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)
143 * @param md metadata to manipulate
144 * @param type type of the item to remove
145 * @param data specific value to remove, NULL to remove all
146 * entries of the given type
147 * @return GNUNET_OK on success, GNUNET_SYSERR if the item does not exist in md
150 GNUNET_CONTAINER_meta_data_delete (struct GNUNET_CONTAINER_MetaData *md,
151 EXTRACTOR_KeywordType type,
155 int ret = GNUNET_SYSERR;
156 for (idx = 0; idx < md->itemCount; idx++)
158 if ((md->items[idx].type == type) &&
159 ((data == NULL) || (0 == strcmp (md->items[idx].data, data))))
161 GNUNET_free (md->items[idx].data);
162 md->items[idx] = md->items[md->itemCount - 1];
163 GNUNET_array_grow (md->items, md->itemCount, md->itemCount - 1);
176 * Iterate over MD entries, excluding thumbnails.
178 * @param md metadata to inspect
179 * @param iter function to call on each entry
180 * @param iter_cls closure for iterator
181 * @return number of entries
184 GNUNET_CONTAINER_meta_data_get_contents (const struct
185 GNUNET_CONTAINER_MetaData *md,
186 GNUNET_CONTAINER_MetaDataProcessor
187 iter, void *iter_cls)
193 for (i = 0; i < md->itemCount; i++)
195 if (!EXTRACTOR_isBinaryType (md->items[i].type))
197 if ((iter != NULL) &&
198 (GNUNET_OK != iter (iter_cls,
201 return GNUNET_SYSERR;
206 return (int) (md->itemCount - sub);
210 * Iterate over MD entries
212 * @return number of entries
215 GNUNET_CONTAINER_meta_data_get_by_type (const struct GNUNET_CONTAINER_MetaData
216 *md, EXTRACTOR_KeywordType type)
220 for (i = 0; i < md->itemCount; i++)
221 if (type == md->items[i].type)
222 return GNUNET_strdup (md->items[i].data);
227 * Iterate over MD entries
229 * @return number of entries
232 GNUNET_CONTAINER_meta_data_get_first_by_types (const struct
233 GNUNET_CONTAINER_MetaData *md,
238 EXTRACTOR_KeywordType type;
244 type = va_arg (args, EXTRACTOR_KeywordType);
247 ret = GNUNET_CONTAINER_meta_data_get_by_type (md, type);
256 * Get a thumbnail from the meta-data (if present).
258 * @param md metadata to get the thumbnail from
259 * @param thumb will be set to the thumbnail data. Must be
260 * freed by the caller!
261 * @return number of bytes in thumbnail, 0 if not available
264 GNUNET_CONTAINER_meta_data_get_thumbnail (const struct
265 GNUNET_CONTAINER_MetaData * md,
266 unsigned char **thumb)
273 GNUNET_CONTAINER_meta_data_get_by_type (md, EXTRACTOR_THUMBNAIL_DATA);
276 if (strlen (encoded) == 0)
278 GNUNET_free (encoded);
279 return 0; /* invalid */
282 ret = EXTRACTOR_binaryDecode (encoded, thumb, &size);
283 GNUNET_free (encoded);
290 * Duplicate struct GNUNET_CONTAINER_MetaData.
292 * @param md what to duplicate
293 * @return duplicate meta-data container
295 struct GNUNET_CONTAINER_MetaData *
296 GNUNET_CONTAINER_meta_data_duplicate (const struct GNUNET_CONTAINER_MetaData
300 struct GNUNET_CONTAINER_MetaData *ret;
304 ret = GNUNET_CONTAINER_meta_data_create ();
305 for (i = 0; i < md->itemCount; i++)
306 GNUNET_CONTAINER_meta_data_insert (ret, md->items[i].type,
312 * Extract meta-data from a file.
314 * @return GNUNET_SYSERR on error, otherwise the number
315 * of meta-data items obtained
318 GNUNET_CONTAINER_meta_data_extract_from_file (struct GNUNET_CONTAINER_MetaData
319 *md, const char *filename,
320 EXTRACTOR_ExtractorList *
323 EXTRACTOR_KeywordList *head;
324 EXTRACTOR_KeywordList *pos;
327 if (filename == NULL)
328 return GNUNET_SYSERR;
329 if (extractors == NULL)
331 head = EXTRACTOR_getKeywords (extractors, filename);
332 head = EXTRACTOR_removeDuplicateKeywords (head,
333 EXTRACTOR_DUPLICATES_REMOVE_UNKNOWN);
339 GNUNET_CONTAINER_meta_data_insert (md, pos->keywordType,
344 EXTRACTOR_freeKeywords (head);
350 tryCompression (char *data, unsigned int oldSize)
356 dlen = compressBound (oldSize);
358 dlen = oldSize + (oldSize / 100) + 20;
359 /* documentation says 100.1% oldSize + 12 bytes, but we
360 should be able to overshoot by more to be safe */
362 tmp = GNUNET_malloc (dlen);
363 if (Z_OK == compress2 ((Bytef *) tmp,
364 &dlen, (const Bytef *) data, oldSize, 9))
368 memcpy (data, tmp, dlen);
378 * Decompress input, return the decompressed data
379 * as output, set outputSize to the number of bytes
382 * @return NULL on error
385 decompress (const char *input,
386 unsigned int inputSize, unsigned int outputSize)
392 output = GNUNET_malloc (olen);
393 if (Z_OK == uncompress ((Bytef *) output,
394 &olen, (const Bytef *) input, inputSize))
400 GNUNET_free (output);
406 * Flag in 'version' that indicates compressed meta-data.
