import torrus 1.0.9

[freeside.git] / FS / FS / Misc.pm

diff --git a/FS/FS/Misc.pm b/FS/FS/Misc.pm
index 0e8d92b..fe8ac60 100644 (file)
--- a/FS/FS/Misc.pm
+++ b/FS/FS/Misc.pm
@@ -20,6 +20,7 @@ use Tie::IxHash;
                  pkg_freqs
                  generate_ps generate_pdf do_print
                  csv_from_fixed
+                 ocr_image
                );
 
 $DEBUG = 0;
@@ -850,6 +851,41 @@ sub csv_from_fixed {
   '';
 }
 
+=item ocr_image IMAGE_SCALAR
+
+Runs OCR on the provided image data and returns a list of text lines.
+
+=cut
+
+sub ocr_image {
+  my $logo_data = shift;
+
+  #XXX use conf dir location from Makefile
+  my $dir = $FS::UID::conf_dir. "/cache.". $FS::UID::datasrc;
+  my $fh = new File::Temp(
+    TEMPLATE => 'bizcard.XXXXXXXX',
+    SUFFIX   => '.png', #XXX assuming, but should handle jpg, gif, etc. too
+    DIR      => $dir,
+    UNLINK   => 0,
+  ) or die "can't open temp file: $!\n";
+
+  my $filename = $fh->filename;
+
+  print $fh $logo_data;
+  close $fh;
+
+  run( [qw(ocroscript recognize), $filename], '>'=>"$filename.hocr" )
+    or die "ocroscript recognize failed\n";
+
+  run( [qw(ocroscript hocr-to-text), "$filename.hocr"], '>pipe'=>\*OUT )
+    or die "ocroscript hocr-to-text failed\n";
+
+  my @lines = split(/\n/, <OUT> );
+
+  foreach (@lines) { s/\.c0m\s*$/.com/; }
+
+  @lines;
+}
 
 =back