Update to that last patch with exif enabled.
On 9/26/06, Edward Duffy <[EMAIL PROTECTED]> wrote:
Here's the patch for moving everything to tracker-extract.
On 9/26/06, Jamie McCracken <[EMAIL PROTECTED]> wrote:
> Edward Duffy wrote:
> > Is there any advantage to keeping them internal? Should we just stick
> > with tracker-extract for all types?
>
> I suppose it makes writing metadata easier having them internal although
> we could use pipes for that too.
>
> Its also nice to have a seperate app to extract as you might find it
> useful to use elsewhere or just for testing.
>
> Feel free to move your stuff to tracker-extract for the time being.
>
>
> --
> Mr Jamie McCracken
> http://jamiemcc.livejournal.com/
>
Index: configure.in
===================================================================
RCS file: /cvs/gnome/tracker/configure.in,v
retrieving revision 1.18
diff -u -p -r1.18 configure.in
--- configure.in 26 Sep 2006 00:52:38 -0000 1.18
+++ configure.in 27 Sep 2006 03:07:13 -0000
@@ -347,6 +347,34 @@ AC_SUBST(THEORA_LIBS)
test "$have_theora" = "yes" && AC_DEFINE(HAVE_THEORA, [], [Define if we have libtheora])
+##################################################################
+# check for libpng
+##################################################################
+
+LIBPNG_REQUIRED=1.2
+
+PKG_CHECK_MODULES(LIBPNG, [libpng >= $LIBPNG_REQUIRED], [have_libpng=yes] , [have_png=no])
+
+AM_CONDITIONAL(HAVE_LIBPNG, test "$have_libpng" = "yes")
+AC_SUBST(LIBPNG_CFLAGS)
+AC_SUBST(LIBPNG_LIBS)
+test "$have_libpng" = "yes" && AC_DEFINE(HAVE_LIBPNG, [], [Define if we have libpng])
+
+
+##################################################################
+# check for libexif
+##################################################################
+
+LIBEXIF_REQUIRED=0.6
+
+PKG_CHECK_MODULES(LIBEXIF, [libexif >= $LIBEXIF_REQUIRED], [have_libexif=yes] , [have_libexif=no])
+
+AM_CONDITIONAL(HAVE_LIBEXIF, test "$have_libexif" = "yes")
+AC_SUBST(LIBEXIF_CFLAGS)
+AC_SUBST(LIBEXIF_LIBS)
+test "$have_libexif" = "yes" && AC_DEFINE(HAVE_LIBEXIF, [], [Define if we have libexif])
+
+
#####################################################
AM_CONFIG_HEADER(src/trackerd/config.h)
@@ -395,5 +423,7 @@ Metadata extractors:
pdf : $have_poppler
ogg/vorbis : $have_vorbis
ogg/theora : $have_theora
+ png : $have_libpng
+ exif (jpeg) : $have_libexif
"
Index: src/trackerd/Makefile.am
===================================================================
RCS file: /cvs/gnome/tracker/src/trackerd/Makefile.am,v
retrieving revision 1.11
diff -u -p -r1.11 Makefile.am
--- src/trackerd/Makefile.am 26 Sep 2006 00:52:39 -0000 1.11
+++ src/trackerd/Makefile.am 27 Sep 2006 03:07:25 -0000
@@ -15,9 +15,6 @@ INCLUDES = \
$(FAM_CFLAGS) \
$(DBUS_CFLAGS) \
$(MYSQL_CFLAGS) \
- $(POPPLER_GLIB_CFLAGS) \
- $(VORBIS_CFLAGS) \
- $(THEORA_CFLAGS) \
$(additional_mysql_flags) \
$(CFLAGS) \
-g
@@ -101,11 +98,6 @@ trackerd_SOURCES = \
tracker-mbox.h \
tracker-metadata.c \
tracker-metadata.h \
- tracker-metadata-oasis.c \
- tracker-metadata-ps.c \
- tracker-metadata-pdf.c \
- tracker-metadata-abw.c \
- tracker-metadata-vorbis.c \
tracker-rdf-query.c \
tracker-rdf-query.h \
tracker-stemmer-english.c \
@@ -138,9 +130,6 @@ trackerd_LDADD = $(GLIB2_LIBS) \
$(GMIME_LIBS) \
$(QDBM_LIBS) \
$(SQLITE3_LIBS) \
- $(POPPLER_GLIB_LIBS) \
- $(VORBIS_LIBS) \
- $(THEORA_LIBS) \
-lbz2 \
-lstdc++
Index: src/trackerd/trackerd.c
===================================================================
RCS file: /cvs/gnome/tracker/src/trackerd/trackerd.c,v
retrieving revision 1.38
diff -u -p -r1.38 trackerd.c
--- src/trackerd/trackerd.c 26 Sep 2006 00:52:39 -0000 1.38
