2 This file is part of GNUnet.
3 (C) 2003, 2004, 2005, 2006, 2008, 2009 Christian Grothoff (and other contributing authors)
5 GNUnet is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published
7 by the Free Software Foundation; either version 2, or (at your
8 option) any later version.
10 GNUnet is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with GNUnet; see the file COPYING. If not, write to the
17 Free Software Foundation, Inc., 59 Temple Place - Suite 330,
18 Boston, MA 02111-1307, USA.
22 * @file util/container_meta_data.c
23 * @brief Storing of meta data
24 * @author Christian Grothoff
28 #include "gnunet_common.h"
29 #include "gnunet_container_lib.h"
30 #include "gnunet_strings_lib.h"
31 #include "gnunet_time_lib.h"
32 #include <extractor.h>
35 #define EXTRA_CHECKS ALLOW_EXTRA_CHECKS
39 EXTRACTOR_KeywordType type;
44 * Meta data to associate with a file, directory or namespace.
46 struct GNUNET_CONTAINER_MetaData
53 * Create a fresh struct CONTAINER_MetaData token.
55 struct GNUNET_CONTAINER_MetaData *
56 GNUNET_CONTAINER_meta_data_create ()
58 struct GNUNET_CONTAINER_MetaData *ret;
59 ret = GNUNET_malloc (sizeof (struct GNUNET_CONTAINER_MetaData));
69 GNUNET_CONTAINER_meta_data_destroy (struct GNUNET_CONTAINER_MetaData *md)
75 for (i = 0; i < md->itemCount; i++)
76 GNUNET_free (md->items[i].data);
77 GNUNET_array_grow (md->items, md->itemCount, 0);
82 * Add the current time as the publication date
86 GNUNET_CONTAINER_meta_data_add_publication_date (struct
87 GNUNET_CONTAINER_MetaData
91 struct GNUNET_TIME_Absolute t;
93 t = GNUNET_TIME_absolute_get ();
94 GNUNET_CONTAINER_meta_data_delete (md, EXTRACTOR_PUBLICATION_DATE, NULL);
95 dat = GNUNET_STRINGS_absolute_time_to_string (t);
96 GNUNET_CONTAINER_meta_data_insert (md, EXTRACTOR_PUBLICATION_DATE, dat);
102 * @return GNUNET_OK on success, GNUNET_SYSERR if this entry already exists
105 GNUNET_CONTAINER_meta_data_insert (struct GNUNET_CONTAINER_MetaData *md,
106 EXTRACTOR_KeywordType type,
112 GNUNET_assert (data != NULL);
113 for (idx = 0; idx < md->itemCount; idx++)
115 if ((md->items[idx].type == type) &&
116 (0 == strcmp (md->items[idx].data, data)))
117 return GNUNET_SYSERR;
120 GNUNET_array_grow (md->items, md->itemCount, md->itemCount + 1);
121 md->items[idx].type = type;
122 md->items[idx].data = p = GNUNET_strdup (data);
124 /* change OS native dir separators to unix '/' and others to '_' */
125 if (type == EXTRACTOR_FILENAME)
129 if (*p == DIR_SEPARATOR)
142 * @return GNUNET_OK on success, GNUNET_SYSERR if the item does not exist in md
145 GNUNET_CONTAINER_meta_data_delete (struct GNUNET_CONTAINER_MetaData *md,
146 EXTRACTOR_KeywordType type,
150 int ret = GNUNET_SYSERR;
151 for (idx = 0; idx < md->itemCount; idx++)
153 if ((md->items[idx].type == type) &&
154 ((data == NULL) || (0 == strcmp (md->items[idx].data, data))))
156 GNUNET_free (md->items[idx].data);
157 md->items[idx] = md->items[md->itemCount - 1];
158 GNUNET_array_grow (md->items, md->itemCount, md->itemCount - 1);
171 * Iterate over MD entries, excluding thumbnails.
173 * @return number of entries
176 GNUNET_CONTAINER_meta_data_get_contents (const struct
177 GNUNET_CONTAINER_MetaData *md,
178 GNUNET_CONTAINER_MetaDataProcessor
179 iterator, void *closure)
185 for (i = 0; i < md->itemCount; i++)
187 if (!EXTRACTOR_isBinaryType (md->items[i].type))
189 if ((iterator != NULL) &&
190 (GNUNET_OK != iterator (closure,
193 return GNUNET_SYSERR;
198 return (int) (md->itemCount - sub);
202 * Iterate over MD entries
204 * @return number of entries
207 GNUNET_CONTAINER_meta_data_get_by_type (const struct GNUNET_CONTAINER_MetaData
208 *md, EXTRACTOR_KeywordType type)
212 for (i = 0; i < md->itemCount; i++)
213 if (type == md->items[i].type)
214 return GNUNET_strdup (md->items[i].data);
219 * Iterate over MD entries
221 * @return number of entries
224 GNUNET_CONTAINER_meta_data_get_first_by_types (const struct
225 GNUNET_CONTAINER_MetaData *md,
230 EXTRACTOR_KeywordType type;
236 type = va_arg (args, EXTRACTOR_KeywordType);
239 ret = GNUNET_CONTAINER_meta_data_get_by_type (md, type);
248 * Get a thumbnail from the meta-data (if present).
250 * @param md metadata to get the thumbnail from
251 * @param thumb will be set to the thumbnail data. Must be
252 * freed by the caller!
253 * @return number of bytes in thumbnail, 0 if not available
256 GNUNET_CONTAINER_meta_data_get_thumbnail (const struct
257 GNUNET_CONTAINER_MetaData * md,
258 unsigned char **thumb)
265 GNUNET_CONTAINER_meta_data_get_by_type (md, EXTRACTOR_THUMBNAIL_DATA);
268 if (strlen (encoded) == 0)
270 GNUNET_free (encoded);
271 return 0; /* invalid */
274 ret = EXTRACTOR_binaryDecode (encoded, thumb, &size);
275 GNUNET_free (encoded);
282 * Duplicate struct GNUNET_CONTAINER_MetaData.
284 struct GNUNET_CONTAINER_MetaData *
285 GNUNET_CONTAINER_meta_data_duplicate (const struct GNUNET_CONTAINER_MetaData
289 struct GNUNET_CONTAINER_MetaData *ret;
293 ret = GNUNET_CONTAINER_meta_data_create ();
294 for (i = 0; i < md->itemCount; i++)
295 GNUNET_CONTAINER_meta_data_insert (ret, md->items[i].type,
301 * Extract meta-data from a file.
303 * @return GNUNET_SYSERR on error, otherwise the number
304 * of meta-data items obtained
307 GNUNET_CONTAINER_meta_data_extract_from_file (struct GNUNET_CONTAINER_MetaData
308 *md, const char *filename,
309 EXTRACTOR_ExtractorList *
312 EXTRACTOR_KeywordList *head;
313 EXTRACTOR_KeywordList *pos;
316 if (filename == NULL)
317 return GNUNET_SYSERR;
318 if (extractors == NULL)
320 head = EXTRACTOR_getKeywords (extractors, filename);
321 head = EXTRACTOR_removeDuplicateKeywords (head,
322 EXTRACTOR_DUPLICATES_REMOVE_UNKNOWN);
328 GNUNET_CONTAINER_meta_data_insert (md, pos->keywordType,
333 EXTRACTOR_freeKeywords (head);
339 tryCompression (char *data, unsigned int oldSize)
345 dlen = compressBound (oldSize);
347 dlen = oldSize + (oldSize / 100) + 20;
348 /* documentation says 100.1% oldSize + 12 bytes, but we
349 should be able to overshoot by more to be safe */
351 tmp = GNUNET_malloc (dlen);
352 if (Z_OK == compress2 ((Bytef *) tmp,
353 &dlen, (const Bytef *) data, oldSize, 9))
357 memcpy (data, tmp, dlen);
367 * Decompress input, return the decompressed data
368 * as output, set outputSize to the number of bytes
371 * @return NULL on error
374 decompress (const char *input,
375 unsigned int inputSize, unsigned int outputSize)
381 output = GNUNET_malloc (olen);
382 if (Z_OK == uncompress ((Bytef *) output,
383 &olen, (const Bytef *) input, inputSize))
389 GNUNET_free (output);
395 * Flag in 'version' that indicates compressed meta-data.
397 #define HEADER_COMPRESSED 0x80000000
400 * Bits in 'version' that give the version number.
402 #define HEADER_VERSION_MASK 0x7FFFFFFF
404 struct MetaDataHeader
407 * The version of the MD serialization.
408 * The highest bit is used to indicate
411 * Version 0 is the current version;
412 * Version is 1 for a NULL pointer.
413 * Other version numbers are not yet defined.
418 * How many MD entries are there?
423 * Size of the MD (decompressed)
428 * This is followed by 'entries' values of type 'unsigned int' that
429 * correspond to EXTRACTOR_KeywordTypes. After that, the meta-data
430 * keywords follow (0-terminated). The MD block always ends with
431 * 0-termination, padding with 0 until a multiple of 8 bytes.
437 * Serialize meta-data to target.
439 * @param md metadata to serialize
440 * @param target where to write the serialized metadata
441 * @param max maximum number of bytes available in target
442 * @param part is it ok to just write SOME of the
443 * meta-data to match the size constraint,
444 * possibly discarding some data?
445 * @return number of bytes written on success,
446 * GNUNET_SYSERR on error (typically: not enough
450 GNUNET_CONTAINER_meta_data_serialize (const struct GNUNET_CONTAINER_MetaData
451 *md, char *target, size_t max,
453 GNUNET_CONTAINER_MetaDataSerializationOptions
456 struct MetaDataHeader *hdr;
463 if (max < sizeof (struct MetaDataHeader))
464 return GNUNET_SYSERR; /* far too small */
465 ic = md ? md->itemCount : 0;
469 size = sizeof (struct MetaDataHeader);
470 size += sizeof (uint32_t) * ic;
471 for (i = 0; i < ic; i++)
472 size += 1 + strlen (md->items[i].data);
473 while (size % 8 != 0)
475 hdr = GNUNET_malloc (size);
476 hdr->version = htonl (md == NULL ? 1 : 0);
477 hdr->entries = htonl (ic);
478 for (i = 0; i < ic; i++)
479 ((uint32_t *) &hdr[1])[i] =
480 htonl ((uint32_t) md->items[i].type);
481 pos = sizeof (struct MetaDataHeader);
482 pos += sizeof (unsigned int) * ic;
483 for (i = 0; i < ic; i++)
485 len = strlen (md->items[i].data) + 1;
486 memcpy (&((char *) hdr)[pos], md->items[i].data, len);
490 hdr->size = htonl (size);
491 if ((part & GNUNET_CONTAINER_META_DATA_SERIALIZE_NO_COMPRESS) == 0)
493 pos = tryCompression ((char *) &hdr[1],
494 size - sizeof (struct MetaDataHeader));
498 pos = size - sizeof (struct MetaDataHeader);
500 if (pos < size - sizeof (struct MetaDataHeader))
502 hdr->version = htonl (HEADER_COMPRESSED);
503 size = pos + sizeof (struct MetaDataHeader);
510 if ((part & GNUNET_CONTAINER_META_DATA_SERIALIZE_PART) == 0)
512 return GNUNET_SYSERR; /* does not fit! */
514 /* partial serialization ok, try again with less meta-data */
516 ic = ic * 2 / 3; /* still far too big, make big reductions */
518 ic--; /* small steps, we're close */
520 GNUNET_assert (size <= max);
521 memcpy (target, hdr, size);
523 /* extra check: deserialize! */
526 struct GNUNET_CONTAINER_MetaData *mdx;
527 mdx = GNUNET_CONTAINER_meta_data_deserialize (target, size);
528 GNUNET_assert (NULL != mdx);
529 GNUNET_CONTAINER_meta_data_destroy (mdx);
536 * Estimate (!) the size of the meta-data in
537 * serialized form. The estimate MAY be higher
538 * than what is strictly needed.
540 * @param md metadata to inspect
541 * @param opt is it ok to just write SOME of the
542 * meta-data to match the size constraint,
543 * possibly discarding some data?
