libextractor

commit b39508a7a5fde9f1b27a5ea1b05df1d2de685b54
parent fc2e9074d5f8c076c5f0d0725fee0689baad3be0
Author: Christian Grothoff <christian@grothoff.org>
Date:   Tue,  5 Sep 2006 23:07:07 +0000

fixing Mantis 1127 and 1130 (different PDF problems)

Diffstat:
M
src/plugins/pdf/PDFDoc.h
 | 
4
++--
M
src/plugins/pdf/pdfextractor.cc
 | 
4
++--
M
src/plugins/pdfextractor.c
 | 
21
+++++++++++++--------

3 files changed, 17 insertions(+), 12 deletions(-)
diff --git a/src/plugins/pdf/PDFDoc.h b/src/plugins/pdf/PDFDoc.h
@@ -138,8 +138,8 @@ public:
   GBool isLinearized();
 
   // Return the document's Info dictionary (if any).
-  Object *getDocInfo(Object *obj) { return xref->getDocInfo(obj); }
-  Object *getDocInfoNF(Object *obj) { return xref->getDocInfoNF(obj); }
+  Object *getDocInfo(Object *obj) { return xref != NULL ? xref->getDocInfo(obj) : NULL; }
+  Object *getDocInfoNF(Object *obj) { return xref != NULL ? xref->getDocInfoNF(obj) : NULL; }
 
   // Return the PDF version specified by the file.
   double getPDFVersion() { return pdfVersion; }
diff --git a/src/plugins/pdf/pdfextractor.cc b/src/plugins/pdf/pdfextractor.cc
@@ -233,8 +233,8 @@ extern "C" {
     result = addKeyword(EXTRACTOR_MIMETYPE,
 			strdup("application/pdf"),
 			prev);
-    doc->getDocInfo(&info);
-    if (info.isDict()) {
+    if ( (NULL != doc->getDocInfo(&info)) &&
+	 (info.isDict()) ) {
       result = printInfoString(info.getDict(),
 			       "Title",
 			       EXTRACTOR_TITLE,
diff --git a/src/plugins/pdfextractor.c b/src/plugins/pdfextractor.c
@@ -437,20 +437,25 @@ libextractor_pdf_extract(const char * filename,
 		      pos - spos);
 	if (i == 0) {
 	  dmeta = dateDecode(meta);
-	  mlen = strlen((const char*)dmeta);
+	  if (dmeta != NULL)
+	    mlen = strlen((const char*)dmeta);
+	  else
+	    mlen = 0;
 	} else {
 	  dmeta = stringDecode(meta,
 			       &mlen);
 	}
 	if (meta != NULL)
 	  free(meta);
-	meta = charsetDecode(dmeta, mlen);
-	if (dmeta != NULL)
-	  free(dmeta);
-	if (meta != NULL) {
-	  prev = addKeyword(tagmap[i].type,
-			    meta,
-			    prev);
+	if (dmeta != NULL) {
+	  meta = charsetDecode(dmeta, mlen);
+	  if (dmeta != NULL)
+	    free(dmeta);
+	  if (meta != NULL) {
+	    prev = addKeyword(tagmap[i].type,
+			      meta,
+			      prev);
+	  }
 	}
 	break;
       }