archive_read_support_format_warc.c revision 358088
1/*- 2 * Copyright (c) 2014 Sebastian Freundt 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 15 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 16 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 17 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 18 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 19 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 20 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 21 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 22 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 23 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 24 */ 25 26#include "archive_platform.h" 27__FBSDID("$FreeBSD: stable/11/contrib/libarchive/libarchive/archive_read_support_format_warc.c 358088 2020-02-19 01:50:47Z mm $"); 28 29/** 30 * WARC is standardised by ISO TC46/SC4/WG12 and currently available as 31 * ISO 28500:2009. 32 * For the purposes of this file we used the final draft from: 33 * http://bibnum.bnf.fr/warc/WARC_ISO_28500_version1_latestdraft.pdf 34 * 35 * Todo: 36 * [ ] real-world warcs can contain resources at endpoints ending in / 37 * e.g. http://bibnum.bnf.fr/warc/ 38 * if you're lucky their response contains a Content-Location: header 39 * pointing to a unix-compliant filename, in the example above it's 40 * Content-Location: http://bibnum.bnf.fr/warc/index.html 41 * however, that's not mandated and github for example doesn't follow 42 * this convention. 43 * We need a set of archive options to control what to do with 44 * entries like these, at the moment care is taken to skip them. 45 * 46 **/ 47 48#ifdef HAVE_SYS_STAT_H 49#include <sys/stat.h> 50#endif 51#ifdef HAVE_ERRNO_H 52#include <errno.h> 53#endif 54#ifdef HAVE_STDLIB_H 55#include <stdlib.h> 56#endif 57#ifdef HAVE_STRING_H 58#include <string.h> 59#endif 60#ifdef HAVE_LIMITS_H 61#include <limits.h> 62#endif 63#ifdef HAVE_CTYPE_H 64#include <ctype.h> 65#endif 66#ifdef HAVE_TIME_H 67#include <time.h> 68#endif 69 70#include "archive.h" 71#include "archive_entry.h" 72#include "archive_private.h" 73#include "archive_read_private.h" 74 75typedef enum { 76 WT_NONE, 77 /* warcinfo */ 78 WT_INFO, 79 /* metadata */ 80 WT_META, 81 /* resource */ 82 WT_RSRC, 83 /* request, unsupported */ 84 WT_REQ, 85 /* response, unsupported */ 86 WT_RSP, 87 /* revisit, unsupported */ 88 WT_RVIS, 89 /* conversion, unsupported */ 90 WT_CONV, 91 /* continuation, unsupported at the moment */ 92 WT_CONT, 93 /* invalid type */ 94 LAST_WT 95} warc_type_t; 96 97typedef struct { 98 size_t len; 99 const char *str; 100} warc_string_t; 101 102typedef struct { 103 size_t len; 104 char *str; 105} warc_strbuf_t; 106 107struct warc_s { 108 /* content length ahead */ 109 size_t cntlen; 110 /* and how much we've processed so far */ 111 size_t cntoff; 112 /* and how much we need to consume between calls */ 113 size_t unconsumed; 114 115 /* string pool */ 116 warc_strbuf_t pool; 117 /* previous version */ 118 unsigned int pver; 119 /* stringified format name */ 120 struct archive_string sver; 121}; 122 123static int _warc_bid(struct archive_read *a, int); 124static int _warc_cleanup(struct archive_read *a); 125static int _warc_read(struct archive_read*, const