tracker r2579 - branches/turtle/src/tracker-indexer

[pvanhoof svn gnome org]

Author: pvanhoof
Date: Wed Nov 26 12:44:26 2008
New Revision: 2579
URL: http://svn.gnome.org/viewvc/tracker?rev=2579&view=rev

Log:
Cleaning up

Modified:
   branches/turtle/src/tracker-indexer/tracker-removable-device.c

Modified: branches/turtle/src/tracker-indexer/tracker-removable-device.c
==============================================================================
--- branches/turtle/src/tracker-indexer/tracker-removable-device.c	(original)
+++ branches/turtle/src/tracker-indexer/tracker-removable-device.c	Wed Nov 26 12:44:26 2008
@@ -54,38 +54,34 @@
 #include <libtracker-data/tracker-data-query.h>
 #include <libtracker-data/tracker-data-update.h>
 
-
-typedef void (*executer_func) (const gchar     *subject, 
-			       const gchar     *rdf_type, 
-			       gpointer         info);
-
 typedef struct {
 	const gchar *ttl_file;
 	gchar *last_subject;
 	gchar *base;
 	guint amount;
-	executer_func exec_func;
-	gboolean transactions;
-
-	/* These are only used by the optimizer */
-	raptor_serializer *serializer;
-	gchar *uri, *about_uri;
-	GHashTable *hash;
-
-	/* These are only used by the in-data handler */
 	TrackerIndexer *indexer;
 	TrackerDataMetadata *metadata;
 	gchar *rdf_type;
-} TurtleParseInfo;
+} TurtleStorerInfo;
 
+typedef struct {
+	const gchar *ttl_file;
+	gchar *last_subject;
+	gchar *base;
+	guint amount;
+	gchar *rdf_type;
+	raptor_serializer *serializer;
+	gchar *uri, *about_uri;
+	GHashTable *hash;
+} TurtleOptimizerInfo;
 
 static void
 foreach_in_hash (gpointer key, gpointer value, gpointer user_data)
 {
-	raptor_statement   *statement;
-	TurtleParseInfo    *item = user_data;
-	const gchar        *about_uri = item->uri;
-	raptor_serializer  *serializer = item->serializer;
+	raptor_statement    *statement;
+	TurtleOptimizerInfo *item = user_data;
+	const gchar         *about_uri = item->last_subject;
+	raptor_serializer   *serializer = item->serializer;
 
 	/* TODO: cope with group values by making them ; separated, perhaps by
 	 * reading the type from the TrackerField? Also, numeric values don't
@@ -120,47 +116,48 @@
 	g_free (statement);
 }
 
-static void
-optimizer (const gchar *subject, const gchar *rdf_type, TurtleParseInfo *info)
-{
-	info->about_uri = (gchar *) subject;
-	g_hash_table_foreach (info->hash, 
-			      foreach_in_hash,
-			      info);
-}
 
 static void
-put_in_tracker_data (const gchar *subject, const gchar *rdf_type, TurtleParseInfo *info)
+commit_turtle_parse_info_optimizer (TurtleOptimizerInfo *info)
 {
-	/* We have it as a URI, database api wants Paths. Update this when
-	 * the database api becomes sane and uses URIs everywhere */
+	if (info->last_subject) {
 
-	tracker_data_replace_service (subject + 7, 
-				      rdf_type, 
-				      info->metadata);
+		g_hash_table_foreach (info->hash, 
+				      foreach_in_hash,
+				      info);
+
+		info->amount++;
+
+		g_hash_table_destroy (info->hash);
+
+		g_free (info->last_subject);
+		g_free (info->rdf_type);
+		info->last_subject = NULL;
+		info->hash = NULL;
+		info->rdf_type = NULL;
+	}
 }
 
 static void
-commit_turtle_parse_info_data (TurtleParseInfo *info, gboolean may_flush, executer_func exec_func)
+commit_turtle_parse_info_storer (TurtleStorerInfo *info, gboolean may_flush)
 {
 	if (info->last_subject) {
 
-		exec_func (info->last_subject, 
-			   info->rdf_type, 
-			   info);
+		/* We have it as a URI, database api wants Paths. Update this when
+		 * the database api becomes sane and uses URIs everywhere */
+
+		tracker_data_replace_service (info->last_subject + 7, 
+					      info->rdf_type, 
+					      info->metadata);
 
 		info->amount++;
 
