tracker r2577 - in branches/turtle: . src/tracker-indexer

[pvanhoof svn gnome org]

Author: pvanhoof
Date: Wed Nov 26 12:22:38 2008
New Revision: 2577
URL: http://svn.gnome.org/viewvc/tracker?rev=2577&view=rev

Log:
2008-11-26  Philip Van Hoof  <philip codeminded be>

	* src/tracker-indexer/tracker-removable-device.c: Copying with 
	predicates that we don't know about in our own ontology.



Modified:
   branches/turtle/ChangeLog
   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 12:22:38 2008
@@ -62,22 +62,25 @@
 typedef struct {
 	const gchar *ttl_file;
 	gchar *last_subject;
-	TrackerDataMetadata *metadata;
 	gchar *base;
 	guint amount;
-	TrackerIndexer *indexer;
-	gchar *rdf_type;
 	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;
 
 
 static void
-foreach_in_metadata (TrackerField *field, gpointer value, gpointer user_data)
+foreach_in_hash (gpointer key, gpointer value, gpointer user_data)
 {
 	raptor_statement   *statement;
 	TurtleParseInfo    *item = user_data;
@@ -101,7 +104,7 @@
 	statement->subject = (void *) raptor_new_uri (about_uri);
 	statement->subject_type = RAPTOR_IDENTIFIER_TYPE_RESOURCE;
 
-	statement->predicate = (void *) raptor_new_uri (tracker_field_get_name (field));
+	statement->predicate = (void *) raptor_new_uri (key);
 	statement->predicate_type = RAPTOR_IDENTIFIER_TYPE_RESOURCE;
 
 	statement->object = (unsigned char *) g_strdup (value);
@@ -121,9 +124,9 @@
 optimizer (const gchar *subject, const gchar *rdf_type, TurtleParseInfo *info)
 {
 	info->about_uri = (gchar *) subject;
-	tracker_data_metadata_foreach (info->metadata, 
-				       foreach_in_metadata,
-				       info);
+	g_hash_table_foreach (info->hash, 
+			      foreach_in_hash,
+			      info);
 }
 
 static void
@@ -148,11 +151,16 @@
 
 		info->amount++;
 
-		tracker_data_metadata_free (info->metadata);
+		if (info->metadata)
+			tracker_data_metadata_free (info->metadata);
+		if (info->hash)
+			g_hash_table_destroy (info->hash);
+
 		g_free (info->last_subject);
 		g_free (info->rdf_type);
 		info->last_subject = NULL;
 		info->metadata = NULL;
+		info->hash = NULL;
 		info->rdf_type = NULL;
 	}
 
@@ -223,6 +231,63 @@
 
 }
 
+
+
+static void
+consume_triple_optimizer (void* user_data, const raptor_statement* triple) 
+{
+	TurtleParseInfo *info = user_data;
+	gchar           *subject;
+
+	subject = (gchar *) raptor_uri_as_string ((raptor_uri *) triple->subject);
+
+	if (!info->last_subject || strcmp (subject, info->last_subject) != 0) {
+
+		/* Commit previous subject */
+
+		commit_turtle_parse_info_data (info, TRUE, info->exec_func);
+		info->last_subject = g_strdup (subject);
+		info->hash = g_hash_table_new_full (g_str_hash, g_str_equal,
+						    (GDestroyNotify) g_free,
+						    (GDestroyNotify) g_free);
+	}
+
+	if (triple->object_type == RAPTOR_IDENTIFIER_TYPE_LITERAL) {
+		gchar *predicate;
+
+		predicate = g_strdup ((const gchar *) raptor_uri_as_string ((raptor_uri *) triple->predicate));
+
+		if (strcmp (predicate, "rdf:type") == 0) {
+			g_free (info->rdf_type);
+
+			/* TODO: ontology */
+			/* Change this when Files and Emails becomes File and Email */
+
+			info->rdf_type = g_strdup_printf ("%ss", triple->object);
+		} else
+			g_hash_table_replace (info->hash,
+					      predicate,
+					      g_strdup (triple->object));
+
+	} else if (triple->object_type == RAPTOR_IDENTIFIER_TYPE_RESOURCE) {
+		gchar *key = g_strdup_printf ("file://%s/:", info->base);
+
+		if (strcmp (key, triple->object) == 0) {
+			gchar *predicate;
+
+			predicate = (gchar *) raptor_uri_as_string ((raptor_uri *) triple->predicate);
+
+			g_hash_table_replace (info->hash,
+					      predicate,
+					      g_strdup (triple->object));
+
+		}
+
+		g_free (key);
+	}
+
+}
+
 static void 
 raptor_error (void *user_data, raptor_locator* locator, const char *message)
 {
@@ -284,7 +349,7 @@
 		info->exec_func = (executer_func) optimizer;
 		info->transactions = FALSE;
 
-		raptor_set_statement_handler (parser, info, consume_triple);
+		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);
@@ -292,13 +357,8 @@
 		copy_file = g_strdup (file);
 
 		ptr = strstr (copy_file, "/metadata/metadata.ttl");
-		if (ptr) {
-			/* .cache remains, and will be cut later, just like dummy_file is */
-			*ptr = '\0';
-		} else {
-			g_free (copy_file);
-			copy_file = g_strdup ("/home/pvanhoof/dummy_file");
-		}
+		/* .cache remains, and will be cut later, just like dummy_file is */
+		*ptr = '\0';
 
 		uri_stringa = raptor_uri_filename_to_uri_string (file);
 		uri_stringb = raptor_uri_filename_to_uri_string (copy_file);
@@ -387,14 +447,8 @@
 		copy_file = g_strdup (file);
 
 		ptr = strstr (copy_file, "/metadata/metadata.ttl");
-		if (ptr) {
-			/* .cache remains, and will be cut later, just like dummy_file is */
-			*ptr = '\0';
-		} else {
-			g_free (copy_file);
-			copy_file = g_strdup ("/home/pvanhoof/dummy_file");
-		}
-
+		/* .cache remains, and will be cut later, just like dummy_file is */
+		*ptr = '\0';
 
 		uri_stringa = raptor_uri_filename_to_uri_string (file);
 		uri_stringb = raptor_uri_filename_to_uri_string (copy_file);