void**, size_t*, int64_t*); 126static int _warc_skip(struct archive_read *a); 127static int _warc_rdhdr(struct archive_read *a, struct archive_entry *e); 128 129/* private routines */ 130static unsigned int _warc_rdver(const char buf[10], size_t bsz); 131static unsigned int _warc_rdtyp(const char *buf, size_t bsz); 132static warc_string_t _warc_rduri(const char *buf, size_t bsz); 133static ssize_t _warc_rdlen(const char *buf, size_t bsz); 134static time_t _warc_rdrtm(const char *buf, size_t bsz); 135static time_t _warc_rdmtm(const char *buf, size_t bsz); 136static const char *_warc_find_eoh(const char *buf, size_t bsz); 137static const char *_warc_find_eol(const char *buf, size_t bsz); 138 139int 140archive_read_support_format_warc(struct archive *_a) 141{ 142 struct archive_read *a = (struct archive_read *)_a; 143 struct warc_s *w; 144 int r; 145 146 archive_check_magic(_a, ARCHIVE_READ_MAGIC, 147 ARCHIVE_STATE_NEW, "archive_read_support_format_warc"); 148 149 if ((w = calloc(1, sizeof(*w))) == NULL) { 150 archive_set_error(&a->archive, ENOMEM, 151 "Can't allocate warc data"); 152 return (ARCHIVE_FATAL); 153 } 154 155 r = __archive_read_register_format( 156 a, w, "warc", 157 _warc_bid, NULL, _warc_rdhdr, _warc_read, 158 _warc_skip, NULL, _warc_cleanup, NULL, NULL); 159 160 if (r != ARCHIVE_OK) { 161 free(w); 162 return (r); 163 } 164 return (ARCHIVE_OK); 165} 166 167static int 168_warc_cleanup(struct archive_read *a) 169{ 170 struct warc_s *w = a->format->data; 171 172 if (w->pool.len > 0U) { 173 free(w->pool.str); 174 } 175 archive_string_free(&w->sver); 176 free(w); 177 a->format->data = NULL; 178 return (ARCHIVE_OK); 179} 180 181static int 182_warc_bid(struct archive_read *a, int best_bid) 183{ 184 const char *hdr; 185 ssize_t nrd; 186 unsigned int ver; 187 188 (void)best_bid; /* UNUSED */ 189 190 /* check first line of file, it should be a record already */ 191 if ((hdr = __archive_read_ahead(a, 12U, &nrd)) == NULL) { 192 /* no idea what to do */ 193 return -1; 194 } else if (nrd < 12) { 195 /* nah, not for us, our magic cookie is at least 12 bytes */ 196 return -1; 197 } 198 199 /* otherwise snarf the record's version number */ 200 ver = _warc_rdver(hdr, nrd); 201 if (ver < 1200U || ver > 10000U) { 202 /* we only support WARC 0.12 to 1.0 */ 203 return -1; 204 } 205 206 /* otherwise be confident */ 207 return (64); 208} 209 210static int 211_warc_rdhdr(struct archive_read *a, struct archive_entry *entry) 212{ 213#define HDR_PROBE_LEN (12U) 214 struct warc_s *w = a->format->data; 215 unsigned int ver; 216 const char *buf; 217 ssize_t nrd; 218 const char *eoh; 219 /* for the file name, saves some strndup()'ing */ 220 warc_string_t fnam; 221 /* warc record type, not that we really use it a lot */ 222 warc_type_t ftyp; 223 /* content-length+error monad */ 224 ssize_t cntlen; 225 /* record time is the WARC-Date time we reinterpret it as ctime */ 226 time_t rtime; 227 /* mtime is the Last-Modified time which will be the entry's mtime */ 228 time_t mtime; 229 230start_over: 231 /* just use read_ahead() they keep track of unconsumed 232 * bits and bobs for us; no need to put an extra shift in 233 * and reproduce that functionality here */ 234 buf = __archive_read_ahead(a, HDR_PROBE_LEN, &nrd); 