Update app.py

app.py

@@ -1,13 +1,13 @@
 """
 app.py
-Step 6: Add official links & forms display + Admin upload (TXT) with auto-reindex.
+Step 7: Allow uploading .txt + .pdf + .html files from the Admin panel (then auto-reindex).
 
-New in Step 6 (Objective):
-- After answering, we extract URLs from retrieved chunks, split "forms", and show both as clickable lists.
-- Admin can upload new TXT files from the UI; we save them to data/seed/ and rebuild the index automatically.
+(Objective)
+- Accept multiple files (txt/pdf/html) in the Admin section.
+- Save them into data/seed/ (overwrites by filename).
+- Rebuild index and clear the cached searcher.
 """
 
-import os
 from pathlib import Path
 from typing import List
 
@@ -24,13 +24,12 @@ from rag_search import (
 from evo_inference import synthesize_with_evo
 from utils_lang import SUPPORTED, normalize_lang, L
 
-# Global (Objective) lazy searcher instance
 _searcher = None
 DATA_SEED_DIR = Path("data/seed")
+ALLOWED_EXTS = {".txt", ".pdf", ".html", ".htm", ".xhtml"}  # (Objective)
 
 
 def ensure_searcher():
-    """(Objective) Create/return a global RAGSearcher bound to the current index."""
     global _searcher
     if _searcher is None:
         _searcher = RAGSearcher()
@@ -38,7 +37,6 @@ def ensure_searcher():
 
 
 def on_build_index():
-    """(Objective) Build FAISS index, reset searcher so it reloads next query."""
     try:
         msg = build_index()
         global _searcher
@@ -48,11 +46,10 @@ def on_build_index():
     return msg
 
 
-def save_uploaded_txt(files: List[gr.File]) -> List[str]:
+def _save_files(files: List[gr.File]) -> List[str]:
     """
-    (Objective) Save uploaded .txt files into data/seed/.
-    Gradio may pass objects with .name or file paths as strings; support both.
-    Returns a list of saved filenames.
+    (Objective) Save uploaded files with allowed extensions into data/seed/.
+    Supports different Gradio file object shapes.
     """
     saved = []
     DATA_SEED_DIR.mkdir(parents=True, exist_ok=True)
@@ -61,47 +58,37 @@ def save_uploaded_txt(files: List[gr.File]) -> List[str]:
 
     for f in files:
         try:
-            # Handle both dict-like and path-like objects gracefully
-            path = getattr(f, "name", None) or (f if isinstance(f, str) else None)
-            # Some gradio versions pass dicts with 'name'
+            # Resolve a path from the gradio file object
+            path = None
             if isinstance(f, dict) and "name" in f:
-                path = f["name"]
-            if not path:
+                path = Path(f["name"])
+            elif hasattr(f, "name"):
+                path = Path(getattr(f, "name"))
+            elif isinstance(f, str):
+                path = Path(f)
+            if not path or not path.exists():
                 continue
-            # Read bytes and write to seed folder using the base filename
-            src = Path(path)
-            if src.suffix.lower() != ".txt":
-                continue  # Only .txt for this step
-            content = Path(path).read_bytes()
-            dest = DATA_SEED_DIR / src.name
-            dest.write_bytes(content)
-            saved.append(src.name)
+
+            if path.suffix.lower() not in ALLOWED_EXTS:
+                continue
+
+            dest = DATA_SEED_DIR / path.name
+            dest.write_bytes(path.read_bytes())
+            saved.append(path.name)
         except Exception:
-            # Skip problematic files silently for robustness
             continue
     return saved
 
 
 def on_upload_and_reindex(files) -> str:
-    """
-    (Objective) Save uploaded .txt files and rebuild the index immediately.
-    """
-    saved = save_uploaded_txt(files or [])
+    saved = _save_files(files or [])
     if not saved:
-        return "No valid .txt uploaded."
+        return "No valid .txt/.pdf/.html uploaded."
     status = on_build_index()
     return f"Uploaded: {', '.join(saved)}\n{status}"
 
 
 def on_ask(question: str, top_k: int, lang_code: str, history: list):
-    """
-    (Objective) Handle user question end-to-end:
-    - Ensure index exists
-    - Retrieve top-k hits
-    - Synthesize grounded answer (templated Evo)
-    - Extract and render links & forms
-    - Append to chat history
-    """
     lang = normalize_lang(lang_code)
     if not question or len(question.strip()) < 3:
         return history, L(lang, "intro_err"), f"{L(lang, 'sources')}: (none)", "", ""
@@ -111,12 +98,10 @@ def on_ask(question: str, top_k: int, lang_code: str, history: list):
         hits = searcher.search(question, k=int(top_k))
         answer = synthesize_with_evo(question, lang, hits)
 
-        # Sources (localized header)
         srcs = format_sources(hits)
         if srcs.startswith("Sources:"):
             srcs = srcs.replace("Sources:", f"{L(lang, 'sources')}:")
 
-        # Links & Forms
         all_links = extract_links(hits)
         form_links, other_links = split_form_links(all_links)
         links_md = links_markdown(L(lang, "links"), other_links)
@@ -133,21 +118,24 @@ def on_ask(question: str, top_k: int, lang_code: str, history: list):
 
 
 with gr.Blocks(title=L("en", "title")) as demo:
-    gr.Markdown(f"# 🇲🇺 **{L('en','title')}** — Step 6")
+    gr.Markdown(f"# 🇲🇺 **{L('en','title')}** — Step 7")
 
-    # --- Admin controls
     with gr.Row():
         build_btn = gr.Button(L("en", "build_btn"))
         status = gr.Markdown(L("en", "status_idle"))
     build_btn.click(fn=on_build_index, outputs=status)
 
-    with gr.Accordion("Admin: upload TXT and auto-reindex", open=False):
-        upload = gr.File(file_count="multiple", file_types=["text"], label="Upload .txt files")
+    # Admin: upload txt/pdf/html, then reindex
+    with gr.Accordion("Admin: upload TXT / PDF / HTML and auto-reindex", open=False):
+        upload = gr.File(
+            file_count="multiple",
+            file_types=["file"],  # allow any file; we'll filter by extension
+            label="Upload .txt / .pdf / .html"
+        )
         save_btn = gr.Button("Save & Reindex")
         upload_status = gr.Markdown()
         save_btn.click(fn=on_upload_and_reindex, inputs=upload, outputs=upload_status)
 
-    # --- User Q&A
     gr.Markdown("### Language / Langue / Langaz")
     lang = gr.Dropdown(choices=list(SUPPORTED.keys()), value="en", label="EN / FR / MFE")