feat(backend): read time of my messages from the archive, exact dates and group readers asked from telegram on demand

backend/src/utils/read/chats.py

@@ -10,7 +10,7 @@ from utils.read.models import (
     MessageView,
     Page,
 )
-from utils.read.read_receipts import read_up_to
+from utils.read.read_receipts import read_times, read_up_to
 
 _MESSAGE_COLS = (
     "chat_id, message_id, date, sender_id, text, has_media, is_self_destruct, "
@@ -285,12 +285,18 @@ async def _apply_read_status(
     self_id = await self_user_id(pool, account_id)
     if self_id is None:
         return
-    marker = await read_up_to(pool, account_id, chat_id)
-    if marker is None:
+    own = [view for view in views if view.sender_id == self_id]
+    if not own:
         return
-    for view in views:
-        if view.sender_id == self_id and view.message_id <= marker:
-            view.read = True
+    marker = await read_up_to(pool, account_id, chat_id)
+    times = await read_times(
+        pool, account_id, chat_id, [(view.message_id, view.date) for view in own]
+    )
+    for view in own:
+        view.read_at = times.get(view.message_id)
+        view.read = view.read_at is not None or (
+            marker is not None and view.message_id <= marker
+        )
 
 
 def _build_album(

backend/src/utils/read/models.py

@@ -203,6 +203,7 @@ class MessageView(BaseModel):
     is_sticker: bool = False
     is_animated_emoji: bool = False
     read: bool = False
+    read_at: datetime | None = None
 
 
 class MessageVersionView(BaseModel):
@@ -387,6 +388,29 @@ class PeerView(BaseModel):
     updated_at: datetime
 
 
+class ReaderView(BaseModel):
+    peer_id: int
+    read_at: datetime
+    peer: PeerView | None
+
+
+class ReadInfoView(BaseModel):
+    chat_id: int
+    message_id: int
+    chat_kind: str
+    own: bool
+    voice: bool
+    observed_at: datetime | None
+    listened_at: datetime | None
+    exact_at: datetime | None
+    readers: list[ReaderView]
+    read: bool
+    unread: bool
+    outcome: str | None
+    fetched_at: datetime | None
+    can_fetch: bool
+
+
 class PeerHistoryView(BaseModel):
     observed_at: datetime
     first_name: str | None

backend/src/utils/read/read_receipts.py

@@ -1,5 +1,89 @@
+from datetime import UTC, datetime, timedelta
+
 import asyncpg
 
+from utils.policy.models import ChatKind
+from utils.read.discover import chat_kind
+from utils.read.models import ReaderView, ReadInfoView
+from utils.read.peers import get_peers
+
+FETCH_WINDOW = timedelta(days=7)
+CLOCK_SKEW = timedelta(minutes=1)
+VOICE_KINDS = ["voice", "video_note"]
+
+_READ_TIMES = """
+SELECT m.id,
+    COALESCE(
+        (SELECT min(mr.read_at) FROM message_reads mr
+         WHERE mr.account_id = $1 AND mr.chat_id = $2 AND mr.message_id = m.id),
+        CASE WHEN m.date >= $5 THEN (
+            SELECT r.ts FROM read_receipts r
+            WHERE r.account_id = $1 AND r.chat_id = $2 AND r.kind = 'read'
+              AND r.message_id >= m.id AND r.ts >= m.date - $6::interval
+            ORDER BY r.ts LIMIT 1
+        ) END
+    ) AS read_at
+FROM unnest($3::bigint[], $4::timestamptz[]) AS m(id, date)
+"""
+
+_MESSAGE = """
+SELECT m.sender_id, m.date, a.tg_user_id, a.is_bot,
+    EXISTS (
+        SELECT 1 FROM media md
+        WHERE md.account_id = m.account_id AND md.chat_id = m.chat_id
+          AND md.message_id = m.message_id AND md.kind = ANY($4::text[])
+    ) AS voice
+FROM messages m
+JOIN accounts a ON a.account_id = m.account_id
+WHERE m.account_id = $1 AND m.chat_id = $2 AND m.message_id = $3
+LIMIT 1
+"""
+
+_RECEIPTS = """
+SELECT
+    (SELECT ts FROM read_receipts
+     WHERE account_id = $1 AND chat_id = $2 AND kind = 'read'
+       AND message_id >= $3 AND ts >= $4
+     ORDER BY ts LIMIT 1) AS observed_at,
+    (SELECT min(ts) FROM read_receipts
+     WHERE account_id = $1 AND chat_id = $2 AND kind = 'listened'
+       AND message_id = $3) AS listened_at,
+    (SELECT max(message_id) FROM read_receipts
+     WHERE account_id = $1 AND chat_id = $2 AND kind = 'read') AS watermark
+"""
+
+_READS = """
+SELECT reader_id, read_at, source FROM message_reads
+WHERE account_id = $1 AND chat_id = $2 AND message_id = $3
+ORDER BY read_at
+"""
+
+_FETCH = """
+SELECT outcome, fetched_at, retry_at FROM read_fetches
+WHERE account_id = $1 AND chat_id = $2 AND message_id = $3
+"""
+
+_INHERITED = """
+SELECT outcome FROM read_fetches
+WHERE account_id = $1 AND retry_at > now()
+  AND ((chat_id = $2 AND outcome = 'chat_too_big') OR outcome = 'your_privacy')
+LIMIT 1
+"""
+
+_capture_starts: dict[int, datetime] = {}
+
+
+async def capture_start(pool: asyncpg.Pool, account_id: int) -> datetime | None:
+    cached = _capture_starts.get(account_id)
+    if cached is not None:
+        return cached
+    start = await pool.fetchval(
+        "SELECT min(ts) FROM read_receipts WHERE account_id = $1", account_id
+    )
+    if start is not None:
+        _capture_starts[account_id] = start
+    return start
+
 