+++ src/trackerd/trackerd.c 27 Sep 2006 03:08:05 -0000
@@ -1953,8 +1953,6 @@ main (int argc, char **argv)
sigaction (SIGUSR1, &act, NULL);
sigaction (SIGINT, &act, NULL);
- g_type_init ();
-
if (!g_thread_supported ()) {
g_thread_init (NULL);
}
Index: src/trackerd/tracker-metadata.c
===================================================================
RCS file: /cvs/gnome/tracker/src/trackerd/tracker-metadata.c,v
retrieving revision 1.11
diff -u -p -r1.11 tracker-metadata.c
--- src/trackerd/tracker-metadata.c 26 Sep 2006 10:28:53 -0000 1.11
+++ src/trackerd/tracker-metadata.c 27 Sep 2006 03:08:21 -0000
@@ -127,39 +127,6 @@ char *development_mime_types[] = {
};
-typedef void (*MetadataExtractFunc)(gchar *, GHashTable *);
-typedef struct {
- char *mime;
- MetadataExtractFunc extractor;
-} MimeToExtractor;
-
-void tracker_metadata_extract_oasis (gchar *, GHashTable *);
-void tracker_metadata_extract_ps (gchar *, GHashTable *);
-#ifdef HAVE_POPPLER
-void tracker_metadata_extract_pdf (gchar *, GHashTable *);
-#endif
-void tracker_metadata_extract_abw (gchar *, GHashTable *);
-void tracker_metadata_extract_vorbis (gchar *, GHashTable *);
-
-MimeToExtractor internal_metadata_extractors[] = {
- /* Document extractors */
- { "application/vnd.oasis.opendocument.text", tracker_metadata_extract_oasis },
- { "application/vnd.oasis.opendocument.spreadsheet", tracker_metadata_extract_oasis },
- { "application/vnd.oasis.opendocument.graphics", tracker_metadata_extract_oasis },
- { "application/vnd.oasis.opendocument.presentation", tracker_metadata_extract_oasis },
- { "application/postscript", tracker_metadata_extract_ps },
-#ifdef HAVE_POPPLER
- { "application/pdf", tracker_metadata_extract_pdf },
-#endif
- { "application/x-abiword", tracker_metadata_extract_abw },
- /* Video extractors */
- //{ "video/x-theora+ogg", tracker_metadata_extract_theora },
- /* Audio extractors */
- { "audio/x-vorbis+ogg", tracker_metadata_extract_vorbis },
- /* Image extractors */
- { "", NULL }
-};
-
static MetadataFileType
tracker_get_metadata_type (const char *mime)
{
@@ -431,32 +398,14 @@ tracker_metadata_get_thumbnail (const ch
return NULL;
}
-static void log_metadata_cb (gpointer key, gpointer value, gpointer user_data)
-{
- tracker_log ("%s = %s", (gchar *)key, (gchar *)value);
-}
-
void
tracker_metadata_get_embedded (const char *uri, const char *mime, GHashTable *table)
{
- MimeToExtractor *p;
MetadataFileType meta_type;
- gboolean found;
if (!uri || !mime || !table) {
return;
}
-
- found = FALSE;
- for (p = internal_metadata_extractors; p->extractor; ++p) {
- if (strcmp (p->mime, mime) == 0) {
- found = TRUE;
- (*p->extractor)(uri, table);
- g_hash_table_foreach (table, log_metadata_cb, NULL);
- }
- }
- if (found)
- return;
meta_type = tracker_get_metadata_type (mime);
Index: src/tracker-extract/Makefile.am
===================================================================
RCS file: /cvs/gnome/tracker/src/tracker-extract/Makefile.am,v
retrieving revision 1.4
diff -u -p -r1.4 Makefile.am
--- src/tracker-extract/Makefile.am 21 Aug 2006 20:18:26 -0000 1.4
+++ src/tracker-extract/Makefile.am 27 Sep 2006 03:08:39 -0000
@@ -1,8 +1,20 @@
-INCLUDES = $(GLIB2_CFLAGS) $(CFLAGS) -g
+INCLUDES = $(GLIB2_CFLAGS) $(CFLAGS) -g \
+ $(POPPLER_GLIB_CFLAGS) \
+ $(VORBIS_CFLAGS) \
+ $(THEORA_CFLAGS) \
+ $(LIBPNG_CFLAGS) \
+ $(LIBEXIF_CFLAGS)
bin_PROGRAMS = tracker-extract
-tracker_extract_SOURCES = tracker-extract.c
+tracker_extract_SOURCES = tracker-extract.c \
+ tracker-extract-oasis.c \
+ tracker-extract-ps.c \
+ tracker-extract-pdf.c \
+ tracker-extract-abw.c \
