From 81986b06249e5c162dee56b9adbe42cb18ef1135 Mon Sep 17 00:00:00 2001
From: Dima Gerasimov <karlicoss@gmail.com>
Date: Tue, 14 Apr 2020 23:01:44 +0100
Subject: [PATCH] support likes from twint

---
 my/twitter/all.py     |  6 ++----
 my/twitter/archive.py |  7 ++++++-
 my/twitter/twint.py   | 32 +++++++++++++++++++++++++++++---
 3 files changed, 37 insertions(+), 8 deletions(-)

diff --git a/my/twitter/all.py b/my/twitter/all.py
index f1e39a7..4a1d1be 100644
--- a/my/twitter/all.py
+++ b/my/twitter/all.py
@@ -20,10 +20,8 @@ def merge_tweets(*sources):
 def tweets():
     # NOTE order matters.. twint seems to contain better data
     # todo probably, worthy an investigation..
-    yield from merge_tweets(twint.tweets(), archive.tweets())
+    yield from merge_tweets(twint.likes(), archive.tweets())
 
 
-# TODO not sure, likes vs favoites??
 def likes():
-    yield from merge_tweets(archive.likes())
-    # yield from twint
+    yield from merge_tweets(twint.likes(), archive.likes())
diff --git a/my/twitter/archive.py b/my/twitter/archive.py
index 9f4c7ee..24f9a14 100755
--- a/my/twitter/archive.py
+++ b/my/twitter/archive.py
@@ -87,7 +87,7 @@ class Like(NamedTuple):
         return f'https://twitter.com/{self.screen_name}/status/{self.tid}'
 
     @property
-    def tid(self) -> Tid:
+    def id_str(self) -> Tid:
         return self.raw['tweetId']
 
     @property
@@ -95,6 +95,11 @@ class Like(NamedTuple):
         # ugh. I think none means that tweet was deleted?
         return self.raw.get('fullText')
 
+    # TODO deprecate?
+    @property
+    def tid(self) -> Tid:
+        return self.id_str
+
 
 class ZipExport:
     def __init__(self) -> None:
diff --git a/my/twitter/twint.py b/my/twitter/twint.py
index 2180a8a..45f58fd 100644
--- a/my/twitter/twint.py
+++ b/my/twitter/twint.py
@@ -61,11 +61,37 @@ class Tweet(NamedTuple):
     def __repr__(self):
         return f'Tweet(id_str={self.id_str}, created_at={self.created_at}, text={self.text})'
 
+# https://github.com/twintproject/twint/issues/196
+# ugh. so it dumps everything in tweet table, and there is no good way to tell between fav/original tweet.
+# it might result in some tweets missing from the timeline if you happened to like them...
+# not sure what to do with it
+# alternatively, could ask the user to run separate databases for tweets and favs?
+# TODO think about it
 
-def tweets() -> Iterable[Tweet]:
+_QUERY = '''
+SELECT T.*
+FROM      tweets    as T
+LEFT JOIN favorites as F
+ON    T.id_str = F.tweet_id
+WHERE {where}
+ORDER BY T.created_at
+'''
+
+def _get_db():
     import dataset # type: ignore
     db_path = get_db_path()
     # TODO check that exists?
     db = dataset.connect(f'sqlite:///{db_path}')
-    tdb = db.load_table('tweets')
-    yield from map(Tweet, tdb.all(order_by='created_at'))
+    return db
+
+
+def tweets() -> Iterable[Tweet]:
+    db = _get_db()
+    res = db.query(_QUERY.format(where='F.tweet_id IS NULL'))
+    yield from map(Tweet, res)
+
+
+def likes() -> Iterable[Tweet]:
+    db = _get_db()
+    res = db.query(_QUERY.format(where='F.tweet_id IS NOT NULL'))
+    yield from map(Tweet, res)