb_index_entry *alpha = first, *beta = second;
 292 +
 293 +        return strcmp(alpha->id, beta->id);
 294 +}
 295 +
 296 +
 297 +/* Create the zone identifier index by trawling the filesystem. */
 298 +static void create_zone_index(timelib_tzdb *db)
 299 +{
 300 +       size_t dirstack_size,  dirstack_top;
 301 +       size_t index_size, index_next;
 302 +       timelib_tzdb_index_entry *db_index;
 303 +       char **dirstack;
 304 +
 305 +       /* LIFO stack to hold directory entries to scan; each slot is a
 306 +        * directory name relative to the zoneinfo prefix. */
 307 +       dirstack_size = 32;
 308 +       dirstack = malloc(dirstack_size * sizeof *dirstack);
 309 +       dirstack_top = 1;
 310 +       dirstack[0] = strdup("");
 311 +
 312 +       /* Index array. */
 313 +       index_size = 64;
 314 +       db_index = malloc(index_size * sizeof *db_index);
 315 +       index_next = 0;
 316 +
 317 +       do {
 318 +               struct dirent **ents;
 319 +               char name[PATH_MAX], *top;
 320 +               int count;
 321 +
 322 +               /* Pop the top stack entry, and iterate through its contents. */
 323 +               top = dirstack[--dirstack_top];
 324 +               snprintf(name, sizeof name, ZONEINFO_PREFIX "/%s", top);
 325 +
 326 +               count = php_scandir(name, &ents, index_filter, php_alphasort);
 327 +
 328 +               while (count > 0) {
 329 +                       struct stat st;
 330 +                       const char *leaf = ents[count - 1]->d_name;
 331 +
 332 +                       snprintf(name, sizeof name, ZONEINFO_PREFIX "/%s/%s",
 333 +                                top, leaf);
 334 +
 335 +                       if (strlen(name) && stat(name, &st) == 0) {
 336 +                               /* Name, relative to the zoneinfo prefix. */
 337 +                               const char *root = top;
 338 +
 339 +                               if (root[0] == '/') root++;
 340 +
 341 +                               snprintf(name, sizeof name, "%s%s%s", root,
 342 +                                        *root ? "/": "", leaf);
 343 +
 344 +                               if (S_ISDIR(st.st_mode)) {
 345 +                                       if (dirstack_top == dirstack_size) {
 346 +                                               dirstack_size *= 2;
 347 +                                               dirstack = realloc(dirstack,
 348 +                                                                  dirstack_size * sizeof *dirstack);
 349 +                                       }
 350 +                                       dirstack[dirstack_top++] = strdup(name);
 351 +                               }
 352 +                               else {
 353 +                                       if (index_next == index_size) {
 354 +                                               index_size *= 2;
 355 +                                               db_index = realloc(db_index,
 356 +                                                                  index_size * sizeof *db_index);
 357 +                                       }
 358 +
 359 +                                       db_index[index_next++].id = strdup(name);
 360 +                               }
 361 +                       }
 362 +
 363 +                       free(ents[--count]);
 364 +               }
 365 +
 366 +               if (count != -1) free(ents);
 367 +               free(top);
 368 +       } while (dirstack_top);
 369 +
 370 +        qsort(db_index, index_next, sizeof *db_index, sysdbcmp);
 371 +
 372 +       db->index = db_index;
 373 +       db->index_size = index_next;
 374 +
 375 +       free(dirstack);
 376 +}
 377 +
 378 +#define FAKE_HEADER "1234\0??\1??"