+ tracker-extract-vorbis.c \
+ tracker-extract-png.c \
+ tracker-extract-exif.c
if USING_INTERNAL_LIBEXTRACTOR
extractor_ldadd = $(top_builddir)/src/libextractor/src/main/libextractor.la
@@ -10,4 +22,9 @@ else
extractor_ldadd = -lextractor
endif
-tracker_extract_LDADD = $(GLIB2_LIBS) $(extractor_ldadd)
+tracker_extract_LDADD = $(GLIB2_LIBS) $(extractor_ldadd) \
+ $(POPPLER_GLIB_LIBS) \
+ $(VORBIS_LIBS) \
+ $(THEORA_LIBS) \
+ $(LIBPNG_LIBS) \
+ $(LIBEXIF_LIBS)
Index: src/tracker-extract/tracker-extract.c
===================================================================
RCS file: /cvs/gnome/tracker/src/tracker-extract/tracker-extract.c,v
retrieving revision 1.10
diff -u -p -r1.10 tracker-extract.c
--- src/tracker-extract/tracker-extract.c 20 Sep 2006 23:40:50 -0000 1.10
+++ src/tracker-extract/tracker-extract.c 27 Sep 2006 03:09:19 -0000
@@ -30,6 +30,7 @@
# include "../libextractor/src/include/extractor.h"
#endif
+#include "config.h"
typedef enum {
IGNORE_METADATA,
@@ -116,6 +117,62 @@ static struct metadata_format image_keyw
{"EOF", EXTRACTOR_UNKNOWN}
};
+typedef void (*MetadataExtractFunc)(gchar *, GHashTable *);
+typedef struct {
+ char *mime;
+ MetadataExtractFunc extractor;
+} MimeToExtractor;
+
+void tracker_extract_oasis (gchar *, GHashTable *);
+void tracker_extract_ps (gchar *, GHashTable *);
+#ifdef HAVE_POPPLER
+void tracker_extract_pdf (gchar *, GHashTable *);
+#endif
+void tracker_extract_abw (gchar *, GHashTable *);
+#ifdef HAVE_VORBIS
+void tracker_extract_vorbis (gchar *, GHashTable *);
+#endif
+#ifdef HAVE_LIBPNG
+void tracker_extract_png (gchar *, GHashTable *);
+#endif
+#ifdef HAVE_LIBEXIF
+void tracker_extract_exif (gchar *, GHashTable *);
+#endif
+
+MimeToExtractor extractors[] = {
+ /* Document extractors */
+ { "application/vnd.oasis.opendocument.text", tracker_extract_oasis },
+ { "application/vnd.oasis.opendocument.spreadsheet", tracker_extract_oasis },
+ { "application/vnd.oasis.opendocument.graphics", tracker_extract_oasis },
+ { "application/vnd.oasis.opendocument.presentation", tracker_extract_oasis },
+ { "application/postscript", tracker_extract_ps },
+#ifdef HAVE_POPPLER
+ { "application/pdf", tracker_extract_pdf },
+#endif
+ { "application/x-abiword", tracker_extract_abw },
+
+
+ /* Video extractors */
+#ifdef HAVE_THEORA
+ { "video/x-theora+ogg", tracker_extract_theora },
+#endif
+
+
+ /* Audio extractors */
+#ifdef HAVE_VORBIS
+ { "audio/x-vorbis+ogg", tracker_extract_vorbis },
+#endif
+
+
+ /* Image extractors */
+#ifdef HAVE_LIBPNG
+ { "image/png", tracker_extract_png },
+#endif
+#ifdef HAVE_LIBEXIF
+ { "image/jpeg", tracker_extract_exif },
+#endif
+ { "", NULL }
+};
static MetadataFileType
get_metadata_type (const char *mime)
@@ -386,6 +443,21 @@ tracker_get_file_metadata (const char *u
return NULL;
}
+ meta_table = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, g_free);
+
+ if (mime) {
+ MimeToExtractor *p;
+ for (p = extractors; p->extractor; ++p) {
+ if (strcmp (p->mime, mime) == 0) {
+ (*p->extractor)(uri_in_locale, meta_table);
+ return meta_table;
+ }
+ }
+ fprintf (stderr, "Mime %s not found, falling back on libextractor\n", mime);
+ }
+
+
+ /* Fallback to libextractor */
keywords = EXTRACTOR_getKeywords (plugins, uri_in_locale);
g_free (uri_in_locale);
@@ -410,8 +482,6 @@ tracker_get_file_metadata (const char *u
- meta_table = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, g_free);
-
while (keywords != NULL) {
value = g_locale_to_utf8 (keywords->keyword, -1, NULL, NULL, NULL);
@@ -478,6 +548,7 @@ main (int argc, char **argv)
GHashTable *meta;
char *filename;
+ g_set_application_name ("tracker-extract");
