1 /* Utility functions for writing WARC files.
2 Copyright (C) 2011, 2012 Free Software Foundation, Inc.
4 This file is part of GNU Wget.
6 GNU Wget is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 3 of the License, or (at
9 your option) any later version.
11 GNU Wget is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with Wget. If not, see <http://www.gnu.org/licenses/>.
19 Additional permission under GNU GPL version 3 section 7
21 If you modify this program, or any covered work, by linking or
22 combining it with the OpenSSL project's OpenSSL library (or a
23 modified version of that library), containing parts covered by the
24 terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
25 grants you additional permission to convey the resulting work.
26 Corresponding Source for a non-source form of such a combination
27 shall include the source code for the parts of OpenSSL used as well
28 as that of the covered work. */
49 #include <uuid/uuid.h>
58 extern char *version_string;
60 /* Set by main in main.c */
61 extern char *program_argstring;
64 /* The log file (a temporary file that contains a copy
66 static FILE *warc_log_fp;
68 /* The manifest file (a temporary file that contains the
69 warcinfo uuid of every file in this crawl). */
70 static FILE *warc_manifest_fp;
72 /* The current WARC file (or NULL, if WARC is disabled). */
73 static FILE *warc_current_file;
76 /* The gzip stream for the current WARC file
77 (or NULL, if WARC or gzip is disabled). */
78 static gzFile warc_current_gzfile;
80 /* The offset of the current gzip record in the WARC file. */
81 static off_t warc_current_gzfile_offset;
83 /* The uncompressed size (so far) of the current record. */
84 static off_t warc_current_gzfile_uncompressed_size;
87 /* This is true until a warc_write_* method fails. */
88 static bool warc_write_ok;
90 /* The current CDX file (or NULL, if CDX is disabled). */
91 static FILE *warc_current_cdx_file;
93 /* The record id of the warcinfo record of the current WARC file. */
94 static char *warc_current_warcinfo_uuid_str;
96 /* The file name of the current WARC file. */
97 static char *warc_current_filename;
99 /* The serial number of the current WARC file. This number is
100 incremented each time a new file is opened and is used in the
101 WARC file's filename. */
102 static int warc_current_file_number;
104 /* The table of CDX records, if deduplication is enabled. */
105 struct hash_table * warc_cdx_dedup_table;
107 static bool warc_start_new_file (bool meta);
110 struct warc_cdx_record
114 char digest[SHA1_DIGEST_SIZE];
118 warc_hash_sha1_digest (const void *key)
120 /* We just use some of the first bytes of the digest. */
122 memcpy (&v, key, sizeof (unsigned long));
127 warc_cmp_sha1_digest (const void *digest1, const void *digest2)
129 return !memcmp (digest1, digest2, SHA1_DIGEST_SIZE);
134 /* Writes SIZE bytes from BUFFER to the current WARC file,
135 through gzwrite if compression is enabled.
136 Returns the number of uncompressed bytes written. */
138 warc_write_buffer (const char *buffer, size_t size)
141 if (warc_current_gzfile)
143 warc_current_gzfile_uncompressed_size += size;
144 return gzwrite (warc_current_gzfile, buffer, size);
148 return fwrite (buffer, 1, size, warc_current_file);
151 /* Writes STR to the current WARC file.
152 Returns false and set warc_write_ok to false if there
155 warc_write_string (const char *str)
160 size_t n = strlen (str);
161 if (n != warc_write_buffer (str, n))
162 warc_write_ok = false;
164 return warc_write_ok;
168 #define EXTRA_GZIP_HEADER_SIZE 12
169 #define GZIP_STATIC_HEADER_SIZE 10
170 #define FLG_FEXTRA 0x04
173 /* Starts a new WARC record. Writes the version header.
174 If opt.warc_maxsize is set and the current file is becoming
175 too large, this will open a new WARC file.
177 If compression is enabled, this will start a new
178 gzip stream in the current WARC file.
180 Returns false and set warc_write_ok to false if there
183 warc_write_start_record (void)
188 fflush (warc_current_file);
189 if (opt.warc_maxsize > 0 && ftello (warc_current_file) >= opt.warc_maxsize)
190 warc_start_new_file (false);
193 /* Start a GZIP stream, if required. */
194 if (opt.warc_compression_enabled)
196 /* Record the starting offset of the new record. */
197 warc_current_gzfile_offset = ftello (warc_current_file);
199 /* Reserve space for the extra GZIP header field.
200 In warc_write_end_record we will fill this space
201 with information about the uncompressed and
202 compressed size of the record. */
203 fprintf (warc_current_file, "XXXXXXXXXXXX");
204 fflush (warc_current_file);
206 /* Start a new GZIP stream. */
207 warc_current_gzfile = gzdopen (dup (fileno (warc_current_file)), "wb9");
208 warc_current_gzfile_uncompressed_size = 0;
210 if (warc_current_gzfile == NULL)
212 logprintf (LOG_NOTQUIET,
213 _("Error opening GZIP stream to WARC file.\n"));
214 warc_write_ok = false;
220 warc_write_string ("WARC/1.0\r\n");
221 return warc_write_ok;
224 /* Writes a WARC header to the current WARC record.
225 This method may be run after warc_write_start_record and
226 before warc_write_block_from_file. */
228 warc_write_header (const char *name, const char *value)
232 warc_write_string (name);
233 warc_write_string (": ");
234 warc_write_string (value);
235 warc_write_string ("\r\n");
237 return warc_write_ok;
240 /* Copies the contents of DATA_IN to the WARC record.
