* @file fs/fs_file_information.c
* @brief Manage information for publishing directory hierarchies
* @author Christian Grothoff
- *
- * TODO:
- * - metadata filename clean up code
- * - metadata/ksk generation for directories from contained files
*/
#include "platform.h"
#include <extractor.h>
#include "fs_tree.h"
-/**
- * Add meta data that libextractor finds to our meta data
- * container.
- *
- * @param cls closure, our meta data container
- * @param plugin_name name of the plugin that produced this value;
- * special values can be used (i.e. '<zlib>' for zlib being
- * used in the main libextractor library and yielding
- * meta data).
- * @param type libextractor-type describing the meta data
- * @param format basic format information about data
- * @param data_mime_type mime-type of data (not of the original file);
- * can be NULL (if mime-type is not known)
- * @param data actual meta-data found
- * @param data_len number of bytes in data
- * @return always 0 to continue extracting
- */
-static int
-add_to_md (void *cls, const char *plugin_name, enum EXTRACTOR_MetaType type,
- enum EXTRACTOR_MetaFormat format, const char *data_mime_type,
- const char *data, size_t data_len)
-{
- struct GNUNET_CONTAINER_MetaData *md = cls;
-
- (void) GNUNET_CONTAINER_meta_data_insert (md, plugin_name, type, format,
- data_mime_type, data, data_len);
- return 0;
-}
-
-
-/**
- * Extract meta-data from a file.
- *
- * @return GNUNET_SYSERR on error, otherwise the number
- * of meta-data items obtained
- */
-int
-GNUNET_FS_meta_data_extract_from_file (struct GNUNET_CONTAINER_MetaData *md,
- const char *filename,
- struct EXTRACTOR_PluginList *extractors)
-{
- int old;
-
- if (filename == NULL)
- return GNUNET_SYSERR;
- if (extractors == NULL)
- return 0;
- old = GNUNET_CONTAINER_meta_data_iterate (md, NULL, NULL);
- GNUNET_assert (old >= 0);
- EXTRACTOR_extract (extractors, filename, NULL, 0, &add_to_md, md);
- return (GNUNET_CONTAINER_meta_data_iterate (md, NULL, NULL) - old);
-}
-
-
-
/**
* Obtain the name under which this file information
* structure is stored on disk. Only works for top-level
*bo)
{
struct FileInfo *fi;
- struct stat sbuf;
+ uint64_t fsize;
struct GNUNET_FS_FileInformation *ret;
const char *fn;
const char *ss;
char fn_conv[MAX_PATH];
#endif
- if (0 != STAT (filename, &sbuf))
+ /* FIXME: should includeSymLinks be GNUNET_NO or GNUNET_YES here? */
+ if (GNUNET_OK != GNUNET_DISK_file_size (filename, &fsize, GNUNET_NO, GNUNET_YES))
{
GNUNET_log_strerror_file (GNUNET_ERROR_TYPE_WARNING, "stat", filename);
return NULL;
}
ret =
GNUNET_FS_file_information_create_from_reader (h, client_info,
- sbuf.st_size,
+ fsize,
&GNUNET_FS_data_reader_file_,
fi, keywords, meta,
do_index, bo);
#endif
while (NULL != (ss = strstr (fn, DIR_SEPARATOR_STR)))
fn = ss + 1;
+/* FIXME: If we assume that on other platforms CRT is UTF-8-aware, then
+ * this should be changed to EXTRACTOR_METAFORMAT_UTF8
+ */
#if !WINDOWS
GNUNET_CONTAINER_meta_data_insert (ret->meta, "<gnunet>",
EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME,
}
-/**
- * Closure for "dir_scan_cb".
- */
-struct DirScanCls
-{
- /**
- * Metadata extractors to use.
- */
- struct EXTRACTOR_PluginList *extractors;
-
- /**
- * Master context.
- */
- struct GNUNET_FS_Handle *h;
-
- /**
- * Function to call on each directory entry.
- */
- GNUNET_FS_FileProcessor proc;
-
- /**
- * Closure for proc.
- */
- void *proc_cls;
-
- /**
- * Scanner to use for subdirectories.
- */
- GNUNET_FS_DirectoryScanner scanner;
-
- /**
- * Closure for scanner.
- */
- void *scanner_cls;
-
- /**
- * Set to an error message (if any).
- */
- char *emsg;
-
- /**
- * Block options.
- */
- const struct GNUNET_FS_BlockOptions *bo;
-
- /**
- * Should files be indexed?
- */
- int do_index;
-
-};
-
-
-/**
- * Function called on each entry in a file to cause
- * default-publishing.
- *
- * @param cls closure (struct DirScanCls)
- * @param filename name of the file to be published
- * @return GNUNET_OK on success, GNUNET_SYSERR to abort
- */
-static int
-dir_scan_cb (void *cls, const char *filename)
-{
- struct DirScanCls *dsc = cls;
- struct stat sbuf;
- struct GNUNET_FS_FileInformation *fi;
- struct GNUNET_FS_Uri *keywords;
- struct GNUNET_CONTAINER_MetaData *meta;
-
- if (0 != STAT (filename, &sbuf))
- {
- GNUNET_asprintf (&dsc->emsg, _("`%s' failed on file `%s': %s"), "stat",
- filename, STRERROR (errno));
- return GNUNET_SYSERR;
- }
- if (S_ISDIR (sbuf.st_mode))
- {
- fi = GNUNET_FS_file_information_create_from_directory (dsc->h, NULL,
- filename,
- dsc->scanner,
- dsc->scanner_cls,
- dsc->do_index,
- dsc->bo, &dsc->emsg);
- if (NULL == fi)
- {
- GNUNET_assert (NULL != dsc->emsg);
- return GNUNET_SYSERR;
- }
- }
- else
- {
- meta = GNUNET_CONTAINER_meta_data_create ();
- GNUNET_FS_meta_data_extract_from_file (meta, filename, dsc->extractors);
- keywords = GNUNET_FS_uri_ksk_create_from_meta_data (meta);
- fi = GNUNET_FS_file_information_create_from_file (dsc->h, NULL, filename,
- keywords, meta,
- dsc->do_index, dsc->bo);
- GNUNET_CONTAINER_meta_data_destroy (meta);
- GNUNET_FS_uri_destroy (keywords);
- }
- dsc->proc (dsc->proc_cls, filename, fi);
- return GNUNET_OK;
-}
-
-
-/**
- * Simple, useful default implementation of a directory scanner
- * (GNUNET_FS_DirectoryScanner). This implementation expects to get a
- * UNIX filename, will publish all files in the directory except hidden
- * files (those starting with a "."). Metadata will be extracted
- * using GNU libextractor; the specific list of plugins should be
- * specified in "cls", passing NULL will disable (!) metadata
- * extraction. Keywords will be derived from the metadata and be
- * subject to default canonicalization. This is strictly a
- * convenience function.
- *
- * @param cls must be of type "struct EXTRACTOR_Extractor*"
- * @param h handle to the file sharing subsystem
- * @param dirname name of the directory to scan
- * @param do_index should files be indexed or inserted
- * @param bo block options
- * @param proc function called on each entry
- * @param proc_cls closure for proc
- * @param emsg where to store an error message (on errors)
- * @return GNUNET_OK on success
- */
-int
-GNUNET_FS_directory_scanner_default (void *cls, struct GNUNET_FS_Handle *h,
- const char *dirname, int do_index,
- const struct GNUNET_FS_BlockOptions *bo,
- GNUNET_FS_FileProcessor proc,
- void *proc_cls, char **emsg)
-{
- struct EXTRACTOR_PluginList *ex = cls;
- struct DirScanCls dsc;
-
- dsc.h = h;
- dsc.extractors = ex;
- dsc.proc = proc;
- dsc.proc_cls = proc_cls;
- dsc.scanner = &GNUNET_FS_directory_scanner_default;
- dsc.scanner_cls = cls;
- dsc.do_index = do_index;
- dsc.bo = bo;
- if (-1 == GNUNET_DISK_directory_scan (dirname, &dir_scan_cb, &dsc))
- {
- GNUNET_assert (NULL != dsc.emsg);
- *emsg = dsc.emsg;
- return GNUNET_SYSERR;
- }
- return GNUNET_OK;
-}
-
-
-/**
- * Aggregate information we keep for meta data in each directory.
- */
-struct MetaValueInformation
-{
-
- /**
- * Mime-type of data.
- */
- const char *mime_type;
-
- /**
- * The actual meta data.
- */
- const char *data;
-
- /**
- * Number of bytes in 'data'.
- */
- size_t data_size;
-
- /**
- * Type of the meta data.
- */
- enum EXTRACTOR_MetaType type;
-
- /**
- * Format of the meta data.
- */
- enum EXTRACTOR_MetaFormat format;
-
- /**
- * How often does this meta value occur in this directory?
- */
- unsigned int frequency;
-
-};
-
-
-/**
- * Type of a function that libextractor calls for each
- * meta data item found.
