add image handling and prevent leaking temporary files (ourselves, Archive::Zip might be)

1 files changed, 63 insertions, 14 deletions

diff --git a/lib/HTML/AutoConvert/OpenOffice.pm b/lib/HTML/AutoConvert/OpenOffice.pm
index e09a9e4..7b35595 100644
--- a/lib/HTML/AutoConvert/OpenOffice.pm
+++ b/lib/HTML/AutoConvert/OpenOffice.pm
@@ -34,24 +34,66 @@ use strict;
 use vars qw( %info ); #$slept );
 use IPC::Run qw( run timeout io );
 use File::Slurp qw( slurp );
+use Archive::Zip qw( :ERROR_CODES :CONSTANTS );
 
 %info = (
-  'types'   => [qw( doc rtf odt sxw )],
-  'weight'  => 80,
-  'url'     => 'http://wvware.sourceforge.net/',
+  'types'          => [qw( doc rtf odt sxw )],
+  'weight'         => 10,
+  'returns_images' => 1,
+  'url'            => 'http://www.openoffice.org/',
 );
 
 #$slept = 0;
 
 #sub program { ( 'openoffice', '-headless' ); }
 
-#half-ass using DocumentConverter.py for now
-#need to recode with OpenOffice::UNO
-
 sub html_convert {
   my( $self, $file ) = ( shift, shift );
   my $opt = ref($_[0]) ? shift : { @_ };
 
+  my $outfile = $self->odconvert($file, 'html');
+  my $html = slurp($outfile);
+  unlink($outfile) or warn "can't unlink $outfile";
+
+  return $html unless wantarray;
+
+  my @images = $self->extract_images($file, $opt);
+
+  ( $html, @images );
+
+}
+
+#http://cdriga.kfacts.com/open-source-world/tutorial-extract-original-images-from-ms-word-doc-using-openofficeorg/2007/11/04/
+sub extract_images {
+  my( $self, $file ) = ( shift, shift );
+  my $opt = ref($_[0]) ? shift : { @_ };
+
+  my $zipfile = $self->odconvert($file, 'odt');
+  my $zip = Archive::Zip->new();
+
+  unless ( $zip->read( $zipfile ) == AZ_OK ) {
+    die "error reading $zipfile for images";
+  }
+
+  my @members = $zip->membersMatching( '^Pictures/*' );
+
+  my @images = map {
+                     ( my $filename = $_->fileName ) =~ s/^.*\///;
+                     [ $filename, $zip->contents($_) ];
+                   }
+                   @members;
+
+  unlink($zipfile);
+
+  @images;
+}
+
+#half-ass using DocumentConverter.py for now
+#need to recode with OpenOffice::UNO
+sub odconvert {
+  my( $self, $file, $suffix ) = ( shift, shift, shift );
+  my $opt = ref($_[0]) ? shift : { @_ };
+
   $self->start_openoffice($opt);
 
   my $program = 'DocumentConverter.py';
@@ -59,20 +101,27 @@ sub html_convert {
   my $timeout = 60; #?
 
   use File::Temp qw/ tempfile /;
-  my($fh, $outfile) = tempfile(SUFFIX => '.html');
+  my($fh, $outfile) = tempfile(SUFFIX => ".$suffix");
   #hmm, it gets overwritten so $fh is bunk
 
   my($out, $err) = ( '', '' );
   local($SIG{CHLD}) = sub {};
-  run( [ $program, $file, $outfile ], \undef, \$out, \$err, timeout($timeout) )
-    or die "$program failed with exit status ". ( $? >> 8 ). ": $out\n";
-
-  my $html = slurp($outfile);
-
-  $html;
-
+  eval {
+    run( [ $program, $file, $outfile ], \undef, \$out, \$err, timeout($timeout) )
+      or do {
+              unlink($outfile) or warn "$!\n";
+              die "$program failed with exit status ". ( $? >> 8 ). ": $out\n";
+            };
+  };
+  if ( $@ ) {
+    unlink($outfile) or warn "$!\n";
+    die "$program failed: $@\n";
+  }
+
+  $outfile;
 }
 
+
 sub start_openoffice {
   my( $self ) = ( shift, shift );
   my $opt = ref($_[0]) ? shift : { @_ };