-		if (info->metadata)
-			tracker_data_metadata_free (info->metadata);
-		if (info->hash)
-			g_hash_table_destroy (info->hash);
+		tracker_data_metadata_free (info->metadata);
 
 		g_free (info->last_subject);
 		g_free (info->rdf_type);
 		info->last_subject = NULL;
 		info->metadata = NULL;
-		info->hash = NULL;
 		info->rdf_type = NULL;
 	}
 
@@ -169,20 +166,20 @@
 	 * moment */
 
 	if (may_flush && info->amount > 100) {
-		if (info->transactions)
-			tracker_indexer_commit_transaction (info->indexer);
+		tracker_indexer_commit_transaction (info->indexer);
 		g_main_context_iteration (NULL, FALSE);
-		if (info->transactions)
-			tracker_indexer_open_transaction (info->indexer);
+		tracker_indexer_open_transaction (info->indexer);
 		info->amount = 0;
 	}
 }
 
+
+
 static void
-consume_triple (void* user_data, const raptor_statement* triple) 
+consume_triple_storer (void* user_data, const raptor_statement* triple) 
 {
-	TurtleParseInfo *info = user_data;
-	gchar           *subject;
+	TurtleStorerInfo *info = user_data;
+	gchar            *subject;
 
 	subject = (gchar *) raptor_uri_as_string ((raptor_uri *) triple->subject);
 
@@ -190,7 +187,7 @@
 
 		/* Commit previous subject */
 
-		commit_turtle_parse_info_data (info, TRUE, info->exec_func);
+		commit_turtle_parse_info_storer (info, TRUE);
 		info->last_subject = g_strdup (subject);
 		info->metadata = tracker_data_metadata_new ();
 	}
@@ -232,12 +229,11 @@
 }
 
 
-
 static void
 consume_triple_optimizer (void* user_data, const raptor_statement* triple) 
 {
-	TurtleParseInfo *info = user_data;
-	gchar           *subject;
+	TurtleOptimizerInfo *info = user_data;
+	gchar               *subject;
 
 	subject = (gchar *) raptor_uri_as_string ((raptor_uri *) triple->subject);
 
@@ -245,7 +241,7 @@
 
 		/* Commit previous subject */
 
-		commit_turtle_parse_info_data (info, TRUE, info->exec_func);
+		commit_turtle_parse_info_optimizer (info);
 		info->last_subject = g_strdup (subject);
 		info->hash = g_hash_table_new_full (g_str_hash, g_str_equal,
 						    (GDestroyNotify) g_free,
@@ -291,8 +287,9 @@
 static void 
 raptor_error (void *user_data, raptor_locator* locator, const char *message)
 {
-	TurtleParseInfo *info = user_data;
-	g_message ("RAPTOR parse error: %s for %s\n", message, info->ttl_file);
+	g_message ("RAPTOR parse error: %s for %s\n", 
+		   message, 
+		   (gchar *) user_data);
 }
 
 
@@ -308,14 +305,14 @@
 				 "metadata", "metadata.ttl", NULL);
 
 	if (g_file_test (file, G_FILE_TEST_EXISTS)) {
-		unsigned char   *uri_stringa, *uri_stringb;
-		raptor_uri      *uri, *base_uri, *suri;
-		static gboolean  has_init = FALSE;
-		raptor_parser   *parser;
-		TurtleParseInfo *info;
-		gchar           *copy_file, 
-				*ptr, *tmp_file;
-		FILE            *target_file;
+		unsigned char       *uri_stringa, *uri_stringb;
+		raptor_uri          *uri, *base_uri, *suri;
+		static gboolean      has_init = FALSE;
+		raptor_parser       *parser;
+		TurtleOptimizerInfo *info;
+		gchar               *copy_file, 
+				    *ptr, *tmp_file;
+		FILE                *target_file;
 
 		tmp_file = g_strdup_printf ("%s.tmp", file);
 
@@ -334,7 +331,7 @@
 
 		parser = raptor_new_parser ("turtle");
 