setlocale (LC_ALL, "");
if ((argc == 1) || (argc > 3)) {
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-abw.c 2006-09-26 18:35:58.000000000 -0400
@@ -0,0 +1,47 @@
+
+#include <string.h>
+#include <stdio.h>
+#include <glib.h>
+
+void tracker_extract_abw (gchar *filename, GHashTable *metadata)
+{
+ FILE *f;
+ gchar *line;
+ gsize length = 0;
+
+ if(f = fopen (filename, "r")) {
+ line = NULL;
+ getline (&line, &length, f);
+ while (!feof (f)) {
+ if (g_str_has_suffix (line, "</m>\n")) {
+ line[strlen(line) - 5] = '\0';
+ }
+ if (g_str_has_prefix (line, "<m key=\"dc.title\">")) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Title"), g_strdup (line+18));
+ }
+ else if (g_str_has_prefix (line, "<m key=\"dc.subject\">")) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Subject"), g_strdup (line+20));
+ }
+ else if (g_str_has_prefix (line, "<m key=\"dc.creator\">")) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Author"), g_strdup (line+20));
+ }
+ else if (g_str_has_prefix (line, "<m key=\"abiword.keywords\">")) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Keywords"), g_strdup (line+26));
+ }
+ else if (g_str_has_prefix (line, "<m key=\"dc.description\">")) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Comments"), g_strdup (line+24));
+ }
+ g_free (line);
+ line = NULL;
+ getline (&line, &length, f);
+ }
+ g_free (line);
+ }
+ fclose (f);
+}
+
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-oasis.c 2006-09-26 18:35:58.000000000 -0400
@@ -0,0 +1,181 @@
+
+#include <string.h>
+#include <glib.h>
+
+typedef enum {
+ READ_TITLE,
+ READ_SUBJECT,
+ READ_AUTHOR,
+ READ_KEYWORDS,
+ READ_COMMENTS,
+ READ_STATS,
+ READ_CREATED,
+ READ_FILE_OTHER
+ } tag_type;
+
+typedef struct {
+ GHashTable *metadata;
+ tag_type current;
+} ODTParseInfo;
+
+static void start_element_handler (GMarkupParseContext *context,
+ const gchar *element_name,
+ const gchar **attribute_names,
+ const gchar **attribute_values,
+ gpointer user_data,
+ GError **error);
+
+static void end_element_handler (GMarkupParseContext *context,
+ const gchar *element_name,
+ gpointer user_data,
+ GError **error);
+
+static void text_handler (GMarkupParseContext *context,
+ const gchar *text,
+ gsize text_len,
+ gpointer user_data,
+ GError **error);
+
+void tracker_extract_oasis (gchar *filename, GHashTable *metadata)
+{
+
+ gchar *argv[5];
+ gchar *xml;
+ ODTParseInfo info = { metadata, -1 };
+
+ argv[0] = g_strdup ("unzip");
+ argv[1] = g_strdup ("-p");
+ argv[2] = g_strdup (filename);
+ argv[3] = g_strdup ("meta.xml");
+ argv[4] = NULL;
+
+ if(g_spawn_sync (NULL,
+ argv,
+ NULL,
+ G_SPAWN_SEARCH_PATH | G_SPAWN_STDERR_TO_DEV_NULL,
+ NULL,
+ NULL,
+ &xml,
+ NULL,
+ NULL,
+ NULL)) {
+
+ GMarkupParseContext *context;
+ GMarkupParser parser = {
+ start_element_handler,
+ end_element_handler,
+ text_handler,
+ NULL,
+ NULL
+ };
+
+ context = g_markup_parse_context_new (&parser, 0, &info, NULL);
+ g_markup_parse_context_parse (context, xml, -1, NULL);
+
+ g_markup_parse_context_free (context);
+ g_free (xml);
+ }
+
+ g_free (argv[3]);
+ g_free (argv[2]);
+ g_free (argv[1]);
+ g_free (argv[0]);
+
+}
+
+void start_element_handler (GMarkupParseContext *context,
+ const gchar *element_name,
+ const gchar **attribute_names,
+ const gchar **attribute_values,
+ gpointer user_data,
+ GError **error)
+{
+ if(strcmp(element_name, "dc:title") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_TITLE;
+ }
+ else if(strcmp(element_name, "dc:subject") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_SUBJECT;
+ }
+ else if(strcmp(element_name, "dc:creator") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_AUTHOR;
