2 This file is part of GNUnet.
3 (C) 2012 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 3, 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 src/fs/gnunet-helper-fs-publish.c
23 * @brief Tool to help extract meta data asynchronously
24 * @author Christian Grothoff
26 * This program will scan a directory for files with meta data
27 * and report the results to stdout.
30 #include "gnunet_fs_service.h"
34 * A node of a directory tree.
40 * This is a doubly-linked list
42 struct ScanTreeNode *next;
45 * This is a doubly-linked list
47 struct ScanTreeNode *prev;
50 * Parent of this node, NULL for top-level entries.
52 struct ScanTreeNode *parent;
55 * This is a doubly-linked tree
56 * NULL for files and empty directories
58 struct ScanTreeNode *children_head;
61 * This is a doubly-linked tree
62 * NULL for files and empty directories
64 struct ScanTreeNode *children_tail;
67 * Name of the file/directory
72 * Size of the file (if it is a file), in bytes.
73 * At the moment it is set to 0 for directories.
78 * GNUNET_YES if this is a directory
86 * List of libextractor plugins to use for extracting.
88 static struct EXTRACTOR_PluginList *plugins;
92 * Add meta data that libextractor finds to our meta data
95 * @param cls closure, our meta data container
96 * @param plugin_name name of the plugin that produced this value;
97 * special values can be used (i.e. '<zlib>' for zlib being
98 * used in the main libextractor library and yielding
100 * @param type libextractor-type describing the meta data
101 * @param format basic format information about data
102 * @param data_mime_type mime-type of data (not of the original file);
103 * can be NULL (if mime-type is not known)
104 * @param data actual meta-data found
105 * @param data_len number of bytes in data
106 * @return always 0 to continue extracting
109 add_to_md (void *cls, const char *plugin_name, enum EXTRACTOR_MetaType type,
110 enum EXTRACTOR_MetaFormat format, const char *data_mime_type,
111 const char *data, size_t data_len)
113 struct GNUNET_CONTAINER_MetaData *md = cls;
115 (void) GNUNET_CONTAINER_meta_data_insert (md, plugin_name, type, format,
116 data_mime_type, data, data_len);
122 * Free memory of the 'tree' structure
124 * @param tree tree to free
127 free_tree (struct ScanTreeNode *tree)
129 struct ScanTreeNode *pos;
131 while (NULL != (pos = tree->children_head))
133 if (NULL != tree->parent)
134 GNUNET_CONTAINER_DLL_remove (tree->parent->children_head,
135 tree->parent->children_tail,
137 GNUNET_free (tree->filename);
143 * Write 'size' bytes from 'buf' into 'out'.
145 * @param buf buffer with data to write
146 * @param size number of bytes to write
147 * @return GNUNET_OK on success, GNUNET_SYSERR on error
150 write_all (const void *buf,
153 const char *cbuf = buf;
165 } while ( (wr > 0) && (total < size) );
167 GNUNET_log (GNUNET_ERROR_TYPE_DEBUG,
168 "Failed to write to stdout: %s\n",
170 return (total == size) ? GNUNET_OK : GNUNET_SYSERR;
175 * Write message to the master process.
177 * @param message_type message type to use
178 * @param data data to append, NULL for none
179 * @param data_length number of bytes in data
180 * @return GNUNET_SYSERR to stop scanning (the pipe was broken somehow)
183 write_message (uint16_t message_type,
187 struct GNUNET_MessageHeader hdr;
189 hdr.type = htons (message_type);
190 hdr.size = htons (sizeof (struct GNUNET_MessageHeader) + data_length);
197 return GNUNET_SYSERR;
203 * Function called to (recursively) add all of the files in the
204 * directory to the tree. Called by the directory scanner to initiate
205 * the scan. Does NOT yet add any metadata.
207 * @param filename file or directory to scan
208 * @param dst where to store the resulting share tree item
209 * @return GNUNET_OK on success, GNUNET_SYSERR on error
212 preprocess_file (const char *filename,
213 struct ScanTreeNode **dst);
217 * Closure for the 'scan_callback'
219 struct RecursionContext
222 * Parent to add the files to.
224 struct ScanTreeNode *parent;
227 * Flag to set to GNUNET_YES on serious errors.
234 * Function called by the directory iterator to (recursively) add all
235 * of the files in the directory to the tree. Called by the directory
236 * scanner to initiate the scan. Does NOT yet add any metadata.
238 * @param cls the 'struct RecursionContext'
239 * @param filename file or directory to scan
240 * @return GNUNET_OK on success, GNUNET_SYSERR on error
243 scan_callback (void *cls,
244 const char *filename)
246 struct RecursionContext *rc = cls;
247 struct ScanTreeNode *chld;
250 preprocess_file (filename,
253 rc->stop = GNUNET_YES;
254 return GNUNET_SYSERR;
256 chld->parent = rc->parent;
257 GNUNET_CONTAINER_DLL_insert (rc->parent->children_head,
258 rc->parent->children_tail,
265 * Function called to (recursively) add all of the files in the
266 * directory to the tree. Called by the directory scanner to initiate
267 * the scan. Does NOT yet add any metadata.