-		info = g_slice_new0 (TurtleParseInfo);
+		info = g_slice_new0 (TurtleOptimizerInfo);
 
 		info->serializer = raptor_new_serializer ("turtle");
 		suri = raptor_new_uri ("/");
@@ -342,17 +339,12 @@
 						       suri, target_file);
 
 		info->ttl_file = file;
-		info->indexer = g_object_ref (indexer);
 		info->amount = 0;
 
-		/* The optimizer simply writes all triples to the serializer */
-		info->exec_func = (executer_func) optimizer;
-		info->transactions = FALSE;
-
 		raptor_set_statement_handler (parser, info, consume_triple_optimizer);
-		raptor_set_fatal_error_handler (parser, info, raptor_error);
-		raptor_set_error_handler (parser, info, raptor_error);
-		raptor_set_warning_handler (parser, info, raptor_error);
+		raptor_set_fatal_error_handler (parser, file, raptor_error);
+		raptor_set_error_handler (parser, file, raptor_error);
+		raptor_set_warning_handler (parser, file, raptor_error);
 
 		copy_file = g_strdup (file);
 
@@ -376,15 +368,14 @@
 		raptor_parse_file (parser, uri, base_uri);
 
 		/* Commit final subject (or loop doesn't handle the very last) */
-		commit_turtle_parse_info_data (info, FALSE, info->exec_func);
+		commit_turtle_parse_info_optimizer (info);
 
 		raptor_serialize_end (info->serializer);
 		raptor_free_serializer(info->serializer);
 		fclose (target_file);
 
 		g_free (copy_file);
-		g_object_unref (info->indexer);
-		g_slice_free (TurtleParseInfo, info);
+		g_slice_free (TurtleOptimizerInfo, info);
 
 		raptor_free_parser (parser);
 
@@ -418,31 +409,27 @@
 				 "metadata", "metadata.ttl", NULL);
 
 	if (g_file_test (file, G_FILE_TEST_EXISTS)) {
-		unsigned char   *uri_stringa, *uri_stringb;
-		raptor_uri      *uri, *base_uri;
-		static gboolean  has_init = FALSE;
-		raptor_parser   *parser;
-		TurtleParseInfo *info;
-		gchar           *copy_file, *ptr;
+		unsigned char    *uri_stringa, *uri_stringb;
+		raptor_uri       *uri, *base_uri;
+		static gboolean   has_init = FALSE;
+		raptor_parser    *parser;
+		TurtleStorerInfo *info;
+		gchar            *copy_file, *ptr;
 
 		raptor_init();
 
 		parser = raptor_new_parser ("turtle");
 
-		info = g_slice_new0 (TurtleParseInfo);
+		info = g_slice_new0 (TurtleStorerInfo);
 
 		info->ttl_file = file;
 		info->indexer = g_object_ref (indexer);
 		info->amount = 0;
 
-		/* This handler puts the triples into our store */
-		info->exec_func = (executer_func) put_in_tracker_data;
-		info->transactions = TRUE;
-
-		raptor_set_statement_handler (parser, info, consume_triple);
-		raptor_set_fatal_error_handler (parser, info, raptor_error);
-		raptor_set_error_handler (parser, info, raptor_error);
-		raptor_set_warning_handler (parser, info, raptor_error);
+		raptor_set_statement_handler (parser, info, consume_triple_storer);
+		raptor_set_fatal_error_handler (parser, file, raptor_error);
+		raptor_set_error_handler (parser, file, raptor_error);
+		raptor_set_warning_handler (parser, file, raptor_error);
 
 		copy_file = g_strdup (file);
 
@@ -474,7 +461,7 @@
 		raptor_parse_file (parser, uri, base_uri);
 
 		/* Commit final subject (or loop doesn't handle the very last) */
-		commit_turtle_parse_info_data (info, FALSE, info->exec_func);
+		commit_turtle_parse_info_storer (info, FALSE);
 
 		/* g_timer_stop (timer);
 		g_print ("\nTIME: %f\n", g_timer_elapsed (timer, NULL)); */
@@ -486,7 +473,7 @@
 
 		g_free (copy_file);
 		g_object_unref (info->indexer);
-		g_slice_free (TurtleParseInfo, info);
+		g_slice_free (TurtleStorerInfo, info);
 
 		raptor_free_parser (parser);