fix(application-form-fill): 代理人字段暂用生产企业信息

review_agent/application_form_fill/services/field_extract.py

@@ -17,6 +17,11 @@ from review_agent.regulatory_review.services.text_extract import extract_text
 
 FIELD_ALIASES = {
     "product_name": ["产品名称"],
+    "applicant_name": ["注册人名称", "生产企业名称", "企业名称", "生产企业"],
+    "applicant_address": ["注册人住所", "生产企业住所", "企业住所", "住所"],
+    "manufacturer_address": ["生产地址", "生产企业地址", "生产场所"],
+    "agent_name": ["代理人名称", "生产企业名称", "企业名称", "生产企业", "注册人名称"],
+    "agent_address": ["代理人住所", "生产企业住所", "企业住所", "住所", "注册人住所"],
     "package_specification": ["包装规格", "规格"],
     "main_components": ["主要组成成分", "主要组成", "组成成分"],
     "intended_use": ["预期用途"],
@@ -41,7 +46,7 @@ def collect_document_texts(summary_batch: FileSummaryBatch) -> dict[str, str]:
 def extract_by_rules(texts: dict[str, str], specs: list[TemplateSpec]) -> dict[str, Any]:
     fields: list[dict[str, Any]] = []
     field_defs = _field_defs(specs)
-    labels = [field["label"] for field in field_defs if field.get("label")]
+    labels = _all_field_labels(field_defs)
     for file_name, text in texts.items():
         source_role = detect_source_role(file_name, text)
         for field in field_defs:
@@ -174,6 +179,15 @@ def _field_aliases(field: dict[str, str]) -> list[str]:
     return result
 
 
+def _all_field_labels(fields: list[dict[str, str]]) -> list[str]:
+    labels: list[str] = []
+    for field in fields:
+        for label in _field_aliases(field):
+            if label not in labels:
+                labels.append(label)
+    return labels
+
+
 def _extract_label_value(text: str, label: str, labels: list[str]) -> tuple[str, str]:
     return _extract_colon_label_value(text, label, labels)
 

review_agent/application_form_fill/services/field_merge.py

@@ -81,8 +81,30 @@ def merge_fields(regex_results: dict[str, Any], llm_results: dict[str, Any]) ->
                     "handling": "说明书优先，模板内黄底红字高亮" if rank_source(merged_field.source_file, merged_field.source_file) == 1 else "按来源优先级采用最高优先级字段",
                 }
             )
+    _apply_agent_company_fallbacks(merged)
     return merged, conflicts
 
 
 def _distinct_values(candidates: list[dict[str, Any]]) -> set[str]:
     return {normalize_field_value(str(item.get("value") or "")) for item in candidates if item.get("value")}
+
+
+def _apply_agent_company_fallbacks(merged: dict[str, MergedField]) -> None:
+    fallback_pairs = {
+        "agent_name": ("applicant_name", "代理人名称"),
+        "agent_address": ("applicant_address", "代理人住所"),
+    }
+    for target_key, (source_key, target_label) in fallback_pairs.items():
+        if target_key in merged or source_key not in merged:
+            continue
+        source = merged[source_key]
+        merged[target_key] = MergedField(
+            key=target_key,
+            label=target_label,
+            value=source.value,
+            source_file=source.source_file,
+            evidence=source.evidence,
+            confidence=source.confidence,
+            has_conflict=source.has_conflict,
+            conflict_values=source.conflict_values,
+        )

review_agent/application_form_fill/templates/application_form_templates_v1.yaml

@@ -36,6 +36,24 @@ templates:
         source_roles:
           - 申请表
           - 质量管理体系文件
+      - key: agent_name
+        label: 代理人名称
+        target:
+          type: table_row
+          row_label: 代理人名称
+        source_roles:
+          - 说明书
+          - 企业信息
+          - 申请表
+      - key: agent_address
+        label: 代理人住所
+        target:
+          type: table_row
+          row_label: 代理人住所
+        source_roles:
+          - 说明书
+          - 企业信息
+          - 申请表
       - key: product_name
         label: 产品名称
         target:

tests/test_application_form_fill_field_extract.py

@@ -84,6 +84,25 @@ def test_rule_extracts_bracket_sections_from_instructions():
     assert "-20±5℃" in values["storage_condition_and_validity"]
 
 
+def test_rule_maps_agent_fields_to_manufacturer_company_for_now():
+    texts = {
+        "目标产品说明书.docx": "\n".join(
+            [
+                "生产企业名称：卡尤迪生物科技宜兴有限公司",
+                "生产企业住所：江苏省宜兴经济技术开发区杏里路10号",
+                "生产地址：江苏省宜兴经济技术开发区杏里路10号宜兴光电产业园4幢102室",
+            ]
+        )
+    }
+
+    result = extract_by_rules(texts, _registration_specs())
+
+    values = {field["key"]: field["value"] for field in result["fields"]}
+    assert values["agent_name"] == "卡尤迪生物科技宜兴有限公司"
+    assert values["agent_address"] == "江苏省宜兴经济技术开发区杏里路10号"
+    assert values["manufacturer_address"] == "江苏省宜兴经济技术开发区杏里路10号宜兴光电产业园4幢102室"
+
+
 def test_llm_extract_parses_structured_json(monkeypatch):
     monkeypatch.setattr(
         "review_agent.application_form_fill.services.field_extract.generate_completion",

tests/test_application_form_fill_field_merge.py

@@ -77,3 +77,35 @@ def test_merge_fields_combines_consistent_values_without_conflict():
     assert merged["product_name"].value == "甲胎蛋白检测试剂盒"
     assert merged["product_name"].has_conflict is False
     assert conflicts == []
+
+
+def test_merge_fields_fills_agent_from_applicant_for_now():
+    regex_results = {
+        "fields": [
+            {
+                "key": "applicant_name",
+                "label": "注册人名称",
+                "value": "卡尤迪生物科技宜兴有限公司",
+                "source_file": "目标产品说明书.docx",
+                "source_role": "说明书",
+                "evidence": "生产企业名称：卡尤迪生物科技宜兴有限公司",
+                "confidence": 0.75,
+            },
+            {
+                "key": "applicant_address",
+                "label": "注册人住所",
+                "value": "江苏省宜兴经济技术开发区杏里路10号",
+                "source_file": "目标产品说明书.docx",
+                "source_role": "说明书",
+                "evidence": "生产企业住所：江苏省宜兴经济技术开发区杏里路10号",
+                "confidence": 0.75,
+            },
+        ]
+    }
+
+    merged, conflicts = merge_fields(regex_results, {"fields": []})
+
+    assert merged["agent_name"].value == "卡尤迪生物科技宜兴有限公司"
+    assert merged["agent_name"].label == "代理人名称"
+    assert merged["agent_address"].value == "江苏省宜兴经济技术开发区杏里路10号"
+    assert conflicts == []