fix(text-extraction-error): Fix for when a file is failed to be extra…

…cted
DigitalNZ · Sep 2, 2024 · 8706d1c · 8706d1c
1 parent 787ea80
commit 8706d1c
Show file tree

Hide file tree

Showing 5 changed files with 21 additions and 14 deletions.
diff --git a/app/sidekiq/file_extraction_worker.rb b/app/sidekiq/file_extraction_worker.rb
@@ -55,6 +55,7 @@ def create_transformation_job(page)
 
   def api_record_id(page)
     return nil unless @extraction_job.extraction_definition.enrichment?
+    return nil if @extraction_job.documents[page].file_path.nil?
 
     @extraction_job.documents[page].file_path.match(/__(?<record_id>.+)__/)[:record_id]
   end

diff --git a/app/supplejack/extraction/document.rb b/app/supplejack/extraction/document.rb
@@ -35,7 +35,7 @@ def save(file_path)
     end
 
     def size_in_bytes
-      return if file_path.nil?
+      return 0 if file_path.nil?
 
       File.size(file_path)
     end

diff --git a/app/supplejack/extraction/documents.rb b/app/supplejack/extraction/documents.rb
@@ -18,7 +18,9 @@ def initialize(folder)
 
     def [](key)
       @current_page = key&.to_i || 1
-      return nil if documents_filepath.blank?
+      if documents_filepath.blank?
+        return @documents[@current_page] = Document.new(documents_filepath, body: '{"message":"File does not exist in filesystem"}') 
+      end
 
       @documents[@current_page] = Document.load_from_file(documents_filepath)
     end

diff --git a/app/views/extraction_jobs/show.html.erb b/app/views/extraction_jobs/show.html.erb
@@ -121,10 +121,12 @@
       <li class="list-group-item d-flex justify-content-between align-items-start">
         <div class="ms-2 me-auto">
           <div class="fw-bold">Request Headers</div>
-          <% @document.request_headers.each do |header, value| %>
-              <dt class="d-inline"><%= header %></dt>
-              <dd class="d-inline"><%= value %></dd>
-              <br>
+          <% if @document.request_headers.present? %>
+            <% @document.request_headers.each do |header, value| %>
+                <dt class="d-inline"><%= header %></dt>
+                <dd class="d-inline"><%= value %></dd>
+                <br>
+              <% end %>
             <% end %>
         </div>
       </li>
@@ -141,10 +143,12 @@
       <li class="list-group-item d-flex justify-content-between align-items-start">
         <div class="ms-2 me-auto">
           <div class="fw-bold">Response Headers</div>
-            <% @document.response_headers.each do |header, value| %>
-              <dt class="d-inline"><%= header %></dt>
-              <dd class="d-inline"><%= value %></dd>
-              <br>
+            <% if @document.response_headers.present? %>
+              <% @document.response_headers.each do |header, value| %>
+                <dt class="d-inline"><%= header %></dt>
+                <dd class="d-inline"><%= value %></dd>
+                <br>
+              <% end %>
             <% end %>
           </dl>
         </div>

diff --git a/spec/supplejack/extraction/documents_spec.rb b/spec/supplejack/extraction/documents_spec.rb
@@ -41,10 +41,10 @@
       expect(subject.current_page).to eq 130
     end
 
-    it 'returns nil when index is out of bounds' do
-      expect(subject[200]).to be_nil
-      expect(subject[0]).to be_nil
-      expect(subject[-1]).to be_nil
+    it 'returns a blank document when index is out of bounds' do
+      expect(subject[200]).to be_a(Extraction::Document)
+      expect(subject[0]).to be_a(Extraction::Document)
+      expect(subject[-1]).to be_a(Extraction::Document)
     end
 
     it 'returns pages based on their page number, rather than their order in the file system' do