269 * @param filename file or directory to scan
270 * @param dst where to store the resulting share tree item
271 * @return GNUNET_OK on success, GNUNET_SYSERR on error
274 preprocess_file (const char *filename,
275 struct ScanTreeNode **dst)
277 struct ScanTreeNode *item;
281 if ((0 != STAT (filename, &sbuf)) ||
282 ((!S_ISDIR (sbuf.st_mode)) && (GNUNET_OK != GNUNET_DISK_file_size (
283 filename, &fsize, GNUNET_NO, GNUNET_YES))))
285 /* If the file doesn't exist (or is not stat-able for any other reason)
286 skip it (but report it), but do continue. */
288 write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_SKIP_FILE,
289 filename, strlen (filename) + 1))
290 return GNUNET_SYSERR;
294 /* Report the progress */
296 write_message (S_ISDIR (sbuf.st_mode)
297 ? GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_PROGRESS_DIRECTORY
298 : GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_PROGRESS_FILE,
299 filename, strlen (filename) + 1))
300 return GNUNET_SYSERR;
301 item = GNUNET_malloc (sizeof (struct ScanTreeNode));
302 item->filename = GNUNET_strdup (filename);
303 item->is_directory = (S_ISDIR (sbuf.st_mode)) ? GNUNET_YES : GNUNET_NO;
304 item->file_size = fsize;
305 if (item->is_directory == GNUNET_YES)
307 struct RecursionContext rc;
311 GNUNET_DISK_directory_scan (filename,
314 if ( (rc.stop == GNUNET_YES) ||
316 write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_PROGRESS_DIRECTORY,
320 return GNUNET_SYSERR;
329 * Extract metadata from files.
331 * @param item entry we are processing
332 * @return GNUNET_OK on success, GNUNET_SYSERR on fatal errors
335 extract_files (struct ScanTreeNode *item)
337 struct GNUNET_CONTAINER_MetaData *meta;
341 if (item->is_directory == GNUNET_YES)
343 /* for directories, we simply only descent, no extraction, no
344 progress reporting */
345 struct ScanTreeNode *pos;
347 for (pos = item->children_head; NULL != pos; pos = pos->next)
350 return GNUNET_SYSERR;
354 /* this is the expensive operation, *afterwards* we'll check for aborts */
355 meta = GNUNET_CONTAINER_meta_data_create ();
357 EXTRACTOR_extract (plugins, item->filename, NULL, 0, &add_to_md, meta);
358 slen = strlen (item->filename) + 1;
359 size = GNUNET_CONTAINER_meta_data_get_serialized_size (meta);
363 GNUNET_CONTAINER_meta_data_destroy (meta);
365 write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_META_DATA,
366 item->filename, slen))
367 return GNUNET_SYSERR;
371 char buf[size + slen];
372 char *dst = &buf[slen];
374 memcpy (buf, item->filename, slen);
375 size = GNUNET_CONTAINER_meta_data_serialize (meta,
377 GNUNET_CONTAINER_META_DATA_SERIALIZE_PART);
378 GNUNET_CONTAINER_meta_data_destroy (meta);
380 write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_META_DATA,
383 return GNUNET_SYSERR;
390 * Main function of the helper process to extract meta data.
392 * @param argc should be 3
393 * @param argv [0] our binary name
394 * [1] name of the file or directory to process
395 * [2] "-" to disable extraction, NULL for defaults,
396 * otherwise custom plugins to load from LE
397 * @return 0 on success
402 const char *filename_expanded;
404 struct ScanTreeNode *root;
407 /* We're using stdout to communicate binary data back to the parent; use
410 _setmode (1, _O_BINARY);
413 /* parse command line */
414 if ( (argc != 3) && (argc != 2) )
418 "gnunet-helper-fs-publish needs exactly one or two arguments\n");
421 filename_expanded = argv[1];
424 (0 != strcmp (ex, "-")) )
426 plugins = EXTRACTOR_plugin_add_defaults (EXTRACTOR_OPTION_DEFAULT_POLICY);
428 plugins = EXTRACTOR_plugin_add_config (plugins, ex,
429 EXTRACTOR_OPTION_DEFAULT_POLICY);
432 /* scan tree to find out how much work there is to be done */
433 if (GNUNET_OK != preprocess_file (filename_expanded,
436 (void) write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_ERROR, NULL, 0);
439 /* signal that we're done counting files, so that a percentage of
440 progress can now be calculated */
442 write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_COUNTING_DONE, NULL, 0))
445 extract_files (root))
447 (void) write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_ERROR, NULL, 0);
452 /* enable "clean" shutdown by telling parent that we are done */
453 (void) write_message (GNUNET_MESSAGE_TYPE_FS_PUBLISH_HELPER_FINISHED, NULL, 0);
455 EXTRACTOR_plugin_remove_all (plugins);
460 /* end of gnunet-helper-fs-publish.c */