Changeset 37046 for gs3-extensions/structured-image/trunk/perllib/plugins/GoogleVisionPagedImagePlugin.pm

Timestamp:

2022-12-23T10:19:59+13:00 (16 months ago)

Author:

davidb

Message:

Code extended to now generate Open Annotation (JSON format) of the OCR'd blocks of text; some refinement of the existing Google Vision perl code

File:

• gs3-extensions/structured-image/trunk/perllib/plugins/GoogleVisionPagedImagePlugin.pm (modified) (2 diffs)

gs3-extensions/structured-image/trunk/perllib/plugins/GoogleVisionPagedImagePlugin.pm

@@ -86 +86 @@
 
 
+sub init_new_doc_item
+{    
+    my $self = shift (@_);
+    my ($filename_full_path, $processor, $metadata) = @_;
+
+    my $doc_obj = $self->SUPER::init_new_doc_item(@_);
+
+    $self->{'gv-dococr-json-filename-recs'} = [];
+    
+    return $doc_obj;    
+}
+
+
 sub process_image {
     my $self = shift(@_);
@@ -94 +107 @@
     
     $self->run_gv_convert($filename_full_path,$url_encoded_filename,$doc_obj,$section);
-
+    
     return $self->SUPER::process_image(@_);
 }
 
+sub post_process_doc_obj {
+    my $self = shift (@_);  
+    my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
+
+    my $ret_val_ok = $self->SUPER::post_process_doc_obj(@_);
+
+    if ($ret_val_ok) {
+    $ret_val_ok = $self->opt_run_gen_openannotation($doc_obj);
+    }
+    
+    return $ret_val_ok;
+}
+
+
 1;