codex review fixes: duplicate-post recovery, fsync, fatal corrupt state, narrower comment drop

- create_post: on HTTPError, look for an existing post tagged ledger:<id>
  (insert-then-error recovery) and adopt it before retrying
- save_state: fsync tmp file before rename
- load_state: corrupt state is fatal (auto-reseed would duplicate every
  open item's post)
- comment: drop only on 500/404/410; 401/403/429/502/503 retry next poll

Accepted as-is (with rationale): same-second ordering relies on the feed
replaying the JSONL in append order + Python's stable sort; seed-window
add+close races reduce to the documented 'predates bridge' skip.

Co-Authored-By: Claude Fable 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01YECpkAwQUwgu7NVy91R8fW

bridge.py

@@ -97,7 +97,11 @@ def load_state() -> dict:
         try:
             return json.loads(STATE_FILE.read_text(encoding="utf-8"))
         except json.JSONDecodeError:
-            log(f"WARN state file corrupt, starting fresh: {STATE_FILE}")
+            # Starting fresh here would re-seed every open item -> a duplicate
+            # post per item. Refuse to run until a human inspects the file.
+            log(f"FATAL state file corrupt: {STATE_FILE} — fix or remove it "
+                "deliberately; refusing to reseed automatically")
+            sys.exit(1)
     return {"checkpoint": "", "seen": [], "map": {}, "seeded": False,
             "channel_id": ""}
 
@@ -105,7 +109,10 @@ def load_state() -> dict:
 def save_state(state: dict) -> None:
     STATE_FILE.parent.mkdir(parents=True, exist_ok=True)
     tmp = STATE_FILE.with_suffix(".json.tmp")
-    tmp.write_text(json.dumps(state, indent=1), encoding="utf-8")
+    with open(tmp, "w", encoding="utf-8") as f:
+        json.dump(state, f, indent=1)
+        f.flush()
+        os.fsync(f.fileno())
     os.replace(tmp, STATE_FILE)
 
 
@@ -239,10 +246,36 @@ def send_message(channel_id: str, content: str, thread_id=None) -> dict:
                body)
 
 
+def find_existing_post(state: dict, ledger_id: str):
+    """Recovery lookup: the target API can insert a row and then fail, so a
+    retried create must first check whether a post tagged ledger:<id> already
+    exists. Bounded to the most recent 200 channel messages."""
+    try:
+        msgs = api("GET", f"/api/workspaces/{WORKSPACE_ID}/comms/channels/"
+                          f"{ensure_channel(state)}/messages?limit=200") or []
+    except Exception:
+        return None
+    needle = f"ledger:{ledger_id}"
+    for m in msgs:
+        if not m.get("thread_id") and needle in (m.get("content") or ""):
+            return m.get("id")
+    return None
+
+
 def create_post(state: dict, item: dict) -> None:
     if item["id"] in state["map"]:
         return  # already mirrored (seed/feed overlap)
+    try:
         msg = send_message(ensure_channel(state), post_body(item))
+    except urllib.error.HTTPError:
+        # The insert may have landed before the error (proven insert-then-500
+        # behaviour). Adopt an existing post instead of duplicating on retry.
+        existing = find_existing_post(state, item["id"])
+        if existing:
+            state["map"][item["id"]] = existing
+            log(f"post   {item['id'][:8]} -> adopted existing {existing}")
+            return
+        raise  # genuinely not created; retry next poll
     state["map"][item["id"]] = msg["id"]
     log(f"post   {item['id'][:8]} -> {msg['id']} ({item.get('title', '')[:60]})")
 
@@ -255,11 +288,15 @@ def comment(state: dict, ledger_id: str, content: str, what: str) -> None:
     try:
         msg = send_message(ensure_channel(state), content, thread_id=post_id)
     except urllib.error.HTTPError as e:
-        # mib007 comms can 500 AFTER the insert (proven: mention-path crash),
-        # so retrying a failed comment risks duplicates. Comments are
-        # best-effort: log and move on instead of blocking the checkpoint.
-        log(f"DROP {what} {ledger_id[:8]}: HTTP {e.code} (comment not retried)")
+        # 500: mib007 comms can fail AFTER the insert (proven mention-path
+        # crash) — retrying risks duplicates, so drop. 404/410: the post is
+        # gone — retrying can never succeed. Everything else (401/403/429/
+        # 502/503) is safe to retry next poll.
+        if e.code in (500, 404, 410):
+            log(f"DROP {what} {ledger_id[:8]}: HTTP {e.code} "
+                "(comment not retried)")
             return
+        raise
     log(f"{what:6} {ledger_id[:8]} -> comment {msg['id']}")