241 Adds a Content-Length header to the WARC record.
242 Run this method after warc_write_header,
243 then run warc_write_end_record. */
245 warc_write_block_from_file (FILE *data_in)
247 /* Add the Content-Length header. */
248 char content_length[22];
249 fseeko (data_in, 0L, SEEK_END);
250 number_to_string (content_length, ftello (data_in));
251 warc_write_header ("Content-Length", content_length);
253 /* End of the WARC header section. */
254 warc_write_string ("\r\n");
256 if (fseeko (data_in, 0L, SEEK_SET) != 0)
257 warc_write_ok = false;
259 /* Copy the data in the file to the WARC record. */
262 while (warc_write_ok && (s = fread (buffer, 1, BUFSIZ, data_in)) > 0)
264 if (warc_write_buffer (buffer, s) < s)
265 warc_write_ok = false;
268 return warc_write_ok;
271 /* Run this method to close the current WARC record.
273 If compression is enabled, this method closes the
274 current GZIP stream and fills the extra GZIP header
275 with the uncompressed and compressed length of the
278 warc_write_end_record (void)
280 warc_write_buffer ("\r\n\r\n", 4);
283 /* We start a new gzip stream for each record. */
284 if (warc_write_ok && warc_current_gzfile)
286 if (gzclose (warc_current_gzfile) != Z_OK)
288 warc_write_ok = false;
292 fflush (warc_current_file);
293 fseeko (warc_current_file, 0, SEEK_END);
295 /* The WARC standard suggests that we add 'skip length' data in the
296 extra header field of the GZIP stream.
298 In warc_write_start_record we reserved space for this extra header.
299 This extra space starts at warc_current_gzfile_offset and fills
300 EXTRA_GZIP_HEADER_SIZE bytes. The static GZIP header starts at
301 warc_current_gzfile_offset + EXTRA_GZIP_HEADER_SIZE.
303 We need to do three things:
304 1. Move the static GZIP header to warc_current_gzfile_offset;
305 2. Set the FEXTRA flag in the GZIP header;
306 3. Write the extra GZIP header after the static header, that is,
307 starting at warc_current_gzfile_offset + GZIP_STATIC_HEADER_SIZE.
310 /* Calculate the uncompressed and compressed sizes. */
311 off_t current_offset = ftello (warc_current_file);
312 off_t uncompressed_size = current_offset - warc_current_gzfile_offset;
313 off_t compressed_size = warc_current_gzfile_uncompressed_size;
315 /* Go back to the static GZIP header. */
316 fseeko (warc_current_file, warc_current_gzfile_offset
317 + EXTRA_GZIP_HEADER_SIZE, SEEK_SET);
319 /* Read the header. */
320 char static_header[GZIP_STATIC_HEADER_SIZE];
321 size_t result = fread (static_header, 1, GZIP_STATIC_HEADER_SIZE,
323 if (result != GZIP_STATIC_HEADER_SIZE)
325 warc_write_ok = false;
329 /* Set the FEXTRA flag in the flags byte of the header. */
330 static_header[OFF_FLG] = static_header[OFF_FLG] | FLG_FEXTRA;
332 /* Write the header back to the file, but starting at
333 warc_current_gzfile_offset. */
334 fseeko (warc_current_file, warc_current_gzfile_offset, SEEK_SET);
335 fwrite (static_header, 1, GZIP_STATIC_HEADER_SIZE, warc_current_file);
337 /* Prepare the extra GZIP header. */
338 char extra_header[EXTRA_GZIP_HEADER_SIZE];
339 /* XLEN, the length of the extra header fields. */
340 extra_header[0] = ((EXTRA_GZIP_HEADER_SIZE - 2) & 255);
341 extra_header[1] = ((EXTRA_GZIP_HEADER_SIZE - 2) >> 8) & 255;
342 /* The extra header field identifier for the WARC skip length. */
343 extra_header[2] = 's';
344 extra_header[3] = 'l';
345 /* The size of the uncompressed record. */
346 extra_header[4] = (uncompressed_size & 255);
347 extra_header[5] = (uncompressed_size >> 8) & 255;
348 extra_header[6] = (uncompressed_size >> 16) & 255;
349 extra_header[7] = (uncompressed_size >> 24) & 255;
350 /* The size of the compressed record. */
351 extra_header[8] = (compressed_size & 255);
352 extra_header[9] = (compressed_size >> 8) & 255;
353 extra_header[10] = (compressed_size >> 16) & 255;
354 extra_header[11] = (compressed_size >> 24) & 255;
356 /* Write the extra header after the static header. */
357 fseeko (warc_current_file, warc_current_gzfile_offset
358 + GZIP_STATIC_HEADER_SIZE, SEEK_SET);
359 fwrite (extra_header, 1, EXTRA_GZIP_HEADER_SIZE, warc_current_file);
361 /* Done, move back to the end of the file. */
362 fflush (warc_current_file);
363 fseeko (warc_current_file, 0, SEEK_END);
365 #endif /* HAVE_LIBZ */
367 return warc_write_ok;
371 /* Writes the WARC-Date header for the given timestamp to
372 the current WARC record.
373 If timestamp is NULL, the current time will be used. */
375 warc_write_date_header (const char *timestamp)
377 if (timestamp == NULL)
379 char current_timestamp[21];
380 warc_timestamp (current_timestamp);
381 timestamp = current_timestamp;
383 return warc_write_header ("WARC-Date", timestamp);
386 /* Writes the WARC-IP-Address header for the given IP to
387 the current WARC record. If IP is NULL, no header will
390 warc_write_ip_header (ip_address *ip)
393 return warc_write_header ("WARC-IP-Address", print_address (ip));
395 return warc_write_ok;
399 /* warc_sha1_stream_with_payload is a modified copy of sha1_stream
400 from gnulib/sha1.c. This version calculates two digests in one go.