 379 +#define FAKE_UTC_POS (7 - 4)
 380 +
 381 +/* Create a fake data segment for database 'sysdb'. */
 382 +static void fake_data_segment(timelib_tzdb *sysdb,
 383 +                              struct location_info **info)
 384 +{
 385 +        size_t n;
 386 +        char *data, *p;
 387 +
 388 +        data = malloc(3 * sysdb->index_size + 7);
 389 +
 390 +        p = mempcpy(data, FAKE_HEADER, sizeof(FAKE_HEADER) - 1);
 391 +
 392 +        for (n = 0; n < sysdb->index_size; n++) {
 393 +                const struct location_info *li;
 394 +                timelib_tzdb_index_entry *ent;
 395 +
 396 +                ent = (timelib_tzdb_index_entry *)&sysdb->index[n];
 397 +
 398 +                /* Lookup the timezone name in the hash table. */
 399 +                if (strcmp(ent->id, "UTC") == 0) {
 400 +                        ent->pos = FAKE_UTC_POS;
 401 +                        continue;
 402 +                }
 403 +
 404 +                li = find_zone_info(info, ent->id);
 405 +                if (li) {
 406 +                        /* If found, append the BC byte and the
 407 +                         * country code; set the position for this
 408 +                         * section of timezone data.  */
 409 +                        ent->pos = (p - data) - 4;
 410 +                        *p++ = '\1';
 411 +                        *p++ = li->code[0];
 412 +                        *p++ = li->code[1];
 413 +                }
 414 +                else {
 415 +                        /* If not found, the timezone data can
 416 +                         * point at the header. */
 417 +                        ent->pos = 0;
 418 +                }
 419 +        }
 420 +
 421 +        sysdb->data = (unsigned char *)data;
 422 +}
 423 +
 424 +/* Returns true if the passed-in stat structure describes a
 425 + * probably-valid timezone file. */
 426 +static int is_valid_tzfile(const struct stat *st)
 427 +{
 428 +       return S_ISREG(st->st_mode) && st->st_size > 20;
 429 +}
 430 +
 431 +/* Return the mmap()ed tzfile if found, else NULL.  On success, the
 432 + * length of the mapped data is placed in *length. */
 433 +static char *map_tzfile(const char *timezone, size_t *length)
 434 +{
 435 +       char fname[PATH_MAX];
 436 +       struct stat st;
 437 +       char *p;
 438 +       int fd;
 439 +
 440 +       if (timezone[0] == '\0' || strstr(timezone, "..") != NULL) {
 441 +               return NULL;
 442 +       }
 443 +
 444 +       snprintf(fname, sizeof fname, ZONEINFO_PREFIX "/%s", timezone);
 445 +
 446 +       fd = open(fname, O_RDONLY);
 447 +       if (fd == -1) {
 448 +               return NULL;
 449 +       } else if (fstat(fd, &st) != 0 || !is_valid_tzfile(&st)) {
 450 +               close(fd);
 451 +               return NULL;
 452 +       }
 453 +
 454 +       *length = st.st_size;
 455 +       p = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0);
 456 +       close(fd);
 457 +
 458 +       return p != MAP_FAILED ? p : NULL;
 459 +}
 460 +
 461 +#endif
 462 +
 463 +static int inmem_seek_to_tz_position(const unsigned char **tzf, char *timezone, const timelib_tzdb *tzdb)
 464  {
 465         int left = 0, right = tzdb->index_size - 1;
 466  #ifdef HAVE_SETLOCALE
 467 @@ -292,36 +702,125 @@ static int seek_to_tz_position(const uns
 468         return 0;
 469  }
 470
 471 +static int seek_to_tz_position(const unsigned char **tzf, char *timezone,
 472 +                              char **map, size_t *maplen,
 473 +                              const timelib_tzdb *tzdb)
 474 +{
 475 +       if (tzdb == timezonedb_system) {
 476 +               char *orig;
 477 +
 478 +               orig = map_tzfile(timezone, maplen);
 479 +               if (orig == NULL) {
 480 +                       return 0;
 481 +               }
 482 +
 483 +               (*tzf) = (unsigned char *)orig ;
 484 +               *map = orig;
 485 +
 486 +                return 1;
 487 +       }
 488 +       else {
 489 +               return inmem_seek_to_tz_position(tzf, timezone, tzdb);
 490 +       }
 491 +}
 492 +
 493  const timelib_tzdb *timelib_builtin_db(void)
 494  {
 495 +#ifdef HAVE_SYSTEM_TZDATA
 496 +       if (timezonedb_system == NULL) {
 497 +               timelib_tzdb *tmp = malloc(sizeof *tmp);
 498 +
 499 +               tmp->version = "0.system";
 500 +               tmp->data = NULL;
 501 +               create_zone_index(tmp);