+ }
+ else if(strcmp(element_name, "meta:keyword") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_KEYWORDS;
+ }
+ else if(strcmp(element_name, "dc:description") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_COMMENTS;
+ }
+ else if(strcmp(element_name, "meta:document-statistic") == 0) {
+ GHashTable *metadata = ((ODTParseInfo *)user_data)->metadata;
+ const gchar **a, **v;
+ for(a=attribute_names,v=attribute_values; *a; ++a,++v) {
+ if (strcmp (*a, "meta:word-count") == 0) {
+ g_hash_table_insert (metadata,
+ g_strdup("Doc.WordCount"), g_strdup (*v));
+ }
+ else if (strcmp (*a, "meta:page-count") == 0) {
+ g_hash_table_insert (metadata,
+ g_strdup("Doc.PageCount"), g_strdup (*v));
+ }
+ }
+ ((ODTParseInfo *)user_data)->current = READ_STATS;
+ }
+ else if(strcmp(element_name, "meta:creation-date") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_CREATED;
+ }
+ else if(strcmp(element_name, "meta:generator") == 0) {
+ ((ODTParseInfo *)user_data)->current = READ_FILE_OTHER;
+ }
+ else {
+ ((ODTParseInfo *)user_data)->current = -1;
+ }
+}
+
+void end_element_handler (GMarkupParseContext *context,
+ const gchar *element_name,
+ gpointer user_data,
+ GError **error)
+{
+ ((ODTParseInfo *)user_data)->current = -1;
+}
+
+void text_handler (GMarkupParseContext *context,
+ const gchar *text,
+ gsize text_len,
+ gpointer user_data,
+ GError **error)
+{
+ GHashTable *metadata = ((ODTParseInfo *)user_data)->metadata;
+
+ switch(((ODTParseInfo *)user_data)->current) {
+ case READ_TITLE:
+ g_hash_table_insert (metadata, g_strdup("Doc.Title"), g_strdup (text));
+ break;
+ case READ_SUBJECT:
+ g_hash_table_insert (metadata, g_strdup("Doc.Subject"), g_strdup (text));
+ break;
+ case READ_AUTHOR:
+ g_hash_table_insert (metadata, g_strdup("Doc.Author"), g_strdup (text));
+ break;
+ case READ_KEYWORDS: {
+ gchar *keywords;
+ if (keywords = g_hash_table_lookup (metadata, "Doc.Keywords")) {
+ g_hash_table_replace (metadata, "Doc.Keywords",
+ g_strconcat (keywords, ",", text, NULL));
+ }
+ else {
+ g_hash_table_insert (metadata, g_strdup("Doc.Keywords"), g_strdup (text));
+ }
+ }
+ break;
+ case READ_COMMENTS:
+ g_hash_table_insert (metadata, g_strdup("Doc.Comments"), g_strdup (text));
+ break;
+ case READ_CREATED:
+ g_hash_table_insert (metadata, g_strdup("Doc.Created"), g_strdup (text));
+ break;
+ case READ_FILE_OTHER:
+ g_hash_table_insert (metadata, g_strdup("File.Other"), g_strdup (text));
+ break;
+ }
+}
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-exif.c 2006-09-26 23:03:09.000000000 -0400
@@ -0,0 +1,113 @@
+
+#include "config.h"
+
+#ifdef HAVE_LIBEXIF
+
+#include <exif-data.h>
+#include <stdio.h>
+#include <string.h>
+#include <glib.h>
+
+static gchar *fix_focal_length (gchar *fl)
+{
+ return g_strndup (fl, (strstr (fl, "mm") - fl));
+}
+
+static gchar *fix_flash (gchar *flash)
+{
+ if (g_str_has_prefix (flash, "No"))
+ return g_strdup ("0");
+ else
+ return g_strdup ("1");
+}
+
+static gchar *fix_fnumber (gchar *fn)
+{
+ if (fn && fn[0] == 'F') {
+ fn[0] = ' ';
+ }
+ else if (fn && fn[0] == 'f' && fn[1] == '/') {
+ fn[0] = ' ', fn[1] = ' ';
+ }
+ return fn;
+}
+
+static gchar *fix_exposure_time (gchar *et)
+{
+ gchar *sep = strchr (et, '/');
+
+ if (sep) {
+ gdouble fraction = g_ascii_strtod (sep+1, NULL);
+
+ if (fraction > 0) {
+ gdouble val = 1.0f / fraction;
+ char str_value[30];
+
+ g_ascii_dtostr (str_value, 30, val);
+ return g_strdup (str_value);
+ }
+ }
+ return et;
+}
+
+typedef gchar *(*PostProcessor)(gchar *);
+
+typedef struct {
+ ExifTag tag;
+ gchar *name;
+ PostProcessor post;
+} TagType;