- *
- * @param cls the container multihashmap to update
- * @param plugin_name name of the plugin that produced this value;
- * special values can be used (i.e. '<zlib>' for zlib being
- * used in the main libextractor library and yielding
- * meta data).
- * @param type libextractor-type describing the meta data
- * @param format basic format information about data
- * @param data_mime_type mime-type of data (not of the original file);
- * can be NULL (if mime-type is not known)
- * @param data actual meta-data found
- * @param data_len number of bytes in data
- * @return 0 to continue extracting / iterating
- */
-static int
-update_metamap (void *cls, const char *plugin_name,
- enum EXTRACTOR_MetaType type, enum EXTRACTOR_MetaFormat format,
- const char *data_mime_type, const char *data, size_t data_len)
-{
- struct GNUNET_CONTAINER_MultiHashMap *map = cls;
- GNUNET_HashCode key;
- struct MetaValueInformation *mvi;
-
- GNUNET_CRYPTO_hash (data, data_len, &key);
- mvi = GNUNET_CONTAINER_multihashmap_get (map, &key);
- if (mvi == NULL)
- {
- mvi = GNUNET_malloc (sizeof (struct MetaValueInformation));
- mvi->mime_type = data_mime_type;
- mvi->data = data;
- mvi->data_size = data_len;
- mvi->type = type;
- mvi->format = format;
- GNUNET_CONTAINER_multihashmap_put (map, &key, mvi,
- GNUNET_CONTAINER_MULTIHASHMAPOPTION_UNIQUE_ONLY);
- }
- mvi->frequency++;
- return 0;
-}
-
-
-/**
- * Aggregate information we keep for keywords in each directory.
- */
-struct KeywordInformation
-{
-
- /**
- * Mime-type of keyword.
- */
- const char *keyword;
-
- /**
- * How often does this meta value occur in this directory?
- */
- unsigned int frequency;
-
-};
-
-
-/**
- * Closure for dirproc function.
- */
-struct EntryProcCls
-{
- /**
- * Linked list of directory entries that is being
- * created.
- */
- struct GNUNET_FS_FileInformation *entries;
-
- /**
- * Map describing the meta data for all entries in the
- * directory. Keys are the hash of the meta-value,
- * values are of type 'struct MetaValueInformation'.
- */
- struct GNUNET_CONTAINER_MultiHashMap *metamap;
-
- /**
- * Map describing the keywords for all entries in the
- * directory. Keys are the hash of the keyword,
- * values are of type 'struct KeywordInformation'.
- */
- struct GNUNET_CONTAINER_MultiHashMap *keywordmap;
-
- /**
- * Number of entries in 'entries'.
- */
- unsigned int count;
-
-};
-
-
-/**
- * Function that processes a directory entry that
- * was obtained from the scanner. Adds each entry to
- * the directory and computes directroy meta map.
- *
- * @param cls our closure
- * @param filename name of the file (unused, why there???)
- * @param fi information for publishing the file
- */
-static void
-dirproc_add (void *cls, const char *filename,
- struct GNUNET_FS_FileInformation *fi)
-{
- struct EntryProcCls *dc = cls;
- unsigned int i;
- const char *kw;
- struct KeywordInformation *ki;
- GNUNET_HashCode key;
-
- GNUNET_assert (fi->next == NULL);
- GNUNET_assert (fi->dir == NULL);
- fi->next = dc->entries;
- dc->entries = fi;
- dc->count++;
- if (NULL != fi->meta)
- GNUNET_CONTAINER_meta_data_iterate (fi->meta, &update_metamap, dc->metamap);
- for (i = 0; i < fi->keywords->data.ksk.keywordCount; i++)
- {
- kw = fi->keywords->data.ksk.keywords[i];
- GNUNET_CRYPTO_hash (kw, strlen (kw), &key);
- ki = GNUNET_CONTAINER_multihashmap_get (dc->keywordmap, &key);
- if (ki == NULL)
- {
- ki = GNUNET_malloc (sizeof (struct KeywordInformation));
- ki->keyword = &kw[1];
- GNUNET_CONTAINER_multihashmap_put (dc->keywordmap, &key, ki,
- GNUNET_CONTAINER_MULTIHASHMAPOPTION_UNIQUE_ONLY);
- }
- ki->frequency++;
- }
-}
-
-
-/**
- * Closure for 'compute_directory_metadata'.
- */
-struct ComputeDirectoryMetadataContext
-{
- /**
- * Where to store the extracted keywords.
