my.telegram: initial module

my/config.py

@@ -159,3 +159,10 @@ class export:
         export_path: Paths = ''
     class active_browser:
         export_path: Paths = ''
+
+
+class telegram:
+    class telegram_backup:
+        export_path: PathIsh = ''
+
+

my/core/sqlite.py

@@ -46,6 +46,8 @@ def sqlite_connection(db: PathIsh, *, immutable: bool=False, row_factory: Option
     dbp = f'file:{db}'
     # https://www.sqlite.org/draft/uri.html#uriimmutable
     if immutable:
+        # assert results in nicer error than sqlite3.OperationalError
+        assert Path(db).exists(), db
         dbp = f'{dbp}?immutable=1'
     row_factory_: Any = None
     if row_factory is not None:

my/telegram/telegram_backup.py

@@ -0,0 +1,77 @@
+"""
+Telegram data via [fabianonline/telegram_backup](https://github.com/fabianonline/telegram_backup) tool
+"""
+
+from dataclasses import dataclass
+from datetime import datetime, timezone
+import sqlite3
+from typing import Dict, Iterator
+
+from my.core import datetime_aware, PathIsh
+from my.core.sqlite import sqlite_connection
+
+from my.config import telegram as user_config
+
+
+@dataclass
+class config(user_config.telegram_backup):
+    # path to the export database.sqlite
+    export_path: PathIsh
+
+
+@dataclass
+class Chat:
+    id: str
+    name: str
+    # not sure if need type?
+
+
+@dataclass
+class User:
+    id: str
+    name: str
+
+
+@dataclass
+class Message:
+    id: int
+    time: datetime_aware
+    chat: Chat
+    sender: User
+    text: str
+
+
+Chats = Dict[str, Chat]
+def _message_from_row(r: sqlite3.Row, *, chats: Chats) -> Message:
+    ts = r['time']
+    time = datetime.fromtimestamp(ts, tz=timezone.utc)
+    chat = chats[r['source_id']]
+    sender = chats[r['sender_id']]
+    return Message(
+        id=r['message_id'],
+        time=time,
+        chat=chat,
+        sender=sender,
+        text=r['text'],
+    )
+
+
+def messages() -> Iterator[Message]:
+    with sqlite_connection(config.export_path, immutable=True, row_factory='row') as db:
+
+        chats: Chats = {}
+        for r in db.execute('SELECT * FROM chats'):
+            chat = Chat(id=r['id'], name=r['name'])
+            assert chat.id not in chats
+            chats[chat.id] = chat
+
+        for r in db.execute('SELECT * FROM users'):
+            chat = Chat(id=r['id'], name=f'{r["first_name"]} {r["last_name"]}')
+            assert chat.id not in chats
+            chats[chat.id] = chat
+
+        # TODO order by? not sure
+        for r in db.execute('SELECT * FROM messages WHERE message_type NOT IN ("service_message", "empty_message")'):
+            # seems like the only remaining have message_type = 'message'
+            yield _message_from_row(r, chats=chats)
+

my/tinder/android.py

@@ -13,10 +13,13 @@
 
 from more_itertools import unique_everseen
 
-from my.core import Paths, get_files, Res, assert_never, stat, Stats, datetime_aware
+from my.core import Paths, get_files, Res, assert_never, stat, Stats, datetime_aware, LazyLogger
 from my.core.sqlite import sqlite_connection
 
 
+logger = LazyLogger(__name__)
+
+
 from my.config import tinder as user_config
 @dataclass
 class config(user_config.android):
@@ -38,7 +41,7 @@ class _BaseMatch:
     id: str
 
 
-@dataclass
+@dataclass(unsafe_hash=True)
 class _Match(_BaseMatch):
     person_id: str
 
@@ -58,7 +61,7 @@ class _BaseMessage:
     text: str
 
 
-@dataclass
+@dataclass(unsafe_hash=True)
 class _Message(_BaseMessage):
     match_id: str
     from_id: str
@@ -83,7 +86,9 @@ def inputs() -> Sequence[Path]:
 
 
 def _entities() -> Iterator[Res[_Entity]]:
-    for db_file in inputs():
+    dbs = inputs()
+    for i, db_file in enumerate(dbs):
+        logger.debug(f'processing {db_file} {i}/{len(dbs)}')
         with sqlite_connection(db_file, immutable=True, row_factory='row') as db:
             yield from _handle_db(db)