Added get_random_best code. Added the AB test

Author: ledovsky

pytest.ini

@@ -1,8 +1,11 @@
 [pytest]
 env =
-    DATABASE_URL=postgresql+asyncpg://app:app@app_db/postgres
     SITE_URL=
+    DATABASE_URL=postgresql+asyncpg://app:app@app_db:5432/app
     REDIS_URL=redis://:myStrongPassword@redis:6379
+    # no docker version, useful for debugging
+    # DATABASE_URL=postgresql+asyncpg://app:[email protected]:65432/app
+    # REDIS_URL=redis://:[email protected]:36379
 
     SITE_DOMAIN=127.0.0.1
     SECURE_COOKIES=false

src/recommendations/candidates.py

@@ -399,3 +399,49 @@ async def get_best_memes_from_each_source(
     """
     res = await fetch_all(text(query))
     return res
+
+
+async def get_random_best(
+    user_id: int,
+    limit: int = 10,
+    exclude_meme_ids: list[int] = [],
+):
+    """Selects 'limit' best memes from the top 100 memes obtained using the cleared meme statistics
+    The cleared statistics aggregates reactions from users with less than 200 previous reactions 
+    It's aim is to remove the bias from old users
+    """
+
+    query = f"""
+        SELECT
+            M.id
+            , M.type, M.telegram_file_id, M.caption, M.recommended_by
+        FROM (
+            SELECT
+                M.id
+                , M.type, M.telegram_file_id, M.caption
+                , 'random_best_ab_240422' AS recommended_by
+                , random() rand
+            FROM meme M
+            INNER JOIN meme_stats MS
+                ON MS.meme_id = M.id
+
+            INNER JOIN user_language L
+                ON L.user_id = {user_id}
+                AND L.language_code = M.language_code
+
+            LEFT JOIN user_meme_reaction R
+                    ON R.meme_id = M.id
+                    AND R.user_id = {user_id}
+
+            WHERE 1=1
+                AND M.status = 'ok'
+                AND R.meme_id IS NULL
+                -- 100 ru, 50 en, 50 all lang
+                AND M.id IN (4101086, 4442353, 3755262, 4524041, 914304, 1213657, 3477742, 3850309, 4106545, 3918656, 1976055, 3729527, 4370768, 4031941, 3902467, 3940729, 3966109, 4144377, 4131644, 4720051, 4438220, 943398, 3486879, 3958437, 3193252, 4011185, 3855063, 4261258, 4368086, 4255270, 1194244, 10222, 4818828, 3820043, 758408, 3188657, 4451345, 2050874, 4665040, 4106819, 3798967, 1825631, 3140601, 4840661, 4250457, 10202, 4363045, 3823857, 3755199, 4214428, 3604880, 3759401, 3928967, 3859587, 1240438, 4634391, 4002944, 2914449, 1955395, 1902244, 4256739, 1721327, 1285555, 1901653, 1584871, 3517077, 4493086, 4128512, 3570595, 3975285, 1484762, 1811655, 1071204, 4033401, 2294710, 4236782, 881987, 4180263, 1100991, 3867070, 1859048, 4285721, 1466518, 2262302, 4478289, 1859157, 4232654, 1202886, 978202, 2279188, 1892350, 961273, 4033397, 3513207, 3635346, 4320621, 4558947, 4252321, 1084225, 2350587, 4339982, 3724969, 3613758, 1768655, 4148626, 1285566, 2181541, 1103300, 3516406, 1197518, 4036174, 3537906, 2953444, 13636, 3724910, 3911502, 1988648, 3587199, 1398183, 4166913, 3911320, 1311422, 2153377, 3604881, 3596142, 1006843, 4473556, 4231678, 4856209, 10114, 3520485, 4232460, 1721545, 3747694, 3914292, 4119263, 4033399, 1482707, 4243473, 4336344, 1678337, 3516170, 2279191, 3724979, 3772372, 4763033, 4128276, 463991, 1006837, 1202853, 4101086, 1103300, 4119263, 4357615, 1194244, 3859587, 3630862, 4478289, 4665040, 3798967, 3940785, 10222, 4255187, 1304918, 3823857, 1398183, 16818, 881987, 2005796, 3639651, 4231648, 3902342, 4031503, 4231678, 4166913, 4720051, 3855063, 4370768, 2350587, 758408, 4818828, 4261258, 3587199, 648225, 4716664, 3918656, 4183519, 3600534, 4473556, 3772372, 4243473, 4524041, 943398, 4840661, 4250457, 1825631, 4363045, 4232460, 4148761, 3513207)
+                {exclude_meme_ids_sql_filter(exclude_meme_ids)}
+            ORDER BY rand
+            LIMIT {limit}
+        ) M
+    """
+    res = await fetch_all(text(query))
+    return res

src/recommendations/meme_queue.py

@@ -4,6 +4,7 @@
 from src.recommendations.candidates import (
     classic,
     get_best_memes_from_each_source,
+    get_random_best,
     less_seen_meme_and_source,
     like_spread_and_recent_memes,
 )
@@ -45,9 +46,19 @@ async def generate_cold_start_recommendations(user_id, limit=10):
     memes_in_queue = await redis.get_all_memes_in_queue_by_key(queue_key)
     meme_ids_in_queue = [meme["id"] for meme in memes_in_queue]
 
