| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142 |
- from __future__ import annotations
- from uuid import uuid4
- import pytest
- from acquisition.domain import Query, QueryBatch
- from acquisition.queries.builder import (
- QueryBuildOptions,
- build_creation_query_batch,
- persist_query_batch,
- )
- from core.config import PgConfig, Settings
- def _settings() -> Settings:
- return Settings(
- pg=PgConfig(host="h", port=5432, user="u", password="p", database="d"),
- aiddit_crawler_base_url="http://crawler.test",
- crawler_timeout=30,
- openrouter_timeout_seconds=90,
- openrouter_model="m",
- openrouter_base_url="http://openrouter.test",
- openrouter_api_key="k",
- llm_model="m",
- max_cards=12,
- frames_dir="f",
- douyin_ratio="540p",
- data_dir="",
- )
- class FakeRepo:
- def __init__(self) -> None:
- self.batch_kwargs = None
- self.queries = []
- def create_query_batch(self, **kwargs):
- self.batch_kwargs = kwargs
- return QueryBatch(id=uuid4(), **kwargs)
- def add_query(self, **kwargs):
- self.queries.append(kwargs)
- return Query(id=uuid4(), **kwargs)
- def test_build_creation_query_batch_defaults_to_first_two_families():
- generated = build_creation_query_batch(
- _settings(),
- options=QueryBuildOptions(per=2, batch_n=4, dry=True),
- )
- assert [family["key"] for family in generated["families"]] == ["f1", "f2"]
- assert generated["metadata"]["active_family_keys"] == ["f1", "f2"]
- assert sum(len(family["items"]) for family in generated["families"]) == 4
- def test_build_creation_query_batch_can_explicitly_enable_reserved_families():
- all_keys = (
- "f1",
- "f2",
- "f4",
- "f3",
- "f5",
- "a_shi",
- "a_xing",
- "a_both",
- "a_purpose",
- "a_tail",
- "b_shi",
- "b_xing",
- "b_both",
- "b_purpose",
- "b_tail",
- )
- generated = build_creation_query_batch(
- _settings(),
- options=QueryBuildOptions(
- per=1,
- batch_n=4,
- dry=True,
- active_family_keys=all_keys,
- ),
- )
- assert [family["key"] for family in generated["families"]] == list(all_keys)
- assert generated["metadata"]["active_family_keys"] == list(all_keys)
- def test_build_creation_query_batch_rejects_unknown_family_key():
- with pytest.raises(ValueError, match="unknown query family"):
- build_creation_query_batch(
- _settings(),
- options=QueryBuildOptions(
- dry=True,
- active_family_keys=("f1", "no_such_family"),
- ),
- )
- def test_persist_query_batch_writes_formal_batch_and_query_contract():
- repo = FakeRepo()
- generated = {
- "metadata": {
- "query_filter_prompt_version": "abc123",
- "active_family_keys": ["f1", "f2"],
- },
- "families": [
- {
- "key": "f1",
- "name": "实质 × 模态 × 业务阶段 × 知识类型",
- "axes": ["实质", "模态"],
- "items": [
- {
- "query": "反转 视频 脚本 怎么做",
- "parts": {"实质": "反转", "模态": "视频"},
- "keep": True,
- "valid": 8,
- "relevant": True,
- "reason": "可搜创作方法",
- }
- ],
- }
- ],
- }
- batch, count = persist_query_batch(repo, generated, name="formal-demo")
- assert batch.name == "formal-demo"
- assert count == 1
- assert repo.batch_kwargs["status"] == "ready"
- assert repo.batch_kwargs["target_platforms"] == ["xiaohongshu", "weixin", "douyin"]
- assert repo.batch_kwargs["metadata"]["active_family_keys"] == ["f1", "f2"]
- row = repo.queries[0]
- assert row["batch_id"] == batch.id
- assert row["query_text"] == "反转 视频 脚本 怎么做"
- assert row["axes"] == {"实质": "反转", "模态": "视频"}
- assert row["keep"] is True
- assert row["filter_reason"] == "可搜创作方法"
- assert row["metadata"]["family_key"] == "f1"
- assert row["metadata"]["valid"] == 8
- assert row["metadata"]["query_filter_prompt_version"] == "abc123"
|