235 236 if (nrd < 0) { 237 /* no good */ 238 archive_set_error( 239 &a->archive, ARCHIVE_ERRNO_MISC, 240 "Bad record header"); 241 return (ARCHIVE_FATAL); 242 } else if (buf == NULL) { 243 /* there should be room for at least WARC/bla\r\n 244 * must be EOF therefore */ 245 return (ARCHIVE_EOF); 246 } 247 /* looks good so far, try and find the end of the header now */ 248 eoh = _warc_find_eoh(buf, nrd); 249 if (eoh == NULL) { 250 /* still no good, the header end might be beyond the 251 * probe we've requested, but then again who'd cram 252 * so much stuff into the header *and* be 28500-compliant */ 253 archive_set_error( 254 &a->archive, ARCHIVE_ERRNO_MISC, 255 "Bad record header"); 256 return (ARCHIVE_FATAL); 257 } 258 ver = _warc_rdver(buf, eoh - buf); 259 /* we currently support WARC 0.12 to 1.0 */ 260 if (ver == 0U) { 261 archive_set_error( 262 &a->archive, ARCHIVE_ERRNO_MISC, 263 "Invalid record version"); 264 return (ARCHIVE_FATAL); 265 } else if (ver < 1200U || ver > 10000U) { 266 archive_set_error( 267 &a->archive, ARCHIVE_ERRNO_MISC, 268 "Unsupported record version: %u.%u", 269 ver / 10000, (ver % 10000) / 100); 270 return (ARCHIVE_FATAL); 271 } 272 cntlen = _warc_rdlen(buf, eoh - buf); 273 if (cntlen < 0) { 274 /* nightmare! the specs say content-length is mandatory 275 * so I don't feel overly bad stopping the reader here */ 276 archive_set_error( 277 &a->archive, EINVAL, 278 "Bad content length"); 279 return (ARCHIVE_FATAL); 280 } 281 rtime = _warc_rdrtm(buf, eoh - buf); 282 if (rtime == (time_t)-1) { 283 /* record time is mandatory as per WARC/1.0, 284 * so just barf here, fast and loud */ 285 archive_set_error( 286 &a->archive, EINVAL, 287 "Bad record time"); 288 return (ARCHIVE_FATAL); 289 } 290 291 /* let the world know we're a WARC archive */ 292 a->archive.archive_format = ARCHIVE_FORMAT_WARC; 293 if (ver != w->pver) { 294 /* stringify this entry's version */ 295 archive_string_sprintf(&w->sver, 296 "WARC/%u.%u", ver / 10000, (ver % 10000) / 100); 297 /* remember the version */ 298 w->pver = ver; 299 } 300 /* start off with the type */ 301 ftyp = _warc_rdtyp(buf, eoh - buf); 302 /* and let future calls know about the content */ 303 w->cntlen = cntlen; 304 w->cntoff = 0U; 305 mtime = 0;/* Avoid compiling error on some platform. */ 306 307 switch (ftyp) { 308 case WT_RSRC: 309 case WT_RSP: 310 /* only try and read the filename in the cases that are 311 * guaranteed to have one */ 312 fnam = _warc_rduri(buf, eoh - buf); 313 /* check the last character in the URI to avoid creating 314 * directory endpoints as files, see Todo above */ 315 if (fnam.len == 0 || fnam.str[fnam.len - 1] == '/') { 316 /* break here for now */ 317 fnam.len = 0U; 318 fnam.str = NULL; 319 break; 320 } 321 /* bang to our string pool, so we save a 322 * malloc()+free() roundtrip */ 323 if (fnam.len + 1U > w->pool.len) { 324 w->pool.len = ((fnam.len + 64U) / 64U) * 64U; 325 w->pool.str = realloc(w->pool.str, w->pool.len); 326 } 327 memcpy(w->pool.str, fnam.str, fnam.len); 328 w->pool.str[fnam.len] = '\0'; 329 /* let no one else know about the pool, it's a secret, shhh */ 330 fnam.str = w->pool.str; 331 332 /* snarf mtime or deduce from rtime 333 * this is a custom header added by