Changeset 37010

Timestamp:

2022-12-10T18:19:13+13:00 (12 months ago)

Author:

davidb

Message:

Changes to make it easier to display OCR bounding-box info in PagedImage doc view

File:

• gs3-extensions/structured-image/trunk/perllib/plugins/GoogleVisionAPIConverter.pm (modified) (7 diffs)

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

@@ -133 +133 @@
 
     if ($self->{'enable_image_labelling'}) { push(@vision_type, "enable_image_labelling"); } 
-    if ($self->{'enable_image_ocr'}) { push(@vision_type, "enable_image_ocr"); }
-    if ($self->{'enable_document_ocr'}) { push(@vision_type, "enable_document_ocr"); } 
+    if ($self->{'enable_image_ocr'})       { push(@vision_type, "enable_image_ocr");       }
+    if ($self->{'enable_document_ocr'})    { push(@vision_type, "enable_document_ocr");    } 
 
     my $vision_type_length = @vision_type; # assigning scalar var to array returns length
@@ -156 +156 @@
 
             $self->run_vision($file, $filename, $ofile, $ofilename, $vision_cmd, $vision_type_first, $section, $doc_obj);
-        } elsif ($vision_type_length == 2) {
+        }
+    elsif ($vision_type_length == 2) {
             my $vision_type_first = $vision_type[0];
             my $vision_type_second = $vision_type[1];
@@ -189 +190 @@
         = $self->run_cached_general_cmd($vision_cmd,$filename,$ofilename,$print_info);
 
-    $doc_obj->associate_file($ofilename,$vision_type . $ofile,"application/json",$section);
+    # Need to work a bit harder in setting up the associated JSON file
+    # => strip of 'enable_' in favour of 'gv_'
+    # => add in section number as part of the file name to avoid clashes
+    
+    my $gv_assoc_prefix = $vision_type;
+    $gv_assoc_prefix =~ s/^enable_/gv_/;
+
+    my $section_file_suffix = $section;
+    $section_file_suffix =~ s/\./_/g;
+
+    my $assoc_ofile = $gv_assoc_prefix.$ofile;
+    $assoc_ofile =~ s/\.(.*?)$/$section_file_suffix.$1/;
+    
+    $doc_obj->associate_file($ofilename,$assoc_ofile,"application/json",$section);
 
     my $json_text = do { # read in json file
@@ -198 +212 @@
     };
 
-    #my $cursection = $doc_obj->get_top_section(); # get top section for text append
-    
     my $decoded_json = from_json($json_text); 
     my $ocr_text;
@@ -221 +233 @@
             }
         }
+    
         for (keys %text_and_language) {
             $doc_obj->add_utf8_metadata ($section, "z_" . $_, $text_and_language{$_});
         }
 
-
-    } elsif ($vision_type eq "enable_image_labelling") {
-        $ocr_text = $decoded_json->{labelAnnotations}; 
+    
+    my $assoc_json_metaname = "HasGoogleVision";
+    
+    if ($vision_type eq "enable_document_ocr") {
+        $assoc_json_metaname .= "DocumentOCRJSON";
+
+        $doc_obj->add_utf8_metadata ($section, "GVDocumentOCRJSON",$assoc_ofile);
+    }
+    else {
+        # $vision_type eq "enable_image_ocr")
+        $assoc_json_metaname .= "ImageOCRJSON";
+
+        $doc_obj->add_utf8_metadata ($section, "GVImageOCRJSON",$assoc_ofile);
+    }
+
+    $doc_obj->add_utf8_metadata ($section, $assoc_json_metaname, 1);
+    }
+    elsif ($vision_type eq "enable_image_labelling") {
+        $ocr_text = $decoded_json->{labelAnnotations};  
         foreach my $label (@{ $ocr_text }) {
             # write to metadata : 'description'='Book' 'score'='0.9' 'topicality'='0.9' 'mid'='/m/0123'
@@ -234 +263 @@
             $doc_obj->add_utf8_metadata ($section, "topicality", $label->{topicality});
             $doc_obj->add_utf8_metadata ($section, "mid", $label->{mid});
-            # write to metadata : 'descriptions'='Book' 'Book_score'='0.9' 'Book_topicality'='0.9' 'Book_mid'='/m/0123' 
+
+        # write to metadata : 'descriptions'='Book' 'Book_score'='0.9' 'Book_topicality'='0.9' 'Book_mid'='/m/0123' 
             $doc_obj->add_utf8_metadata ($section, "descriptions", $label->{description});
             $doc_obj->add_utf8_metadata ($section, $label->{description} . "_score", $label->{score});
@@ -241 +271 @@
 
         }
+    
+    $doc_obj->add_utf8_metadata ($section, "HasGoogleVisionImageLabellingJSON", 1);
+    $doc_obj->add_utf8_metadata ($section, "GVImageLabellingJSON",$assoc_ofile);
+
     }
 }