[tracker/pdfmem-for-master] tracker-extract: Avoid a strdup of some memory in PDF extractor

[Philip Van Hoof <pvanhoof src gnome org>]

commit 5faa8162005bee7e5a1bd8e25c58f0d8c9f9ab9c
Author: Philip Van Hoof <philip codeminded be>
Date:   Thu Apr 29 16:34:36 2010 +0200

    tracker-extract: Avoid a strdup of some memory in PDF extractor

 src/tracker-extract/tracker-extract-pdf.cpp |  331 +++++++++++++--------------
 1 files changed, 164 insertions(+), 167 deletions(-)
---
diff --git a/src/tracker-extract/tracker-extract-pdf.cpp b/src/tracker-extract/tracker-extract-pdf.cpp
index 0441569..d812bd4 100644
--- a/src/tracker-extract/tracker-extract-pdf.cpp
+++ b/src/tracker-extract/tracker-extract-pdf.cpp
@@ -499,7 +499,6 @@ extract_pdf (const gchar          *uri,
 	PDFData pd = { 0 }; /* actual data */
 	PDFData md = { 0 }; /* for merging */
 	PDFDoc *document;
-	gchar *xml = NULL;
 	gchar *content;
 	guint n_words;
 	Object obj;
@@ -559,223 +558,221 @@ extract_pdf (const gchar          *uri,
 	}
 	obj.free ();
 
+
 	catalog = document->getCatalog ();
 	if (catalog && catalog->isOk ()) {
 		GooString *s = catalog->readMetadata ();
 		if ( s != NULL ) {
-			xml = g_strdup (s->getCString());
-			delete s;
-		}
-	}
+			const gchar *xml;
 
-	if (xml) {
-		xd = tracker_xmp_new (xml, strlen (xml), uri);
+			xml = s->getCString();
+			xd = tracker_xmp_new (xml, strlen (xml), uri);
 
-		if (!xd) {
-			xd = g_new0 (TrackerXmpData, 1);
-		}
+			if (!xd) {
+				xd = g_new0 (TrackerXmpData, 1);
+			}
 
-		g_free (xml);
-		xml = NULL;
+			delete s;
 
-		/* The casts here are well understood and known */
-		md.title = (gchar *) tracker_coalesce_strip (3, pd.title, xd->title, xd->title2, xd->pdf_title);
-		md.subject = (gchar *) tracker_coalesce_strip (2, pd.subject, xd->subject);
-		md.date = (gchar *) tracker_coalesce_strip (3, pd.creation_date, xd->date, xd->time_original);
-		md.author = (gchar *) tracker_coalesce_strip (2, pd.author, xd->creator);
+			/* The casts here are well understood and known */
+			md.title = (gchar *) tracker_coalesce_strip (3, pd.title, xd->title, xd->title2, xd->pdf_title);
+			md.subject = (gchar *) tracker_coalesce_strip (2, pd.subject, xd->subject);
+			md.date = (gchar *) tracker_coalesce_strip (3, pd.creation_date, xd->date, xd->time_original);
+			md.author = (gchar *) tracker_coalesce_strip (2, pd.author, xd->creator);
 
-		write_pdf_data (md, metadata);
+			write_pdf_data (md, metadata);
 
-		if (xd->keywords) {
-			insert_keywords (metadata, xd->keywords);
-		}
+			if (xd->keywords) {
+				insert_keywords (metadata, xd->keywords);
+			}
 
-		if (xd->pdf_keywords) {
-			insert_keywords (metadata, xd->pdf_keywords);
-		}
+			if (xd->pdf_keywords) {
+				insert_keywords (metadata, xd->pdf_keywords);
+			}
 
