From afdf9d43340ec56ab3ba49ee5fe44103d661d54b Mon Sep 17 00:00:00 2001
From: Dima Gerasimov <karlicoss@gmail.com>
Date: Mon, 7 Feb 2022 09:58:50 +0000
Subject: [PATCH] twitter: initial talon module, processing data from Talon
 android app

---
 my/config.py          |   5 ++
 my/twitter/archive.py |   0
 my/twitter/talon.py   | 113 ++++++++++++++++++++++++++++++++++++++++++
 my/twitter/twint.py   |   3 ++
 4 files changed, 121 insertions(+)
 mode change 100755 => 100644 my/twitter/archive.py
 create mode 100644 my/twitter/talon.py

diff --git a/my/config.py b/my/config.py
index aef90bc..7201a84 100644
--- a/my/config.py
+++ b/my/config.py
@@ -124,3 +124,8 @@ class fbmessenger:
         export_db: PathIsh
     class android:
         export_path: Paths
+
+
+class twitter:
+    class talon:
+        export_path: Paths
diff --git a/my/twitter/archive.py b/my/twitter/archive.py
old mode 100755
new mode 100644
diff --git a/my/twitter/talon.py b/my/twitter/talon.py
new file mode 100644
index 0000000..a369f69
--- /dev/null
+++ b/my/twitter/talon.py
@@ -0,0 +1,113 @@
+"""
+Twitter data from Talon app database (in =/data/data/com.klinker.android.twitter_l/databases/=)
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime
+from typing import Iterator, Sequence, Optional, Dict
+
+
+from my.config import twitter as user_config
+
+
+from ..core import Paths
+@dataclass
+class config(user_config.talon):
+    # paths[s]/glob to the exported sqlite databases
+    export_path: Paths
+
+
+from ..core import get_files
+from pathlib import Path
+def inputs() -> Sequence[Path]:
+    return get_files(config.export_path)
+
+
+
+@dataclass(unsafe_hash=True)
+class Tweet:
+    id_str: str
+    # TODO figure out if utc
+    created_at: datetime
+    screen_name: str
+    text: str
+    urls: Sequence[str]
+
+
+# meh... just wrappers to tell apart tweets from favorites...
+@dataclass(unsafe_hash=True)
+class _IsTweet:
+    tweet: Tweet
+@dataclass(unsafe_hash=True)
+class _IsFavorire:
+    tweet: Tweet
+
+
+from typing import Union
+from ..core.error import Res
+from ..core.dataset import connect_readonly
+Entity = Union[_IsTweet, _IsFavorire]
+def _entities() -> Iterator[Res[Entity]]:
+    for f in inputs():
+        yield from _process_one(f)
+
+
+def _process_one(f: Path) -> Iterator[Res[Entity]]:
+    handlers = {
+        'user_tweets.db'    : _process_user_tweets,
+        'favorite_tweets.db': _process_favorite_tweets,
+    }
+    fname = f.name
+    handler = handlers.get(fname)
+    if handler is None:
+        yield RuntimeError(f"Coulnd't find handler for {fname}")
+        return
+    with connect_readonly(f) as db:
+        yield from handler(db)
+
+
+def _process_user_tweets(db) -> Iterator[Res[Entity]]:
+    # dunno why it's called 'lists'
+    for r in db['lists'].all(order_by='time'):
+        try:
+            yield _IsTweet(_parse_tweet(r))
+        except Exception as e:
+            yield e
+
+
+def _process_favorite_tweets(db) -> Iterator[Res[Entity]]:
+    for r in db['favorite_tweets'].all(order_by='time'):
+        try:
+            yield _IsFavorire(_parse_tweet(r))
+        except Exception as e:
+            yield e
+
+def _parse_tweet(row) -> Tweet:
+    # TODO row['retweeter] if not empty, would be user's name and means retweet?
+    # screen name would be the actual tweet's author
+    return Tweet(
+        id_str=str(row['tweet_id']),
+        created_at=datetime.fromtimestamp(row['time'] / 1000),
+        screen_name=row['screen_name'],
+        text=row['text'],
+        # todo hmm text sometimes is trimmed with ellipsis? at least urls
+        urls=tuple(u for u in row['other_url'].split(' ') if len(u.strip()) > 0),
+    )
+
+
+from more_itertools import unique_everseen
+def tweets() -> Iterator[Res[Tweet]]:
+    for x in unique_everseen(_entities()):
+        if isinstance(x, Exception):
+            yield x
+        elif isinstance(x, _IsTweet):
+            yield x.tweet
+
+def likes() -> Iterator[Res[Tweet]]:
+    for x in unique_everseen(_entities()):
+        if isinstance(x, Exception):
+            yield x
+        elif isinstance(x, _IsFavorire):
+            yield x.tweet
+
diff --git a/my/twitter/twint.py b/my/twitter/twint.py
index 6f70ce3..c8d426e 100644
--- a/my/twitter/twint.py
+++ b/my/twitter/twint.py
@@ -8,6 +8,8 @@ from ..core.common import Paths
 from dataclasses import dataclass
 from my.config import twint as user_config
 
+# TODO move to twitter.twint config structure
+
 @dataclass
 class twint(user_config):
     export_path: Paths # path[s]/glob to the twint Sqlite database
@@ -63,6 +65,7 @@ class Tweet(NamedTuple):
             return []
         return ustr.split(',')
 
+    # TODO move to common
     @property
     def permalink(self) -> str:
         return f'https://twitter.com/{self.screen_name}/status/{self.id_str}'