| """DocumentProfiler 测试（intent-understanding-layer 任务 4）。 |
| |
| 覆盖需求 2.1–2.4 / 2.6：表类型识别、稳定性时序标注、多文档聚合、保留时间防护。 |
| """ |
|
|
| from __future__ import annotations |
|
|
| from kernel.task_sheet import TableType |
| from kernel.understanding import DocumentProfiler |
|
|
|
|
| class _MockLLM: |
| def __init__(self, out): |
| self._out = out |
|
|
| def profile_and_extract(self, text, goal): |
| return self._out |
|
|
|
|
| SL0010_TEXT = """=== File: SL-0010-3.xlsx === |
| 有关物质检测结果 |
| 样品名称 名称 保留时间min RRT 总杂% |
| SL-0010-20μg 右美托咪定 25.440 1.00 0.00 |
| 含量均匀度 |
| 含量% |
| 99.19 |
| """ |
|
|
|
|
| def test_heuristic_profile_no_llm_sl0010(): |
| profiler = DocumentProfiler(llm=None) |
| profile = profiler.profile(SL0010_TEXT, ["SL-0010-3.xlsx"]) |
| types = {t.table_type for t in profile.tables} |
| assert TableType.RELATED_SUBSTANCES in types |
| assert TableType.CONTENT_UNIFORMITY in types |
| assert profile.has_stability_time_series is False |
| assert any("未检测到稳定性" in n for n in profile.notes) |
|
|
|
|
| def test_llm_profile_respects_retention_guard(): |
| |
| llm = _MockLLM({ |
| "tables": [{ |
| "table_type": "related_substances", |
| "title": "有关物质", |
| "source_document": "SL-0010-3.xlsx", |
| "is_stability_time_series": True, |
| "columns": ["保留时间min", "总杂%"], |
| }], |
| "intent": "descriptive_summary", |
| "items": [], |
| }) |
| profile = DocumentProfiler(llm).profile(SL0010_TEXT, ["SL-0010-3.xlsx"]) |
| assert profile.tables[0].is_stability_time_series is False |
| assert profile.has_stability_time_series is False |
|
|
|
|
| def test_multi_document_aggregation(): |
| text = ( |
| "=== File: a.xlsx ===\n有关物质检测结果\n总杂%\n0.1\n" |
| "=== File: b.xlsx ===\n溶出曲线\n累积溶出度%\n95\n" |
| ) |
| profile = DocumentProfiler(llm=None).profile(text, ["a.xlsx", "b.xlsx"]) |
| srcs = {t.source_document for t in profile.tables} |
| assert "a.xlsx" in srcs |
| assert "b.xlsx" in srcs |
|
|
|
|
| def test_stability_time_series_detected_from_llm(): |
| llm = _MockLLM({ |
| "tables": [{ |
| "table_type": "stability_time_series", |
| "title": "长期稳定性", |
| "is_stability_time_series": True, |
| "columns": ["时间点（月）", "总杂%"], |
| }], |
| "intent": "shelf_life_extrapolation", |
| "items": [], |
| }) |
| profile = DocumentProfiler(llm).profile("=== File: s.xlsx ===\n长期稳定性\n", ["s.xlsx"]) |
| assert profile.has_stability_time_series is True |
|
|
