Merge pull request #117 from SuffolkLITLab/docx-labeler

Add an MVP feature that lets you automatically add labels to a DOCX file with OpenAI

docassemble/ALDashboard/data/questions/docx_wrangling.yml

@@ -0,0 +1,158 @@
+---
+include:
+  - nav.yml
+---
+modules:
+  - .docx_wrangling
+---
+objects:
+  - new_docx: DAFile
+---
+id: interview order
+mandatory: True
+code: |
+  docx_file
+  if not started_task.ready():
+    waiting_screen
+  show_stats
+  ask_about_labels
+  save_changes
+  show_final_docx
+---
+event: waiting_screen
+question: |
+  Please wait while we process your file
+subquestion: |
+  <div class="spinner-border" role="status">
+    <span class="visually-hidden">Processing...</span>
+  </div>
+reload: True
+---
+continue button field: show_stats
+question: |
+  Your DOCX file has been processed
+subquestion: |
+  GPT-4 found ${ len(draft_labels) } labels in your DOCX file.
+
+  On the next screen, you can review and make any necessary changes
+  to the draft Jinja2 labels.
+---
+question: |
+  Upload a DOCX file
+subquestion: |
+  We will use GPT-4 to try to add variables in the
+  [AssemblyLine 
+  convention](https://suffolklitlab.org/docassemble-AssemblyLine-documentation/docs/label_variables) 
+  to your DOCX file.
+
+  Your upload can have up to 300 pages, but the result cannot be larger than about 4,000 words. The result will
+  only include the modified paragraphs.
+fields:
+  - DOCX file: docx_file
+    datatype: file
+    accept: |
+      ".docx, application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+  - Include custom people names: include_custom_people_names
+    datatype: yesno
+    default: False
+  - Custom people names: custom_people_names_text
+    datatype: area
+    show if: include_custom_people_names
+    help: |
+      Enter a list of custom nouns to use, one per line. Optionally, add 
+      an explanation after each name, separated by a colon. For example:
+      ```
+      agents: people making decisions in healthcare proxy or power of attorney
+      attorneys_in_fact: people granted power of attorney
+      ```
+
+      These names will be used to label variables
+      in the DOCX file.
+---
+code: |
+  started_task = background_action('task_draft_labels')      
+---
+event: task_draft_labels
+code: |
+  if include_custom_people_names:
+    custom_people_names = [tuple(line.split(':')) for line in custom_people_names_text.split('\n')]
+  else:
+    custom_people_names = None
+  draft_labels = get_labeled_docx_runs(docx_file[0].path(), custom_people_names = custom_people_names)
+  background_response_action('save_draft_labels', draft_labels=draft_labels)
+---
+event: save_draft_labels
+code: |
+  draft_labels = action_argument('draft_labels')
+  background_response()
+---
+objects:
+  - final_labels: DAList.using(object_type=DAObject, auto_gather=False, gathered=True)
+---
+code: |
+  import docx
+  original_doc = docx.Document(docx_file[0].path())
+
+  label_question = []
+  for idx, item in enumerate(draft_labels):
+    new_obj = final_labels.appendObject()
+    new_obj.paragraph = item[0]
+    new_obj.run = item[1]
+    new_obj.draft_label_text = item[2]
+
+    # Results will be a tuple of paragraph number, run, modified text with label
+    label_question.append({
+      'label': original_doc.paragraphs[item[0]].runs[item[1]].text,
+      'field': f'final_labels[{idx}].label',
+      'default': item[2],
+      'label above field': True,
+      'grid': 8,
+      'hide if': f'final_labels[{idx}].leave_unchanged'
+    })
+    label_question.append({
+      'label': 'Leave unchanged',
+      'field': f'final_labels[{idx}].leave_unchanged',
+      'datatype': 'yesno',
+      'grid': {
+        'width': 4,
+        'end': True,
+      },
+      'label above field': True,
+    })
+
+  del docx
+  del original_doc
+---
+code: |
+
+  new_doc_obj = update_docx(
+    docx_file[0].path(), 
+    [
+      (item.paragraph, item.run, item.label, 0) 
+      for item in final_labels 
+      if not item.leave_unchanged
+    ]
+  )
+
+  new_docx.initialize(filename=docx_file[0].filename)
+  new_doc_obj.save(new_docx.path())
+  new_docx.commit()
+
+  del new_doc_obj
+
+  save_changes = True
+---
+continue button field: ask_about_labels
+question: |
+  Review the labels
+subuestion: |
+  On the left is the original text. On the right, you will see the modified
+  text with Jinja2 added. Make any changes you need.
+fields:
+  - code: label_question
+---
+event: show_final_docx
+question: |
+  Here is your new DOCX file
+subquestion: |
+  ${ new_docx }

docassemble/ALDashboard/data/questions/menu.yml

@@ -55,6 +55,8 @@ buttons:
     image: paperclip
   - PDF tools:  ${ interview_url(i=user_info().package + ":pdf_wrangling.yml", reset=1) }
     image: file-pdf
+  - Auto-label DOCX: ${ interview_url(i=user_info().package + ":docx_wrangling.yml", reset=1) }
+    image: file-word
   - Compile Bootstrap theme: ${ interview_url(i=user_info().package + ":compile_bootstrap.yml", reset=1) }
     image: bootstrap
   - Install fonts: ${ interview_url(i=user_info().package + ":install_fonts.yml", reset=1)}