# -*- coding: utf-8 -*-
"""文档抽取（documents）单元测试：分块 / 章节识别 / 引用溯源。"""
import os
import sys
import unittest

sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import _bootstrap  # noqa: F401

from rag_kb.documents import (  # noqa: E402
    KbDocument,
    build_document,
    chunk_document,
)

SOP_TEXT = """# 沸腾氯化炉异常处置SOP

## 2.1 炉温骤升处置
当炉温骤升超过阈值时，立即降低氯气流量，并开启冷却循环水。
同时通知中控室确认炉压状态。

## 2.2 炉压异常
炉压异常升高时，检查排渣系统是否堵塞，必要时切换备用炉。
"""


class ChunkDocumentTest(unittest.TestCase):
    def test_heading_and_sections(self):
        doc = KbDocument(
            doc_id="sop-001",
            title="沸腾氯化炉异常处置SOP",
            text=SOP_TEXT,
            category="sop",
        )
        chunks = chunk_document(doc)
        # 两个小节各至少一个段落
        self.assertGreaterEqual(len(chunks), 2)
        sections = {c.section for c in chunks}
        self.assertIn("2.1 炉温骤升处置", sections)
        self.assertIn("2.2 炉压异常", sections)

    def test_source_traceability(self):
        doc = KbDocument(
            doc_id="sop-001",
            title="沸腾氯化炉异常处置SOP",
            text=SOP_TEXT,
            category="sop",
        )
        chunks = chunk_document(doc)
        first = chunks[0]
        # 溯源串 = 标题 §章节 ¶序号（PRD 5.4 引用溯源）
        self.assertTrue(first.source.startswith("沸腾氯化炉异常处置SOP §2.1 炉温骤升处置 ¶1"))
        self.assertEqual(first.category, "sop")
        # 段落序号全局递增
        seqs = [c.seq for c in chunks]
        self.assertEqual(seqs, sorted(seqs))

    def test_long_paragraph_split_at_sentence_boundary(self):
        text = "第一句。第二句！第三句？第四句；第五句。" * 10  # 200 字
        doc = KbDocument(doc_id="d", title="长文档", text=text)
        chunks = chunk_document(doc, max_chars=120)
        # 每段不超过阈值（硬切兜底除外）
        for c in chunks:
            self.assertLessEqual(len(c.text), 120 + 20)  # 句边界允许略超
        # 拼接后应还原全文（顺序无损）
        joined = "".join(c.text for c in chunks)
        self.assertIn("第一句。第二句！", joined)

    def test_pure_heading_skipped(self):
        doc = KbDocument(doc_id="d", title="只有标题", text="# 第一章\n\n# 第二章\n")
        chunks = chunk_document(doc)
        self.assertEqual(chunks, [])  # 纯标题行不产生段落

    def test_empty_text_rejected(self):
        with self.assertRaises(ValueError):
            KbDocument(doc_id="d", title="空文档", text="  \n  ")


class BuildDocumentTest(unittest.TestCase):
    def test_doc_id_slug(self):
        doc = build_document("沸腾氯化工艺规范", "正文内容")
        self.assertEqual(doc.doc_id, "沸腾氯化工艺规范")
        self.assertEqual(doc.category, "process")

    def test_custom_doc_id(self):
        doc = build_document("沸腾氯化工艺规范", "正文内容", doc_id="spec-001")
        self.assertEqual(doc.doc_id, "spec-001")


if __name__ == "__main__":
    unittest.main()