544 * @return number of bytes needed for serialization, -1 on error
547 GNUNET_CONTAINER_meta_data_get_serialized_size (const struct
548 GNUNET_CONTAINER_MetaData *md,
550 GNUNET_CONTAINER_MetaDataSerializationOptions
553 struct MetaDataHeader *hdr;
560 ic = md ? md->itemCount : 0;
561 size = sizeof (struct MetaDataHeader);
562 size += sizeof (uint32_t) * ic;
563 for (i = 0; i < ic; i++)
564 size += 1 + strlen (md->items[i].data);
565 while (size % 8 != 0)
567 hdr = GNUNET_malloc (size);
568 hdr->version = htonl (md == NULL ? 1 : 0);
569 hdr->entries = htonl (ic);
570 for (i = 0; i < ic; i++)
571 ((uint32_t *) &hdr[1])[i] = htonl ((uint32_t) md->items[i].type);
572 pos = sizeof (struct MetaDataHeader);
573 pos += sizeof (uint32_t) * ic;
574 for (i = 0; i < ic; i++)
576 len = strlen (md->items[i].data) + 1;
577 memcpy (&((char *) hdr)[pos], md->items[i].data, len);
580 if ((opt & GNUNET_CONTAINER_META_DATA_SERIALIZE_NO_COMPRESS) == 0)
583 tryCompression ((char *) &hdr[1],
584 size - sizeof (struct MetaDataHeader));
588 pos = size - sizeof (struct MetaDataHeader);
590 if (pos < size - sizeof (struct MetaDataHeader))
591 size = pos + sizeof (struct MetaDataHeader);
598 * Deserialize meta-data. Initializes md.
600 * @param input buffer with the serialized metadata
601 * @param size number of bytes available in input
602 * @return MD on success, NULL on error (i.e.
605 struct GNUNET_CONTAINER_MetaData *
606 GNUNET_CONTAINER_meta_data_deserialize (const char *input, size_t size)
608 struct GNUNET_CONTAINER_MetaData *md;
609 const struct MetaDataHeader *hdr;
620 if (size < sizeof (struct MetaDataHeader))
622 hdr = (const struct MetaDataHeader *) input;
623 version = ntohl (MAKE_UNALIGNED (hdr->version)) & HEADER_VERSION_MASK;
625 return NULL; /* null pointer */
628 GNUNET_break_op (0); /* unsupported version */
631 ic = ntohl (MAKE_UNALIGNED (hdr->entries));
633 (ntohl (MAKE_UNALIGNED (hdr->version)) & HEADER_COMPRESSED) != 0;
637 ntohl (MAKE_UNALIGNED (hdr->size)) - sizeof (struct MetaDataHeader);
638 if (dataSize > 2 * 1042 * 1024)
641 return NULL; /* only 2 MB allowed [to make sure we don't blow
642 our memory limit because of a mal-formed
646 decompress ((const char *) &input[sizeof (struct MetaDataHeader)],
647 size - sizeof (struct MetaDataHeader), dataSize);
658 cdata = (const char *) &hdr[1];
659 dataSize = size - sizeof (struct MetaDataHeader);
660 if (size != ntohl (MAKE_UNALIGNED (hdr->size)))
667 if ((sizeof (uint32_t) * ic + ic) > dataSize)
672 if ((ic > 0) && (cdata[dataSize - 1] != '\0'))
678 md = GNUNET_CONTAINER_meta_data_create ();
679 GNUNET_array_grow (md->items, md->itemCount, ic);
681 pos = sizeof (uint32_t) * ic;
682 while ((pos < dataSize) && (i < ic))
684 len = strlen (&cdata[pos]) + 1;
685 md->items[i].type = (EXTRACTOR_KeywordType)
686 ntohl (MAKE_UNALIGNED (((const uint32_t *) cdata)[i]));
687 md->items[i].data = GNUNET_strdup (&cdata[pos]);
693 GNUNET_CONTAINER_meta_data_destroy (md);
696 GNUNET_free_non_null (data);
699 GNUNET_free_non_null (data);
700 return NULL; /* size too small */
704 * Test if two MDs are equal.
706 * @param md1 first value to check
707 * @param md2 other value to check
708 * @return GNUNET_YES if they are equal
711 GNUNET_CONTAINER_meta_data_test_equal (const struct GNUNET_CONTAINER_MetaData
713 const struct GNUNET_CONTAINER_MetaData
720 if (md1->itemCount != md2->itemCount)
722 for (i = 0; i < md1->itemCount; i++)
725 for (j = 0; j < md2->itemCount; j++)
726 if ((md1->items[i].type == md2->items[j].type) &&
727 (0 == strcmp (md1->items[i].data, md2->items[j].data)))
732 if (found == GNUNET_NO)
739 /* end of container_meta_data.c */