402 Compute SHA1 message digests for bytes read from STREAM. The
403 digest of the complete file will be written into the 16 bytes
404 beginning at RES_BLOCK.
406 If payload_offset >= 0, a second digest will be calculated of the
407 portion of the file starting at payload_offset and continuing to
408 the end of the file. The digest number will be written into the
409 16 bytes beginning ad RES_PAYLOAD. */
411 warc_sha1_stream_with_payload (FILE *stream, void *res_block, void *res_payload,
412 off_t payload_offset)
414 #define BLOCKSIZE 32768
416 struct sha1_ctx ctx_block;
417 struct sha1_ctx ctx_payload;
421 char *buffer = malloc (BLOCKSIZE + 72);
425 /* Initialize the computation context. */
426 sha1_init_ctx (&ctx_block);
427 if (payload_offset >= 0)
428 sha1_init_ctx (&ctx_payload);
432 /* Iterate over full file contents. */
435 /* We read the file in blocks of BLOCKSIZE bytes. One call of the
436 computation function processes the whole buffer so that with the
437 next round of the loop another block can be read. */
441 /* Read block. Take care for partial reads. */
444 n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream);
449 if (sum == BLOCKSIZE)
454 /* Check for the error flag IFF N == 0, so that we don't
455 exit the loop after a partial read due to e.g., EAGAIN
462 goto process_partial_block;
465 /* We've read at least one byte, so ignore errors. But always
466 check for EOF, since feof may be true even though N > 0.
467 Otherwise, we could end up calling fread after EOF. */
469 goto process_partial_block;
472 /* Process buffer with BLOCKSIZE bytes. Note that
475 sha1_process_block (buffer, BLOCKSIZE, &ctx_block);
476 if (payload_offset >= 0 && payload_offset < pos)
478 /* At least part of the buffer contains data from payload. */
479 off_t start_of_payload = payload_offset - (pos - BLOCKSIZE);
480 if (start_of_payload <= 0)
481 /* All bytes in the buffer belong to the payload. */
482 start_of_payload = 0;
484 /* Process the payload part of the buffer.
485 Note: we can't use sha1_process_block here even if we
486 process the complete buffer. Because the payload doesn't
487 have to start with a full block, there may still be some
488 bytes left from the previous buffer. Therefore, we need
489 to continue with sha1_process_bytes. */
490 sha1_process_bytes (buffer + start_of_payload,
491 BLOCKSIZE - start_of_payload, &ctx_payload);
495 process_partial_block:;
497 /* Process any remaining bytes. */
500 sha1_process_bytes (buffer, sum, &ctx_block);
501 if (payload_offset >= 0 && payload_offset < pos)
503 /* At least part of the buffer contains data from payload. */
504 off_t start_of_payload = payload_offset - (pos - sum);
505 if (start_of_payload <= 0)
506 /* All bytes in the buffer belong to the payload. */
507 start_of_payload = 0;
509 /* Process the payload part of the buffer. */
510 sha1_process_bytes (buffer + start_of_payload,
511 sum - start_of_payload, &ctx_payload);
515 /* Construct result in desired memory. */
516 sha1_finish_ctx (&ctx_block, res_block);
517 if (payload_offset >= 0)
518 sha1_finish_ctx (&ctx_payload, res_payload);
525 /* Converts the SHA1 digest to a base32-encoded string.
526 "sha1:DIGEST\0" (Allocates a new string for the response.) */
528 warc_base32_sha1_digest (char *sha1_digest)
530 /* length: "sha1:" + digest + "\0" */
531 char *sha1_base32 = malloc (BASE32_LENGTH(SHA1_DIGEST_SIZE) + 1 + 5 );
532 base32_encode (sha1_digest, SHA1_DIGEST_SIZE, sha1_base32 + 5,
533 BASE32_LENGTH(SHA1_DIGEST_SIZE) + 1);
534 memcpy (sha1_base32, "sha1:", 5);
535 sha1_base32[BASE32_LENGTH(SHA1_DIGEST_SIZE) + 5] = '\0';
540 /* Sets the digest headers of the record.
541 This method will calculate the block digest and, if payload_offset >= 0,
542 will also calculate the payload digest of the payload starting at the
545 warc_write_digest_headers (FILE *file, long payload_offset)
547 if (opt.warc_digests_enabled)
549 /* Calculate the block and payload digests. */
550 char sha1_res_block[SHA1_DIGEST_SIZE];
551 char sha1_res_payload[SHA1_DIGEST_SIZE];
554 if (warc_sha1_stream_with_payload (file, sha1_res_block,
555 sha1_res_payload, payload_offset) == 0)
559 digest = warc_base32_sha1_digest (sha1_res_block);
560 warc_write_header ("WARC-Block-Digest", digest);
563 if (payload_offset >= 0)
565 digest = warc_base32_sha1_digest (sha1_res_payload);
566 warc_write_header ("WARC-Payload-Digest", digest);
574 /* Fills timestamp with the current time and date.
575 The UTC time is formatted following ISO 8601, as required
576 for use in the WARC-Date header.
577 The timestamp will be 21 characters long. */
579 warc_timestamp (char *timestamp)
582 struct tm * timeinfo;
584 timeinfo = gmtime (&rawtime);
585 strftime (timestamp, 21, "%Y-%m-%dT%H:%M:%SZ", timeinfo);
589 /* Fills urn_str with a UUID in the format required
590 for the WARC-Record-Id header.