-		if (xd->publisher) {
-			tracker_sparql_builder_predicate (metadata, "nco:publisher");
-			tracker_sparql_builder_object_blank_open (metadata);
-			tracker_sparql_builder_predicate (metadata, "a");
-			tracker_sparql_builder_object (metadata, "nco:Contact");
-			tracker_sparql_builder_predicate (metadata, "nco:fullname");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->publisher);
-			tracker_sparql_builder_object_blank_close (metadata);
-		}
+			if (xd->publisher) {
+				tracker_sparql_builder_predicate (metadata, "nco:publisher");
+				tracker_sparql_builder_object_blank_open (metadata);
+				tracker_sparql_builder_predicate (metadata, "a");
+				tracker_sparql_builder_object (metadata, "nco:Contact");
+				tracker_sparql_builder_predicate (metadata, "nco:fullname");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->publisher);
+				tracker_sparql_builder_object_blank_close (metadata);
+			}
 
-		if (xd->type) {
-			tracker_sparql_builder_predicate (metadata, "dc:type");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->type);
-		}
+			if (xd->type) {
+				tracker_sparql_builder_predicate (metadata, "dc:type");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->type);
+			}
 
-		if (xd->format) {
-			tracker_sparql_builder_predicate (metadata, "dc:format");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->format);
-		}
+			if (xd->format) {
+				tracker_sparql_builder_predicate (metadata, "dc:format");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->format);
+			}
 
-		if (xd->identifier) {
-			tracker_sparql_builder_predicate (metadata, "dc:identifier");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->identifier);
-		}
+			if (xd->identifier) {
+				tracker_sparql_builder_predicate (metadata, "dc:identifier");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->identifier);
+			}
 
-		if (xd->source) {
-			tracker_sparql_builder_predicate (metadata, "dc:source");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->source);
-		}
+			if (xd->source) {
+				tracker_sparql_builder_predicate (metadata, "dc:source");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->source);
+			}
 
-		if (xd->language) {
-			tracker_sparql_builder_predicate (metadata, "dc:language");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->language);
-		}
+			if (xd->language) {
+				tracker_sparql_builder_predicate (metadata, "dc:language");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->language);
+			}
 
-		if (xd->relation) {
-			tracker_sparql_builder_predicate (metadata, "dc:relation");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->relation);
-		}
+			if (xd->relation) {
+				tracker_sparql_builder_predicate (metadata, "dc:relation");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->relation);
+			}
 
-		if (xd->coverage) {
-			tracker_sparql_builder_predicate (metadata, "dc:coverage");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->coverage);
-		}
+			if (xd->coverage) {
+				tracker_sparql_builder_predicate (metadata, "dc:coverage");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->coverage);
+			}
 
-		if (xd->license) {
-			tracker_sparql_builder_predicate (metadata, "nie:license");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->license);
-		}
+			if (xd->license) {
+				tracker_sparql_builder_predicate (metadata, "nie:license");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->license);
+			}
 
-		if (xd->make || xd->model) {
-			gchar *camera;
+			if (xd->make || xd->model) {
+				gchar *camera;
 
-			if ((xd->make == NULL || xd->model == NULL) ||
-			    (xd->make && xd->model && strstr (xd->model, xd->make) == NULL)) {
-				camera = tracker_merge_const (" ", 2, xd->make, xd->model);
-			} else {
-				camera = g_strdup (xd->model);
+				if ((xd->make == NULL || xd->model == NULL) ||
+					(xd->make && xd->model && strstr (xd->model, xd->make) == NULL)) {
+					camera = tracker_merge_const (" ", 2, xd->make, xd->model);
+				} else {
+					camera = g_strdup (xd->model);
+				}
+
+				tracker_sparql_builder_predicate (metadata, "nmm:camera");
+				tracker_sparql_builder_object_unvalidated (metadata, camera);
+				g_free (camera);
 			}
 
-			tracker_sparql_builder_predicate (metadata, "nmm:camera");
-			tracker_sparql_builder_object_unvalidated (metadata, camera);
-			g_free (camera);
-		}
+			if (xd->orientation) {
+				tracker_sparql_builder_predicate (metadata, "nfo:orientation");
+				tracker_sparql_builder_object (metadata, xd->orientation);
+			}
 
-		if (xd->orientation) {
-			tracker_sparql_builder_predicate (metadata, "nfo:orientation");
-			tracker_sparql_builder_object (metadata, xd->orientation);
-		}
+			if (xd->rights) {
+				tracker_sparql_builder_predicate (metadata, "nie:copyright");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->rights);
+			}
 
