Update app.py

app.py

@@ -5,6 +5,7 @@ from docx import Document
 import tempfile
 import os
 import traceback
+import io
 
 def setup_logging():
     """Initialize logging configuration."""
@@ -651,27 +652,14 @@ def process_file(file_obj, doc_type, template_type):
         return "Please upload a document first."
     
     try:
-        # Debug information about the uploaded file
-        print(f"Received file: {type(file_obj)}")
-        print(f"File attributes: {dir(file_obj)}")
-        
-        # Check if it's a file-like object
-        if hasattr(file_obj, 'name'):
-            print(f"File name: {file_obj.name}")
-        
-        # For Gradio file upload, the file object should have a name attribute
-        # that contains the path to the temporary uploaded file
-        if isinstance(file_obj, str):
-            # If file_obj is a string (filepath)
-            file_path = file_obj
+        # Convert bytes to BytesIO object that Document can read
+        if isinstance(file_obj, bytes):
+            doc_bytes = io.BytesIO(file_obj)
         else:
-            # If file_obj is a file-like object
-            file_path = file_obj.name
+            doc_bytes = io.BytesIO(file_obj.read())
             
-        print(f"File path: {file_path}")
-        
         # Process the document and get results
-        results = process_document(file_obj, doc_type, template_type)
+        results = process_document(doc_bytes, doc_type, template_type)
         return results
         
     except Exception as e:
@@ -692,6 +680,53 @@ Technical details: {str(e)}"""
         
         return error_message
 
+def process_document(file_obj, doc_type, template_type):
+    """Process the document and perform checks based on document type and template type."""
+    try:
+        # Read the Word document directly from BytesIO
+        doc = Document(file_obj)
+        print("Document read successfully.")
+        
+        # Get required headings for document type and template type
+        checks = get_document_checks(doc_type, template_type)
+        required_headings = checks.get("required_headings", [])
+        
+        # Step 2: Perform all checks
+        heading_valid, headings_found = heading_title_check(doc, required_headings)
+        acronyms_valid, undefined_acronyms = acronym_check(doc)
+        legal_valid, incorrect_legal_references = legal_check(doc)
+        table_valid, incorrect_captions = table_caption_check(doc, doc_type)
+        figure_valid, incorrect_fig_captions = figure_caption_check(doc, doc_type)
+        references_valid, incorrect_table_figure_references = table_figure_reference_check(doc, doc_type)
+        title_style_valid, incorrect_titles = document_title_check(doc, doc_type)  # Modified to use doc object directly
+        double_period_valid, incorrect_sentences = double_period_check(doc)
+        spacing_valid, incorrect_spacing = spacing_check(doc)
+        abbreviation_issues = check_abbreviation_usage(doc)
+        date_issues = check_date_formats(doc)
+        placeholder_issues = check_placeholders(doc)
+
+        # Format results for Gradio
+        results = format_results_for_gradio(
+            heading_valid, headings_found,
+            acronyms_valid, undefined_acronyms,
+            legal_valid, incorrect_legal_references,
+            table_valid, incorrect_captions,
+            figure_valid, incorrect_fig_captions,
+            references_valid, incorrect_table_figure_references,
+            title_style_valid, incorrect_titles,
+            required_headings, doc_type,
+            double_period_valid, incorrect_sentences,
+            spacing_valid, incorrect_spacing,
+            abbreviation_issues, date_issues,
+            placeholder_issues
+        )
+        
+        return results
+    
+    except Exception as e:
+        print(f"Error in process_document: {str(e)}")
+        raise
+
 # Create the interface with simplified layout for Hugging Face Spaces
 demo = gr.Blocks(title="FAA Document Checker")
 
@@ -749,24 +784,6 @@ with demo:
         inputs=[file_input, doc_type, template_type],
         outputs=[output]
     )
-    
-    gr.Markdown("""
-    ## Instructions
-    1. Upload your Word document (.docx format)
-    2. Select the document type
-    3. If it's an Advisory Circular, select the template type
-    4. Click 'Check Document' to process
-    5. Review the results in the output panel
-    
-    The checker will verify:
-    - Required headings
-    - Acronym definitions
-    - Legal terminology
-    - Table and figure captions
-    - Document title styling
-    - Spacing and formatting
-    - And more...
-    """)
 
 # For Hugging Face Spaces, we just need to expose the 'demo'
 demo.launch()