fix(application-form-fill): 抽取说明书章节和表格字段

tests/test_application_form_fill_field_extract.py

@@ -48,6 +48,42 @@ def test_rule_extracts_registration_certificate_fields():
     assert values["package_specification"]["extractor"] == "rule"
 
 
+def test_rule_extracts_bracket_sections_from_instructions():
+    texts = {
+        "目标产品说明书.docx": "\n".join(
+            [
+                "【产品名称】",
+                "新型冠状病毒2019-nCoV核酸检测试剂盒（荧光PCR法）",
+                "【包装规格】",
+                "规格A：24人份/盒、48人份/盒、96人份/盒。",
+                "规格B：24人份/盒、48人份/盒、96人份/盒。",
+                "【预期用途】",
+                "本试剂盒用于体外定性检测咽拭子、痰液样本中新型冠状病毒（2019-nCoV）ORF1ab和N基因。",
+                "【检测原理】",
+                "本段不应进入预期用途。",
+                "【主要组成成分】",
+                "表1 规格A大包装试剂盒组成成分",
+                "组分\t规格\t数量",
+                "PCR反应液\t24人份/盒\t1管",
+                "【储存条件及有效期】",
+                "-20±5℃的避光条件，有效期12个月。",
+                "反复冻融次数不得超过4次。",
+                "【样本要求】",
+                "适用样本类型：咽拭子、痰液。",
+            ]
+        )
+    }
+
+    result = extract_by_rules(texts, _registration_specs())
+
+    values = {field["key"]: field["value"] for field in result["fields"]}
+    assert values["product_name"] == "新型冠状病毒2019-nCoV核酸检测试剂盒（荧光PCR法）"
+    assert "规格A" in values["package_specification"]
+    assert "检测原理" not in values["intended_use"]
+    assert "PCR反应液" in values["main_components"]
+    assert "-20±5℃" in values["storage_condition_and_validity"]
+
+
 def test_llm_extract_parses_structured_json(monkeypatch):
     monkeypatch.setattr(
         "review_agent.application_form_fill.services.field_extract.generate_completion",

tests/test_regulatory_text_extract.py

@@ -37,3 +37,25 @@ def test_extract_text_reports_unsupported_file(tmp_path):
 
     assert result.status == "unsupported"
     assert result.text == ""
+
+
+def test_extract_text_from_docx_preserves_table_text(tmp_path):
+    from docx import Document
+
+    path = tmp_path / "说明书.docx"
+    document = Document()
+    document.add_paragraph("【主要组成成分】")
+    table = document.add_table(rows=2, cols=2)
+    table.rows[0].cells[0].text = "组分"
+    table.rows[0].cells[1].text = "数量"
+    table.rows[1].cells[0].text = "PCR反应液"
+    table.rows[1].cells[1].text = "1管"
+    document.add_paragraph("【储存条件及有效期】")
+    document.add_paragraph("-20±5℃保存，有效期12个月。")
+    document.save(path)
+
+    result = extract_text(path)
+
+    assert result.status == "success"
+    assert "组分\t数量" in result.text
+    assert result.text.index("PCR反应液") < result.text.index("【储存条件及有效期】")