-		if (xd->rights) {
-			tracker_sparql_builder_predicate (metadata, "nie:copyright");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->rights);
-		}
+			if (xd->white_balance) {
+				tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance");
+				tracker_sparql_builder_object (metadata, xd->white_balance);
+			}
 
-		if (xd->white_balance) {
-			tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance");
-			tracker_sparql_builder_object (metadata, xd->white_balance);
-		}
+			if (xd->fnumber) {
+				gdouble value;
 
-		if (xd->fnumber) {
-			gdouble value;
+				value = g_strtod (xd->fnumber, NULL);
+				tracker_sparql_builder_predicate (metadata, "nmm:fnumber");
+				tracker_sparql_builder_object_double (metadata, value);
+			}
 
-			value = g_strtod (xd->fnumber, NULL);
-			tracker_sparql_builder_predicate (metadata, "nmm:fnumber");
-			tracker_sparql_builder_object_double (metadata, value);
-		}
+			if (xd->flash) {
+				tracker_sparql_builder_predicate (metadata, "nmm:flash");
+				tracker_sparql_builder_object (metadata, xd->flash);
+			}
 
-		if (xd->flash) {
-			tracker_sparql_builder_predicate (metadata, "nmm:flash");
-			tracker_sparql_builder_object (metadata, xd->flash);
-		}
+			if (xd->focal_length) {
+				gdouble value;
 
-		if (xd->focal_length) {
-			gdouble value;
+				value = g_strtod (xd->focal_length, NULL);
+				tracker_sparql_builder_predicate (metadata, "nmm:focalLength");
+				tracker_sparql_builder_object_double (metadata, value);
+			}
 
-			value = g_strtod (xd->focal_length, NULL);
-			tracker_sparql_builder_predicate (metadata, "nmm:focalLength");
-			tracker_sparql_builder_object_double (metadata, value);
-		}
+			/* Question: Shouldn't xd->Artist be merged with md.author instead? */
 
-		/* Question: Shouldn't xd->Artist be merged with md.author instead? */
+			if (xd->artist || xd->contributor) {
+				const gchar *artist;
 
-		if (xd->artist || xd->contributor) {
-			const gchar *artist;
+				artist = tracker_coalesce_strip (2, xd->artist, xd->contributor);
+				tracker_sparql_builder_predicate (metadata, "nco:contributor");
+				tracker_sparql_builder_object_blank_open (metadata);
+				tracker_sparql_builder_predicate (metadata, "a");
+				tracker_sparql_builder_object (metadata, "nco:Contact");
+				tracker_sparql_builder_predicate (metadata, "nco:fullname");
+				tracker_sparql_builder_object_unvalidated (metadata, artist);
+				tracker_sparql_builder_object_blank_close (metadata);
+			}
 
-			artist = tracker_coalesce_strip (2, xd->artist, xd->contributor);
-			tracker_sparql_builder_predicate (metadata, "nco:contributor");
-			tracker_sparql_builder_object_blank_open (metadata);
-			tracker_sparql_builder_predicate (metadata, "a");
-			tracker_sparql_builder_object (metadata, "nco:Contact");
-			tracker_sparql_builder_predicate (metadata, "nco:fullname");
-			tracker_sparql_builder_object_unvalidated (metadata, artist);
-			tracker_sparql_builder_object_blank_close (metadata);
-		}
+			if (xd->exposure_time) {
+				gdouble value;
 
-		if (xd->exposure_time) {
-			gdouble value;
+				value = g_strtod (xd->exposure_time, NULL);
+				tracker_sparql_builder_predicate (metadata, "nmm:exposureTime");
+				tracker_sparql_builder_object_double (metadata, value);
+			}
 
-			value = g_strtod (xd->exposure_time, NULL);
-			tracker_sparql_builder_predicate (metadata, "nmm:exposureTime");
-			tracker_sparql_builder_object_double (metadata, value);
-		}
+			if (xd->iso_speed_ratings) {
+				gdouble value;
 
-		if (xd->iso_speed_ratings) {
-			gdouble value;
+				value = g_strtod (xd->iso_speed_ratings, NULL);
+				tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed");
+				tracker_sparql_builder_object_double (metadata, value);
+			}
 
