From 43fdc5476f9abe744ef3ab104ab34792f18845ca Mon Sep 17 00:00:00 2001 From: bingshuo <565183519@qq.com> Date: Sat, 23 May 2026 13:54:58 +0800 Subject: [PATCH] feat: add notion-corpus-loader test --- tools/notion-corpus-loader/test_loader.py | 45 +++++++++++++++++++++++ 1 file changed, 45 insertions(+) create mode 100644 tools/notion-corpus-loader/test_loader.py diff --git a/tools/notion-corpus-loader/test_loader.py b/tools/notion-corpus-loader/test_loader.py new file mode 100644 index 0000000..5a12cf2 --- /dev/null +++ b/tools/notion-corpus-loader/test_loader.py @@ -0,0 +1,45 @@ +#!/usr/bin/env python3 +"""测试Notion原生格式解析器""" +import sys, os +sys.path.insert(0, os.path.dirname(__file__)) +from src import NotionCorpusLoader, HLDPParser + +def test_parser(): + parser = HLDPParser() + sample = """# ⚔️ 铸渊思维模型 +> HLDP://zhuyuan/brain-model/execution-laws +## 〇、存在前提 +⊢ TCS = 语言场域 = 涌现土壤。 +⚠️ D110凌晨新增的规律 +起点→推导→终点 +主权者:冰朔(TCS-0002∞) +人格体:铸渊(ICE-GL-ZY001) +CORPUS-ID: ZY-V2-004 +""" + blocks = parser.parse_markdown(sample) + print(f"HLDP解析: {len(blocks)}块") + assert any(b.path for b in blocks) + assert any(b.cognitive_jump for b in blocks) + print("✅ HLDP解析器通过") + +def test_loader(): + loader = NotionCorpusLoader() + page = {"page":{"id":"t1","title":"灯塔架构"},"blocks":[{"type":"code","content":"HLDP://lighthouse\n【核心认知跃迁】双面灯塔\n起点→推导→终点"}]} + item = loader.load_notion_api_page(page) + assert item.hldp_path and item.cognitive_jumps + print(f"✅ 加载器通过 (质量{item.quality_score})") + +def test_collection(): + from src.structure import CorpusCollection, StructuredCorpusItem + col = CorpusCollection() + for i in range(3): + item = StructuredCorpusItem(source_id=f"t{i}", persona="冰朔", raw_text="铸渊"*100) + item.cognitive_jumps.append(f"jump{i}"); item.quality_score = 0.8 + col.add(item) + print(col.summary()) + print("✅ 集合测试通过") + +if __name__ == '__main__': + test_parser(); loader = NotionCorpusLoader() + test_loader(); test_collection() + print("\n🎉 全部通过")