-    candidates = await get_best_memes_from_each_source(
-        user_id, limit=limit, exclude_meme_ids=meme_ids_in_queue
-    )
+    candidates = []
+
+    # AB test
+    if user_id % 100 < 50:
+        candidates = await get_random_best(
+            user_id, limit=limit, exclude_meme_ids=meme_ids_in_queue
+        )
+
+    if len(candidates) == 0:
+        candidates = await get_best_memes_from_each_source(
+            user_id, limit=limit, exclude_meme_ids=meme_ids_in_queue
+        )
+
     if len(candidates) == 0:
         return
 
@@ -63,10 +74,18 @@ async def generate_recommendations(user_id, limit):
 
     r = random.random()
 
+    candidates = []
+
     if user_info["nmemes_sent"] < 30:
-        candidates = await get_best_memes_from_each_source(
-            user_id, limit=limit, exclude_meme_ids=meme_ids_in_queue
-        )
+        # AB test
+        if user_id % 100 < 50:
+            candidates = await get_random_best(
+                user_id, limit=limit, exclude_meme_ids=meme_ids_in_queue
+            )
+        if len(candidates) == 0:
+            candidates = await get_best_memes_from_each_source(
+                user_id, limit=limit, exclude_meme_ids=meme_ids_in_queue
+            )
 
     elif user_info["nmemes_sent"] < 100:
         if r < 0.5:

tests/recommendations/test_random_best.py

@@ -0,0 +1,100 @@
+from datetime import datetime
+
+import pytest
+import pytest_asyncio
+from sqlalchemy import delete, insert, select
+from sqlalchemy.ext.asyncio import AsyncConnection
+
+from src import redis
+from src.database import (engine, meme, meme_source, meme_stats, user,
+                          user_language, user_meme_reaction)
+from src.recommendations.candidates import get_random_best
+from src.recommendations.meme_queue import generate_cold_start_recommendations
+
+
+@pytest_asyncio.fixture()
+async def conn():
+    async with engine.connect() as conn:
+
+        await conn.execute(
+            insert(user),
+            [{'id': 1, 'type': "user"}, {'id': 51, 'type': "user"}]
+        )
+        await conn.execute(
+            insert(meme_source),
+            {'id': 1, 'type': 'telegram', 'url': '111', 'status': 'parsing_enabled', 'created_at': datetime(2024, 1, 1)}
+        )
+        meme_basic = {
+            'raw_meme_id': 1, 'type': 'image', 'telegram_image_id': '111', 'caption': '111', 'meme_source_id': 1,
+            'published_at': datetime(2024, 1, 1), 'status': 'ok', 'language_code': 'ru',
+        }
+        good_meme_1 = 4101086
+        good_meme_2 = 4442353
+        seen_meme = 3755262
+        bad_meme = 1
+        meme_ids = [good_meme_1, good_meme_2, seen_meme, bad_meme]
+
+        await conn.execute(
+            insert(meme),
+            [meme_basic.copy() | {'id': meme_id, 'raw_meme_id': meme_id} for meme_id in meme_ids]
+        )
+        await conn.execute(
+            insert(meme_stats),
+        [{'meme_id': meme_id} for meme_id in meme_ids],
+        )
+        await conn.execute(
+            insert(user_language),
+            [
+                {'user_id': 1, 'language_code': 'ru', 'created_at': datetime(2024, 1, 1)},
+                {'user_id': 51, 'language_code': 'ru', 'created_at': datetime(2024, 1, 1)}
+            ]
+        )
+        await conn.execute(
+            insert(user_meme_reaction),
+            [
+                {'user_id': 1, 'meme_id': seen_meme, 'reaction_id': 1, 'recommended_by': '111', 'sent_at': datetime(2024, 1, 1)},
+                {'user_id': 51, 'meme_id': seen_meme, 'reaction_id': 1, 'recommended_by': '111', 'sent_at': datetime(2024, 1, 1)}
+            ]
+        )
+
+        await conn.commit()
+        yield conn
+
+        await conn.execute(delete(user_meme_reaction))
+        await conn.execute(delete(user_language))
+        await conn.execute(delete(meme_stats))
+        await conn.execute(delete(meme))
+        await conn.execute(delete(meme_source))
+        await conn.execute(delete(user))
+        await conn.commit()
+
+        queue_key = redis.get_meme_queue_key(1)
+        await redis.delete_by_key(queue_key)
+        queue_key = redis.get_meme_queue_key(51)
+        await redis.delete_by_key(queue_key)
+
+        # TODO: redis sends the runtime error after the test succeeds
+        # RuntimeError: Event loop is closed
+
+
+@pytest.mark.asyncio
+async def test_random_best(conn: AsyncConnection):
+    recs = await get_random_best(1, 10)
+    assert len(recs) == 2
+
+
+@pytest.mark.asyncio
+async def test_random_best_meme_queue(conn: AsyncConnection):
+    user_id = 1
+    await generate_cold_start_recommendations(user_id)
+    queue_key = redis.get_meme_queue_key(user_id)
+    recs = await redis.get_all_memes_in_queue_by_key(queue_key)
+    assert len(recs) == 2
+    assert recs[0]['recommended_by'] == 'random_best_ab_240422'
+
+    user_id = 51
+    await generate_cold_start_recommendations(user_id)
+    queue_key = redis.get_meme_queue_key(user_id)
+    recs = await redis.get_all_memes_in_queue_by_key(queue_key)
+    assert len(recs) == 1
+    assert recs[0]['recommended_by'] == 'best_meme_from_each_source'