content-find-agent-new/tests/test_query_prompt_config.py

from pathlib import Path

from content_agent.integrations.query_prompt_config import (
    DEFAULT_PROFILE,
    QUERY_PROMPTS_PATH,
    load_profile,
)

ROOT = Path(__file__).resolve().parents[1]


def test_load_profile_reads_douyin_v1_from_json():
    profile = load_profile("douyin", "V1", root_dir=ROOT)

    assert profile["prompt_version"] == "query_variant.v1"
    assert profile["temperature"] == 0.4
    assert profile["max_tokens"] == 64
    assert profile["variants_per_seed"] == 1
    assert "generic_filter" in profile


def test_load_profile_missing_file_returns_default(tmp_path):
    profile = load_profile("douyin", "V1", root_dir=tmp_path)

    assert profile == DEFAULT_PROFILE
    profile["temperature"] = 1.0
    assert DEFAULT_PROFILE["temperature"] == 0.4


def test_load_profile_unknown_key_returns_default():
    assert load_profile("xhs", "V9", root_dir=ROOT) == DEFAULT_PROFILE


def test_default_profile_matches_json_profile():
    assert load_profile("douyin", "V1", root_dir=ROOT) == DEFAULT_PROFILE


def test_default_profile_preserves_prompt_sampling_evidence_and_generic_filter():
    profile = load_profile("douyin", "V1", root_dir=ROOT)

    assert QUERY_PROMPTS_PATH.as_posix() == "product_documents/配置/query_prompts.v1.json"
    assert profile["system"].startswith("You generate one concise Chinese short-video search query.")
    assert "{seed_term}" in profile["user"]
    assert "{evidence_context}" in profile["user"]
    assert profile["evidence_fields"] == [
        "seed_term",
        "seed_terms",
        "existing_search_queries",
        "source_field",
        "source_index",
        "itemset_items",
        "category_bindings",
        "element_bindings",
        "pattern_source_system",
        "pattern_execution_id",
        "mining_config_id",
        "source_post_id",
        "matched_post_ids",
        "itemset_ids",
        "support",
        "absolute_support",
        "confidence",
    ]
    assert profile["generic_filter"]["queries"] == [
        "内容",
        "视频",
        "热门",
        "推荐",
        "短视频",
        "热门视频",
        "推荐视频",
        "热门内容",
        "推荐内容",
        "相关视频",
        "相关内容",
        "热点视频",
        "热点内容",
    ]
    assert profile["generic_filter"]["tokens"] == [
        "短视频",
        "热门",
        "推荐",
        "相关",
        "热点",
        "内容",
        "视频",
        "素材",
        "资料",
        "信息",
        "话题",
    ]