feat(application-form-fill): 实现字段抽取与冲突合并

tests/test_application_form_fill_field_merge.py

@@ -0,0 +1,79 @@
+import pytest
+
+from review_agent.application_form_fill.services.field_merge import merge_fields, normalize_field_value, rank_source
+
+
+def test_normalize_field_value_removes_whitespace():
+    assert normalize_field_value(" 2-8℃ 保存 \n 有效期12个月 ") == "2-8℃保存有效期12个月"
+
+
+def test_rank_source_prefers_instructions():
+    assert rank_source("说明书") < rank_source("产品技术要求")
+
+
+def test_merge_fields_prefers_instructions_and_marks_conflict():
+    regex_results = {
+        "fields": [
+            {
+                "key": "storage_condition_and_validity",
+                "label": "产品储存条件及有效期",
+                "value": "2-8℃保存，有效期12个月",
+                "source_file": "说明书.txt",
+                "source_role": "说明书",
+                "evidence": "产品储存条件及有效期：2-8℃保存，有效期12个月",
+                "confidence": 0.75,
+            },
+            {
+                "key": "storage_condition_and_validity",
+                "label": "产品储存条件及有效期",
+                "value": "-20℃保存",
+                "source_file": "产品技术要求.txt",
+                "source_role": "产品技术要求",
+                "evidence": "产品储存条件及有效期：-20℃保存",
+                "confidence": 0.8,
+            },
+        ]
+    }
+
+    merged, conflicts = merge_fields(regex_results, {"fields": []})
+
+    field = merged["storage_condition_and_validity"]
+    assert field.value == "2-8℃保存，有效期12个月"
+    assert field.has_conflict is True
+    assert conflicts[0]["selected_value"] == "2-8℃保存，有效期12个月"
+    assert conflicts[0]["conflict_values"][0]["value"] == "-20℃保存"
+
+
+def test_merge_fields_combines_consistent_values_without_conflict():
+    regex_results = {
+        "fields": [
+            {
+                "key": "product_name",
+                "label": "产品名称",
+                "value": "甲胎蛋白检测试剂盒",
+                "source_file": "说明书.txt",
+                "source_role": "说明书",
+                "evidence": "产品名称：甲胎蛋白检测试剂盒",
+                "confidence": 0.75,
+            }
+        ]
+    }
+    llm_results = {
+        "fields": [
+            {
+                "key": "product_name",
+                "label": "产品名称",
+                "value": "甲胎蛋白 检测试剂盒",
+                "source_file": "产品技术要求.txt",
+                "source_role": "产品技术要求",
+                "evidence": "产品名称：甲胎蛋白 检测试剂盒",
+                "confidence": 0.9,
+            }
+        ]
+    }
+
+    merged, conflicts = merge_fields(regex_results, llm_results)
+
+    assert merged["product_name"].value == "甲胎蛋白检测试剂盒"
+    assert merged["product_name"].has_conflict is False
+    assert conflicts == []