591 The string will be 47 characters long. */
593 warc_uuid_str (char *urn_str)
598 uuid_generate (record_id);
599 uuid_unparse (record_id, uuid_str);
601 sprintf (urn_str, "<urn:uuid:%s>", uuid_str);
604 /* Fills urn_str with a UUID based on random numbers in the format
605 required for the WARC-Record-Id header.
606 (See RFC 4122, UUID version 4.)
608 Note: this is a fallback method, it is much better to use the
609 methods provided by libuuid.
611 The string will be 47 characters long. */
613 warc_uuid_str (char *urn_str)
615 // RFC 4122, a version 4 UUID with only random numbers
617 unsigned char uuid_data[16];
620 uuid_data[i] = random_number (255);
622 // Set the four most significant bits (bits 12 through 15) of the
623 // time_hi_and_version field to the 4-bit version number
624 uuid_data[6] = (uuid_data[6] & 0x0F) | 0x40;
626 // Set the two most significant bits (bits 6 and 7) of the
627 // clock_seq_hi_and_reserved to zero and one, respectively.
628 uuid_data[8] = (uuid_data[8] & 0xBF) | 0x80;
631 "<urn:uuid:%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x>",
632 uuid_data[0], uuid_data[1], uuid_data[2], uuid_data[3], uuid_data[4],
633 uuid_data[5], uuid_data[6], uuid_data[7], uuid_data[8], uuid_data[9],
634 uuid_data[10], uuid_data[11], uuid_data[12], uuid_data[13], uuid_data[14],
639 /* Write a warcinfo record to the current file.
640 Updates warc_current_warcinfo_uuid_str. */
642 warc_write_warcinfo_record (char *filename)
644 /* Write warc-info record as the first record of the file. */
645 /* We add the record id of this info record to the other records in the
647 warc_current_warcinfo_uuid_str = (char *) malloc (48);
648 warc_uuid_str (warc_current_warcinfo_uuid_str);
651 warc_timestamp (timestamp);
653 char *filename_copy, *filename_basename;
654 filename_copy = strdup (filename);
655 filename_basename = strdup (basename (filename_copy));
657 warc_write_start_record ();
658 warc_write_header ("WARC-Type", "warcinfo");
659 warc_write_header ("Content-Type", "application/warc-fields");
660 warc_write_header ("WARC-Date", timestamp);
661 warc_write_header ("WARC-Record-ID", warc_current_warcinfo_uuid_str);
662 warc_write_header ("WARC-Filename", filename_basename);
664 /* Create content. */
665 FILE *warc_tmp = warc_tempfile ();
666 if (warc_tmp == NULL)
668 free (filename_copy);
669 free (filename_basename);
673 fprintf (warc_tmp, "software: Wget/%s (%s)\r\n", version_string, OS_TYPE);
674 fprintf (warc_tmp, "format: WARC File Format 1.0\r\n");
676 "conformsTo: http://bibnum.bnf.fr/WARC/WARC_ISO_28500_version1_latestdraft.pdf\r\n");
677 fprintf (warc_tmp, "robots: %s\r\n", (opt.use_robots ? "classic" : "off"));
678 fprintf (warc_tmp, "wget-arguments: %s\r\n", program_argstring);
679 /* Add the user headers, if any. */
680 if (opt.warc_user_headers)
683 for (i = 0; opt.warc_user_headers[i]; i++)
684 fprintf (warc_tmp, "%s\r\n", opt.warc_user_headers[i]);
686 fprintf(warc_tmp, "\r\n");
688 warc_write_digest_headers (warc_tmp, -1);
689 warc_write_block_from_file (warc_tmp);
690 warc_write_end_record ();
693 logprintf (LOG_NOTQUIET, _("Error writing warcinfo record to WARC file.\n"));
695 free (filename_copy);
696 free (filename_basename);
698 return warc_write_ok;
701 /* Opens a new WARC file.
702 If META is true, generates a filename ending with 'meta.warc.gz'.
705 1. close the current WARC file (if there is one);
706 2. increment warc_current_file_number;
707 3. open a new WARC file;
708 4. write the initial warcinfo record.
710 Returns true on success, false otherwise.
713 warc_start_new_file (bool meta)
715 if (opt.warc_filename == NULL)
718 if (warc_current_file != NULL)
719 fclose (warc_current_file);
720 if (warc_current_warcinfo_uuid_str)
721 free (warc_current_warcinfo_uuid_str);
722 if (warc_current_filename)
723 free (warc_current_filename);
725 warc_current_file_number++;
727 int base_filename_length = strlen (opt.warc_filename);
728 /* filename format: base + "-" + 5 digit serial number + ".warc.gz" */
729 char *new_filename = malloc (base_filename_length + 1 + 5 + 8 + 1);
730 warc_current_filename = new_filename;
733 const char *extension = (opt.warc_compression_enabled ? "warc.gz" : "warc");
735 const char *extension = "warc";
738 /* If max size is enabled, we add a serial number to the file names. */
740 sprintf (new_filename, "%s-meta.%s", opt.warc_filename, extension);
741 else if (opt.warc_maxsize > 0)
743 sprintf (new_filename, "%s-%05d.%s", opt.warc_filename,
744 warc_current_file_number, extension);
747 sprintf (new_filename, "%s.%s", opt.warc_filename, extension);
749 logprintf (LOG_VERBOSE, _("Opening WARC file %s.\n\n"), quote (new_filename));
751 /* Open the WARC file. */
752 warc_current_file = fopen (new_filename, "wb+");
753 if (warc_current_file == NULL)
755 logprintf (LOG_NOTQUIET, _("Error opening WARC file %s.\n"),
756 quote (new_filename));
760 if (! warc_write_warcinfo_record (new_filename))
763 /* Add warcinfo uuid to manifest. */
764 if (warc_manifest_fp)
765 fprintf (warc_manifest_fp, "%s\n", warc_current_warcinfo_uuid_str);
770 /* Opens the CDX file for output. */
772 warc_start_cdx_file (void)
774 int filename_length = strlen (opt.warc_filename);
775 char *cdx_filename = alloca (filename_length + 4 + 1);
776 memcpy (cdx_filename, opt.warc_filename, filename_length);
777 memcpy (cdx_filename + filename_length, ".cdx", 5);
778 warc_current_cdx_file = fopen (cdx_filename, "a+");
779 if (warc_current_cdx_file == NULL)
782 /* Print the CDX header.