408 #define HEADER_COMPRESSED 0x80000000
411 * Bits in 'version' that give the version number.
413 #define HEADER_VERSION_MASK 0x7FFFFFFF
415 struct MetaDataHeader
418 * The version of the MD serialization.
419 * The highest bit is used to indicate
422 * Version 0 is the current version;
423 * Version is 1 for a NULL pointer.
424 * Other version numbers are not yet defined.
429 * How many MD entries are there?
434 * Size of the MD (decompressed)
439 * This is followed by 'entries' values of type 'unsigned int' that
440 * correspond to EXTRACTOR_KeywordTypes. After that, the meta-data
441 * keywords follow (0-terminated). The MD block always ends with
442 * 0-termination, padding with 0 until a multiple of 8 bytes.
448 * Serialize meta-data to target.
450 * @param md metadata to serialize
451 * @param target where to write the serialized metadata
452 * @param max maximum number of bytes available in target
453 * @param opt is it ok to just write SOME of the
454 * meta-data to match the size constraint,
455 * possibly discarding some data?
456 * @return number of bytes written on success,
457 * GNUNET_SYSERR on error (typically: not enough
461 GNUNET_CONTAINER_meta_data_serialize (const struct GNUNET_CONTAINER_MetaData
462 *md, char *target, size_t max,
464 GNUNET_CONTAINER_MetaDataSerializationOptions
467 struct MetaDataHeader *hdr;
474 if (max < sizeof (struct MetaDataHeader))
475 return GNUNET_SYSERR; /* far too small */
476 ic = md ? md->itemCount : 0;
480 size = sizeof (struct MetaDataHeader);
481 size += sizeof (uint32_t) * ic;
482 for (i = 0; i < ic; i++)
483 size += 1 + strlen (md->items[i].data);
484 while (size % 8 != 0)
486 hdr = GNUNET_malloc (size);
487 hdr->version = htonl (md == NULL ? 1 : 0);
488 hdr->entries = htonl (ic);
489 for (i = 0; i < ic; i++)
490 ((uint32_t *) &hdr[1])[i] =
491 htonl ((uint32_t) md->items[i].type);
492 pos = sizeof (struct MetaDataHeader);
493 pos += sizeof (unsigned int) * ic;
494 for (i = 0; i < ic; i++)
496 len = strlen (md->items[i].data) + 1;
497 memcpy (&((char *) hdr)[pos], md->items[i].data, len);
501 hdr->size = htonl (size);
502 if ((opt & GNUNET_CONTAINER_META_DATA_SERIALIZE_NO_COMPRESS) == 0)
504 pos = tryCompression ((char *) &hdr[1],
505 size - sizeof (struct MetaDataHeader));
509 pos = size - sizeof (struct MetaDataHeader);
511 if (pos < size - sizeof (struct MetaDataHeader))
513 hdr->version = htonl (HEADER_COMPRESSED);
514 size = pos + sizeof (struct MetaDataHeader);
521 if ((opt & GNUNET_CONTAINER_META_DATA_SERIALIZE_PART) == 0)
523 return GNUNET_SYSERR; /* does not fit! */
525 /* partial serialization ok, try again with less meta-data */
527 ic = ic * 2 / 3; /* still far too big, make big reductions */
529 ic--; /* small steps, we're close */
531 GNUNET_assert (size <= max);
532 memcpy (target, hdr, size);
534 /* extra check: deserialize! */
537 struct GNUNET_CONTAINER_MetaData *mdx;
538 mdx = GNUNET_CONTAINER_meta_data_deserialize (target, size);
539 GNUNET_assert (NULL != mdx);
540 GNUNET_CONTAINER_meta_data_destroy (mdx);
547 * Estimate (!) the size of the meta-data in
548 * serialized form. The estimate MAY be higher
549 * than what is strictly needed.
551 * @param md metadata to inspect
552 * @param opt is it ok to just write SOME of the
553 * meta-data to match the size constraint,
554 * possibly discarding some data?
