feat: 1-to-1 message render + web data-lake backend

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

@@ -23,28 +23,71 @@ INSERT INTO messages
      has_media, is_self_destruct, edited_at)
 VALUES ($1, $2, $3, $4, $5, $6, $7::jsonb, $8, $9, now())
 ON CONFLICT (account_id, chat_id, message_id, date) DO UPDATE SET
+    text = EXCLUDED.text,
+    raw = EXCLUDED.raw,
+    has_media = EXCLUDED.has_media,
+    is_self_destruct = EXCLUDED.is_self_destruct,
     edited_at = now()
 """
 
 _INSERT_VERSION = """
 INSERT INTO message_versions
     (account_id, chat_id, message_id, observed_at, edit_date, text, raw)
-VALUES ($1, $2, $3, now(), $4, $5, $6::jsonb)
+VALUES ($1, $2, $3, clock_timestamp(), $4, $5, $6::jsonb)
 ON CONFLICT DO NOTHING
 """
 
+_SNAPSHOT_ORIGINAL = """
+INSERT INTO message_versions
+    (account_id, chat_id, message_id, observed_at, edit_date, text, raw)
+SELECT account_id, chat_id, message_id, clock_timestamp(), NULL, text, raw
+FROM messages m
+WHERE m.account_id = $1 AND m.chat_id = $2 AND m.message_id = $3
+  AND NOT EXISTS (
+    SELECT 1 FROM message_versions v
+    WHERE v.account_id = m.account_id AND v.chat_id = m.chat_id
+      AND v.message_id = m.message_id
+  )
+ON CONFLICT DO NOTHING
+"""
+
+_CURRENT_CONTENT = """
+SELECT
+    m.text AS text,
+    (SELECT d.unique_id FROM media d
+     WHERE d.account_id = m.account_id AND d.chat_id = m.chat_id
+       AND d.message_id = m.message_id) AS media_unique_id
+FROM messages m
+WHERE m.account_id = $1 AND m.chat_id = $2 AND m.message_id = $3
+ORDER BY m.date DESC LIMIT 1
+"""
+
+_CURRENT_MEDIA = """
+SELECT unique_id, storage_key, file_size, downloaded FROM media
+WHERE account_id = $1 AND chat_id = $2 AND message_id = $3
+"""
+
+_INSERT_MEDIA_VERSION = """
+INSERT INTO media_versions
+    (account_id, chat_id, message_id, observed_at, kind, storage_key,
+     file_size, mime, ttl_seconds, unique_id)
+VALUES ($1, $2, $3, clock_timestamp(), $4, $5, $6, $7, $8, $9)
+ON CONFLICT (account_id, chat_id, message_id, storage_key) DO NOTHING
+"""
+
 _INSERT_MEDIA = """
 INSERT INTO media
     (account_id, chat_id, message_id, kind, storage_key, file_size, mime,
-     ttl_seconds, downloaded)
-VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9)
+     ttl_seconds, downloaded, unique_id)
+VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10)
 ON CONFLICT (account_id, chat_id, message_id) DO UPDATE SET
     kind = EXCLUDED.kind,
     storage_key = EXCLUDED.storage_key,
     file_size = EXCLUDED.file_size,
     mime = EXCLUDED.mime,
     ttl_seconds = EXCLUDED.ttl_seconds,
-    downloaded = EXCLUDED.downloaded
+    downloaded = EXCLUDED.downloaded,
+    unique_id = EXCLUDED.unique_id
 """
 
 
@@ -111,11 +154,20 @@ async def add_version(  # noqa: PLR0913
     text: str | None,
     raw: str,
     edit_date: datetime | None,
+    media_unique_id: str | None,
     *,
     has_media: bool,
     is_self_destruct: bool,
-) -> None:
+) -> bool:
     async with pool.acquire() as conn, conn.transaction():
+        current = await conn.fetchrow(_CURRENT_CONTENT, account_id, chat_id, message_id)
+        text_changed = current is None or current["text"] != text
+        media_changed = (
+            current is not None and current["media_unique_id"] != media_unique_id
+        )
+        if not (text_changed or media_changed):
+            return False
+        await conn.execute(_SNAPSHOT_ORIGINAL, account_id, chat_id, message_id)
         await conn.execute(
             _TOUCH_EDITED,
             account_id,
@@ -131,6 +183,13 @@ async def add_version(  # noqa: PLR0913
         await conn.execute(
             _INSERT_VERSION, account_id, chat_id, message_id, edit_date, text, raw
         )
+    return True
+
+
+async def current_media(
+    pool: asyncpg.Pool, account_id: int, chat_id: int, message_id: int
+) -> asyncpg.Record | None:
+    return await pool.fetchrow(_CURRENT_MEDIA, account_id, chat_id, message_id)
 
 
 async def insert_media(  # noqa: PLR0913
@@ -143,21 +202,37 @@ async def insert_media(  # noqa: PLR0913
     file_size: int | None,
     mime: str | None,
     ttl_seconds: int | None,
+    unique_id: str | None,
     *,
     downloaded: bool,
 ) -> None:
-    await pool.execute(
-        _INSERT_MEDIA,
-        account_id,
-        chat_id,
-        message_id,
-        kind,
-        storage_key,
-        file_size,
-        mime,
-        ttl_seconds,
-        downloaded,
-    )
+    async with pool.acquire() as conn, conn.transaction():
+        await conn.execute(
+            _INSERT_MEDIA,
+            account_id,
+            chat_id,
+            message_id,
+            kind,
+            storage_key,
+            file_size,
+            mime,
+            ttl_seconds,
+            downloaded,
+            unique_id,
+        )
+        if storage_key is not None:
+            await conn.execute(
+                _INSERT_MEDIA_VERSION,
+                account_id,
+                chat_id,
+                message_id,
+                kind,
+                storage_key,
+                file_size,
+                mime,
+                ttl_seconds,
+                unique_id,
+            )
 
 
 async def insert_callbacks(