+
+TagType tags[] = {
+ { EXIF_TAG_PIXEL_Y_DIMENSION, "Image.Height", NULL },
+ { EXIF_TAG_PIXEL_X_DIMENSION, "Image.Width", NULL },
+ { EXIF_TAG_RELATED_IMAGE_WIDTH, "Image.Width", NULL },
+ { EXIF_TAG_DOCUMENT_NAME, "Image.Title", NULL },
+ /* { -1, "Image.Album", NULL }, */
+ { EXIF_TAG_DATE_TIME, "Image.Date", NULL },
+ /* { -1, "Image.Keywords", NULL }, */
+ { EXIF_TAG_ARTIST, "Image.Creator", NULL },
+ { EXIF_TAG_USER_COMMENT, "Image.Comments", NULL },
+ { EXIF_TAG_IMAGE_DESCRIPTION, "Image.Description", NULL },
+ { EXIF_TAG_SOFTWARE, "Image.Software", NULL },
+ { EXIF_TAG_MAKE, "Image.CameraMake", NULL },
+ { EXIF_TAG_MODEL, "Image.CameraModel", NULL },
+ { EXIF_TAG_ORIENTATION, "Image.Orientation", NULL },
+ { EXIF_TAG_EXPOSURE_PROGRAM, "Image.ExposureProgram", NULL },
+ { EXIF_TAG_EXPOSURE_TIME, "Image.ExposureTime", fix_exposure_time },
+ { EXIF_TAG_FNUMBER, "Image.Fnumber", fix_fnumber },
+ { EXIF_TAG_FLASH, "Image.Flash", fix_flash },
+ { EXIF_TAG_FOCAL_LENGTH, "Image.FocalLength", fix_focal_length },
+ { EXIF_TAG_ISO_SPEED_RATINGS, "Image.ISOSpeed", NULL },
+ { EXIF_TAG_METERING_MODE, "Image.MeteringMode", NULL },
+ { EXIF_TAG_WHITE_BALANCE, "Image.WhiteBalance", NULL },
+ { EXIF_TAG_COPYRIGHT, "Image.Copyright", NULL },
+ { -1, NULL, NULL }
+};
+
+void
+tracker_extract_exif (gchar *filename, GHashTable *metadata)
+{
+ ExifData *exif;
+ ExifEntry *entry;
+ char buffer[1024];
+ TagType *p;
+
+ exif = exif_data_new_from_file (filename);
+ for (p = tags; p->name; ++p) {
+ entry = exif_data_get_entry (exif, p->tag);
+ if (entry) {
+ exif_entry_get_value (entry, buffer, 1024);
+ if (p->post)
+ g_hash_table_insert (metadata, g_strdup (p->name),
+ g_strdup ((*p->post)(buffer)));
+ else
+ g_hash_table_insert (metadata, g_strdup (p->name),
+ g_strdup (buffer));
+ }
+ }
+}
+
+#else
+#warning "Not building EXIF metadata extractor."
+#endif /* HAVE_LIBEXIF */
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-pdf.c 2006-09-26 18:35:58.000000000 -0400
@@ -0,0 +1,64 @@
+
+#include "config.h"
+
+#ifdef HAVE_POPPLER
+
+#include <poppler.h>
+#include <string.h>
+#include <glib.h>
+
+void tracker_extract_pdf (gchar *filename, GHashTable *metadata)
+{
+ PopplerDocument *document;
+ gchar *tmp;
+ gchar *title;
+ gchar *author;
+ gchar *subject;
+ gchar *keywords;
+ GTime creation_date;
+ GError *error = NULL;
+
+ tmp = g_strconcat ("file://", filename, NULL);
+ document = poppler_document_new_from_file (tmp, NULL, &error);
+ g_free (tmp);
+ if (document == NULL || error)
+ return;
+
+ g_object_get (document,
+ "title", &title,
+ "author", &author,
+ "subject", &subject,
+ "keywords", &keywords,
+ "creation-date", &creation_date,
+ NULL);
+
+ if (title && strlen (title))
+ g_hash_table_insert (metadata, g_strdup ("Doc.Title"), g_strdup (title));
+ if (author && strlen (author))
+ g_hash_table_insert (metadata, g_strdup ("Doc.Author"), g_strdup (author));
+ if (subject && strlen (subject))
+ g_hash_table_insert (metadata, g_strdup ("Doc.Subject"), g_strdup (subject));
+ if (keywords && strlen (keywords))
+ g_hash_table_insert (metadata, g_strdup ("Doc.Keywords"), g_strdup (keywords));
+
+#if 0
+ GTimeVal creation_date_val = { creation_date, 0 };
+ g_hash_table_insert (metadata, g_strdup ("Doc.Created"),
+ g_time_val_to_iso8601 (creation_date_val));
+#endif
+
+ g_hash_table_insert (metadata, g_strdup ("Doc.PageCount"),
+ g_strdup_printf ("%d", poppler_document_get_n_pages (document)));
+
+ g_free (title);
+ g_free (author);
+ g_free (subject);
+ g_free (keywords);
+ g_object_unref (document);
+}
+
+#else
+#warning "Not building PDF metadata extractor."
