Update document_converter.py

document_converter.py

@@ -97,14 +97,10 @@ class DocumentConverter:
     def _detect_input_format(self, content: str, force_latex: bool) -> str:
         """
         Decide whether Pandoc should treat the text as 'latex' or 'markdown'.
-
-        In our use case, we generally force LaTeX, but this function
-        is kept flexible.
         """
         if force_latex:
             return 'latex'
 
-        # Heuristics to detect LaTeX
         latex_markers = [
             '\\documentclass',
             '\\begin{document}',
@@ -119,7 +115,6 @@ class DocumentConverter:
         if any(marker in content for marker in latex_markers):
             return 'latex'
 
-        # Default to markdown if no strong LaTeX signs
         return 'markdown'
 
     # =====================
@@ -137,8 +132,7 @@ class DocumentConverter:
         Convert content (LaTeX/Markdown) directly to DOCX/PDF/etc using Pandoc.
 
         - If include_latex=True, treat content as LaTeX.
-        - Strips ```latex ... ``` or ``` ... ``` fenced code blocks first,
-          because Pandoc would otherwise render them as literal code.
+        - Strips ```latex ... ``` or ``` ... ``` fenced code blocks first.
 
         Returns: bytes of the generated file (DOCX/PDF/whatever you requested).
         """
@@ -150,42 +144,47 @@ class DocumentConverter:
         input_format = self._detect_input_format(content, force_latex=include_latex)
 
         # 3) Decide the Pandoc output format from the extension
-        #    e.g. ".docx" -> "docx", ".pdf" -> "pdf"
         target = output_format.lstrip('.').lower() or 'docx'
 
         try:
-            # Use correct suffix for temp file: .docx, .pdf, etc.
+            # Create temp file for Pandoc output
             with tempfile.NamedTemporaryFile(suffix=f".{target}", delete=False) as temp_output:
                 output_filename = temp_output.name
 
-            # Run Pandoc conversion
-                    try:
-            # Run Pandoc conversion
-            pypandoc.convert_text(
-                content,
-                target,
-                format=input_format,
-                outputfile=output_filename,
-                extra_args=['--standalone']
-            )
-
-        except OSError:
-            # Try to download pandoc automatically (useful in local dev)
-            import pypandoc as _pd
-            _pd.download_pandoc()
-            # Retry once
-            pypandoc.convert_text(
-                content,
-                target,
-                format=input_format,
-                outputfile=output_filename,
-                extra_args=['--standalone']
-            )
+            # First attempt to run Pandoc
+            try:
+                pypandoc.convert_text(
+                    content,
+                    target,
+                    format=input_format,
+                    outputfile=output_filename,
+                    extra_args=['--standalone']
+                )
+            except OSError:
+                # Pandoc binary not found – try to download (useful in local dev; may not work on HF)
+                import pypandoc as _pd
+                _pd.download_pandoc()
+                # Retry once
+                pypandoc.convert_text(
+                    content,
+                    target,
+                    format=input_format,
+                    outputfile=output_filename,
+                    extra_args=['--standalone']
+                )
+
+            # Read the generated file back into bytes
+            with open(output_filename, "rb") as f:
+                file_bytes = f.read()
+
+            # Cleanup temp file
+            os.unlink(output_filename)
+
+            return file_bytes
 
         except Exception as e:
             raise RuntimeError(f"Conversion failed: {str(e)}")
 
-
     # =====================
     # ADD SIGNATURE TO DOCX
     # =====================
@@ -199,11 +198,6 @@ class DocumentConverter:
     ) -> bytes:
         """
         Add a signature image (base64) and optional signer name to a DOCX document.
-
-        - file_content: existing DOCX bytes
-        - signature_data: base64 image string (with or without `data:image/...;base64,` prefix)
-        - position: 'left', 'center', 'right' / 'bottom-right' / 'bottom-center' etc (only horizontal used)
-        - signer_name: optional text below the signature
         """
         try:
             doc = Document(io.BytesIO(file_content))