788 * k - new style checksum
791 * V - compressed arc file offset
795 fprintf (warc_current_cdx_file, " CDX a b a m s k r M V g u\n");
796 fflush (warc_current_cdx_file);
801 #define CDX_FIELDSEP " \t\r\n"
803 /* Parse the CDX header and find the field numbers of the original url,
804 checksum and record ID fields. */
806 warc_parse_cdx_header (char *lineptr, int *field_num_original_url,
807 int *field_num_checksum, int *field_num_record_id)
809 *field_num_original_url = -1;
810 *field_num_checksum = -1;
811 *field_num_record_id = -1;
815 token = strtok_r (lineptr, CDX_FIELDSEP, &save_ptr);
817 if (token != NULL && strcmp (token, "CDX") == 0)
820 while (token != NULL)
822 token = strtok_r (NULL, CDX_FIELDSEP, &save_ptr);
828 *field_num_original_url = field_num;
831 *field_num_checksum = field_num;
834 *field_num_record_id = field_num;
842 return *field_num_original_url != -1
843 && *field_num_checksum != -1
844 && *field_num_record_id != -1;
847 /* Parse the CDX record and add it to the warc_cdx_dedup_table hash table. */
849 warc_process_cdx_line (char *lineptr, int field_num_original_url,
850 int field_num_checksum, int field_num_record_id)
852 char *original_url = NULL;
853 char *checksum = NULL;
854 char *record_id = NULL;
858 token = strtok_r (lineptr, CDX_FIELDSEP, &save_ptr);
860 /* Read this line to get the fields we need. */
862 while (token != NULL)
865 if (field_num == field_num_original_url)
867 else if (field_num == field_num_checksum)
869 else if (field_num == field_num_record_id)
875 *val = strdup (token);
877 token = strtok_r (NULL, CDX_FIELDSEP, &save_ptr);
881 if (original_url != NULL && checksum != NULL && record_id != NULL)
883 /* For some extra efficiency, we decode the base32 encoded
884 checksum value. This should produce exactly SHA1_DIGEST_SIZE
888 base32_decode_alloc (checksum, strlen (checksum), &checksum_v,
892 if (checksum_v != NULL && checksum_l == SHA1_DIGEST_SIZE)
894 /* This is a valid line with a valid checksum. */
895 struct warc_cdx_record *rec;
896 rec = malloc (sizeof (struct warc_cdx_record));
897 rec->url = original_url;
898 rec->uuid = record_id;
899 memcpy (rec->digest, checksum_v, SHA1_DIGEST_SIZE);
900 hash_table_put (warc_cdx_dedup_table, rec->digest, rec);
906 if (checksum_v != NULL)
913 xfree_null(checksum);
914 xfree_null(original_url);
915 xfree_null(record_id);
919 /* Loads the CDX file from opt.warc_cdx_dedup_filename and fills
920 the warc_cdx_dedup_table. */
922 warc_load_cdx_dedup_file (void)
924 FILE *f = fopen (opt.warc_cdx_dedup_filename, "r");
928 int field_num_original_url = -1;
929 int field_num_checksum = -1;
930 int field_num_record_id = -1;
932 char *lineptr = NULL;
936 /* The first line should contain the CDX header.
937 Format: " CDX x x x x x"
938 where x are field type indicators. For our purposes, we only
939 need 'a' (the original url), 'k' (the SHA1 checksum) and
940 'u' (the WARC record id). */
941 line_length = getline (&lineptr, &n, f);
942 if (line_length != -1)
943 warc_parse_cdx_header (lineptr, &field_num_original_url,
944 &field_num_checksum, &field_num_record_id);
946 /* If the file contains all three fields, read the complete file. */
947 if (field_num_original_url == -1
948 || field_num_checksum == -1
949 || field_num_record_id == -1)
951 if (field_num_original_url == -1)
952 logprintf (LOG_NOTQUIET,
953 _("CDX file does not list original urls. (Missing column 'a'.)\n"));
954 if (field_num_checksum == -1)
955 logprintf (LOG_NOTQUIET,
956 _("CDX file does not list checksums. (Missing column 'k'.)\n"));
957 if (field_num_record_id == -1)
958 logprintf (LOG_NOTQUIET,
959 _("CDX file does not list record ids. (Missing column 'u'.)\n"));
963 /* Initialize the table. */
964 warc_cdx_dedup_table = hash_table_new (1000, warc_hash_sha1_digest,
965 warc_cmp_sha1_digest);
969 line_length = getline (&lineptr, &n, f);
970 if (line_length != -1)
972 warc_process_cdx_line (lineptr, field_num_original_url,
973 field_num_checksum, field_num_record_id);
977 while (line_length != -1);
980 int nrecords = hash_table_count (warc_cdx_dedup_table);
981 logprintf (LOG_VERBOSE, ngettext ("Loaded %d record from CDX.\n\n",
982 "Loaded %d records from CDX.\n\n",
994 /* Returns the existing duplicate CDX record for the given url and payload
995 digest. Returns NULL if the url is not found or if the payload digest
996 does not match, or if CDX deduplication is disabled. */
997 static struct warc_cdx_record *
998 warc_find_duplicate_cdx_record (char *url, char *sha1_digest_payload)
1000 if (warc_cdx_dedup_table == NULL)
1003 struct warc_cdx_record *rec_existing
1004 = hash_table_get (warc_cdx_dedup_table, sha1_digest_payload);
1006 if (rec_existing && strcmp (rec_existing->url, url) == 0)
1007 return rec_existing;
1012 /* Initializes the WARC writer (if opt.warc_filename is set).
