Update app.py

app.py

@@ -66,6 +66,22 @@ AUTO_CONT_NEW_TOKENS = 256     # tokens per continuation step
 def dlog(tag, msg):
     if DEBUG: print(f"[{tag}] {msg}")
 
+    
+# --- Logging/upload config (OFF by default unless env says "1")
+PUSH_TRANSCRIPTS  = os.environ.get("PUSH_TRANSCRIPTS", "0") == "1"
+PUSH_FEEDBACK     = os.environ.get("PUSH_FEEDBACK", "0") == "1"
+HF_WRITE_TOKEN    = os.environ.get("HF_WRITE_TOKEN", os.environ.get("HF_TOKEN", ""))
+LOGS_DATASET_ID   = os.environ.get("LOGS_DATASET_ID", "")
+
+def _log_path(prefix: str, ext: str) -> str:
+    # rotate daily, e.g., transcripts-20250929.jsonl
+    today = datetime.datetime.utcnow().strftime("%Y%m%d")
+    return f"{prefix}-{today}.{ext}"
+
+# rotate by default if user didn't override via env
+TRANSCRIPT_PATH = os.environ.get("TRANSCRIPT_PATH", _log_path("transcripts", "jsonl"))
+FEEDBACK_PATH   = os.environ.get("FEEDBACK_PATH",   _log_path("feedback",   "csv"))
+
 # ================== HEAVY IMPORTS ==================
 import faiss
 from sentence_transformers import SentenceTransformer
@@ -916,53 +932,80 @@ def enter_app(first_name, last_name, state):
     state["last_name"]  = last_name
     return gr.update(visible=False), gr.update(visible=True), state, f"Welcome, {first_name}! You can start chatting."
 
-def _push_file_to_hub(local_path: str, repo_path: str) -> None:
-    if not PUSH_FEEDBACK:
-        return
+from huggingface_hub import HfApi
+
+def _push_file_to_dataset(local_path: str, repo_path: str) -> None:
+    """
+    Upload a local file to a *dataset* repo so the Space doesn't rebuild.
+    Requires: HF_WRITE_TOKEN, LOGS_DATASET_ID
+    """
     if not os.path.exists(local_path):
-        dlog("UPLOAD", f"Skip: {local_path} does not exist")
-        return
-    if not SPACE_REPO_ID:
-        dlog("UPLOAD", "Skip: SPACE_REPO_ID not set")
-        return
+        dlog("UPLOAD", f"Skip: {local_path} does not exist"); return
+    if not LOGS_DATASET_ID:
+        dlog("UPLOAD", "Skip: LOGS_DATASET_ID not set"); return
     if not HF_WRITE_TOKEN:
-        dlog("UPLOAD", "Skip: HF_WRITE_TOKEN not set")
-        return
+        dlog("UPLOAD", "Skip: HF_WRITE_TOKEN not set"); return
     try:
         api = HfApi(token=HF_WRITE_TOKEN)
         api.upload_file(
             path_or_fileobj=local_path,
-            path_in_repo=repo_path,
-            repo_id=SPACE_REPO_ID,
-            repo_type="space",
+            path_in_repo=repo_path,      # keep short names at the root
+            repo_id=LOGS_DATASET_ID,
+            repo_type="dataset",         # <- key change: dataset, not space
             commit_message=f"Update {repo_path}",
         )
-        dlog("UPLOAD", f"Uploaded {repo_path} to Hub")
+        dlog("UPLOAD", f"Uploaded {repo_path} to dataset {LOGS_DATASET_ID}")
     except Exception as e:
         dlog("UPLOAD", f"Upload failed: {e}")
 
+
 def _push_feedback_to_hub() -> None:
-    _push_file_to_hub(FEEDBACK_PATH, "analytics/feedback.csv")
+    """
+    Back-compat name, but uploads to a *dataset* repo so the Space won't rebuild.
+    Requires: PUSH_FEEDBACK=1, HF_WRITE_TOKEN, LOGS_DATASET_ID, and _push_file_to_dataset().
+    """
+    if not PUSH_FEEDBACK:
+        return
+    try:
+        # Use the rotated filename if you enabled daily rotation
+        remote_name = os.path.basename(FEEDBACK_PATH)
+        _push_file_to_dataset(FEEDBACK_PATH, remote_name)
+        dlog("UPLOAD", f"Feedback uploaded: {remote_name}")
+    except Exception as e:
+        dlog("UPLOAD", f"Feedback upload failed: {e}")
+
 
 def _log_turn(state: Dict[str, Any], question: str, answer: str):
+    """
+    Append locally first, then (optionally) upload to a *dataset* repo.
+    This avoids commits to the Space repo and prevents rebuilds after each turn.
+    """
     rec = {
         "timestamp_utc": _now_iso(),
         "session_id": state.get("session_id", ""),
-        "first_name": state.get("first_name", ""),
-        "last_name":  state.get("last_name", ""),
-        "question": question,
-        "answer":   answer,
+        "first_name":  state.get("first_name", ""),
+        "last_name":   state.get("last_name", ""),
+        "question":    question,
+        "answer":      answer,
     }
+
+    # Local append (ensure dir exists)
     try:
+        os.makedirs(os.path.dirname(TRANSCRIPT_PATH) or ".", exist_ok=True)
         with open(TRANSCRIPT_PATH, "a", encoding="utf-8") as f:
             f.write(json.dumps(rec, ensure_ascii=False) + "\n")
-    except Exception:
-        pass
-    try:
-        if PUSH_TRANSCRIPTS:
-            _push_file_to_hub(TRANSCRIPT_PATH, "analytics/transcripts.jsonl")
-    except Exception:
-        pass
+    except Exception as e:
+        dlog("LOG", f"Failed to write transcript locally: {e}")
+
+    # Optional remote upload (dataset repo -> no Space rebuild)
+    if PUSH_TRANSCRIPTS:
+        try:
+            remote_name = os.path.basename(TRANSCRIPT_PATH)
+            _push_file_to_dataset(TRANSCRIPT_PATH, remote_name)
+            dlog("UPLOAD", f"Transcript uploaded: {remote_name}")
+        except Exception as e:
+            dlog("UPLOAD", f"Transcript upload failed: {e}")
+
 
 def save_feedback(rating, comment, state):
     if rating is None: