tracker r2074 - in branches/indexer-split: . src/libtracker-common src/libtracker-db src/tracker-extract src/tracker-indexer src/tracker-utils src/trackerd tests/libtracker-common tests/libtracker-db
- From: mr svn gnome org
- To: svn-commits-list gnome org
- Subject: tracker r2074 - in branches/indexer-split: . src/libtracker-common src/libtracker-db src/tracker-extract src/tracker-indexer src/tracker-utils src/trackerd tests/libtracker-common tests/libtracker-db
- Date: Thu, 14 Aug 2008 15:25:30 +0000 (UTC)
Author: mr
Date: Thu Aug 14 15:25:30 2008
New Revision: 2074
URL: http://svn.gnome.org/viewvc/tracker?rev=2074&view=rev
Log:
* src/libtracker-db/tracker-db-index.c: (tracker_db_index_flush):
If the index is not open when we try to flush, try and open it
first.
* src/tracker-extract/tracker-extract-tiff.c:
(tracker_extract_tiff):
* tests/libtracker-db/tracker-db-dbus-test.c:
(test_dbus_query_result_to_ptr_array): Fixed compiler warnings.
* src/tracker-indexer/tracker-indexer.c: Fixed a nasty crash I
introduced last commit and clean up the merge function for words
tables. Plus make sure we free basename and dirname in
delete_item(), this was quite a leak.
* src/tracker-utils/tracker-files.c: (main): Fixed printing a %d
as a %s.
* src/trackerd/tracker-db.c: (update_metadata_index):
* tests/libtracker-common/tracker-parser-test.c: Use
g_hash_table_unref() instead of the custom function we had for the
parser for freeing hash tables.
* src/trackerd/tracker-processor.c: (tracker_processor_stop):
Fixed a warning where the trackerd would stop before it started
processing (due to Ctrl+C) and the private->timer was NULL, so
using it caused warnings.
Modified:
branches/indexer-split/ChangeLog
branches/indexer-split/src/libtracker-common/tracker-parser.c
branches/indexer-split/src/libtracker-common/tracker-parser.h
branches/indexer-split/src/libtracker-db/tracker-db-index.c
branches/indexer-split/src/tracker-extract/tracker-extract-tiff.c
branches/indexer-split/src/tracker-indexer/tracker-indexer.c
branches/indexer-split/src/tracker-utils/tracker-files.c
branches/indexer-split/src/trackerd/tracker-db.c
branches/indexer-split/src/trackerd/tracker-processor.c
branches/indexer-split/tests/libtracker-common/tracker-parser-test.c
branches/indexer-split/tests/libtracker-db/tracker-db-dbus-test.c
Modified: branches/indexer-split/src/libtracker-common/tracker-parser.c
==============================================================================
--- branches/indexer-split/src/libtracker-common/tracker-parser.c (original)
+++ branches/indexer-split/src/libtracker-common/tracker-parser.c Thu Aug 14 15:25:30 2008
@@ -166,7 +166,7 @@
gboolean filter_words,
gboolean filter_numbers,
gboolean delimit_hyphen,
- const gchar **index_word)
+ gchar **index_word)
{
TrackerParserWordType word_type;
gunichar word[64];
@@ -286,60 +286,59 @@
word[length -1] = c;
}
- if (is_valid) {
- if (word_type == TRACKER_PARSER_WORD_NUM) {
- if (!filter_numbers || length >= INDEX_NUMBER_MIN_LENGTH) {
- *index_word = g_ucs4_to_utf8 (word, length, NULL, NULL, NULL);
- }
+ if (!is_valid) {
+ return p;
+ }
+
+ if (word_type == TRACKER_PARSER_WORD_NUM) {
+ if (!filter_numbers || length >= INDEX_NUMBER_MIN_LENGTH) {
+ *index_word = g_ucs4_to_utf8 (word, length, NULL, NULL, NULL);
+ }
+ } else if (length >= min_word_length) {
+ const gchar *stem_word;
+ gchar *stripped_word;
+ gchar *str;
+ gchar *utf8;
+ guint32 len;
+
+ utf8 = g_ucs4_to_utf8 (word, length, NULL, &bytes, NULL);
+
+ if (!utf8) {
+ return p;
+ }
+
+ if (do_strip) {
+ stripped_word = strip_word (utf8, bytes, &len);
} else {
- if (length >= min_word_length) {
- gchar *str = NULL;
- gchar *tmp;
- guint32 len;
- gchar *utf8;
-
- utf8 = g_ucs4_to_utf8 (word, length, NULL, &bytes, NULL);
-
- if (!utf8) {
- return p;
- }
-
- if (do_strip) {
- str = strip_word (utf8, bytes, &len);
- }
-
- if (!str) {
- tmp = g_utf8_normalize (utf8, bytes, G_NORMALIZE_NFC);
- } else {
- tmp = g_utf8_normalize (str, len, G_NORMALIZE_NFC);
- g_free (str);
- }
-
- g_free (utf8);
-
- *index_word = tracker_language_stem_word (language,
- tmp,
- strlen (tmp));
- g_free (tmp);
-
- if (filter_words && is_stop_word (language, *index_word)) {
- *index_word = NULL;
- }
- }
+ stripped_word = NULL;
}
- }
+
+ if (!stripped_word) {
+ str = g_utf8_normalize (utf8,
+ bytes,
+ G_NORMALIZE_NFC);
+ } else {
+ str = g_utf8_normalize (stripped_word,
+ len,
+ G_NORMALIZE_NFC);
+ g_free (stripped_word);
+ }
+
+ g_free (utf8);
+
+ stem_word = tracker_language_stem_word (language,
+ str,
+ strlen (str));
+ g_free (str);
+
+ if (!filter_words || !is_stop_word (language, stem_word)) {
+ *index_word = g_strdup (stem_word);
+ }
+ }
return p;
}
-static void
-delete_words (gpointer key,
- gpointer value,
- gpointer user_data)
-{
- g_free (key);
-}
-
gchar *
tracker_parser_text_to_string (const gchar *txt,
TrackerLanguage *language,
@@ -425,8 +424,8 @@
parsed_text = g_string_free (strs, FALSE);
return g_strstrip (parsed_text);
} else {
- GString *str;
- const gchar *word;
+ GString *str;
+ gchar *word;
str = g_string_new (" ");
@@ -444,6 +443,7 @@
if (word) {
g_string_append (str, word);
g_string_append_c (str, ' ');
+ g_free (word);
}
if (!p || !*p) {
@@ -494,7 +494,10 @@
/* Use this for already processed text only */
if (!word_table) {
- word_table = g_hash_table_new (g_str_hash, g_str_equal);
+ word_table = g_hash_table_new_full (g_str_hash,
+ g_str_equal,
+ g_free,
+ NULL);
}
if (!txt || weight == 0) {
@@ -525,11 +528,11 @@
}
static gboolean
-word_table_increment (GHashTable *word_table,
- const gchar *index_word,
- gint weight,
- gint total_words,
- gint max_words_to_index)
+word_table_increment (GHashTable *word_table,
+ gchar *index_word,
+ gint weight,
+ gint total_words,
+ gint max_words_to_index)
{
gboolean update_count;
@@ -542,10 +545,11 @@
p = g_hash_table_lookup (word_table, index_word);
count = GPOINTER_TO_INT (p);
- /* Take a copy the first time */
- g_hash_table_insert (word_table,
- count == 0 ? g_strdup (index_word) : (gchar*) index_word,
- GINT_TO_POINTER (count + weight));
+ g_hash_table_replace (word_table,
+ index_word,
+ GINT_TO_POINTER (count + weight));
+ } else {
+ g_free (index_word);
}
return update_count;
@@ -571,7 +575,10 @@
g_return_val_if_fail (language != NULL, NULL);
if (!word_table) {
- word_table = g_hash_table_new (g_str_hash, g_str_equal);
+ word_table = g_hash_table_new_full (g_str_hash,
+ g_str_equal,
+ g_free,
+ NULL);
total_words = 0;
} else {
total_words = g_hash_table_size (word_table);
@@ -612,31 +619,29 @@
end_word = g_utf8_offset_to_pointer (txt, i);
if (start_word != end_word) {
- gchar *s;
+ gchar *str;
gchar *index_word;
gboolean was_updated;
/* Normalize word */
- s = g_utf8_casefold (start_word, end_word - start_word);
- if (!s) {
+ str = g_utf8_casefold (start_word, end_word - start_word);
+ if (!str) {
continue;
}
- index_word = g_utf8_normalize (s, -1, G_NORMALIZE_NFC);
- g_free (s);
+ index_word = g_utf8_normalize (str, -1, G_NORMALIZE_NFC);
+ g_free (str);
if (!index_word) {
continue;
}
-
- total_words++;
+ total_words++;
was_updated = word_table_increment (word_table,
index_word,
weight,
total_words,
max_words_to_index);
- g_free (index_word);
if (!was_updated) {
break;
@@ -653,7 +658,7 @@
g_free (attrs);
} else {
- const gchar *word;
+ gchar *word;
while (TRUE) {
i++;
@@ -686,12 +691,3 @@
return word_table;
}
-
-void
-tracker_parser_text_free (GHashTable *table)
-{
- if (table) {
- g_hash_table_foreach (table, delete_words, NULL);
- g_hash_table_destroy (table);
- }
-}
Modified: branches/indexer-split/src/libtracker-common/tracker-parser.h
==============================================================================
--- branches/indexer-split/src/libtracker-common/tracker-parser.h (original)