555 * @return number of bytes needed for serialization, -1 on error
558 GNUNET_CONTAINER_meta_data_get_serialized_size (const struct
559 GNUNET_CONTAINER_MetaData *md,
561 GNUNET_CONTAINER_MetaDataSerializationOptions
564 struct MetaDataHeader *hdr;
571 ic = md ? md->itemCount : 0;
572 size = sizeof (struct MetaDataHeader);
573 size += sizeof (uint32_t) * ic;
574 for (i = 0; i < ic; i++)
575 size += 1 + strlen (md->items[i].data);
576 while (size % 8 != 0)
578 hdr = GNUNET_malloc (size);
579 hdr->version = htonl (md == NULL ? 1 : 0);
580 hdr->entries = htonl (ic);
581 for (i = 0; i < ic; i++)
582 ((uint32_t *) &hdr[1])[i] = htonl ((uint32_t) md->items[i].type);
583 pos = sizeof (struct MetaDataHeader);
584 pos += sizeof (uint32_t) * ic;
585 for (i = 0; i < ic; i++)
587 len = strlen (md->items[i].data) + 1;
588 memcpy (&((char *) hdr)[pos], md->items[i].data, len);
591 if ((opt & GNUNET_CONTAINER_META_DATA_SERIALIZE_NO_COMPRESS) == 0)
594 tryCompression ((char *) &hdr[1],
595 size - sizeof (struct MetaDataHeader));
599 pos = size - sizeof (struct MetaDataHeader);
601 if (pos < size - sizeof (struct MetaDataHeader))
602 size = pos + sizeof (struct MetaDataHeader);
609 * Deserialize meta-data. Initializes md.
611 * @param input buffer with the serialized metadata
612 * @param size number of bytes available in input
613 * @return MD on success, NULL on error (i.e.
616 struct GNUNET_CONTAINER_MetaData *
617 GNUNET_CONTAINER_meta_data_deserialize (const char *input, size_t size)
619 struct GNUNET_CONTAINER_MetaData *md;
620 const struct MetaDataHeader *hdr;
631 if (size < sizeof (struct MetaDataHeader))
633 hdr = (const struct MetaDataHeader *) input;
634 version = ntohl (MAKE_UNALIGNED (hdr->version)) & HEADER_VERSION_MASK;
636 return NULL; /* null pointer */
639 GNUNET_break_op (0); /* unsupported version */
642 ic = ntohl (MAKE_UNALIGNED (hdr->entries));
644 (ntohl (MAKE_UNALIGNED (hdr->version)) & HEADER_COMPRESSED) != 0;
648 ntohl (MAKE_UNALIGNED (hdr->size)) - sizeof (struct MetaDataHeader);
649 if (dataSize > 2 * 1042 * 1024)
652 return NULL; /* only 2 MB allowed [to make sure we don't blow
653 our memory limit because of a mal-formed
657 decompress ((const char *) &input[sizeof (struct MetaDataHeader)],
658 size - sizeof (struct MetaDataHeader), dataSize);
669 cdata = (const char *) &hdr[1];
670 dataSize = size - sizeof (struct MetaDataHeader);
671 if (size != ntohl (MAKE_UNALIGNED (hdr->size)))
678 if ((sizeof (uint32_t) * ic + ic) > dataSize)
683 if ((ic > 0) && (cdata[dataSize - 1] != '\0'))
689 md = GNUNET_CONTAINER_meta_data_create ();
690 GNUNET_array_grow (md->items, md->itemCount, ic);
692 pos = sizeof (uint32_t) * ic;
693 while ((pos < dataSize) && (i < ic))
695 len = strlen (&cdata[pos]) + 1;
696 md->items[i].type = (EXTRACTOR_KeywordType)
697 ntohl (MAKE_UNALIGNED (((const uint32_t *) cdata)[i]));
698 md->items[i].data = GNUNET_strdup (&cdata[pos]);
704 GNUNET_CONTAINER_meta_data_destroy (md);
707 GNUNET_free_non_null (data);
710 GNUNET_free_non_null (data);
711 return NULL; /* size too small */
715 * Test if two MDs are equal.
717 * @param md1 first value to check
718 * @param md2 other value to check
719 * @return GNUNET_YES if they are equal
722 GNUNET_CONTAINER_meta_data_test_equal (const struct GNUNET_CONTAINER_MetaData
724 const struct GNUNET_CONTAINER_MetaData
731 if (md1->itemCount != md2->itemCount)
733 for (i = 0; i < md1->itemCount; i++)
736 for (j = 0; j < md2->itemCount; j++)
737 if ((md1->items[i].type == md2->items[j].type) &&
738 (0 == strcmp (md1->items[i].data, md2->items[j].data)))
743 if (found == GNUNET_NO)
750 /* end of container_meta_data.c */