Changeset 8278 for trunk/gsdl/perllib/plugins/PDFPlug.pm

Timestamp:

2004-10-12T10:51:13+13:00 (20 years ago)

Author:

jrm21

Message:

sanity check for a valid date before trying to add it as metadata, otherwise
there might be uninitialised variables

File:

• trunk/gsdl/perllib/plugins/PDFPlug.pm (modified) (1 diff)

trunk/gsdl/perllib/plugins/PDFPlug.pm

@@ -219 +219 @@
 
     # We're just interested in the date bit, not the time
-    $datemeta =~ /(\d\d\d\d)-(\d\d)-(\d\d).*/;
-    $doc_obj->add_utf8_metadata($cursection, "Date", "$1$2$3");
+    # some pdf creators (eg "Acrobat 5.0 Scan Plug-in for Windows")
+    # set a /CreationDate, and set /ModDate to 000000000. pdftohtml
+    # extracts the ModDate, so it is 0...
+    $datemeta =~ /(\d+)-(\d+)-(\d+)/;
+    my ($year, $month, $day) = ($1,$2,$3);
+    if (defined($year) && defined($month) && defined($day)) {
+        if ($year == 0) {next}
+        if ($year < 100) {$year += 1900} # just to be safe
+        if ($month =~ /^\d$/) {$month="0$month"} # single digit
+        if ($day =~ /^\d$/) {$day="0$day"} # single digit
+        my $date="$year$month$day";
+        $doc_obj->add_utf8_metadata($cursection, "Date", $date);
+    }
     }