feat: add topic tagging, semantic duplicate detection, and per-provider vision toggle

tests/test_semantic_dedup_and_vision.py

@@ -0,0 +1,120 @@
+import os
+import asyncio
+from db.database import init_db, close_db_pool
+from db.repository import Repository
+from db.models import Post, TargetChannel, SourceChannel
+from core.llm import LLMClient
+from services.ai_processor import AIProcessor
+from services.collector import CollectorService
+
+
+async def test_provider_vision_toggle():
+    await init_db()
+    repo = Repository()
+
+    prov_id = await repo.add_provider_profile(
+        name="Test Vision Model",
+        provider_type="openai",
+        model="gpt-4o",
+        supports_vision=False
+    )
+    assert prov_id is not None
+
+    p = await repo.get_provider_profile_by_id(prov_id)
+    assert p.supports_vision is False
+
+    await repo.update_provider_vision(prov_id, True)
+    p_updated = await repo.get_provider_profile_by_id(prov_id)
+    assert p_updated.supports_vision is True
+
+    await repo.update_provider_vision(prov_id, False)
+    p_reverted = await repo.get_provider_profile_by_id(prov_id)
+    assert p_reverted.supports_vision is False
+
+    await repo.delete_provider_profile(prov_id)
+
+
+async def test_find_candidate_posts_by_tags():
+    await init_db()
+    repo = Repository()
+
+    import time
+    msg_id = int(time.time() * 1000) % 100000000
+    post1_id = await repo.create_raw_post(
+        source_channel_id=-100111222,
+        source_message_id=msg_id,
+        raw_text="قیمت بیت‌کوین به ۷۰ هزار دلار رسید و رکورد جدیدی ثبت کرد.",
+        tags=["بیت_کوین", "ارز_دیجیتال", "اقتصاد"],
+        subject="رکوردشکنی بیت‌کوین",
+    )
+    assert post1_id is not None
+
+    candidates = await repo.find_candidate_posts_by_tags(["بیت_کوین", "رمزارز"])
+    assert len(candidates) >= 1
+    found_ids = [c.id for c in candidates]
+    assert post1_id in found_ids
+
+    no_match = await repo.find_candidate_posts_by_tags(["ورزش", "فوتبال"])
+    assert post1_id not in [c.id for c in no_match]
+
+
+async def test_semantic_duplicate_detection():
+    await init_db()
+    repo = Repository()
+
+    class FakeDuplicateLLM:
+        def __init__(self):
+            self.supports_vision = False
+            self.last_used_model = "fake"
+            self.last_used_provider = "fake"
+
+        async def generate_json(self, prompt, system_prompt=None, action_name=None, image_path=None):
+            if action_name == "extract_tags_and_subject":
+                return {
+                    "subject": "رشد بیت‌کوین",
+                    "tags": ["بیت_کوین", "کریپتو", "قیمت"]
+                }
+            elif action_name == "check_semantic_duplicate":
+                return {
+                    "is_duplicate": True,
+                    "duplicate_of_id": 55,
+                    "reason": "پوشش یکسان خبر افزایش نرخ بیت‌کوین"
+                }
+            return {}
+
+    fake_llm = FakeDuplicateLLM()
+    processor = AIProcessor(repo=repo, llm=fake_llm)
+
+    tags, subject = await processor.extract_tags_and_subject("بیت‌کوین ۷۰ هزار دلار شد")
+    assert tags == ["بیت_کوین", "کریپتو", "قیمت"]
+    assert subject == "رشد بیت‌کوین"
+
+    candidates = [
+        Post(
+            id=55,
+            source_channel_id=-100123,
+            source_message_id=10,
+            raw_text="بیت کوین به ۷۰۰۰۰ دلار رسید",
+            tags=["بیت_کوین"],
+            subject="افزایش قیمت بیت‌کوین"
+        )
+    ]
+
+    is_dup, dup_id, reason = await processor.check_semantic_duplicate(
+        "بیت‌کوین به مرز هفتاد هزار دلار دست یافت",
+        candidates
+    )
+    assert is_dup is True
+    assert dup_id == 55
+    assert "پوشش یکسان" in reason
+
+
+async def main():
+    await test_provider_vision_toggle()
+    await test_find_candidate_posts_by_tags()
+    await test_semantic_duplicate_detection()
+    print("All semantic deduplication and vision configuration tests passed successfully!")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())