devplacepy/database/schema.py

@@ -30,6 +30,26 @@ def migrate_bug_tables_to_issue_tables() -> None:
         logger.info("Dropped table %s after migration", source_name)
 
 
+def _add_message_unique_index() -> None:
+    if "messages" not in db.tables:
+        return
+    with db:
+        db.query(
+            """
+            DELETE FROM messages WHERE id NOT IN (
+                SELECT MIN(id) FROM messages GROUP BY sender_uid, receiver_uid, content
+            )
+            """
+        )
+    _index(
+        db,
+        "messages",
+        "idx_messages_unique_sender_receiver_content",
+        ["sender_uid", "receiver_uid", "content"],
+        unique=True,
+    )
+
+
 def init_db():
     tables = db.tables
     _index(db, "users", "idx_users_username", ["username"])
@@ -131,6 +151,7 @@ def init_db():
         "idx_messages_conversation_rev",
         ["receiver_uid", "sender_uid"],
     )
+    _add_message_unique_index()
     _index(db, "notifications", "idx_notifications_user", ["user_uid"])
     _index(db, "notifications", "idx_notifications_user_read", ["user_uid", "read"])
     _index(db, "push_registration", "idx_push_registration_user", ["user_uid"])

devplacepy/services/correction.py

@@ -88,7 +88,7 @@ def gateway_complete(
             {"role": "system", "content": system},
             {"role": "user", "content": text},
         ],
-        "temperature": 0.1,
+        "temperature": 0.0,
     }
     headers = {
         "Content-Type": "application/json",

devplacepy/services/messaging/persist.py

@@ -1,6 +1,9 @@
 # retoor <retoor@molodetz.nl>
 
+import hashlib
 import logging
+import time
+from collections import OrderedDict
 from datetime import datetime, timezone
 from typing import Any, Optional
 
@@ -15,12 +18,16 @@ from devplacepy.utils import (
     track_action,
 )
 from devplacepy.services.audit import record as audit
+from sqlalchemy.exc import IntegrityError
+
 from devplacepy.services.correction import schedule_correction
 from devplacepy.services.ai_modifier import schedule_modification
 
 logger = logging.getLogger("messaging.persist")
 
 MAX_CONTENT_LENGTH = 2000
+DEDUP_WINDOW_SECONDS = 3
+_content_cache: dict[str, tuple[float, str]] = OrderedDict()
 
 
 def _slim_attachment(attachment: dict[str, Any]) -> dict[str, Any]:
@@ -81,19 +88,60 @@ def persist_message(
 
     sender_uid = sender["uid"]
     sender_username = sender.get("username", "")
+
+    content_hash = hashlib.sha256(
+        f"{sender_uid}:{receiver_uid}:{content}".encode()
+    ).hexdigest()[:16]
+    now = time.time()
+    last_seen, cached_uid = _content_cache.get(content_hash, (0.0, None))
+    if now - last_seen < DEDUP_WINDOW_SECONDS and cached_uid is not None:
+        logger.debug(
+            "Dedup hit for message hash %s (original uid %s)", content_hash, cached_uid
+        )
+        cached = get_table("messages").find_one(uid=cached_uid)
+        if cached:
+            return {
+                "uid": cached["uid"],
+                "sender_uid": cached["sender_uid"],
+                "receiver_uid": cached["receiver_uid"],
+                "content": cached["content"],
+                "read": cached.get("read", False),
+                "created_at": cached["created_at"],
+            }
+
     messages_table = get_table("messages")
     msg_uid = generate_uid()
     created_at = datetime.now(timezone.utc).isoformat()
-    messages_table.insert(
-        {
-            "uid": msg_uid,
-            "sender_uid": sender_uid,
-            "receiver_uid": receiver_uid,
-            "content": content,
-            "read": False,
-            "created_at": created_at,
+
+    try:
+        messages_table.insert(
+            {
+                "uid": msg_uid,
+                "sender_uid": sender_uid,
+                "receiver_uid": receiver_uid,
+                "content": content,
+                "read": False,
+                "created_at": created_at,
+            }
+        )
+    except IntegrityError:
+        existing = messages_table.find_one(
+            sender_uid=sender_uid, receiver_uid=receiver_uid, content=content
+        )
+        if not existing:
+            raise
+        logger.debug(
+            "Dedup via unique constraint for message (uid %s)", existing["uid"]
+        )
+        _content_cache[content_hash] = (time.time(), existing["uid"])
+        return {
+            "uid": existing["uid"],
+            "sender_uid": existing["sender_uid"],
+            "receiver_uid": existing["receiver_uid"],
+            "content": existing["content"],
+            "read": existing.get("read", False),
+            "created_at": existing["created_at"],
         }
-    )
 
     link_attachments(attachment_uids, "message", msg_uid)
     schedule_correction(sender, "messages", msg_uid, request)
@@ -114,6 +162,8 @@ def persist_message(
     )
     track_action(sender_uid, "message")
 
+    _content_cache[content_hash] = (time.time(), msg_uid)
+
     logger.info(
         "Message %s sent from %s to %s via %s",
         msg_uid,

tests/api/messages/send.py

@@ -166,3 +166,23 @@ def test_send_attachment_only_empty_content_succeeds(seeded_db):
     refresh_snapshot()
     row = get_table("messages").find_one(uid=msg["uid"])
     assert row["content"] == ""
+
+
+def test_duplicate_message_returns_same_uid(seeded_db):
+    s, _ = _member()
+    receiver = _db_user("bob_test")["uid"]
+    content = _unique("dupmsg")
+
+    first = s.post(
+        f"{BASE_URL}/messages/send",
+        headers=JSON_audit_log,
+        data={"content": content, "receiver_uid": receiver},
+    ).json()["data"]
+
+    second = s.post(
+        f"{BASE_URL}/messages/send",
+        headers=JSON_audit_log,
+        data={"content": content, "receiver_uid": receiver},
+    ).json()["data"]
+
+    assert first["uid"] == second["uid"], "duplicate messages should return the same uid"