1013 This should be called before any WARC record is written. */
1017 warc_write_ok = true;
1019 if (opt.warc_filename != NULL)
1021 if (opt.warc_cdx_dedup_filename != NULL)
1023 if (! warc_load_cdx_dedup_file ())
1025 logprintf (LOG_NOTQUIET,
1026 _("Could not read CDX file %s for deduplication.\n"),
1027 quote (opt.warc_cdx_dedup_filename));
1032 warc_manifest_fp = warc_tempfile ();
1033 if (warc_manifest_fp == NULL)
1035 logprintf (LOG_NOTQUIET,
1036 _("Could not open temporary WARC manifest file.\n"));
1040 if (opt.warc_keep_log)
1042 warc_log_fp = warc_tempfile ();
1043 if (warc_log_fp == NULL)
1045 logprintf (LOG_NOTQUIET,
1046 _("Could not open temporary WARC log file.\n"));
1049 log_set_warc_log_fp (warc_log_fp);
1052 warc_current_file_number = -1;
1053 if (! warc_start_new_file (false))
1055 logprintf (LOG_NOTQUIET, _("Could not open WARC file.\n"));
1059 if (opt.warc_cdx_enabled)
1061 if (! warc_start_cdx_file ())
1063 logprintf (LOG_NOTQUIET,
1064 _("Could not open CDX file for output.\n"));
1071 /* Writes metadata (manifest, configuration, log file) to the WARC file. */
1073 warc_write_metadata (void)
1075 /* If there are multiple WARC files, the metadata should be written to a separate file. */
1076 if (opt.warc_maxsize > 0)
1077 warc_start_new_file (true);
1079 char manifest_uuid [48];
1080 warc_uuid_str (manifest_uuid);
1082 fflush (warc_manifest_fp);
1083 warc_write_resource_record (manifest_uuid,
1084 "metadata://gnu.org/software/wget/warc/MANIFEST.txt",
1085 NULL, NULL, NULL, "text/plain",
1086 warc_manifest_fp, -1);
1087 /* warc_write_resource_record has closed warc_manifest_fp. */
1089 FILE * warc_tmp_fp = warc_tempfile ();
1090 if (warc_tmp_fp == NULL)
1092 logprintf (LOG_NOTQUIET, _("Could not open temporary WARC file.\n"));
1095 fflush (warc_tmp_fp);
1096 fprintf (warc_tmp_fp, "%s\n", program_argstring);
1098 warc_write_resource_record (manifest_uuid,
1099 "metadata://gnu.org/software/wget/warc/wget_arguments.txt",
1100 NULL, NULL, NULL, "text/plain",
1102 /* warc_write_resource_record has closed warc_tmp_fp. */
1104 if (warc_log_fp != NULL)
1106 warc_write_resource_record (NULL,
1107 "metadata://gnu.org/software/wget/warc/wget.log",
1108 NULL, manifest_uuid, NULL, "text/plain",
1110 /* warc_write_resource_record has closed warc_log_fp. */
1113 log_set_warc_log_fp (NULL);
1117 /* Finishes the WARC writing.
1118 This should be called at the end of the program. */
1122 if (warc_current_file != NULL)
1124 warc_write_metadata ();
1125 free (warc_current_warcinfo_uuid_str);
1126 fclose (warc_current_file);
1128 if (warc_current_cdx_file != NULL)
1129 fclose (warc_current_cdx_file);
1130 if (warc_log_fp != NULL)
1132 fclose (warc_log_fp);
1133 log_set_warc_log_fp (NULL);
1137 /* Creates a temporary file for writing WARC output.
1138 The temporary file will be created in opt.warc_tempdir.
1139 Returns the pointer to the temporary file, or NULL. */
1141 warc_tempfile (void)
1144 if (path_search (filename, 100, opt.warc_tempdir, "wget", true) == -1)
1147 int fd = mkstemp (filename);
1151 if (unlink (filename) < 0)
1154 return fdopen (fd, "wb+");
1158 /* Writes a request record to the WARC file.
1159 url is the target uri of the request,
1160 timestamp_str is the timestamp of the request (generated with warc_timestamp),
1161 record_uuid is the uuid of the request (generated with warc_uuid_str),
1162 body is a pointer to a file containing the request headers and body.
