fix(backend): profile history gets a row only when a field really changed, partial snapshots keep the known phone and username

backend/src/userbot/modules/profiles/parse.py

@@ -1,4 +1,4 @@
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 
 from pyrogram import Client, raw
 from pyrogram.types import Chat, User
@@ -12,6 +12,11 @@ class ProfileFields:
     phone: str | None = None
     photo_unique_id: str | None = None
     is_deleted_account: bool = False
+    unknown: frozenset[str] = field(default=frozenset(), compare=False)
+
+
+MIN_UNKNOWN = frozenset({"username", "phone"})
+CHAT_UNKNOWN = frozenset({"phone", "is_deleted_account"})
 
 
 def active_username(usernames: list[raw.types.Username]) -> str | None:
@@ -35,6 +40,7 @@ async def snapshot_from_user(
         phone=user.phone_number if user else None,
         photo_unique_id=photo_unique_id,
         is_deleted_account=bool(getattr(raw_user, "deleted", False)),
+        unknown=MIN_UNKNOWN if getattr(raw_user, "min", False) else frozenset(),
     )
     return fields, photo_file_id, photo_unique_id
 
@@ -52,6 +58,7 @@ def snapshot_from_high_level(
         phone=user.phone_number,
         photo_unique_id=photo_unique_id,
         is_deleted_account=bool(user.is_deleted),
+        unknown=MIN_UNKNOWN if user.is_min else frozenset(),
     )
     return fields, photo_file_id, photo_unique_id
 
@@ -64,8 +71,7 @@ def snapshot_from_chat(chat: Chat) -> tuple[ProfileFields, str | None, str | Non
         first_name=chat.first_name,
         last_name=chat.last_name,
         username=chat.username,
-        phone=None,
         photo_unique_id=photo_unique_id,
-        is_deleted_account=False,
+        unknown=CHAT_UNKNOWN,
     )
     return fields, photo_file_id, photo_unique_id

backend/src/userbot/modules/profiles/repository.py

@@ -9,26 +9,46 @@ WHERE account_id = $1 AND peer_id = $2
 """
 
 _UPSERT_PEER = """
-INSERT INTO peers
+INSERT INTO peers AS p
     (account_id, peer_id, first_name, last_name, username, phone,
      photo_unique_id, is_deleted_account, raw, updated_at)
 VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9::jsonb, now())
 ON CONFLICT (account_id, peer_id) DO UPDATE SET
     first_name = EXCLUDED.first_name,
     last_name = EXCLUDED.last_name,
-    username = EXCLUDED.username,
-    phone = EXCLUDED.phone,
+    username = CASE WHEN 'username' = ANY($10::text[])
+        THEN COALESCE(EXCLUDED.username, p.username) ELSE EXCLUDED.username END,
+    phone = CASE WHEN 'phone' = ANY($10::text[])
+        THEN COALESCE(EXCLUDED.phone, p.phone) ELSE EXCLUDED.phone END,
     photo_unique_id = EXCLUDED.photo_unique_id,
-    is_deleted_account = EXCLUDED.is_deleted_account,
+    is_deleted_account = CASE WHEN 'is_deleted_account' = ANY($10::text[])
+        THEN p.is_deleted_account ELSE EXCLUDED.is_deleted_account END,
     raw = EXCLUDED.raw,
     updated_at = now()
 """
 
-_INSERT_HISTORY = """
+_INSERT_CHANGE = """
 INSERT INTO peer_history
     (account_id, peer_id, observed_at, first_name, last_name, username, phone,
      photo_unique_id, is_deleted_account, raw)
-VALUES ($1, $2, now(), $3, $4, $5, $6, $7, $8, $9::jsonb)
+SELECT account_id, peer_id, clock_timestamp(), first_name, last_name, username,
+       phone, photo_unique_id, is_deleted_account, raw
+FROM peers p
+WHERE account_id = $1 AND peer_id = $2
+  AND NOT EXISTS (
+    SELECT 1 FROM (
+        SELECT first_name, last_name, username, phone, photo_unique_id,
+               is_deleted_account
+        FROM peer_history
+        WHERE account_id = $1 AND peer_id = $2
+        ORDER BY observed_at DESC LIMIT 1
+    ) latest
+    WHERE (latest.first_name, latest.last_name, latest.username, latest.phone,
+           latest.photo_unique_id, latest.is_deleted_account)
+          IS NOT DISTINCT FROM
+          (p.first_name, p.last_name, p.username, p.phone,
+           p.photo_unique_id, p.is_deleted_account)
+  )
 ON CONFLICT DO NOTHING
 """
 
@@ -57,8 +77,12 @@ async def write_profile(  # noqa: PLR0913
     raw: str,
     *,
     history: bool,
-) -> None:
-    args = (
+) -> bool:
+    async with pool.acquire() as conn, conn.transaction():
+        await conn.execute(
+            _UPSERT_PEER,
+            account_id,
+            peer_id,
             fields.first_name,
             fields.last_name,
             fields.username,
@@ -66,8 +90,9 @@ async def write_profile(  # noqa: PLR0913
             fields.photo_unique_id,
             fields.is_deleted_account,
             raw,
+            sorted(fields.unknown),
         )
-    async with pool.acquire() as conn, conn.transaction():
-        await conn.execute(_UPSERT_PEER, account_id, peer_id, *args)
-        if history:
-            await conn.execute(_INSERT_HISTORY, account_id, peer_id, *args)
+        if not history:
+            return False
+        status = await conn.execute(_INSERT_CHANGE, account_id, peer_id)
+    return status == "INSERT 0 1"