| 1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253 |
- """query demo 离线测：①②纯逻辑 + 树采样 + parse_suggest（不碰网络/LLM）。"""
- from __future__ import annotations
- import pytest
- from acquisition.query import (
- CARRIER_POS, CARRIER_POS_GROUPED, KTYPE_SUFFIX, _nonleaf_d4, sample_nodes,
- tactic_multiaxis,
- )
- from acquisition.suggest import SuggestError, parse_suggest
- def test_sample_nodes():
- topics = sample_nodes("实质", limit=8)
- forms = sample_nodes("形式", limit=8)
- assert 0 < len(topics) <= 8 and 0 < len(forms) <= 8
- assert all(isinstance(x, str) and x for x in topics + forms)
- def test_carrier_pos_and_form_sampling():
- # ② 的载体位置 = 载体×位置交叉；文章无封面；形式树（架构分支）可采样
- assert "短视频开头" in CARRIER_POS and "图片封面" in CARRIER_POS
- assert "文章封面" not in CARRIER_POS # 文章无封面
- assert CARRIER_POS_GROUPED["文章"] == ["开头", "中间", "收尾"]
- forms = sample_nodes("形式", under="架构", depths=(4,), limit=8)
- assert 0 < len(forms) <= 8 and all(isinstance(f, str) and f for f in forms)
- def test_parse_suggest_mines_tags_and_titles():
- resp = {"code": 0, "data": {"data": [
- {"title": "健身脚本怎么写 #健身 #跟练脚本", "body_text": "教程 #减脂",
- "topic_list": [{"name": "健身博主"}]},
- ]}}
- out = parse_suggest(resp, limit=10)
- assert "健身" in out and "跟练脚本" in out and "减脂" in out and "健身博主" in out
- def test_multiaxis_assembly():
- rows = tactic_multiaxis(n=12)
- assert len(rows) == 12
- r = rows[0]
- assert {"实质", "形式", "阶段", "动作", "作用", "知识类型", "query"} <= set(r)
- # 形式限架构：不应混进 剪辑/后期 等制作节点
- forms = _nonleaf_d4("形式", 6, under="架构")
- assert "剪辑组接" not in forms and "后期处理" not in forms
- # 组合 query = 各轴机械拼接，句尾是知识类型后缀
- assert r["query"].startswith(r["实质"] + " " + r["形式"])
- assert r["query"].endswith(KTYPE_SUFFIX[r["知识类型"]])
- def test_parse_suggest_business_error():
- with pytest.raises(SuggestError):
- parse_suggest({"code": 10000, "msg": "x"})
|
