Capture router-rejected uploads into private review queue (no data lost)

game/datastore.py

@@ -19,7 +19,8 @@ import datetime
 import json
 import os
 
-DATASET_REPO = "HomesteaderLabs/forager-sightings"
+DATASET_REPO = "HomesteaderLabs/forager-sightings"            # public: in-domain finds
+REVIEW_REPO = "HomesteaderLabs/forager-sightings-review"      # private: router-rejected quarantine
 LICENSE = "CC-BY-4.0"
 
 _TOKEN = os.environ.get("HF_TOKEN")
@@ -120,21 +121,18 @@ def _hamming(a: str, b: str) -> int:
     return bin(int(a, 16) ^ int(b, 16)).count("1")
 
 
-def append_sighting(image, user_label: str, machine: dict, contributor: str) -> str:
-    """
-    Commit one contributed photo + metadata row to the dataset, with near-dup
-    rejection. `machine` carries the model's call.
-    Returns: "stored" | "duplicate" | "disabled".
-    """
+def _store(repo: str, image, base_row: dict, contributor: str, msg: str) -> str:
+    """Dedup against `repo`'s metadata.jsonl, then commit image + metadata row.
+    Returns "stored" | "duplicate" | "disabled". Used by both the public sightings
+    write and the private review-queue write."""
     if not _TOKEN:
         return "disabled"
     from io import BytesIO
     from huggingface_hub import CommitOperationAdd, HfApi, hf_hub_download
 
-    # read existing metadata (+ its phashes) once
     existing, rows = "", []
     try:
-        with open(hf_hub_download(DATASET_REPO, "metadata.jsonl", repo_type="dataset",
+        with open(hf_hub_download(repo, "metadata.jsonl", repo_type="dataset",
                                   token=_TOKEN, force_download=True)) as f:
             txt = f.read()
         existing = txt.rstrip("\n")
@@ -152,24 +150,45 @@ def append_sighting(image, user_label: str, machine: dict, contributor: str) ->
     fname = f"images/{contributor}_{ts.replace(':', '').replace('-', '')}.jpg"
     buf = BytesIO()
     image.convert("RGB").save(buf, format="JPEG", quality=90)
-    row = {
-        "file_name": fname, "user_label": user_label,
+    row = {"file_name": fname, **base_row, "contributor": contributor,
+           "consent": True, "license": LICENSE, "timestamp": ts, "phash": ph}
+    new_meta = (existing + "\n" if existing else "") + json.dumps(row) + "\n"
+    try:
+        HfApi(token=_TOKEN).create_commit(
+            repo_id=repo, repo_type="dataset", commit_message=msg,
+            operations=[
+                CommitOperationAdd(path_in_repo=fname, path_or_fileobj=buf.getvalue()),
+                CommitOperationAdd(path_in_repo="metadata.jsonl",
+                                   path_or_fileobj=new_meta.encode("utf-8")),
+            ],
+        )
+    except Exception:
+        # e.g. the Space token lacks write scope on this repo — degrade gracefully
+        # so the UI shows a friendly message instead of crashing the handler.
+        return "disabled"
+    return "stored"
+
+
+def append_sighting(image, user_label: str, machine: dict, contributor: str) -> str:
+    """In-domain find -> public dataset. Returns stored/duplicate/disabled."""
+    return _store(DATASET_REPO, image, {
+        "user_label": user_label,
         "machine_prediction": machine.get("species", "unknown"),
         "machine_confidence": round(float(machine.get("confidence", 0.0)), 4),
         "machine_abstained": bool(machine.get("abstained", True)),
         "machine_safety": machine.get("safety", "UNKNOWN"),
         "routed_domain": machine.get("domain", "unknown"),
-        "contributor": contributor, "consent": True, "license": LICENSE,
-        "timestamp": ts, "phash": ph,
-    }
-    new_meta = (existing + "\n" if existing else "") + json.dumps(row) + "\n"
-    HfApi(token=_TOKEN).create_commit(
-        repo_id=DATASET_REPO, repo_type="dataset",
-        commit_message=f"sighting: {row['machine_prediction']} by {contributor}",
-        operations=[
-            CommitOperationAdd(path_in_repo=fname, path_or_fileobj=buf.getvalue()),
-            CommitOperationAdd(path_in_repo="metadata.jsonl",
-                               path_or_fileobj=new_meta.encode("utf-8")),
-        ],
-    )
-    return "stored"
+    }, contributor, f"sighting: {machine.get('species', 'unknown')} by {contributor}")
+
+
+def append_unrouted(image, user_label: str, router: dict, contributor: str) -> str:
+    """Router-rejected (out-of-domain) find -> PRIVATE review queue for later triage.
+    Captures the model's blind spots (real forageables the router fumbles). Returns
+    stored/duplicate/disabled."""
+    return _store(REVIEW_REPO, image, {
+        "status": "unrouted",
+        "user_label": user_label,
+        "router_domain": router.get("domain", "unknown"),
+        "router_confidence": round(float(router.get("domain_confidence", 0.0)), 4),
+        "reason": router.get("reason", ""),
+    }, contributor, f"review: {user_label} by {contributor}")