1163 ip is the ip address of the server (or NULL),
1164 Calling this function will close body.
1165 Returns true on success, false on error. */
1167 warc_write_request_record (char *url, char *timestamp_str, char *record_uuid,
1168 ip_address *ip, FILE *body, off_t payload_offset)
1170 warc_write_start_record ();
1171 warc_write_header ("WARC-Type", "request");
1172 warc_write_header ("WARC-Target-URI", url);
1173 warc_write_header ("Content-Type", "application/http;msgtype=request");
1174 warc_write_date_header (timestamp_str);
1175 warc_write_header ("WARC-Record-ID", record_uuid);
1176 warc_write_ip_header (ip);
1177 warc_write_header ("WARC-Warcinfo-ID", warc_current_warcinfo_uuid_str);
1178 warc_write_digest_headers (body, payload_offset);
1179 warc_write_block_from_file (body);
1180 warc_write_end_record ();
1184 return warc_write_ok;
1187 /* Writes a response record to the CDX file.
1188 url is the target uri of the request/response,
1189 timestamp_str is the timestamp of the request that generated this response,
1190 (generated with warc_timestamp),
1191 mime_type is the mime type of the response body (will be printed to CDX),
1192 response_code is the HTTP response code (will be printed to CDX),
1193 payload_digest is the sha1 digest of the payload,
1194 redirect_location is the contents of the Location: header, or NULL (will be printed to CDX),
1195 offset is the position of the WARC record in the WARC file,
1196 warc_filename is the filename of the WARC,
1197 response_uuid is the uuid of the response.
1198 Returns true on success, false on error. */
1200 warc_write_cdx_record (const char *url, const char *timestamp_str,
1201 const char *mime_type, int response_code,
1202 const char *payload_digest, const char *redirect_location,
1203 off_t offset, const char *warc_filename,
1204 const char *response_uuid)
1206 /* Transform the timestamp. */
1207 char timestamp_str_cdx [15];
1208 memcpy (timestamp_str_cdx , timestamp_str , 4); /* "YYYY" "-" */
1209 memcpy (timestamp_str_cdx + 4, timestamp_str + 5, 2); /* "mm" "-" */
1210 memcpy (timestamp_str_cdx + 6, timestamp_str + 8, 2); /* "dd" "T" */
1211 memcpy (timestamp_str_cdx + 8, timestamp_str + 11, 2); /* "HH" ":" */
1212 memcpy (timestamp_str_cdx + 10, timestamp_str + 14, 2); /* "MM" ":" */
1213 memcpy (timestamp_str_cdx + 12, timestamp_str + 17, 2); /* "SS" "Z" */
1214 timestamp_str_cdx[14] = '\0';
1216 /* Rewrite the checksum. */
1217 const char *checksum;
1218 if (payload_digest != NULL)
1219 checksum = payload_digest + 5; /* Skip the "sha1:" */
1223 if (mime_type == NULL || strlen(mime_type) == 0)
1225 if (redirect_location == NULL || strlen(redirect_location) == 0)
1226 redirect_location = "-";
1228 char offset_string[22];
1229 number_to_string (offset_string, offset);
1231 /* Print the CDX line. */
1232 fprintf (warc_current_cdx_file, "%s %s %s %s %d %s %s - %s %s %s\n", url,
1233 timestamp_str_cdx, url, mime_type, response_code, checksum,
1234 redirect_location, offset_string, warc_current_filename,
1236 fflush (warc_current_cdx_file);
1241 /* Writes a revisit record to the WARC file.
1242 url is the target uri of the request/response,
1243 timestamp_str is the timestamp of the request that generated this response
1244 (generated with warc_timestamp),
1245 concurrent_to_uuid is the uuid of the request for that generated this response
1246 (generated with warc_uuid_str),
1247 refers_to_uuid is the uuid of the original response
1248 (generated with warc_uuid_str),
1249 payload_digest is the sha1 digest of the payload,
1250 ip is the ip address of the server (or NULL),
1251 body is a pointer to a file containing the response headers (without payload).
1252 Calling this function will close body.
1253 Returns true on success, false on error. */
1255 warc_write_revisit_record (char *url, char *timestamp_str,
1256 char *concurrent_to_uuid, char *payload_digest,
1257 char *refers_to, ip_address *ip, FILE *body)
1259 char revisit_uuid [48];
1260 warc_uuid_str (revisit_uuid);
1262 char *block_digest = NULL;
1263 char sha1_res_block[SHA1_DIGEST_SIZE];
1264 sha1_stream (body, sha1_res_block);
1265 block_digest = warc_base32_sha1_digest (sha1_res_block);
1267 warc_write_start_record ();
1268 warc_write_header ("WARC-Type", "revisit");
1269 warc_write_header ("WARC-Record-ID", revisit_uuid);
1270 warc_write_header ("WARC-Warcinfo-ID", warc_current_warcinfo_uuid_str);
1271 warc_write_header ("WARC-Concurrent-To", concurrent_to_uuid);
1272 warc_write_header ("WARC-Refers-To", refers_to);
1273 warc_write_header ("WARC-Profile", "http://netpreserve.org/warc/1.0/revisit/identical-payload-digest");
1274 warc_write_header ("WARC-Truncated", "length");
1275 warc_write_header ("WARC-Target-URI", url);
1276 warc_write_date_header (timestamp_str);
1277 warc_write_ip_header (ip);
1278 warc_write_header ("Content-Type", "application/http;msgtype=response");
1279 warc_write_header ("WARC-Block-Digest", block_digest);
1280 warc_write_header ("WARC-Payload-Digest", payload_digest);
1281 warc_write_block_from_file (body);
1282 warc_write_end_record ();
1285 free (block_digest);
1287 return warc_write_ok;
1290 /* Writes a response record to the WARC file.
1291 url is the target uri of the request/response,
1292 timestamp_str is the timestamp of the request that generated this response
1293 (generated with warc_timestamp),
1294 concurrent_to_uuid is the uuid of the request for that generated this response
1295 (generated with warc_uuid_str),
1296 ip is the ip address of the server (or NULL),
1297 body is a pointer to a file containing the response headers and body.
1298 mime_type is the mime type of the response body (will be printed to CDX),
1299 response_code is the HTTP response code (will be printed to CDX),
1300 redirect_location is the contents of the Location: header, or NULL (will be printed to CDX),
1301 Calling this function will close body.
1302 Returns true on success, false on error. */
1304 warc_write_response_record (char *url, char *timestamp_str,
1305 char *concurrent_to_uuid, ip_address *ip,
1306 FILE *body, off_t payload_offset, char *mime_type,
1307 int response_code, char *redirect_location)
1309 char *block_digest = NULL;
1310 char *payload_digest = NULL;
1311 char sha1_res_block[SHA1_DIGEST_SIZE];
1312 char sha1_res_payload[SHA1_DIGEST_SIZE];
1314 if (opt.warc_digests_enabled)
1316 /* Calculate the block and payload digests. */
1318 if (warc_sha1_stream_with_payload (body, sha1_res_block, sha1_res_payload,
1319 payload_offset) == 0)
1321 /* Decide (based on url + payload digest) if we have seen this
1323 struct warc_cdx_record *rec_existing;
1324 rec_existing = warc_find_duplicate_cdx_record (url, sha1_res_payload);
1325 if (rec_existing != NULL)
1329 /* Found an existing record. */
1330 logprintf (LOG_VERBOSE,
1331 _("Found exact match in CDX file. Saving revisit record to WARC.\n"));
1333 /* Remove the payload from the file. */
1334 if (payload_offset > 0)
1336 if (ftruncate (fileno (body), payload_offset) == -1)
1340 /* Send the original payload digest. */
1341 payload_digest = warc_base32_sha1_digest (sha1_res_payload);
1342 result = warc_write_revisit_record (url, timestamp_str,
1343 concurrent_to_uuid, payload_digest, rec_existing->uuid,
1345 free (payload_digest);
1350 block_digest = warc_base32_sha1_digest (sha1_res_block);
1351 payload_digest = warc_base32_sha1_digest (sha1_res_payload);
1355 /* Not a revisit, just store the record. */
1357 char response_uuid [48];
1358 warc_uuid_str (response_uuid);
1360 fseeko (warc_current_file, 0L, SEEK_END);
1361 off_t offset = ftello (warc_current_file);
1363 warc_write_start_record ();
1364 warc_write_header ("WARC-Type", "response");
1365 warc_write_header ("WARC-Record-ID", response_uuid);
1366 warc_write_header ("WARC-Warcinfo-ID", warc_current_warcinfo_uuid_str);
1367 warc_write_header ("WARC-Concurrent-To", concurrent_to_uuid);
1368 warc_write_header ("WARC-Target-URI", url);
1369 warc_write_date_header (timestamp_str);
1370 warc_write_ip_header (ip);
1371 warc_write_header ("WARC-Block-Digest", block_digest);
1372 warc_write_header ("WARC-Payload-Digest", payload_digest);
1373 warc_write_header ("Content-Type", "application/http;msgtype=response");
1374 warc_write_block_from_file (body);
1375 warc_write_end_record ();
1379 if (warc_write_ok && opt.warc_cdx_enabled)
1381 /* Add this record to the CDX. */
1382 warc_write_cdx_record (url, timestamp_str, mime_type, response_code,
1383 payload_digest, redirect_location, offset, warc_current_filename,
1388 free (block_digest);
1390 free (payload_digest);
1392 return warc_write_ok;
1395 /* Writes a resource record to the WARC file.
1396 resource_uuid is the uuid of the resource (or NULL),
1397 url is the target uri of the resource,
1398 timestamp_str is the timestamp (generated with warc_timestamp),
1399 concurrent_to_uuid is the uuid of the request for that generated this
1400 resource (generated with warc_uuid_str) or NULL,
1401 ip is the ip address of the server (or NULL),
1402 content_type is the mime type of the body (or NULL),
1403 body is a pointer to a file containing the resource data.
1404 Calling this function will close body.
1405 Returns true on success, false on error. */
1407 warc_write_resource_record (char *resource_uuid, const char *url,
1408 const char *timestamp_str, const char *concurrent_to_uuid,
1409 ip_address *ip, const char *content_type, FILE *body,
1410 off_t payload_offset)
1412 if (resource_uuid == NULL)
1414 resource_uuid = alloca (48);
1415 warc_uuid_str (resource_uuid);
1418 if (content_type == NULL)
1419 content_type = "application/octet-stream";
1421 warc_write_start_record ();
1422 warc_write_header ("WARC-Type", "resource");
1423 warc_write_header ("WARC-Record-ID", resource_uuid);
1424 warc_write_header ("WARC-Warcinfo-ID", warc_current_warcinfo_uuid_str);
1425 warc_write_header ("WARC-Concurrent-To", concurrent_to_uuid);
1426 warc_write_header ("WARC-Target-URI", url);
1427 warc_write_date_header (timestamp_str);
1428 warc_write_ip_header (ip);
1429 warc_write_digest_headers (body, payload_offset);
1430 warc_write_header ("Content-Type", content_type);
1431 warc_write_block_from_file (body);
1432 warc_write_end_record ();
1436 return warc_write_ok;