added safeguards to help prevent crashers when indexing

svn path=/trunk/; revision=786

4 files changed, 35 insertions, 19 deletions

diff --git a/ChangeLog b/ChangeLog
index 461166e31..c131ed095 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,7 +1,8 @@
 2007-08-19 Patch from Marcus Fritzsch <fritschy at googlemail com>
 	
 	* Added option to skip mounted directories when indexing
-
+	* added safeguards to html and imagemagick extractors to prevent crashers
+	* added safeguard to app indexing
 
 
 2007-08-19 Michael Biebl <mbiebl at gmail com>
diff --git a/src/tracker-extract/tracker-extract-html.c b/src/tracker-extract/tracker-extract-html.c
index 99292ea87..1231f6873 100644
--- a/src/tracker-extract/tracker-extract-html.c
+++ b/src/tracker-extract/tracker-extract-html.c
@@ -26,8 +26,8 @@
 #include <libxml/HTMLparser.h>
 
 typedef enum {
-		READ_TITLE,
-	} tag_type;
+	READ_TITLE,
+} tag_type;
 
 typedef struct {
 	GHashTable *metadata;
@@ -37,27 +37,31 @@ typedef struct {
 gboolean
 has_attribute( const xmlChar ** atts, const char *attr, const char*val )
 {
-        if (atts == NULL || attr == NULL || val == NULL)
-        return FALSE;
+	if (! (atts && attr && val)) {
+		return FALSE;
+	}
 
 	int i;
-	for ( i = 0; atts[i]; i+=2 )
-	{
+	for ( i = 0; atts[i] && atts[i+1]; i+=2 ) {
 		if ( strcasecmp((char*)atts[i],attr) == 0 ) {
-			if ( !val || strcasecmp((char*)atts[i+1],val) == 0 ) {
+			if (strcasecmp((char*)atts[i+1],val) == 0) {
 				return TRUE;
 			}
 		}
 	}
+
 	return FALSE;
 }
 
 const xmlChar *
 lookup_attribute( const xmlChar **atts, const char *attr )
 {
+	if (!atts || !attr) {
+		return NULL;
+	}
+
 	int i;
-	for ( i = 0; atts[i]; i+=2 )
-	{
+	for ( i = 0; atts[i] && atts[i+1]; i+=2 ) {
 		if ( strcasecmp((char*)atts[i],attr) == 0 ) {
 			return atts[i+1];
 		}
@@ -69,27 +73,40 @@ lookup_attribute( const xmlChar **atts, const char *attr )
 void
 startElement (void * info, const xmlChar * name, const xmlChar ** atts)
 {
+	if (! (info && name && atts)) {
+		return;
+	}
+
 	/* Look for RDFa triple describing the license */
 	if ( strcasecmp((char*)name,"a") == 0 ) {
+
 		/* This tag is a license.  Ignore, however, if it is referring to another document */
 		if ( has_attribute(atts,"rel","license") && !has_attribute(atts,"about",NULL) ) {
+
 			const xmlChar *href = lookup_attribute(atts,"href");
 			if ( href ) {
 				g_hash_table_insert (((HTMLParseInfo *)info)->metadata, g_strdup ("File:License"),
 				                     g_strdup( (char*)href ));
 			}
 		}
+
 	} else if ( strcasecmp((char*)name,"title") == 0 ) {
+
 		((HTMLParseInfo *)info)->current = READ_TITLE;
+
 	} else if ( strcasecmp((char*)name,"meta") == 0 ) {
+
 		if ( has_attribute(atts,"name","Author") ) {
+
 			const xmlChar *author = lookup_attribute(atts,"content");
 			if ( author ) {
 				g_hash_table_insert (((HTMLParseInfo *)info)->metadata, g_strdup ("Doc:Author"),
 				                     g_strdup( (char*)author ));
 			}
 		}
+
 		if ( has_attribute(atts,"name","DC.Description") ) {
+
 			const xmlChar *desc = lookup_attribute(atts,"content");
 			if ( desc ) {
 				g_hash_table_insert (((HTMLParseInfo *)info)->metadata, g_strdup ("Doc:Comments"),
diff --git a/src/tracker-extract/tracker-extract-imagemagick.c b/src/tracker-extract/tracker-extract-imagemagick.c
index 8eec802f4..2db24cdfe 100644
--- a/src/tracker-extract/tracker-extract-imagemagick.c
+++ b/src/tracker-extract/tracker-extract-imagemagick.c
@@ -34,6 +34,8 @@ tracker_extract_imagemagick (gchar *filename, GHashTable *metadata)
 	gchar **lines;
 	gint  exit_status;
 
+	g_return_if_fail (filename != NULL);
+
 	/* imagemagick crashes trying to extract from xcf files */
 	if (g_str_has_suffix (filename, ".xcf")) {
 		return;
diff --git a/src/trackerd/tracker-apps.c b/src/trackerd/tracker-apps.c
index 9253cbac6..e4706dff3 100644
--- a/src/trackerd/tracker-apps.c
+++ b/src/trackerd/tracker-apps.c
@@ -92,7 +92,7 @@ tracker_db_index_application (DBConnection *db_con, FileInfo *info)
 	GKeyFile *key_file = NULL;
 
 	gchar *type = NULL;
-	gboolean is_hidden = TRUE;
+	gboolean is_hidden = FALSE;
 	char *hidden = NULL;
 	gchar *tmp_str = NULL;
 	gchar desktop_entry[] = { "Desktop Entry" };
@@ -125,9 +125,10 @@ tracker_db_index_application (DBConnection *db_con, FileInfo *info)
 		if (hidden) {
 			is_hidden = (strcasecmp (hidden, "true") == 0);
 			g_free (hidden);
+			hidden = NULL;
 		}
 
-		if (!hidden && type && (strcasecmp (type, "Application") == 0)) {
+		if (!is_hidden && type && (strcasecmp (type, "Application") == 0)) {
 
 			meta_table = g_hash_table_new (g_str_hash, g_str_equal);
 
@@ -163,9 +164,7 @@ tracker_db_index_application (DBConnection *db_con, FileInfo *info)
 					tracker_add_metadata_to_table  (meta_table, app_mime_type, g_strdup (*array));
 				}
 
-				if (mimes) {
-					g_strfreev (mimes);
-				}
+				g_strfreev (mimes);
 			} 
 
 			if ((categories = g_key_file_get_string_list (key_file, desktop_entry, "Categories", NULL, NULL)) != NULL) {
@@ -175,10 +174,7 @@ tracker_db_index_application (DBConnection *db_con, FileInfo *info)
 					tracker_add_metadata_to_table  (meta_table, app_categories, g_strdup (*array));
 				}
 
-
-				if (categories) {
-					g_strfreev (categories);
-				}
+				g_strfreev (categories);
 			} 
 
 			char *fname = g_path_get_basename (info->uri);