Changeset 37000

Timestamp:

2022-12-09T00:16:02+13:00 (12 months ago)

Author:

davidb

Message:

Changes after testing

Location:

gs3-extensions/structured-image/trunk/perllib/plugins

Files:

• GoogleVisionAPIConverter.pm (modified) (7 diffs)
• GoogleVisionImagePlugin.pm (modified) (1 diff)
• GoogleVisionPagedImagePlugin.pm (modified) (1 diff)

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

@@ -50 +50 @@
 }
 
-my $arguments =
-    [ { 'name' => "enable_image_labelling",
-    'desc' => "{GoogleVisionAPIConverter.enable_image_labelling}",
-    'type' => "flag",
-    'reqd' => "no" },
+my $arguments = [
+    { 'name' => "google_application_credentials",
+      'desc' => "{GoogleVisionAPIConverter.google_applicatio_credentials}",
+      'type' => "string",
+      'reqd' => "no",
+      'deft' => "google-sa-credentials-key.json"
+    },
+    { 'name' => "enable_image_labelling",
+      'desc' => "{GoogleVisionAPIConverter.enable_image_labelling}",
+      'type' => "flag",
+      'reqd' => "no" },
     { 'name' => "enable_image_ocr",
-    'desc' => "{GoogleVisionAPIConverter.enable_image_ocr}",
-    'type' => "flag",
-    'reqd' => "no" },
+      'desc' => "{GoogleVisionAPIConverter.enable_image_ocr}",
+      'type' => "flag",
+      'reqd' => "no" },
     { 'name' => "enable_document_ocr",
-    'desc' => "{GoogleVisionAPIConverter.enable_document_ocr}",
-    'type' => "flag",
-    'reqd' => "no" }
+      'desc' => "{GoogleVisionAPIConverter.enable_document_ocr}",
+      'type' => "flag",
+      'reqd' => "no" }
 ];
 
@@ -113 +119 @@
 }
 
-sub run_convert {
+sub run_gv_convert {
     my $self = shift (@_);
-    my ($filename,$file,$doc_obj) = @_;
-    
-    my $section = $doc_obj->get_top_section();
+    my ($filename,$file,$doc_obj,$opt_section) = @_;
+    
+    my $section = (defined $opt_section) ? $opt_section : $doc_obj->get_top_section();
     
     my $verbosity = $self->{'verbosity'};
     my $outhandle = $self->{'outhandle'};
-    print $outhandle "----- GoogleVisionAPIConveter run_convert -----\n";
+    print $outhandle "----- GoogleVisionAPIConveter run_gv_convert -----\n";
     # print STDERR "*****", $self->{'enable_document_ocr'}, "\n";
 
@@ -143 +149 @@
         my $ofilename = &FileUtils::filenameConcatenate($cached_image_dir,"google-vision-output.json");
         my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
-        my $credentials_filename = &FileUtils::filenameConcatenate($collect_dir, "etc", "atea-storage-cd63a39dfeb5.json");
+        my $credentials_filename = &FileUtils::filenameConcatenate($collect_dir, "etc", $self->{'google_application_credentials'});
 
         if ($vision_type_length == 1) {
@@ -192 +198 @@
     };
 
-    my $cursection = $doc_obj->get_top_section(); # get top section for text append
+    #my $cursection = $doc_obj->get_top_section(); # get top section for text append
     
     my $decoded_json = from_json($json_text); 
@@ -198 +204 @@
     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( $cursection, $ocr_text); # append text to section
+        $doc_obj->add_utf8_text( $section, $ocr_text); # append text to section
 
         my $blocks = $decoded_json->{fullTextAnnotation}{pages}[0]{blocks};
@@ -216 +222 @@
         }
         for (keys %text_and_language) {
-            $doc_obj->add_utf8_metadata ($cursection, "z_" . $_, $text_and_language{$_});
+            $doc_obj->add_utf8_metadata ($section, "z_" . $_, $text_and_language{$_});
         }
 
@@ -224 +230 @@
         foreach my $label (@{ $ocr_text }) {
             # write to metadata : 'description'='Book' 'score'='0.9' 'topicality'='0.9' 'mid'='/m/0123'
-            $doc_obj->add_utf8_metadata ($cursection, "description", $label->{description});
-            $doc_obj->add_utf8_metadata ($cursection, "score", $label->{score});
-            $doc_obj->add_utf8_metadata ($cursection, "topicality", $label->{topicality});
-            $doc_obj->add_utf8_metadata ($cursection, "mid", $label->{mid});
+            $doc_obj->add_utf8_metadata ($section, "description", $label->{description});
+            $doc_obj->add_utf8_metadata ($section, "score", $label->{score});
+            $doc_obj->add_utf8_metadata ($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' 
-            $doc_obj->add_utf8_metadata ($cursection, "descriptions", $label->{description});
-            $doc_obj->add_utf8_metadata ($cursection, $label->{description} . "_score", $label->{score});
-            $doc_obj->add_utf8_metadata ($cursection, $label->{description} . "_topicality", $label->{topicality});
-            $doc_obj->add_utf8_metadata ($cursection, $label->{description} . "_mid", $label->{mid});
+            $doc_obj->add_utf8_metadata ($section, "descriptions", $label->{description});
+            $doc_obj->add_utf8_metadata ($section, $label->{description} . "_score", $label->{score});
+            $doc_obj->add_utf8_metadata ($section, $label->{description} . "_topicality", $label->{topicality});
+            $doc_obj->add_utf8_metadata ($section, $label->{description} . "_mid", $label->{mid});
 
         }

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

@@ -83 +83 @@
     my $url_encoded_filename = &util::rename_file($utf8_filename_no_path, $self->{'file_rename_method'});
     
-    $self->run_convert($filename_full_path,$url_encoded_filename,$doc_obj);
+    $self->run_gv_convert($filename_full_path,$url_encoded_filename,$doc_obj);
     
     $self->SUPER::process(@_);

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

@@ -93 +93 @@
     my $url_encoded_filename = &util::rename_file($utf8_filename_no_path, $self->{'file_rename_method'});
     
-    $self->run_convert($filename_full_path,$url_encoded_filename,$doc_obj);
+    $self->run_gv_convert($filename_full_path,$url_encoded_filename,$doc_obj,$section);
 
     return $self->SUPER::process_image(@_);