 502 +               system_location_table = create_location_table();
 503 +                fake_data_segment(tmp, system_location_table);
 504 +               timezonedb_system = tmp;
 505 +       }
 506 +
 507 +
 508 +       return timezonedb_system;
 509 +#else
 510         return &timezonedb_builtin;
 511 +#endif
 512  }
 513
 514  const timelib_tzdb_index_entry *timelib_timezone_builtin_identifiers_list(int *count)
 515  {
 516 +#ifdef HAVE_SYSTEM_TZDATA
 517 +       *count = timezonedb_system->index_size;
 518 +       return timezonedb_system->index;
 519 +#else
 520         *count = sizeof(timezonedb_idx_builtin) / sizeof(*timezonedb_idx_builtin);
 521         return timezonedb_idx_builtin;
 522 +#endif
 523  }
 524
 525  int timelib_timezone_id_is_valid(char *timezone, const timelib_tzdb *tzdb)
 526  {
 527         const unsigned char *tzf;
 528 -       return (seek_to_tz_position(&tzf, timezone, tzdb));
 529 +
 530 +#ifdef HAVE_SYSTEM_TZDATA
 531 +        if (tzdb == timezonedb_system) {
 532 +            char fname[PATH_MAX];
 533 +            struct stat st;
 534 +
 535 +            if (timezone[0] == '\0' || strstr(timezone, "..") != NULL) {
 536 +               return 0;
 537 +            }
 538 +
 539 +            snprintf(fname, sizeof fname, ZONEINFO_PREFIX "/%s", timezone);
 540 +
 541 +            return stat(fname, &st) == 0 && is_valid_tzfile(&st);
 542 +        }
 543 +#endif
 544 +
 545 +       return (inmem_seek_to_tz_position(&tzf, timezone, tzdb));
 546  }
 547
 548  timelib_tzinfo *timelib_parse_tzfile(char *timezone, const timelib_tzdb *tzdb)
 549  {
 550         const unsigned char *tzf;
 551 +       char *memmap = NULL;
 552 +       size_t maplen;
 553         timelib_tzinfo *tmp;
 554
 555 -       if (seek_to_tz_position(&tzf, timezone, tzdb)) {
 556 +       if (seek_to_tz_position(&tzf, timezone, &memmap, &maplen, tzdb)) {
 557                 tmp = timelib_tzinfo_ctor(timezone);
 558
 559                 read_preamble(&tzf, tmp);
 560                 read_header(&tzf, tmp);
 561                 read_transistions(&tzf, tmp);
 562                 read_types(&tzf, tmp);
 563 -               read_location(&tzf, tmp);
 564 +
 565 +#ifdef HAVE_SYSTEM_TZDATA
 566 +               if (memmap) {
 567 +                       const struct location_info *li;
 568 +
 569 +                       /* TZif-style - grok the location info from the system database,
 570 +                        * if possible. */
 571 +
 572 +                       if ((li = find_zone_info(system_location_table, timezone)) != NULL) {
 573 +                               tmp->location.comments = strdup(li->comment);
 574 +                                strncpy(tmp->location.country_code, li->code, 2);
 575 +                               tmp->location.longitude = li->longitude;
 576 +                               tmp->location.latitude = li->latitude;
 577 +                               tmp->bc = 1;
 578 +                       }
 579 +                       else {
 580 +                               strcpy(tmp->location.country_code, "??");
 581 +                               tmp->bc = 0;
 582 +                               tmp->location.comments = strdup("");
 583 +                       }
 584 +
 585 +                       /* Now done with the mmap segment - discard it. */
 586 +                       munmap(memmap, maplen);
 587 +#endif
 588 +               }
 589 +               else {
 590 +                       /* PHP-style - use the embedded info. */
 591 +                       read_location(&tzf, tmp);
 592 +               }
 593         } else {
 594                 tmp = NULL;
 595         }
 596 --- php-5.3.1/ext/date/lib/timelib.m4.systzdata
 597 +++ php-5.3.1/ext/date/lib/timelib.m4
 598 @@ -78,3 +78,17 @@ stdlib.h
 599
 600  dnl Check for strtoll, atoll
 601  AC_CHECK_FUNCS(strtoll atoll strftime)
 602 +
 603 +PHP_ARG_WITH(system-tzdata, for use of system timezone data,
 604 +[  --with-system-tzdata[=DIR]      to specify use of system timezone data],
 605 +no, no)
 606 +
 607 +if test "$PHP_SYSTEM_TZDATA" != "no"; then
 608 +   AC_DEFINE(HAVE_SYSTEM_TZDATA, 1, [Define if system timezone data is used])
 609 +
 610 +   if test "$PHP_SYSTEM_TZDATA" != "yes"; then
 611 +      AC_DEFINE_UNQUOTED(HAVE_SYSTEM_TZDATA_PREFIX, "$PHP_SYSTEM_TZDATA",
 612 +                         [Define for location of system timezone data])
 613 +   fi
 614 +fi
 615 +