feedback: synchronous dataset upload (ZeroGPU-safe)

app.py

@@ -48,40 +48,43 @@ from huggingface_hub import hf_hub_download
 from llama_cpp import Llama
 
 # ---- feedback logging (JSONL, synced to a private HF dataset) ----
+# NB: huggingface_hub.CommitScheduler's background thread breaks under ZeroGPU's
+# process forking ("Invalid file descriptor: -1"), so we append locally and push
+# the file synchronously from the main process instead.
+from huggingface_hub import HfApi
+
 FEEDBACK_REPO = os.environ.get("FEEDBACK_REPO", "AlexWortega/my-pi-agent-feedback")
 _FB_DIR = Path("feedback")
 _FB_DIR.mkdir(exist_ok=True)
 _FB_FILE = _FB_DIR / f"log_{uuid.uuid4().hex}.jsonl"
-_scheduler = None
-try:
-    if os.environ.get("HF_TOKEN"):
-        from huggingface_hub import CommitScheduler
-
-        _scheduler = CommitScheduler(
-            repo_id=FEEDBACK_REPO,
-            repo_type="dataset",
-            folder_path=_FB_DIR,
-            path_in_repo="data",
-            every=5,
-            token=os.environ["HF_TOKEN"],
-        )
-        print("feedback CommitScheduler -> ", FEEDBACK_REPO, flush=True)
-except Exception as e:  # noqa: BLE001
-    print("feedback scheduler disabled:", repr(e)[:160], flush=True)
+_FB_PATH_IN_REPO = f"data/{_FB_FILE.name}"
+_HF_TOKEN = os.environ.get("HF_TOKEN")
+_api = HfApi(token=_HF_TOKEN) if _HF_TOKEN else None
+print("feedback ->", FEEDBACK_REPO if _api else "(local only, no HF_TOKEN)", flush=True)
 
 
 def _log(record):
-    record = {"ts": datetime.datetime.utcnow().isoformat() + "Z", **record}
+    record = {
+        "ts": datetime.datetime.now(datetime.timezone.utc).isoformat(),
+        **record,
+    }
     try:
-        if _scheduler is not None:
-            with _scheduler.lock:
-                with _FB_FILE.open("a", encoding="utf-8") as f:
-                    f.write(json.dumps(record, ensure_ascii=False) + "\n")
-        else:
-            with _FB_FILE.open("a", encoding="utf-8") as f:
-                f.write(json.dumps(record, ensure_ascii=False) + "\n")
+        with _FB_FILE.open("a", encoding="utf-8") as f:
+            f.write(json.dumps(record, ensure_ascii=False) + "\n")
     except Exception as e:  # noqa: BLE001
-        print("log failed:", repr(e)[:120], flush=True)
+        print("local log failed:", repr(e)[:120], flush=True)
+        return
+    if _api is not None:
+        try:
+            _api.upload_file(
+                path_or_fileobj=str(_FB_FILE),
+                path_in_repo=_FB_PATH_IN_REPO,
+                repo_id=FEEDBACK_REPO,
+                repo_type="dataset",
+                commit_message="feedback log",
+            )
+        except Exception as e:  # noqa: BLE001
+            print("dataset upload failed:", repr(e)[:160], flush=True)
 
 # ---- model (GGUF pulled from the Hub at startup, runs on ZeroGPU) ----
 GGUF_REPO = os.environ.get("GGUF_REPO", "AlexWortega/qwen35-4b-soyuz-merged-gguf")
@@ -243,9 +246,9 @@ def respond(message, history, system_prompt, temperature, max_tokens, meta):
         "code": doc,
         "reaction": None,
     }
-    _log(record)
     meta = meta + [record]
     yield history, "", raw, (thinking or _NO_REASONING), doc, _iframe(doc), meta
+    _log(record)  # after yield: upload doesn't block the visible response
 
 
 def on_like(meta, evt: gr.LikeData):