Changeset 32280 for main/trunk/greenstone2/perllib/plugins/ConvertBinaryFile.pm

Timestamp:

2018-07-17T20:40:57+12:00 (6 years ago)

Author:

ak19

Message:

Implementing PDFv2paged_text (with pdfbox)

File:

• main/trunk/greenstone2/perllib/plugins/ConvertBinaryFile.pm (modified) (1 diff)

main/trunk/greenstone2/perllib/plugins/ConvertBinaryFile.pm

@@ -170 +170 @@
     $self->{'convert_to_plugin'} = "StructuredHTMLPlugin";
     $self->{'convert_to_ext'} = "html";
-    } elsif ($convert_to =~ /^pagedimg/) {
+    } elsif ($convert_to =~ /^pagedimg/ || $convert_to eq "paged_text") {
     $self->{'convert_to_plugin'} = "PagedImagePlugin";
-    my ($convert_to_ext) = $convert_to =~ /pagedimg\_(jpg|gif|png)/i;
-    $convert_to_ext = 'jpg' unless defined $convert_to_ext;
-    $self->{'convert_to_ext'} = $convert_to_ext;
+    if($convert_to eq "paged_text") {
+        $self->{'convert_to_ext'} = "txt";
+    } else {
+        my ($convert_to_ext) = $convert_to =~ /pagedimg(?:txt)?\_(jpg|gif|png)/i; # the ?: prefix avoids capturing or else discards the optional 'txt' in 'pagedimgtxt',
+        # so that we can consider the actual portion we want to capture: the img type
+        $convert_to_ext = 'jpg' unless defined $convert_to_ext;
+        $self->{'convert_to_ext'} = $convert_to_ext;
+    }
     }
 }