Update app.py

app.py

@@ -10,6 +10,7 @@ from pathlib import Path # <-- Add this import at the top of your file with the
 import re
 
 
+
 # --- Agent Imports & Safe Fallbacks ---
 try:
     from alz_companion.agent import (
@@ -83,26 +84,67 @@ CONFIG = {
 
 # --- File Management & Vector Store Logic ---
 
-# --- FIX: Anchor all paths to the script's location for reliability on HF Spaces ---
-SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
-INDEX_BASE = os.path.join(SCRIPT_DIR, 'data')
-PERSONAL_DATA_BASE = os.path.join(SCRIPT_DIR, 'personal_data')
-# --- END FIX ---
 
+# --- Persistent storage root --- CG5
+def _storage_root() -> Path:
+    """
+    Choose a durable home for runtime artefacts:
+    1) $SPACE_STORAGE -> custom mount if you set it
+    2) /data          -> Hugging Face Spaces persistent volume
+    3) ~/.cache/alz_companion -> portable fallback
+    """
+    candidates = [
+        Path(os.getenv("SPACE_STORAGE", "")),
+        Path("/data"),
+        Path.home() / ".cache" / "alz_companion",
+    ]
+    for p in candidates:
+        if not p:
+            continue
+        try:
+            p.mkdir(parents=True, exist_ok=True)
+            probe = p / ".write_test"
+            with open(probe, "w") as f:
+                f.write("ok")
+            probe.unlink(missing_ok=True)
+            return p
+        except Exception:
+            continue
+    # Last resort: temp (not persistent, but avoids crashing)
+    tmp = Path(tempfile.gettempdir()) / "alz_companion"
+    tmp.mkdir(parents=True, exist_ok=True)
+    return tmp
+
+STORAGE_ROOT = _storage_root()
+
+# --- File Management & Vector Store Logic (persistent) --- CG5
+INDEX_BASE = str(STORAGE_ROOT / "index")
+PERSONAL_DATA_BASE = str(STORAGE_ROOT / "personal")
 UPLOADS_BASE = os.path.join(INDEX_BASE, "uploads")
 PERSONAL_INDEX_PATH = os.path.join(PERSONAL_DATA_BASE, "personal_faiss_index")
-THEME_PATHS = {t: os.path.join(INDEX_BASE, f"faiss_index_{t.replace(' ', '').lower()}") for t in CONFIG["themes"]}
+THEME_PATHS = {
+    t: os.path.join(INDEX_BASE, f"faiss_index_{t.replace(' ', '').lower()}")
+    for t in CONFIG["themes"]
+}
 
-# Ensure all necessary directories are created on startup
 os.makedirs(UPLOADS_BASE, exist_ok=True)
 os.makedirs(os.path.dirname(PERSONAL_INDEX_PATH), exist_ok=True)
+for p in THEME_PATHS.values():
+    os.makedirs(p, exist_ok=True)
 
-
-THEME_PATHS = {t: os.path.join(INDEX_BASE, f"faiss_index_{t.replace(' ', '').lower()}") for t in CONFIG["themes"]}
 vectorstores = {}
 personal_vectorstore = None
 test_fixtures = [] # <-- ADD THIS LINE
 
+# --- Load existing personal index if present --- CG5
+try:
+    personal_vectorstore = build_or_load_vectorstore([], PERSONAL_INDEX_PATH, is_personal=True)
+except Exception:
+    # stay graceful if the index is missing/corrupt; user can rebuild by adding memories
+    personal_vectorstore = None
+
+
+
 def canonical_theme(tk: str) -> str: return tk if tk in CONFIG["themes"] else "All"
 def theme_upload_dir(theme: str) -> str:
     p = os.path.join(UPLOADS_BASE, f"theme_{canonical_theme(theme).replace(' ', '').lower()}")
@@ -265,6 +307,19 @@ def handle_add_knowledge(title, text_input, file_input, image_input, yt_url, set
             yt = YouTube(yt_url)
             video_title = yt.title
             final_title = title.strip() if title and title.strip() else video_title
+
+            # --- suggested as optional by CG5
+            # media_dir = STORAGE_ROOT / "media"
+            # media_dir.mkdir(parents=True, exist_ok=True)
+            # temp_audio_path = str(media_dir / f"yt_{datetime.now().strftime('%Y%m%d_%H%M%S')}.mp4")
+
+            # audio_stream = yt.streams.get_audio_only()
+            # audio_stream.download(filename=temp_audio_path)
+
+            # content_text = transcribe_audio(temp_audio_path)
+            # content_source = f"YouTube: {video_title} ({temp_audio_path})"
+            # If you truly don't want to keep files, you may still remove it later via a UI control.
+            
             audio_stream = yt.streams.get_audio_only()
             with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as temp_audio_file:
                 audio_stream.download(filename=temp_audio_file.name)
@@ -272,6 +327,7 @@ def handle_add_knowledge(title, text_input, file_input, image_input, yt_url, set
             content_text = transcribe_audio(temp_audio_path)
             content_source = f"YouTube: {video_title}"
             os.remove(temp_audio_path)
+            
             full_content = f"Title: {final_title}\n\nContent: {content_text}"
             docs_to_add = parse_and_tag_entries(full_content, content_source, settings=settings)
         except Exception as e: