db: extend schema for categories, channel profiles, and error logs

tests/test_repository.py

@@ -0,0 +1,78 @@
+"""Integration tests for the repository layer. Requires a reachable Postgres."""
+import asyncio
+import sys
+
+sys.path.insert(0, "/app")
+
+from db.database import init_db, close_db_pool
+from db.repository import Repository
+
+TEST_SOURCE_ID = -1009999000001
+TEST_TARGET_ID = -1009999000002
+# Deliberately hostile title: quotes and a backslash must survive the JSONB round-trip.
+TEST_TARGET_TITLE = 'News "Daily" \\ Channel'
+
+
+async def _cleanup(repo: Repository):
+    pool = await repo._get_pool()
+    async with pool.acquire() as conn:
+        await conn.execute("DELETE FROM posts WHERE source_channel_id = $1;", TEST_SOURCE_ID)
+        await conn.execute("DELETE FROM sources WHERE channel_id = $1;", TEST_SOURCE_ID)
+        await conn.execute("DELETE FROM targets WHERE channel_id = $1;", TEST_TARGET_ID)
+
+
+async def run_tests():
+    await init_db()
+    repo = Repository()
+    await _cleanup(repo)
+
+    # 1. Sources are looked up by their Telegram channel_id, not the surrogate row id.
+    await repo.add_source(TEST_SOURCE_ID, "Source Tech", "source_tech")
+    source = await repo.get_source_by_channel_id(TEST_SOURCE_ID)
+    assert source is not None, "get_source_by_channel_id returned None for a registered source"
+    assert source.channel_id == TEST_SOURCE_ID
+    assert source.title == "Source Tech"
+    assert await repo.get_source_by_channel_id(-1000000000000) is None
+
+    # 2. Deduplication lookup by content hash.
+    post_id = await repo.create_raw_post(
+        source_channel_id=TEST_SOURCE_ID,
+        source_message_id=101,
+        raw_text="Breaking news: AI update released!",
+        content_hash="hash_12345",
+    )
+    assert post_id is not None
+    dup = await repo.find_duplicate_post("hash_12345")
+    assert dup is not None and dup.id == post_id, "find_duplicate_post did not match a stored hash"
+    assert await repo.find_duplicate_post("no_such_hash") is None
+
+    # Re-inserting the same source message is rejected by the unique constraint.
+    assert await repo.create_raw_post(TEST_SOURCE_ID, 101, "dupe") is None
+
+    # 3. Queueing records the target without prematurely marking the post published.
+    target_id = await repo.add_target(TEST_TARGET_ID, TEST_TARGET_TITLE, "target_chan", post_interval_min=15)
+    await repo.record_post_queued_to_target(post_id, target_id, TEST_TARGET_TITLE)
+    post = await repo.get_post_by_id(post_id)
+    assert post.status == "pending_review", f"queueing must not publish, got {post.status}"
+    assert len(post.published_to) == 1, f"expected 1 queue entry, got {post.published_to}"
+    assert post.published_to[0]["target_title"] == TEST_TARGET_TITLE, "title was mangled in JSONB"
+    assert post.published_to[0]["published_at"] is None
+
+    # 4. Publishing flips status and stamps a real timestamp on the existing entry.
+    await repo.record_post_published_to_target(post_id, target_id, TEST_TARGET_TITLE)
+    post = await repo.get_post_by_id(post_id)
+    assert post.status == "published", f"expected published, got {post.status}"
+    assert len(post.published_to) == 1, f"publishing must not duplicate the entry, got {post.published_to}"
+    stamped = post.published_to[0]["published_at"]
+    assert stamped and "class" not in str(stamped), f"published_at is not a timestamp: {stamped!r}"
+
+    # 5. Counting by status must not require loading every row.
+    assert await repo.count_posts_by_status("published") >= 1
+
+    await _cleanup(repo)
+    await close_db_pool()
+    print("All repository tests passed successfully!")
+
+
+if __name__ == "__main__":
+    asyncio.run(run_tests())