- */
- struct GNUNET_FS_Uri *ksk;
-
- /**
- * Where to store the extracted meta data.
- */
- struct GNUNET_CONTAINER_MetaData *meta;
-
- /**
- * Threshold to apply for adding meta data.
- */
- unsigned int threshold;
-};
-
-
-/**
- * Add metadata that occurs in more than the threshold entries of the
- * directory to the directory itself. For example, if most files in a
- * directory are of the same mime-type, the directory should have that
- * mime-type as a keyword.
- *
- * @param cls the 'struct ComputeDirectoryMetadataContext'
- * @param key unused
- * @param value the 'struct MetaValueInformation' (to be freed as well)
- * @return GNUNET_OK
- */
-static int
-compute_directory_metadata (void *cls, const GNUNET_HashCode * key, void *value)
-{
- struct ComputeDirectoryMetadataContext *cdmc = cls;
- struct MetaValueInformation *mvi = value;
-
- if (mvi->frequency > cdmc->threshold)
- {
- if (mvi->type != EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME)
- (void) GNUNET_CONTAINER_meta_data_insert (cdmc->meta, "<children>",
- mvi->type, mvi->format,
- mvi->mime_type, mvi->data,
- mvi->data_size);
- if ((mvi->format == EXTRACTOR_METAFORMAT_UTF8) ||
- (mvi->format == EXTRACTOR_METAFORMAT_C_STRING))
- GNUNET_FS_uri_ksk_add_keyword (cdmc->ksk, mvi->data, GNUNET_NO);
- }
- GNUNET_free (mvi);
- return GNUNET_OK;
-}
-
-
-/**
- * Add keywords that occur in more than the threshold entries of the
- * directory to the directory itself.
- *
- * @param cls the 'struct ComputeDirectoryMetadataContext'
- * @param key unused
- * @param value the 'struct Keywordnformation' (to be freed as well)
- * @return GNUNET_OK
- */
-static int
-compute_directory_keywords (void *cls, const GNUNET_HashCode * key, void *value)
-{
- struct ComputeDirectoryMetadataContext *cdmc = cls;
- struct KeywordInformation *ki = value;
-
- if (ki->frequency > cdmc->threshold)
- (void) GNUNET_FS_uri_ksk_add_keyword (cdmc->ksk, ki->keyword, GNUNET_NO);
- GNUNET_free (ki);
- return GNUNET_OK;
-}
-
-
-/**
- * Create a publish-structure from an existing file hierarchy, inferring
- * and organizing keywords and metadata as much as possible. This
- * function primarily performs the recursive build and re-organizes
- * keywords and metadata; for automatically getting metadata
- * extraction, scanning of directories and creation of the respective
- * GNUNET_FS_FileInformation entries the default scanner should be
- * passed (GNUNET_FS_directory_scanner_default). This is strictly a
- * convenience function.
- *
- * @param h handle to the file sharing subsystem
- * @param client_info initial value for the client-info value for this entry
- * @param filename name of the top-level file or directory
- * @param scanner function used to get a list of files in a directory
- * @param scanner_cls closure for scanner
- * @param do_index should files in the hierarchy be indexed?
- * @param bo block options
- * @param emsg where to store an error message
- * @return publish structure entry for the directory, NULL on error
- */
-struct GNUNET_FS_FileInformation *
-GNUNET_FS_file_information_create_from_directory (struct GNUNET_FS_Handle *h,
- void *client_info,
- const char *filename,
- GNUNET_FS_DirectoryScanner
- scanner, void *scanner_cls,
- int do_index,
- const struct
- GNUNET_FS_BlockOptions *bo,
- char **emsg)
-{
- struct GNUNET_FS_FileInformation *ret;
- struct ComputeDirectoryMetadataContext cdmc;
- struct EntryProcCls dc;
- const char *fn;
- const char *ss;
- char *dn;
- struct GNUNET_FS_FileInformation *epos;
- unsigned int i;
- const char *kw;
-
- dc.entries = NULL;
- dc.count = 0;
- dc.metamap = GNUNET_CONTAINER_multihashmap_create (64);