+#endif /* HAVE_POPPLER */
+
+
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-png.c 2006-09-26 18:35:58.000000000 -0400
@@ -0,0 +1,51 @@
+
+#include "config.h"
+
+#ifdef HAVE_LIBPNG
+
+#include <stdio.h>
+#include <glib.h>
+#include <png.h>
+
+void
+tracker_extract_png (gchar *filename, GHashTable *metadata)
+{
+ FILE *png;
+ png_structp png_ptr;
+ png_infop info_ptr;
+ png_uint_32 width, height;
+
+ int bit_depth, color_type;
+ int interlace_type, compression_type, filter_type;
+
+ if(png = fopen(filename, "r")) {
+ png_ptr = png_create_read_struct (PNG_LIBPNG_VER_STRING,
+ NULL, NULL, NULL);
+ if (!png_ptr) {
+ fclose (png);
+ return;
+ }
+ info_ptr = png_create_info_struct (png_ptr);
+ if (!info_ptr) {
+ png_destroy_read_struct (&png_ptr, &info_ptr, NULL);
+ fclose (png);
+ return;
+ }
+ png_init_io (png_ptr, png);
+ png_read_info (png_ptr, info_ptr);
+ if (png_get_IHDR(png_ptr, info_ptr, &width, &height, &bit_depth,
+ &color_type, &interlace_type, &compression_type, &filter_type)) {
+ g_hash_table_insert (metadata, g_strdup ("Image.Width"),
+ g_strdup_printf ("%d", width));
+ g_hash_table_insert (metadata, g_strdup ("Image.Height"),
+ g_strdup_printf ("%d", height));
+ }
+
+ png_destroy_read_struct (&png_ptr, &info_ptr, NULL);
+ fclose (png);
+ }
+}
+
+#else
+#warning "Not building PNG metadata extractor."
+#endif /* HAVE_LIBPNG */
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-ps.c 2006-09-26 18:35:58.000000000 -0400
@@ -0,0 +1,58 @@
+
+#include <stdio.h>
+#include <string.h>
+#include <glib.h>
+
+void tracker_extract_ps (gchar *filename, GHashTable *metadata)
+{
+ FILE *f;
+ gchar *line;
+ gsize length = 0;
+ gboolean pageno_atend = FALSE;
+ gboolean header_finished = FALSE;
+
+ if(f = fopen (filename, "r")) {
+ line = NULL;
+ getline (&line, &length, f);
+ while (!feof (f)) {
+ line[strlen(line) - 1] = '\0'; /* overwrite \n char */
+ if (!header_finished
+ && strncmp (line, "%%Copyright:", 12) == 0) {
+ g_hash_table_insert (metadata,
+ g_strdup ("File.Other"), g_strdup (line+13));
+ }
+ else if (!header_finished
+ && strncmp (line, "%%Title:", 8) == 0) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Title"), g_strdup (line+9));
+ }
+ else if (!header_finished
+ && strncmp (line, "%%Creator:", 10) == 0) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Author"), g_strdup (line+11));
+ }
+ else if (!header_finished
+ && strncmp (line, "%%CreationDate:", 15) == 0) {
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.Created"), g_strdup (line+16));
+ }
+ else if (strncmp (line, "%%Pages:", 8) == 0) {
+ if (strcmp (line+9, "(atend)") == 0)
+ pageno_atend = TRUE;
+ else
+ g_hash_table_insert (metadata,
+ g_strdup ("Doc.PageCount"), g_strdup (line+9));
+ }
+ else if (strncmp (line, "%%EndComments", 14) == 0) {
+ header_finished = TRUE;
+ if (!pageno_atend)
+ break;
+ }
+ g_free (line);
+ line = NULL;
+ getline (&line, &length, f);
+ }
+ g_free (line);
+ }
+ fclose (f);
+}
--- /dev/null 2006-08-05 19:53:54.000000000 -0400
+++ src/tracker-extract/tracker-extract-vorbis.c 2006-09-26 18:35:58.000000000 -0400
@@ -0,0 +1,170 @@
+
+#include "config.h"
+
+#ifdef HAVE_VORBIS
+
+#include <stdio.h>
+#include <string.h>
+#include <glib.h>
+#include <vorbis/vorbisfile.h>
+
+/*#include "tracker-utils.h"*/
+
+static struct {
+ char * name;
+ char *meta_name;
+ gboolean writable;
+} tags[] = {
+ {"title", "Audio.Title", FALSE},