+++ branches/indexer-split/src/libtracker-common/tracker-parser.h Thu Aug 14 15:25:30 2008
@@ -66,7 +66,6 @@
TrackerLanguage *language,
gint max_word_length,
gint min_word_length);
-void tracker_parser_text_free (GHashTable *table);
G_END_DECLS
Modified: branches/indexer-split/src/libtracker-db/tracker-db-index.c
==============================================================================
--- branches/indexer-split/src/libtracker-db/tracker-db-index.c (original)
+++ branches/indexer-split/src/libtracker-db/tracker-db-index.c Thu Aug 14 15:25:30 2008
@@ -821,6 +821,13 @@
g_mutex_lock (priv->mutex);
+ if (!priv->index) {
+ g_debug ("Index was not open for flush, opening first...");
+ g_mutex_unlock (priv->mutex);
+ tracker_db_index_open (index);
+ g_mutex_lock (priv->mutex);
+ }
+
if (priv->index) {
size = g_hash_table_size (priv->cache);
g_debug ("Flushing index with %d items in cache", size);
@@ -829,8 +836,7 @@
cache_flush_foreach,
priv->index);
} else {
- g_warning ("Flushing index while closed, "
- "this indicates a problem in the software");
+ g_warning ("Could not open index, cache was not flushed");
size = 0;
}
Modified: branches/indexer-split/src/tracker-extract/tracker-extract-tiff.c
==============================================================================
--- branches/indexer-split/src/tracker-extract/tracker-extract-tiff.c (original)
+++ branches/indexer-split/src/tracker-extract/tracker-extract-tiff.c Thu Aug 14 15:25:30 2008
@@ -99,8 +99,6 @@
{
TIFF *image;
long exifOffset;
- gchar *xmpOffset;
- uint32 size;
TiffTag *tag;
Modified: branches/indexer-split/src/tracker-indexer/tracker-indexer.c
==============================================================================
--- branches/indexer-split/src/tracker-indexer/tracker-indexer.c (original)
+++ branches/indexer-split/src/tracker-indexer/tracker-indexer.c Thu Aug 14 15:25:30 2008
@@ -881,7 +881,7 @@
}
if (full_parsing) {
- parsed = tracker_parser_text (parsed,
+ parsed = tracker_parser_text (NULL,
text,
weight_factor,
indexer->private->language,
@@ -894,13 +894,13 @@
/* We dont know the exact property weight.
Big value works.
*/
- parsed = tracker_parser_text_fast (parsed,
+ parsed = tracker_parser_text_fast (NULL,
text,
weight_factor);
}
g_hash_table_iter_init (&iter, parsed);
-
+
while (g_hash_table_iter_next (&iter, &key, &value)) {
tracker_db_index_add_word (indexer->private->index,
key,
@@ -908,8 +908,8 @@
service_type,
GPOINTER_TO_INT (value));
}
-
- tracker_parser_text_free (parsed);
+
+ g_hash_table_unref (parsed);
}
static void
@@ -1014,28 +1014,34 @@
gpointer user_data)
{
GHashTable *new_table;
- gpointer k = NULL;
- gpointer v = NULL;
+ gpointer k;
+ gpointer v;
gchar *word;
- gint score;
+ gint new_score;
word = key;
- score = GPOINTER_TO_INT (value);
+ new_score = GPOINTER_TO_INT (value);
new_table = user_data;
- if (!g_hash_table_lookup_extended (new_table, word, &k, &v)) {
- g_hash_table_insert (new_table,
- g_strdup (word),
- GINT_TO_POINTER (0 - score));
- } else {
- if ((GPOINTER_TO_INT (v) - score) != 0) {
+ if (g_hash_table_lookup_extended (new_table, word, &k, &v)) {
+ gint old_score;
+ gint calculated_score;
+
+ old_score = GPOINTER_TO_INT (v);
+ calculated_score = old_score - new_score;
+
+ if (calculated_score != 0) {
g_hash_table_insert (new_table,
- (gchar *) word,
- GINT_TO_POINTER (GPOINTER_TO_INT (v) - score));
+ g_strdup (word),
+ GINT_TO_POINTER (calculated_score));
} else {
/* The word is the same in old and new text */
g_hash_table_remove (new_table, word);
- }
+ }
+ } else {
+ g_hash_table_insert (new_table,
+ g_strdup (word),
+ GINT_TO_POINTER (0 - new_score));
}
}
@@ -1152,8 +1158,8 @@
tracker_service_get_id (service_def),
new_words);
- tracker_parser_text_free (old_words);
- tracker_parser_text_free (new_words);
+ g_hash_table_unref (old_words);