- dc.keywordmap = GNUNET_CONTAINER_multihashmap_create (64);
- /* update children to point to directory and generate statistics
- * on all meta data in children */
- scanner (scanner_cls, h, filename, do_index, bo, &dirproc_add, &dc, emsg);
- cdmc.meta = GNUNET_CONTAINER_meta_data_create ();
- cdmc.ksk = GNUNET_malloc (sizeof (struct GNUNET_FS_Uri));
- cdmc.ksk->type = ksk;
- cdmc.threshold = 1 + dc.count / 2; /* 50% threshold for now */
- GNUNET_FS_meta_data_make_directory (cdmc.meta);
- GNUNET_CONTAINER_multihashmap_iterate (dc.metamap,
- &compute_directory_metadata, &cdmc);
- GNUNET_CONTAINER_multihashmap_iterate (dc.keywordmap,
- &compute_directory_keywords, &cdmc);
- GNUNET_CONTAINER_multihashmap_destroy (dc.metamap);
- GNUNET_CONTAINER_multihashmap_destroy (dc.keywordmap);
-
- /* remove keywords in children that are already in the
- * parent */
- for (epos = dc.entries; NULL != epos; epos = epos->next)
- {
- for (i = 0; i < cdmc.ksk->data.ksk.keywordCount; i++)
- {
- kw = cdmc.ksk->data.ksk.keywords[i];
- GNUNET_FS_uri_ksk_remove_keyword (epos->keywords, &kw[1]);
- }
- }
- GNUNET_FS_uri_ksk_add_keyword (cdmc.ksk, GNUNET_FS_DIRECTORY_MIME, GNUNET_NO);
- ret =
- GNUNET_FS_file_information_create_empty_directory (h, client_info, cdmc.ksk,
- cdmc.meta, bo, filename);
- GNUNET_CONTAINER_meta_data_destroy (cdmc.meta);
- GNUNET_FS_uri_destroy (cdmc.ksk);
- ret->data.dir.entries = dc.entries;
- while (dc.entries != NULL)
- {
- dc.entries->dir = ret;
- dc.entries = dc.entries->next;
- }
- fn = filename;
- while ((NULL != (ss = strstr (fn, DIR_SEPARATOR_STR))) && (strlen (ss) > 1))
- fn = ss + 1;
- GNUNET_asprintf (&dn, "%s/", fn);
-#if !WINDOWS
- GNUNET_CONTAINER_meta_data_insert (ret->meta, "<gnunet>",
- EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME,
- EXTRACTOR_METAFORMAT_C_STRING,
- "text/plain", dn, strlen (dn) + 1);
-#else
- GNUNET_CONTAINER_meta_data_insert (ret->meta, "<gnunet>",
- EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME,
- EXTRACTOR_METAFORMAT_UTF8,
- "text/plain", dn, strlen (dn) + 1);
-#endif
- GNUNET_free (dn);
- return ret;
-}
-
-
/**
* Test if a given entry represents a directory.
*
GNUNET_FS_file_information_add (struct GNUNET_FS_FileInformation *dir,
struct GNUNET_FS_FileInformation *ent)
{
- if ((ent->dir != NULL) || (ent->next != NULL) || (!dir->is_directory))
+ if ((ent->dir != NULL) || (ent->next != NULL) || (dir->is_directory != GNUNET_YES))
{
GNUNET_break (0);
return GNUNET_SYSERR;
no = GNUNET_NO;
if (GNUNET_OK !=
proc (proc_cls, dir,
- (dir->is_directory) ? dir->data.dir.dir_size : dir->data.
- file.file_size, dir->meta, &dir->keywords, &dir->bo,
- (dir->is_directory) ? &no : &dir->data.file.do_index,
+ (dir->is_directory == GNUNET_YES) ? dir->data.dir.dir_size : dir->data.
+ file.file_size,
+ dir->meta, &dir->keywords, &dir->bo,
+ (dir->is_directory == GNUNET_YES) ? &no : &dir->data.file.do_index,
&dir->client_info))
return;
- if (!dir->is_directory)
+ if (dir->is_directory != GNUNET_YES)
return;
pos = dir->data.dir.entries;
while (pos != NULL)
no = GNUNET_NO;
if (GNUNET_OK !=
proc (proc_cls, pos,
- (pos->is_directory) ? pos->data.dir.dir_size : pos->data.
+ (pos->is_directory == GNUNET_YES) ? pos->data.dir.dir_size : pos->data.
file.file_size, pos->meta, &pos->keywords, &pos->bo,
- (dir->is_directory) ? &no : &dir->data.file.do_index,
+ (pos->is_directory == GNUNET_YES) ? &no : &pos->data.file.do_index,
&pos->client_info))
break;
pos = pos->next;
int no;
no = GNUNET_NO;
- if (fi->is_directory)
+ if (fi->is_directory == GNUNET_YES)
{
/* clean up directory */
while (NULL != (pos = fi->data.dir.entries))