Changeset 25993

Timestamp:

2012-07-19T18:10:15+12:00 (11 years ago)

Author:

ak19

Message:

Generalised the create_itemfile() subroutine's code so that PDFBoxConverter can reuse this, as the PDFBoxConverter is currently being modified to convert a PDF to images, when the -pagedimage_IMGTYPE flag is specified. After this commit, the updated create_itemfile() will be moved into util for common access.

File:

• main/trunk/greenstone2/bin/script/pdfpstoimg.pl (modified) (3 diffs)

main/trunk/greenstone2/bin/script/pdfpstoimg.pl

@@ -118 +118 @@
     my ($output_dir, $convert_basename, $convert_to) = @_;
     opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";  
-    my $item_file = $output_dir."/".$convert_basename.".item";
-    open(FILE,">$item_file");
-    
-    print FILE "<PagedDocument>\n";
 
     my $page_num = "";
-    @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
+    my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
 
     # Sort files in the directory by page_num    
     sub page_number {
     my ($dir) = @_;
-    my ($pagenum) =($dir =~ m/^.*[-\.](\d+)(\.(jpg|gif|png))?$/i);
-    
+    my ($pagenum) =($dir =~ m/^.*[-\.]?(\d+)(\.(jpg|gif|png))?$/i);
+
     $pagenum = 1 unless defined $pagenum;
     return $pagenum;
@@ -138 +134 @@
     @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;
 
+    # work out if the numbering of the now sorted image files starts at 0 or not
+    # by checking the number of the first _image_ file (skipping item files)
+    my $starts_at_0 = 0;
+    my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
+    if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
+    $starts_at_0 = 1;
+    }
+
+    my $item_file = &util::filename_cat($output_dir, $convert_basename.".item");
+    open(FILE,">$item_file");    
+    print FILE "<PagedDocument>\n";
+
     foreach my $file (@dir_files){
-    $page_num = page_number($file)+1; # image numbers start at 0, so add 1
     if ($file !~ /\.item/i){
+        $page_num = page_number($file);
+        $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
         print FILE "   <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
     } 
@@ -147 +156 @@
     print FILE "</PagedDocument>\n";
     closedir DIR;
-    return "";
+    return $item_file;
 }