-			value = g_strtod (xd->iso_speed_ratings, NULL);
-			tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed");
-			tracker_sparql_builder_object_double (metadata, value);
-		}
+			if (xd->description) {
+				tracker_sparql_builder_predicate (metadata, "nie:description");
+				tracker_sparql_builder_object_unvalidated (metadata, xd->description);
+			}
 
-		if (xd->description) {
-			tracker_sparql_builder_predicate (metadata, "nie:description");
-			tracker_sparql_builder_object_unvalidated (metadata, xd->description);
-		}
+			if (xd->metering_mode) {
+				tracker_sparql_builder_predicate (metadata, "nmm:meteringMode");
+				tracker_sparql_builder_object (metadata, xd->metering_mode);
+			}
 
-		if (xd->metering_mode) {
-			tracker_sparql_builder_predicate (metadata, "nmm:meteringMode");
-			tracker_sparql_builder_object (metadata, xd->metering_mode);
-		}
+			if (xd->address || xd->country || xd->city) {
+				tracker_sparql_builder_predicate (metadata, "mlo:location");
 
-		if (xd->address || xd->country || xd->city) {
-			tracker_sparql_builder_predicate (metadata, "mlo:location");
+				tracker_sparql_builder_object_blank_open (metadata);
+				tracker_sparql_builder_predicate (metadata, "a");
+				tracker_sparql_builder_object (metadata, "mlo:GeoPoint");
 
-			tracker_sparql_builder_object_blank_open (metadata);
-			tracker_sparql_builder_predicate (metadata, "a");
-			tracker_sparql_builder_object (metadata, "mlo:GeoPoint");
+				if (xd->address) {
+					tracker_sparql_builder_predicate (metadata, "mlo:address");
+					tracker_sparql_builder_object_unvalidated (metadata, xd->address);
+				}
 
-			if (xd->address) {
-				tracker_sparql_builder_predicate (metadata, "mlo:address");
-				tracker_sparql_builder_object_unvalidated (metadata, xd->address);
-			}
+				if (xd->state) {
+					tracker_sparql_builder_predicate (metadata, "mlo:state");
+					tracker_sparql_builder_object_unvalidated (metadata, xd->state);
+				}
 
-			if (xd->state) {
-				tracker_sparql_builder_predicate (metadata, "mlo:state");
-				tracker_sparql_builder_object_unvalidated (metadata, xd->state);
-			}
+				if (xd->city) {
+					tracker_sparql_builder_predicate (metadata, "mlo:city");
+					tracker_sparql_builder_object_unvalidated (metadata, xd->city);
+				}
 
-			if (xd->city) {
-				tracker_sparql_builder_predicate (metadata, "mlo:city");
-				tracker_sparql_builder_object_unvalidated (metadata, xd->city);
-			}
+				if (xd->country) {
+					tracker_sparql_builder_predicate (metadata, "mlo:country");
+					tracker_sparql_builder_object_unvalidated (metadata, xd->country);
+				}
 
-			if (xd->country) {
-				tracker_sparql_builder_predicate (metadata, "mlo:country");
-				tracker_sparql_builder_object_unvalidated (metadata, xd->country);
+				tracker_sparql_builder_object_blank_close (metadata);
 			}
 
-			tracker_sparql_builder_object_blank_close (metadata);
-		}
-
-		/* PDF keywords aren't used ATM (why not?) */
-		g_free (pd.keywords);
+			/* PDF keywords aren't used ATM (why not?) */
+			g_free (pd.keywords);
 
-		g_free (pd.title);
-		g_free (pd.subject);
-		g_free (pd.creation_date);
-		g_free (pd.author);
-		g_free (pd.date);
+			g_free (pd.title);
+			g_free (pd.subject);
+			g_free (pd.creation_date);
+			g_free (pd.author);
+			g_free (pd.date);
 
-		tracker_xmp_free (xd);
+			tracker_xmp_free (xd);
+		}
 	} else {
 		/* So if we are here we have NO XMP data and we just
 		 * write what we know from Poppler.