our writer, it's quite 334 * hard to believe anyone else would go through with it 335 * (apart from being part of some http responses of course) */ 336 if ((mtime = _warc_rdmtm(buf, eoh - buf)) == (time_t)-1) { 337 mtime = rtime; 338 } 339 break; 340 default: 341 fnam.len = 0U; 342 fnam.str = NULL; 343 break; 344 } 345 346 /* now eat some of those delicious buffer bits */ 347 __archive_read_consume(a, eoh - buf); 348 349 switch (ftyp) { 350 case WT_RSRC: 351 case WT_RSP: 352 if (fnam.len > 0U) { 353 /* populate entry object */ 354 archive_entry_set_filetype(entry, AE_IFREG); 355 archive_entry_copy_pathname(entry, fnam.str); 356 archive_entry_set_size(entry, cntlen); 357 archive_entry_set_perm(entry, 0644); 358 /* rtime is the new ctime, mtime stays mtime */ 359 archive_entry_set_ctime(entry, rtime, 0L); 360 archive_entry_set_mtime(entry, mtime, 0L); 361 break; 362 } 363 /* FALLTHROUGH */ 364 default: 365 /* consume the content and start over */ 366 _warc_skip(a); 367 goto start_over; 368 } 369 return (ARCHIVE_OK); 370} 371 372static int 373_warc_read(struct archive_read *a, const void **buf, size_t *bsz, int64_t *off) 374{ 375 struct warc_s *w = a->format->data; 376 const char *rab; 377 ssize_t nrd; 378 379 if (w->cntoff >= w->cntlen) { 380 eof: 381 /* it's our lucky day, no work, we can leave early */ 382 *buf = NULL; 383 *bsz = 0U; 384 *off = w->cntoff + 4U/*for \r\n\r\n separator*/; 385 w->unconsumed = 0U; 386 return (ARCHIVE_EOF); 387 } 388 389 if (w->unconsumed) { 390 __archive_read_consume(a, w->unconsumed); 391 w->unconsumed = 0U; 392 } 393 394 rab = __archive_read_ahead(a, 1U, &nrd); 395 if (nrd < 0) { 396 *bsz = 0U; 397 /* big catastrophe */ 398 return (int)nrd; 399 } else if (nrd == 0) { 400 goto eof; 401 } else if ((size_t)nrd > w->cntlen - w->cntoff) { 402 /* clamp to content-length */ 403 nrd = w->cntlen - w->cntoff; 404 } 405 *off = w->cntoff; 406 *bsz = nrd; 407 *buf = rab; 408 409 w->cntoff += nrd; 410 w->unconsumed = (size_t)nrd; 411 return (ARCHIVE_OK); 412} 413 414static int 415_warc_skip(struct archive_read *a) 416{ 417 struct warc_s *w = a->format->data; 418 419 __archive_read_consume(a, w->cntlen + 4U/*\r\n\r\n separator*/); 420 w->cntlen = 0U; 421 w->cntoff = 0U; 422 return (ARCHIVE_OK); 423} 424 425 426/* private routines */ 427static void* 428deconst(const void *c) 429{ 430 return (char *)0x1 + (((const char *)c) - (const char *)0x1); 431} 432 433static char* 434xmemmem(const char *hay, const size_t haysize, 435 const char *needle, const size_t needlesize) 436{ 437 const char *const eoh = hay + haysize; 438 const char *const eon = needle + needlesize; 439 const char *hp; 440 const char *np; 441 const char *cand; 442 unsigned int hsum; 443 unsigned int nsum; 444 unsigned int eqp; 445 446 /* trivial checks first 447 * a 0-sized needle is defined to be found anywhere in haystack 448 * then run strchr() to find a candidate in HAYSTACK (i.e. a portion 449 * that happens to begin with *NEEDLE) */ 450 if (needlesize == 0UL) { 451 return deconst(hay); 452 } else if ((hay = memchr(hay, *needle, haysize)) == NULL) { 453 /* trivial */ 454 return NULL; 455 } 456 457 /* First characters of haystack and needle are the same now. Both are 458 * guaranteed to be at least