"""三元组文本化 + 上下文构建的单元测试。"""

from app.module.kg_graphrag.context_builder import (
    RELATION_TEMPLATES,
    build_context,
    textualize_subgraph,
)
from app.module.kg_graphrag.models import (
    EntitySummary,
    RelationSummary,
    VectorChunk,
)


# ---------------------------------------------------------------------------
# textualize_subgraph 测试
# ---------------------------------------------------------------------------


class TestTextualizeSubgraph:
    """textualize_subgraph 函数的测试。"""

    def test_single_relation(self):
        entities = [
            EntitySummary(id="1", name="用户行为数据", type="Dataset"),
            EntitySummary(id="2", name="user_id", type="Field"),
        ]
        relations = [
            RelationSummary(
                source_name="用户行为数据",
                source_type="Dataset",
                target_name="user_id",
                target_type="Field",
                relation_type="HAS_FIELD",
            ),
        ]
        result = textualize_subgraph(entities, relations)
        assert "Dataset'用户行为数据'包含字段Field'user_id'" in result

    def test_multiple_relations(self):
        entities = [
            EntitySummary(id="1", name="用户行为数据", type="Dataset"),
            EntitySummary(id="2", name="清洗管道", type="Workflow"),
        ]
        relations = [
            RelationSummary(
                source_name="清洗管道",
                source_type="Workflow",
                target_name="用户行为数据",
                target_type="Dataset",
                relation_type="USES_DATASET",
            ),
            RelationSummary(
                source_name="用户行为数据",
                source_type="Dataset",
                target_name="外部系统",
                target_type="DataSource",
                relation_type="SOURCED_FROM",
            ),
        ]
        result = textualize_subgraph(entities, relations)
        assert "Workflow'清洗管道'使用了数据集Dataset'用户行为数据'" in result
        assert "Dataset'用户行为数据'的知识来源于DataSource'外部系统'" in result

    def test_all_relation_templates(self):
        """验证所有 10 种关系模板都能正确生成。"""
        for rel_type, template in RELATION_TEMPLATES.items():
            relations = [
                RelationSummary(
                    source_name="A",
                    source_type="TypeA",
                    target_name="B",
                    target_type="TypeB",
                    relation_type=rel_type,
                ),
            ]
            result = textualize_subgraph([], relations)
            assert "TypeA'A'" in result
            assert "TypeB'B'" in result
            assert result  # 非空

    def test_unknown_relation_type(self):
        """未知关系类型使用通用模板。"""
        relations = [
            RelationSummary(
                source_name="X",
                source_type="T1",
                target_name="Y",
                target_type="T2",
                relation_type="CUSTOM_REL",
            ),
        ]
        result = textualize_subgraph([], relations)
        assert "T1'X'与T2'Y'存在CUSTOM_REL关系" in result

    def test_orphan_entity_with_description(self):
        """无关系的独立实体（有描述）。"""
        entities = [
            EntitySummary(id="1", name="孤立实体", type="Dataset", description="这是一个测试实体"),
        ]
        result = textualize_subgraph(entities, [])
        assert "Dataset'孤立实体': 这是一个测试实体" in result

    def test_orphan_entity_without_description(self):
        """无关系的独立实体（无描述）。"""
        entities = [
            EntitySummary(id="1", name="孤立实体", type="Dataset"),
        ]
        result = textualize_subgraph(entities, [])
        assert "存在Dataset'孤立实体'" in result

    def test_empty_inputs(self):
        result = textualize_subgraph([], [])
        assert result == ""

    def test_entity_with_relation_not_orphan(self):
        """有关系的实体不应出现在独立实体部分。"""
        entities = [
            EntitySummary(id="1", name="A", type="Dataset"),
            EntitySummary(id="2", name="B", type="Field"),
            EntitySummary(id="3", name="C", type="Workflow"),
        ]
        relations = [
            RelationSummary(
                source_name="A",
                source_type="Dataset",
                target_name="B",
                target_type="Field",
                relation_type="HAS_FIELD",
            ),
        ]
        result = textualize_subgraph(entities, relations)
        # A 和 B 有关系，不应作为独立实体出现
        # C 无关系，应出现
        assert "存在Workflow'C'" in result
        lines = result.strip().split("\n")
        assert len(lines) == 2  # 一条关系 + 一个独立实体


# ---------------------------------------------------------------------------
# build_context 测试
# ---------------------------------------------------------------------------


class TestBuildContext:
    """build_context 函数的测试。"""

    def test_both_vector_and_graph(self):
        chunks = [
            VectorChunk(id="1", text="文档片段一", score=0.9),
            VectorChunk(id="2", text="文档片段二", score=0.8),
        ]
        graph_text = "Dataset'用户数据'包含字段Field'user_id'"
        result = build_context(chunks, graph_text)
        assert "## 相关文档" in result
        assert "[1] 文档片段一" in result
        assert "[2] 文档片段二" in result
        assert "## 知识图谱上下文" in result
        assert graph_text in result

    def test_vector_only(self):
        chunks = [VectorChunk(id="1", text="文档片段", score=0.9)]
        result = build_context(chunks, "")
        assert "## 相关文档" in result
        assert "## 知识图谱上下文" not in result

    def test_graph_only(self):
        result = build_context([], "图谱内容")
        assert "## 知识图谱上下文" in result
        assert "## 相关文档" not in result

    def test_empty_both(self):
        result = build_context([], "")
        assert "未检索到相关上下文信息" in result

    def test_context_section_order(self):
        """验证文档在图谱之前。"""
        chunks = [VectorChunk(id="1", text="doc", score=0.9)]
        result = build_context(chunks, "graph")
        doc_pos = result.index("## 相关文档")
        graph_pos = result.index("## 知识图谱上下文")
        assert doc_pos < graph_pos