+ g_hash_table_unref (new_words);
}
g_free (old_text);
@@ -1198,6 +1204,8 @@
if (service_type_id == 0) {
/* File didn't exist, nothing to delete */
+ g_free (dirname);
+ g_free (basename);
return;
}
Modified: branches/indexer-split/src/tracker-utils/tracker-files.c
==============================================================================
--- branches/indexer-split/src/tracker-utils/tracker-files.c (original)
+++ branches/indexer-split/src/tracker-utils/tracker-files.c Thu Aug 14 15:25:30 2008
@@ -111,7 +111,7 @@
if (error) {
g_printerr ("%s:'%s', %s\n",
_("Could not get files by service type"),
- type,
+ service,
error->message);
g_error_free (error);
Modified: branches/indexer-split/src/trackerd/tracker-db.c
==============================================================================
--- branches/indexer-split/src/trackerd/tracker-db.c (original)
+++ branches/indexer-split/src/trackerd/tracker-db.c Thu Aug 14 15:25:30 2008
@@ -261,8 +261,8 @@
tracker_db_update_differential_index (old_table, new_table, id, sid);
#endif
- tracker_parser_text_free (old_table);
- tracker_parser_text_free (new_table);
+ g_hash_table_unref (old_table);
+ g_hash_table_unref (new_table);
}
static gchar *
Modified: branches/indexer-split/src/trackerd/tracker-processor.c
==============================================================================
--- branches/indexer-split/src/trackerd/tracker-processor.c (original)
+++ branches/indexer-split/src/trackerd/tracker-processor.c Thu Aug 14 15:25:30 2008
@@ -1305,6 +1305,8 @@
void
tracker_processor_stop (TrackerProcessor *processor)
{
+ gdouble elapsed;
+
g_return_if_fail (TRACKER_IS_PROCESSOR (processor));
if (processor->private->interrupted) {
@@ -1323,10 +1325,15 @@
g_message ("Process %s\n",
processor->private->finished ? "has finished" : "been stopped");
- g_timer_stop (processor->private->timer);
+ if (processor->private->timer) {
+ g_timer_stop (processor->private->timer);
+ elapsed = g_timer_elapsed (processor->private->timer, NULL);
+ } else {
+ elapsed = 0;
+ }
g_message ("Total time taken : %4.4f seconds",
- g_timer_elapsed (processor->private->timer, NULL));
+ elapsed);
g_message ("Total directories: %d (%d ignored)",
processor->private->directories_found,
processor->private->directories_ignored);
Modified: branches/indexer-split/tests/libtracker-common/tracker-parser-test.c
==============================================================================
--- branches/indexer-split/tests/libtracker-common/tracker-parser-test.c (original)
+++ branches/indexer-split/tests/libtracker-common/tracker-parser-test.c Thu Aug 14 15:25:30 2008
@@ -51,7 +51,7 @@
g_assert_cmpint (g_hash_table_size (result), ==, 5);
- tracker_parser_text_free (result);
+ g_hash_table_unref (result);
}
/*
@@ -75,7 +75,7 @@
g_hash_table_foreach (result, assert_key_length, GINT_TO_POINTER (max_length));
g_assert_cmpint (g_hash_table_size (result), ==, 8);
- tracker_parser_text_free (result);
+ g_hash_table_unref (result);
}
/*
@@ -100,7 +100,7 @@
g_assert_cmpint (g_hash_table_size (result), ==, 4);
- tracker_parser_text_free (result);
+ g_hash_table_unref (result);
result = NULL;
/* No filter */
@@ -117,7 +117,7 @@
g_assert (g_hash_table_lookup (result, "12345678"));
- tracker_parser_text_free (result);
+ g_hash_table_unref (result);
result = NULL;
}
Modified: branches/indexer-split/tests/libtracker-db/tracker-db-dbus-test.c
==============================================================================
--- branches/indexer-split/tests/libtracker-db/tracker-db-dbus-test.c (original)
+++ branches/indexer-split/tests/libtracker-db/tracker-db-dbus-test.c Thu Aug 14 15:25:30 2008
@@ -120,7 +120,6 @@
{
TrackerDBResultSet *result_set = NULL;
GPtrArray *result = NULL;
- gint count;
/* NULL */
result = tracker_dbus_query_result_to_ptr_array (result_set);
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]