 async def read_up_to(pool: asyncpg.Pool, account_id: int, chat_id: int) -> int | None:
     return await pool.fetchval(
@@ -8,3 +92,97 @@ async def read_up_to(pool: asyncpg.Pool, account_id: int, chat_id: int) -> int |
         account_id,
         chat_id,
     )
+
+
+async def read_times(
+    pool: asyncpg.Pool,
+    account_id: int,
+    chat_id: int,
+    messages: list[tuple[int, datetime]],
+) -> dict[int, datetime]:
+    if not messages:
+        return {}
+    start = await capture_start(pool, account_id) or datetime.max.replace(tzinfo=UTC)
+    rows = await pool.fetch(
+        _READ_TIMES,
+        account_id,
+        chat_id,
+        [message_id for message_id, _ in messages],
+        [date for _, date in messages],
+        start,
+        CLOCK_SKEW,
+    )
+    return {row["id"]: row["read_at"] for row in rows if row["read_at"] is not None}
+
+
+async def read_info(
+    pool: asyncpg.Pool, account_id: int, chat_id: int, message_id: int
+) -> ReadInfoView | None:
+    message = await pool.fetchrow(
+        _MESSAGE, account_id, chat_id, message_id, VOICE_KINDS
+    )
+    if message is None:
+        return None
+    kind = await chat_kind(pool, account_id, chat_id)
+    date: datetime = message["date"]
+    receipts = await pool.fetchrow(
+        _RECEIPTS, account_id, chat_id, message_id, date - CLOCK_SKEW
+    )
+    reads = await pool.fetch(_READS, account_id, chat_id, message_id)
+    fetch = await pool.fetchrow(_FETCH, account_id, chat_id, message_id)
+    inherited = await pool.fetchval(_INHERITED, account_id, chat_id)
+    start = await capture_start(pool, account_id)
+
+    observed_at = receipts["observed_at"] if start and date >= start else None
+    exact_at = next(
+        (row["read_at"] for row in reads if row["source"] == "outbox_date"), None
+    )
+    reader_rows = [row for row in reads if row["source"] == "participants"]
+    known = {
+        peer.peer_id: peer
+        for peer in await get_peers(
+            pool, account_id, [row["reader_id"] for row in reader_rows]
+        )
+    }
+    readers = [
+        ReaderView(
+            peer_id=row["reader_id"],
+            read_at=row["read_at"],
+            peer=known.get(row["reader_id"]),
+        )
+        for row in reader_rows
+    ]
+    outcome = fetch["outcome"] if fetch else inherited
+
+    now = datetime.now(UTC)
+    own = (
+        message["sender_id"] is not None
+        and message["sender_id"] == (message["tg_user_id"])
+    )
+    applicable = own and kind != ChatKind.CHANNEL and chat_id != message["tg_user_id"]
+    watermark = receipts["watermark"]
+    covered = watermark is not None and message_id <= watermark
+    seen = covered or any((exact_at, readers, receipts["listened_at"]))
+    evidence = watermark is not None or outcome in {"ok", "not_read_yet"}
+    due = fetch is None or fetch["retry_at"] <= now
+    return ReadInfoView(
+        chat_id=chat_id,
+        message_id=message_id,
+        chat_kind=kind.value,
+        own=own,
+        voice=message["voice"],
+        observed_at=observed_at,
+        listened_at=receipts["listened_at"],
+        exact_at=exact_at,
+        readers=readers,
+        read=applicable and seen,
+        unread=applicable and not seen and evidence,
+        outcome=outcome,
+        fetched_at=fetch["fetched_at"] if fetch else None,
+        can_fetch=applicable
+        and not message["is_bot"]
+        and now - date < FETCH_WINDOW
+        and exact_at is None
+        and inherited is None
+        and due,
+    )