one character long. Now computes the sum 459 * of characters values of needle together with the sum of the first 460 * needle_len characters of haystack. */ 461 for (hp = hay + 1U, np = needle + 1U, hsum = *hay, nsum = *hay, eqp = 1U; 462 hp < eoh && np < eon; 463 hsum ^= *hp, nsum ^= *np, eqp &= *hp == *np, hp++, np++); 464 465 /* HP now references the (NEEDLESIZE + 1)-th character. */ 466 if (np < eon) { 467 /* haystack is smaller than needle, :O */ 468 return NULL; 469 } else if (eqp) { 470 /* found a match */ 471 return deconst(hay); 472 } 473 474 /* now loop through the rest of haystack, 475 * updating the sum iteratively */ 476 for (cand = hay; hp < eoh; hp++) { 477 hsum ^= *cand++; 478 hsum ^= *hp; 479 480 /* Since the sum of the characters is already known to be 481 * equal at that point, it is enough to check just NEEDLESIZE - 1 482 * characters for equality, 483 * also CAND is by design < HP, so no need for range checks */ 484 if (hsum == nsum && memcmp(cand, needle, needlesize - 1U) == 0) { 485 return deconst(cand); 486 } 487 } 488 return NULL; 489} 490 491static int 492strtoi_lim(const char *str, const char **ep, int llim, int ulim) 493{ 494 int res = 0; 495 const char *sp; 496 /* we keep track of the number of digits via rulim */ 497 int rulim; 498 499 for (sp = str, rulim = ulim > 10 ? ulim : 10; 500 res * 10 <= ulim && rulim && *sp >= '0' && *sp <= '9'; 501 sp++, rulim /= 10) { 502 res *= 10; 503 res += *sp - '0'; 504 } 505 if (sp == str) { 506 res = -1; 507 } else if (res < llim || res > ulim) { 508 res = -2; 509 } 510 *ep = (const char*)sp; 511 return res; 512} 513 514static time_t 515time_from_tm(struct tm *t) 516{ 517#if HAVE_TIMEGM 518 /* Use platform timegm() if available. */ 519 return (timegm(t)); 520#elif HAVE__MKGMTIME64 521 return (_mkgmtime64(t)); 522#else 523 /* Else use direct calculation using POSIX assumptions. */ 524 /* First, fix up tm_yday based on the year/month/day. */ 525 if (mktime(t) == (time_t)-1) 526 return ((time_t)-1); 527 /* Then we can compute timegm() from first principles. */ 528 return (t->tm_sec 529 + t->tm_min * 60 530 + t->tm_hour * 3600 531 + t->tm_yday * 86400 532 + (t->tm_year - 70) * 31536000 533 + ((t->tm_year - 69) / 4) * 86400 534 - ((t->tm_year - 1) / 100) * 86400 535 + ((t->tm_year + 299) / 400) * 86400); 536#endif 537} 538 539static time_t 540xstrpisotime(const char *s, char **endptr) 541{ 542/** like strptime() but strictly for ISO 8601 Zulu strings */ 543 struct tm tm; 544 time_t res = (time_t)-1; 545 546 /* make sure tm is clean */ 547 memset(&tm, 0, sizeof(tm)); 548 549 /* as a courtesy to our callers, and since this is a non-standard 550 * routine, we skip leading whitespace */ 551 while (*s == ' ' || *s == '\t') 552 ++s; 553 554 /* read year */ 555 if ((tm.tm_year = strtoi_lim(s, &s, 1583, 4095)) < 0 || *s++ != '-') { 556 goto out; 557 } 558 /* read month */ 559 if ((tm.tm_mon = strtoi_lim(s, &s, 1, 12)) < 0 || *s++ != '-') { 560 goto out; 561 } 562 /* read day-of-month */ 563 if ((tm.tm_mday = strtoi_lim(s, &s, 1, 31)) < 0 || *s++ != 'T') { 564 goto out; 565 } 566 /* read hour */ 567 if ((tm.tm_hour = strtoi_lim(s, &s, 0, 23)) < 0 || *s++ != ':') { 568 goto out; 569 } 570 /* read minute */ 571 if ((tm.tm_min = strtoi_lim(s, &s, 0, 59)) < 0 || *s++ != ':') { 572 goto out; 573 } 574 /* read second */ 575 if ((tm.tm_sec = strtoi_lim(s, &s, 0, 60)) < 0 || *s++ != 'Z') { 576 goto out; 577 } 578 579 /* massage TM to fulfill some of POSIX' constraints */ 580 tm.tm_year -= 1900; 581 tm.tm_mon--; 582 583 /* now convert our custom tm struct to a unix stamp using UTC */ 584 res = time_from_tm(&tm); 585 586out: 587 if (endptr != NULL) { 588 *endptr = deconst(s); 589 } 590 return res; 591} 592 593static unsigned int 594_warc_rdver(const char *buf, size_t bsz) 595{ 596 static const char magic[] = "WARC/"; 597 const char *c; 598 unsigned int ver = 0U; 599 unsigned int end = 0U; 600 601 if (bsz < 12 || memcmp(buf, magic, sizeof(magic) - 1U) != 0) { 602 /* buffer too small or invalid magic */ 603 return ver; 604 } 605 /* looks good so far, read the version number for a laugh */ 606 buf += sizeof(magic) - 1U; 607 608 if (isdigit((unsigned char)buf[0U]) && (buf[1U] == '.') && 609 isdigit((unsigned char)buf[2U])) { 610 /* we support a maximum of 2 digits in the minor version */ 611 if (isdigit((unsigned char)buf[3U])) 612 end = 1U; 613 /* set up major version */ 614 ver = (buf[0U] - '0') * 10000U; 615 /* set up minor version */ 616 if (end == 1U) { 617 ver += (buf[2U] - '0') * 1000U; 618 ver += (buf[3U] - '0') * 100U; 619 } else 620 ver += (buf[2U] - '0') * 100U; 621 /* 622 * WARC below version 0.12 has a space-separated header 623 * WARC 0.12 and above terminates the version with a CRLF 624 */ 625 c = buf + 3U + end; 626 if (ver >= 1200U) { 627 if (memcmp(c, "\r\n", 2U) != 0) 628 ver = 0U; 629 } else { 630 /* ver < 1200U */ 631 if (*c != ' ' && *c != '\t') 632 ver = 0U; 633 } 634 } 635 return ver; 636} 637 638static unsigned int 639_warc_rdtyp(const char *buf, size_t bsz) 640{ 641 static const char _key[] = "\r\nWARC-Type:"; 642 const char *val, *eol; 643 644 if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { 645 /* no bother */ 646 return WT_NONE; 647 } 648 val += sizeof(_key) - 1U; 649 if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { 650 /* no end of line */ 651 return WT_NONE; 652 } 653 654 /* overread whitespace */ 655 while (val < eol && (*val == ' ' || *val == '\t')) 656 ++val; 657 658 if (val + 8U == eol) { 659 if (memcmp(val, "resource", 8U) == 0) 660 return WT_RSRC; 661 else if (memcmp(val, "response", 8U) == 0) 662 return WT_RSP; 663 } 664 return WT_NONE; 665} 666 667static warc_string_t 668_warc_rduri(const char *buf, size_t bsz) 669{ 670 static const char _key[] = "\r\nWARC-Target-URI:"; 671 const char *val, *uri, *eol, *p; 672 warc_string_t res = {0U, NULL}; 673 674 if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { 675 /* no bother */ 676 return res; 677 } 678 /* overread whitespace */ 679 val += sizeof(_key) - 1U; 680 if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { 681 /* no end of line */ 682 return res; 683 } 684 685 while (val < eol && (*val == ' ' || *val == '\t')) 686 ++val; 687 688 /* overread URL designators */ 689 if ((uri = xmemmem(val, eol - val, "://", 3U)) == NULL) { 690 /* not touching that! */ 691 return res; 692 } 693 694 /* spaces inside uri are not allowed, CRLF should follow */ 695 for (p = val; p < eol; p++) { 696 if (isspace((unsigned char)*p)) 697 return res; 698 } 699 700 /* there must be at least space for ftp */ 701 if (uri < (val + 3U)) 702 return res; 703 704 /* move uri to point to after :// */ 705 uri += 3U; 706 707 /* now then, inspect the URI */ 708 if (memcmp(val, "file", 4U) == 0) { 709 /* perfect, nothing left to do here */ 710 711 } else if (memcmp(val, "http", 4U) == 0 || 712 memcmp(val, "ftp", 3U) == 0) { 713 /* overread domain, and the first / */ 714 while (uri < eol && *uri++ != '/'); 715 } else { 716 /* not sure what to do? best to bugger off */ 717 return res; 718 } 719 res.str = uri; 720 res.len = eol - uri; 721 return res; 722} 723 724static ssize_t 725_warc_rdlen(const char *buf, size_t bsz) 726{ 727 static const char _key[] = "\r\nContent-Length:"; 728 const char *val, *eol; 729 char *on = NULL; 730 long int len; 731 732 if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { 733 /* no bother */ 734 return -1; 735 } 736 val += sizeof(_key) - 1U; 737 if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { 738 /* no end of line */ 739 return -1; 740 } 741 742 /* skip leading whitespace */ 743 while (val < eol && (*val == ' ' || *val == '\t')) 744 val++; 745 /* there must be at least one digit */ 746 if (!isdigit((unsigned char)*val)) 747 return -1; 748 errno = 0; 749 len = strtol(val, &on, 10); 750 if (errno != 0 || on != eol) { 751 /* line must end here */ 752 return -1; 753 } 754 755 return (size_t)len; 756} 757 758static time_t 759_warc_rdrtm(const char *buf, size_t bsz) 760{ 761 static const char _key[] = "\r\nWARC-Date:"; 762 const char *val, *eol; 763 char *on = NULL; 764 time_t res; 765 766 if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { 767 /* no bother */ 768 return (time_t)-1; 769 } 770 val += sizeof(_key) - 1U; 771 if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL ) { 772 /* no end of line */ 773 return -1; 774 } 775 776 /* xstrpisotime() kindly overreads whitespace for us, so use that */ 777 res = xstrpisotime(val, &on); 778 if (on != eol) { 779 /* line must end here */ 780 return -1; 781 } 782 return res; 783} 784 785static time_t 786_warc_rdmtm(const char *buf, size_t bsz) 787{ 788 static const char _key[] = "\r\nLast-Modified:"; 789 const char *val, *eol; 790 char *on = NULL; 791 time_t res; 792 793 if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { 794 /* no bother */ 795 return (time_t)-1; 796 } 797 val += sizeof(_key) - 1U; 798 if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL ) { 799 /* no end of line */ 800 return -1; 801 } 802 803 /* xstrpisotime() kindly overreads whitespace for us, so use that */ 804 res = xstrpisotime(val, &on); 805 if (on != eol) { 806 /* line must end here */ 807 return -1; 808 } 809 return res; 810} 811 812static const char* 813_warc_find_eoh(const char *buf, size_t bsz) 814{ 815 static const char _marker[] = "\r\n\r\n"; 816 const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); 817 818 if (hit != NULL) { 819 hit += sizeof(_marker) - 1U; 820 } 821 return hit; 822} 823 824static const char* 825_warc_find_eol(const char *buf, size_t bsz) 826{ 827 static const char _marker[] = "\r\n"; 828 const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); 829 830 return hit; 831} 832/* archive_read_support_format_warc.c ends here */ 833