+ {"artist", "Audio.Artist", FALSE},
+ {"album", "Audio.Album", FALSE},
+ {"albumartist", "Audio.AlbumArtist", FALSE},
+ {"trackcount", "Audio.AlbumTrackCount", FALSE},
+ {"tracknumber", "Audio.TrackNo", FALSE},
+ {"DiscNo", "Audio.DiscNo", FALSE},
+ {"Performer", "Audio.Performer", FALSE},
+ {"TrackGain", "Audio.TrackGain", FALSE},
+ {"TrackPeakGain", "Audio.TrackPeakGain", FALSE},
+ {"AlbumGain", "Audio.AlbumGain", FALSE},
+ {"AlbumPeakGain", "Audio.AlbumPeakGain", FALSE},
+ {"date", "Audio.ReleaseDate", FALSE},
+ {"comment", "Audio.Comment", FALSE},
+ {"genre", "Audio.Genre", FALSE},
+ {"Codec", "Audio.Codec", FALSE},
+ {"CodecVersion", "Audio.CodecVersion", FALSE},
+ {"Samplerate", "Audio.Samplerate", FALSE},
+ {"Channels", "Audio.Channels", FALSE},
+ {"MBAlbumID", "Audio.MBAlbumID", FALSE},
+ {"MBArtistID", "Audio.MBArtistID", FALSE},
+ {"MBAlbumArtistID", "Audio.MBAlbumArtistID", FALSE},
+ {"MBTrackID", "Audio.MBTrackID", FALSE},
+ {"Lyrics", "Audio.Lyrics", FALSE},
+ {"Copyright", "File.Copyright", FALSE},
+ {"License", "File.License", FALSE},
+ {"Organization", "File.Organization", FALSE},
+ {"Location", "File.Location", FALSE},
+ {"Publisher", "File.Publisher", FALSE},
+ {NULL, NULL, FALSE},
+};
+
+
+static char*
+get_comment (vorbis_comment *vc, char *label)
+{
+ char *tag;
+ char *utf_tag;
+
+ if (vc && (tag = vorbis_comment_query (vc, label, 0)) != NULL) {
+
+ utf_tag = g_locale_to_utf8 (tag, -1, NULL, NULL, NULL);
+
+ /*g_free (tag);*/
+
+ return utf_tag;
+
+ } else {
+ return NULL;
+ }
+
+}
+
+gboolean
+tracker_metadata_ogg_is_writable (const char *meta)
+{
+ int i;
+
+ i = 0;
+ while (tags[i].name != NULL) {
+
+ if (strcmp (tags[i].meta_name, meta) == 0) {
+ return tags[i].writable;
+ }
+
+ i++;
+ }
+
+ return FALSE;
+
+}
+
+
+gboolean
+tracker_metadata_ogg_write (const char *meta_name, const char *value)
+{
+ /* to do */
+ return FALSE;
+}
+
+
+void
+tracker_extract_vorbis (const char *filename, GHashTable *metadata)
+{
+ FILE *oggFile;
+ OggVorbis_File vf;
+ int i;
+
+ oggFile = fopen (filename,"r");
+
+ if (!oggFile) {
+ return;
+ }
+
+ if ( ov_open (oggFile, &vf, NULL, 0) < 0 ) {
+ fclose (oggFile);
+ return;
+ }
+
+ char *tmpComment;
+
+ vorbis_comment *comment;
+
+ if ((comment = ov_comment (&vf, -1)) == NULL) {
+ ov_clear (&vf);
+ return;
+ }
+
+ i = 0;
+ while (tags[i].name != NULL) {
+ tmpComment = get_comment (comment, tags[i].name);
+
+ if (tmpComment) {
+ g_hash_table_insert (metadata, g_strdup (tags[i].meta_name), tmpComment);
+ }
+
+ i++;
+ }
+
+ vorbis_comment_clear(comment);
+
+ /* Bitrate */
+
+ vorbis_info *vi;
+ unsigned int bitrate;
+ char *str_bitrate;
+
+ if ( ( vi = ov_info(&vf, 0)) != NULL ) {
+ bitrate = vi->bitrate_nominal/1000;
+ str_bitrate = g_strdup_printf ("%d", bitrate);
+ g_hash_table_insert (metadata, g_strdup ("Audio.Bitrate"), str_bitrate);
+ }
+
+
+
+ /* Duration */
+
+ int time;
+ char *str_time;
+ if ( ( time = ov_time_total(&vf, -1) ) != OV_EINVAL ) {
+ str_time = g_strdup_printf ("%d", time);
+ g_hash_table_insert (metadata, g_strdup ("Audio.Duration"), str_time);
+ }
+
+ g_hash_table_insert (metadata, g_strdup ("Audio.Codec"), g_strdup ("vorbis"));
+
+ ov_clear(&vf);
+
+}
+
+#else
+#warning "Not building ogg/vorbis metadata extractor"
+#endif /* HAVE_VORBIS */
_______________________________________________
tracker-list mailing list
[email protected]
http://mail.gnome.org/mailman/listinfo/tracker-list
