Changeset 37117 for gs3-extensions/structured-image/trunk/perllib/plugins/GoogleVisionAPIConverter.pm

Timestamp:

2023-01-07T11:42:21+13:00 (16 months ago)

Author:

davidb

Message:

Allow for a page having now OCR'd text found

File:

• gs3-extensions/structured-image/trunk/perllib/plugins/GoogleVisionAPIConverter.pm (modified) (1 diff)

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

@@ -261 +261 @@
     my $ocr_text;
     if ($vision_type eq "enable_document_ocr" || $vision_type eq "enable_image_ocr") {
-        $ocr_text = $decoded_json->{'textAnnotations'}[0]{'description'}; # access full ocr content
-        $doc_obj->add_utf8_text($section, $ocr_text); # append text to section
-
-        my $blocks = $decoded_json->{'fullTextAnnotation'}{'pages'}[0]{'blocks'};
-        my %text_and_language;
-
-        foreach my $block (@{$blocks}) {
-            foreach my $paragraph (@{$block->{'paragraphs'}}) {
-                foreach my $word (@{$paragraph->{'words'}}) {
-                    my $detected_language = $word->{'property'}{'detectedLanguages'}[0]{'languageCode'} || "no_lang";
-                    my $word_text = "";
-                    foreach my $letter (@{$word->{'symbols'}}) {
-                        $word_text .= $letter->{'text'};
-                    }
-                    $text_and_language{$detected_language} .= $word_text . " "; 
-                }
-            }
-        }
-    
-        for (keys %text_and_language) {
-            $doc_obj->add_utf8_metadata($section, "z_" . $_, $text_and_language{$_});
-        }
-
+
+    if (defined $decoded_json->{'textAnnotations'}) {
+        $ocr_text = $decoded_json->{'textAnnotations'}->[0]->{'description'}; # access full ocr content
+        $doc_obj->add_utf8_text($section, $ocr_text); # append text to section
+        
+        my $blocks = $decoded_json->{'fullTextAnnotation'}->{'pages'}->[0]->{'blocks'};
+        my %text_and_language;
+
+        foreach my $block (@{$blocks}) {
+        foreach my $paragraph (@{$block->{'paragraphs'}}) {
+            foreach my $word (@{$paragraph->{'words'}}) {
+            my $detected_language = $word->{'property'}->{'detectedLanguages'}->[0]->{'languageCode'} || "no_lang";
+            my $word_text = "";
+            foreach my $letter (@{$word->{'symbols'}}) {
+                $word_text .= $letter->{'text'};
+            }
+            $text_and_language{$detected_language} .= $word_text . " "; 
+            }
+        }
+        }
+    
+        for (keys %text_and_language) {
+        $doc_obj->add_utf8_metadata($section, "z_" . $_, $text_and_language{$_});
+        }
+        
+    }
+
+    # Note: Even if there is no actual OCR'd text detected (if test above),
+    # stil set metadata that show that we applied the Google Vision API seeking text
     
     my $assoc_json_metaname = "HasGoogleVision";