Update app.py

app.py

@@ -9,29 +9,38 @@ import gradio as gr
 
 from pipeline import run_pipeline
 
-# ---- Defaults / Constants ----
-
 DEFAULT_CONFIG: Dict[str, Any] = {
     "model": os.getenv("OPENAI_MODEL", "gpt-4o-mini"),
     "rewrite": False,
     "projects": [{"name": "STANDARD", "description": "Generic scoring"}],
+    # OCR knobs (optional)
+    "ocr_max_pages": 8,
+    "ocr_dpi": 200,
 }
 
 TMP_ROOT = Path("/tmp/resume_evaluator").resolve()
 UPLOAD_DIR = TMP_ROOT / "input_uploads"
-OUTPUT_DIR = TMP_ROOT / "output"
+OUTPUT_DIR = TMP_ROOT / "output"          # IMPORTANT: do not delete if you want dedupe persistence
 RESULTS_ZIP = TMP_ROOT / "results.zip"
 
 
-# ---- Helpers ----
-
-def _clean_tmp() -> None:
-    if TMP_ROOT.exists():
-        shutil.rmtree(TMP_ROOT, ignore_errors=True)
+def _ensure_dirs() -> None:
+    TMP_ROOT.mkdir(parents=True, exist_ok=True)
     UPLOAD_DIR.mkdir(parents=True, exist_ok=True)
     OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
 
 
+def _clean_upload_staging_only() -> None:
+    """
+    DO NOT delete OUTPUT_DIR if you want index dedupe to persist across runs.
+    Only clear uploads staging.
+    """
+    _ensure_dirs()
+    if UPLOAD_DIR.exists():
+        shutil.rmtree(UPLOAD_DIR, ignore_errors=True)
+    UPLOAD_DIR.mkdir(parents=True, exist_ok=True)
+
+
 def _parse_config(config_text: str) -> Dict[str, Any]:
     text = (config_text or "").strip()
     if not text:
@@ -42,17 +51,22 @@ def _parse_config(config_text: str) -> Dict[str, Any]:
     except json.JSONDecodeError as e:
         raise ValueError(f"Config JSON is invalid: {e}") from e
 
-    # Merge user config over defaults (shallow merge is enough for your current shape)
+    if not isinstance(cfg, dict):
+        raise ValueError("Config JSON must be an object (dict).")
+
     merged = dict(DEFAULT_CONFIG)
-    merged.update(cfg if isinstance(cfg, dict) else {})
+    merged.update(cfg)
     return merged
 
 
 def _stage_inputs(uploaded_files: Optional[List[str]]) -> List[str]:
     """
-    Gradio gives file paths when using type='filepath'.
-    We copy into /tmp for stable paths. If ZIP is uploaded, extract PDFs.
+    Gradio gives file paths (type='filepath').
+    Copy into UPLOAD_DIR for stable paths.
+    Supports PDFs and ZIPs containing PDFs.
     """
+    _clean_upload_staging_only()
+
     staged: List[str] = []
     uploaded_files = uploaded_files or []
 
@@ -61,28 +75,25 @@ def _stage_inputs(uploaded_files: Optional[List[str]]) -> List[str]:
         if not src.exists():
             continue
 
-        # ZIP case
         if src.suffix.lower() == ".zip":
             with zipfile.ZipFile(src, "r") as z:
                 z.extractall(UPLOAD_DIR)
             staged.extend([str(p.resolve()) for p in sorted(UPLOAD_DIR.rglob("*.pdf"))])
             continue
 
-        # PDF case
         if src.suffix.lower() == ".pdf":
             dst = UPLOAD_DIR / src.name
             shutil.copy2(src, dst)
             staged.append(str(dst.resolve()))
 
-    # Deduplicate while preserving order
+    # dedupe while preserving order
     seen = set()
-    unique: List[str] = []
+    out: List[str] = []
     for p in staged:
         if p not in seen:
             seen.add(p)
-            unique.append(p)
-
-    return unique
+            out.append(p)
+    return out
 
 
 def _zip_dir(src_dir: Path, zip_path: Path) -> None:
@@ -95,32 +106,27 @@ def _zip_dir(src_dir: Path, zip_path: Path) -> None:
                 z.write(p, arcname=str(p.relative_to(src_dir)))
 
 
-# ---- Gradio handler ----
-
 def process(uploaded_files: Optional[List[str]], config_text: str) -> Tuple[str, Optional[str]]:
     """
-    Returns:
-      (status_text, results_zip_path_or_none)
+    Returns (status, results_zip_path_or_none).
     """
-    _clean_tmp()
+    _ensure_dirs()
 
     cfg = _parse_config(config_text)
     pdf_paths = _stage_inputs(uploaded_files)
 
     if not pdf_paths:
-        return ("No PDFs found. Upload one or more PDFs, or a ZIP containing PDFs.", None)
-
-    # Run pipeline
-try:
-    run_pipeline(
-        input_files=pdf_paths,
-        config=cfg,
-        output_dir=str(OUTPUT_DIR),
-    )
-except Exception as e:
-    return (f"Pipeline failed: {type(e).__name__}: {e}", None)
-
-    # Zip results
+        return ("No PDFs found. Upload PDFs or a ZIP containing PDFs.", None)
+
+    try:
+        run_pipeline(
+            input_files=pdf_paths,
+            config=cfg,
+            output_dir=str(OUTPUT_DIR),
+        )
+    except Exception as e:
+        return (f"Pipeline failed: {type(e).__name__}: {e}", None)
+
     try:
         _zip_dir(OUTPUT_DIR, RESULTS_ZIP)
     except Exception as e:
@@ -129,44 +135,32 @@ except Exception as e:
     return (f"Processed {len(pdf_paths)} file(s). Results are ready.", str(RESULTS_ZIP))
 
 
-# ---- UI ----
-
 def build_ui() -> gr.Blocks:
     with gr.Blocks(title="Resume Evaluator") as demo:
         gr.Markdown("# Resume Evaluator")
 
-        with gr.Group():
-            files = gr.File(
-                label="Upload PDF(s) or a ZIP",
-                file_count="multiple",
-                type="filepath",
-            )
+        files = gr.File(
+            label="Upload PDF(s) or a ZIP",
+            file_count="multiple",
+            type="filepath",
+        )
 
-        with gr.Group():
-            config = gr.Code(
-                label="Config JSON (optional)",
-                language="json",
-                value=json.dumps(DEFAULT_CONFIG, indent=2),
-            )
+        config = gr.Code(
+            label="Config JSON (optional)",
+            language="json",
+            value=json.dumps(DEFAULT_CONFIG, indent=2),
+        )
 
-        with gr.Row():
-            btn = gr.Button("Process", variant="primary")
+        btn = gr.Button("Process", variant="primary")
 
         status = gr.Textbox(label="Status", interactive=False)
-
         results = gr.File(label="Download Results ZIP", interactive=False)
 
-        btn.click(
-            fn=process,
-            inputs=[files, config],
-            outputs=[status, results],
-        )
-
+        btn.click(fn=process, inputs=[files, config], outputs=[status, results])
     return demo
 
 
 demo = build_ui()
 
 if __name__ == "__main__":
-    # Keep it explicit for HF Spaces
     demo.launch(server_name="0.0.0.0", server_port=7860)