Update app.py

app.py

@@ -1,32 +1,42 @@
-import os, shutil, subprocess, tempfile, json
+import os
+import shutil
+import subprocess
+import tempfile
+import json
 import gradio as gr
 
 def process_upload(pdf_file, output_format):
-    # Create temp output directory
     out_dir = tempfile.mkdtemp()
-    # Run Marker CLI: outputs files marker-0000.md or marker-0000.json
-    fmt = "markdown" if output_format=="markdown" else "json"
+    fmt = "markdown" if output_format == "markdown" else "json"
     cmd = [
         "marker_single",
         pdf_file.name,
         "--output_format", fmt,
         "--output_dir", out_dir,
-        "--paginate_output"  # page separators
+        "--paginate_output"
     ]
     subprocess.run(cmd, check=True)
-    # Read and combine results
-    results = []
-    for fname in sorted(os.listdir(out_dir)):
-        path = os.path.join(out_dir, fname)
+
+    # Recursively find only .md/.json files
+    collected = []
+    for root, _, files in os.walk(out_dir):
+        for fname in sorted(files):
+            if fmt == "markdown" and fname.lower().endswith(".md"):
+                collected.append(os.path.join(root, fname))
+            elif fmt == "json" and fname.lower().endswith(".json"):
+                collected.append(os.path.join(root, fname))
+
+    pages = []
+    for path in collected:
         with open(path, 'r', encoding='utf-8') as f:
-            results.append(f.read())
-    # Cleanup
+            pages.append(f.read())
+
     shutil.rmtree(out_dir)
+
     if output_format == "markdown":
-        return "\n\n---\n\n".join(results)
+        return "\n\n---\n\n".join(pages)
     else:
-        # If JSON, combine into list of pages
-        return json.dumps({"pages": results}, indent=2, ensure_ascii=False)
+        return json.dumps({"pages": pages}, indent=2, ensure_ascii=False)
 
 demo = gr.Interface(
     fn=process_upload,
@@ -34,12 +44,11 @@ demo = gr.Interface(
         gr.File(label="Upload PDF", file_types=[".pdf"]),
         gr.Radio(["markdown","json"], value="markdown", label="Output format")
     ],
-    outputs=gr.Code(label="Output"),
+    outputs=gr.Code(label="Converted Output"),
     title="PDF → Markdown/JSON with LaTeX Support",
     description=(
-        "Uploads a PDF and uses Marker to extract text, structure, and LaTeX math. "
-        "Choose Markdown to get a single .md with `$...$`/`$$...$$` math, "
-        "or JSON for a page-by-page array."
+        "Upload a PDF and get back Markdown or structured JSON, "
+        "with math preserved as LaTeX."
     )
 )