Update app.py

app.py

@@ -1,7 +1,8 @@
 """
-app.py — Step 8
-Adds a simple toggle to choose Extractive (safe) or Generative (Evo),
-plus temperature and max_tokens controls.
+app.py — Step 8 (refined)
+Adds:
+- Auto-build of the index on first question (prevents "Index not found" error)
+- Same Extractive/Generative toggle + temperature/max_tokens controls
 """
 
 from pathlib import Path
@@ -9,7 +10,13 @@ from typing import List
 import gradio as gr
 
 from indexer import build_index
-from rag_search import RAGSearcher, format_sources, extract_links, split_form_links, links_markdown
+from rag_search import (
+    RAGSearcher,
+    format_sources,
+    extract_links,
+    split_form_links,
+    links_markdown,
+)
 from evo_inference import synthesize_with_evo
 from utils_lang import SUPPORTED, normalize_lang, L
 
@@ -17,83 +24,140 @@ _searcher = None
 DATA_SEED_DIR = Path("data/seed")
 ALLOWED_EXTS = {".txt", ".pdf", ".html", ".htm", ".xhtml"}
 
+
 def ensure_searcher():
+    """Create (or return) a global searcher bound to the current index."""
     global _searcher
-    if _searcher is None: _searcher = RAGSearcher()
+    if _searcher is None:
+        _searcher = RAGSearcher()  # may raise RuntimeError if index missing
     return _searcher
 
+
 def on_build_index():
+    """Build the FAISS index and reset the cached searcher."""
     try:
         msg = build_index()
-        global _searcher; _searcher = None
+        global _searcher
+        _searcher = None  # force reload next time
     except Exception as e:
         msg = f"Error while building index: {e}"
     return msg
 
+
 def _save_files(files: List[gr.File]) -> List[str]:
-    saved=[]; DATA_SEED_DIR.mkdir(parents=True, exist_ok=True)
-    if not files: return saved
+    """Save uploaded files (txt/pdf/html) to data/seed/."""
+    saved = []
+    DATA_SEED_DIR.mkdir(parents=True, exist_ok=True)
+    if not files:
+        return saved
     for f in files:
         try:
             path = None
-            if isinstance(f, dict) and "name" in f: 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
-            if path.suffix.lower() not in ALLOWED_EXTS: continue
+            if isinstance(f, dict) and "name" in f:
+                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
+            if path.suffix.lower() not in ALLOWED_EXTS:
+                continue
             (DATA_SEED_DIR / path.name).write_bytes(path.read_bytes())
             saved.append(path.name)
-        except: continue
+        except Exception:
+            continue
     return saved
 
+
 def on_upload_and_reindex(files):
+    """Save uploaded files then rebuild the index."""
     saved = _save_files(files or [])
-    if not saved: return "No valid .txt/.pdf/.html uploaded."
+    if not saved:
+        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, mode_label: str, temp: float, max_tokens: int, history: list):
+
+def on_ask(
+    question: str,
+    top_k: int,
+    lang_code: str,
+    mode_label: str,
+    temp: float,
+    max_tokens: int,
+    history: list,
+):
+    """
+    Handle a user question end-to-end:
+    - Ensure index exists (auto-build on first use)
+    - Retrieve top-k chunks
+    - Synthesize answer (Extractive or Generative)
+    - Show sources + extracted 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)", "", ""
+        return history, L(lang, "intro_err"), f"{L(lang, 'sources')}: (none)", "", ""
+
     try:
-        searcher = ensure_searcher()
+        # Ensure searcher; if index missing, build once automatically then retry
+        try:
+            searcher = ensure_searcher()
+        except RuntimeError:
+            on_build_index()
+            searcher = ensure_searcher()
+
         hits = searcher.search(question, k=int(top_k))
-        # Map UI label to internal arg
+
+        # Map UI label to internal mode flag
         mode = "generative" if mode_label.startswith("Generative") else "extractive"
 
         answer = synthesize_with_evo(
-            question, lang, hits,
+            question,
+            lang,
+            hits,
             mode=mode,
             max_new_tokens=int(max_tokens),
             temperature=float(temp),
         )
 
+        # Localized "Sources" header
         srcs = format_sources(hits)
-        if srcs.startswith("Sources:"): srcs = srcs.replace("Sources:", f"{L(lang,'sources')}:")
+        if srcs.startswith("Sources:"):
+            srcs = srcs.replace("Sources:", f"{L(lang, 'sources')}:")
+
+        # Extract URLs and form links
         all_links = extract_links(hits)
         form_links, other_links = split_form_links(all_links)
-        links_md = links_markdown(L(lang,"links"), other_links)
-        forms_md = links_markdown(L(lang,"forms"), form_links)
+        links_md = links_markdown(L(lang, "links"), other_links)
+        forms_md = links_markdown(L(lang, "forms"), form_links)
+
     except Exception as e:
-        answer = f"Error: {e}\n\n{L(lang,'tip_build')}"
-        srcs = f"{L(lang,'sources')}: (none)"
-        links_md = f"**{L(lang,'links')}:** (none)"
-        forms_md = f"**{L(lang,'forms')}:** (none)"
+        answer = f"Error: {e}\n\n{L(lang, 'tip_build')}"
+        srcs = f"{L(lang, 'sources')}: (none)"
+        links_md = f"**{L(lang, 'links')}:** (none)"
+        forms_md = f"**{L(lang, 'forms')}:** (none)"
+
     history = history + [(question, answer)]
     return history, answer, srcs, links_md, forms_md
 
-with gr.Blocks(title=L("en","title")) as demo:
+
+with gr.Blocks(title=L("en", "title")) as demo:
     gr.Markdown(f"# 🇲🇺 **{L('en','title')}** — Step 8")
 
-    # Admin
+    # Admin: build index / upload & reindex
     with gr.Row():
-        build_btn = gr.Button(L("en","build_btn"))
-        status = gr.Markdown(L("en","status_idle"))
+        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 / PDF / HTML and auto-reindex", open=False):
-        upload = gr.File(file_count="multiple", file_types=["file"], label="Upload .txt / .pdf / .html")
+        upload = gr.File(
+            file_count="multiple",
+            file_types=["file"],
+            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)
@@ -107,15 +171,15 @@ with gr.Blocks(title=L("en","title")) as demo:
         mode = gr.Radio(
             choices=["Extractive (safe)", "Generative (Evo)"],
             value="Extractive (safe)",
-            label="Mode"
+            label="Mode",
         )
         temp = gr.Slider(0.0, 1.2, value=0.4, step=0.05, label="Temperature (gen)")
         max_tokens = gr.Slider(64, 384, value=192, step=16, label="Max new tokens (gen)")
 
     gr.Markdown("### Ask a question / Posez une question / Poz enn kestyon")
     with gr.Row():
-        q = gr.Textbox(placeholder=L("en","ask_placeholder"), label=L("en","ask_label"))
-        k = gr.Slider(3, 12, step=1, value=6, label=L("en","k_label"))
+        q = gr.Textbox(placeholder=L("en", "ask_placeholder"), label=L("en", "ask_label"))
+        k = gr.Slider(3, 12, step=1, value=6, label=L("en", "k_label"))
         ask_btn = gr.Button("Ask / Demander / Rode")
 
     chat = gr.Chatbot(label="Assistant", height=360)