修复采购规则跨项目人员串入问题

ccdi-project/src/main/resources/mapper/ccdi/project/CcdiBankTagAnalysisMapper.xml

@@ -30,6 +30,17 @@ PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
         NULL AS reasonDetail
     </sql>
 
+    <sql id="projectScopedDirectStaffSql">
+        select distinct
+            cast(staff.staff_id as char) as staffId,
+            staff.id_card as idCard
+        from ccdi_bank_statement bs
+        inner join ccdi_base_staff staff on staff.id_card = bs.cret_no
+        where bs.project_id = #{projectId}
+          and bs.cret_no is not null
+          and trim(bs.cret_no) != ''
+    </sql>
+
     <sql id="cashDepositPredicate">
         (
             (
@@ -911,8 +922,9 @@ PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
                 pt.supplier_name AS supplierName,
                 pt.actual_amount AS actualAmount
             from ccdi_purchase_transaction pt
-            inner join ccdi_base_staff staff
-                on CAST(staff.staff_id AS CHAR) = pt.applicant_id
+            inner join (
+                <include refid="projectScopedDirectStaffSql"/>
+            ) project_staff on project_staff.staffId = pt.applicant_id
             where IFNULL(pt.actual_amount, 0) > 100000
             union
             select distinct
@@ -921,8 +933,9 @@ PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
                 pt.supplier_name AS supplierName,
                 pt.actual_amount AS actualAmount
             from ccdi_purchase_transaction pt
-            inner join ccdi_base_staff staff
-                on CAST(staff.staff_id AS CHAR) = pt.purchase_leader_id
+            inner join (
+                <include refid="projectScopedDirectStaffSql"/>
+            ) project_staff on project_staff.staffId = pt.purchase_leader_id
             where pt.purchase_leader_id is not null
               and IFNULL(pt.actual_amount, 0) > 100000
         ) t
@@ -955,24 +968,28 @@ PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
                     SUM(source.actualAmount) / NULLIF(total_amount.totalAmount, 0) AS supplierRatio
                 from (
                     select distinct
-                        staff.id_card AS objectKey,
+                        project_staff.idCard AS objectKey,
                         pt.purchase_id AS purchaseId,
                         pt.supplier_name AS supplierName,
                         IFNULL(pt.actual_amount, 0) AS actualAmount
                     from ccdi_purchase_transaction pt
-                    inner join ccdi_base_staff staff on CAST(staff.staff_id AS CHAR) = pt.applicant_id
+                    inner join (
+                        <include refid="projectScopedDirectStaffSql"/>
+                    ) project_staff on project_staff.staffId = pt.applicant_id
                     where IFNULL(pt.actual_amount, 0) > 0
                       and IFNULL(pt.supplier_name, '') &lt;&gt; ''
 
                     union
 
                     select distinct
-                        staff.id_card AS objectKey,
+                        project_staff.idCard AS objectKey,
                         pt.purchase_id AS purchaseId,
                         pt.supplier_name AS supplierName,
                         IFNULL(pt.actual_amount, 0) AS actualAmount
                     from ccdi_purchase_transaction pt
-                    inner join ccdi_base_staff staff on CAST(staff.staff_id AS CHAR) = pt.purchase_leader_id
+                    inner join (
+                        <include refid="projectScopedDirectStaffSql"/>
+                    ) project_staff on project_staff.staffId = pt.purchase_leader_id
                     where pt.purchase_leader_id is not null
                       and IFNULL(pt.actual_amount, 0) > 0
                       and IFNULL(pt.supplier_name, '') &lt;&gt; ''
@@ -983,21 +1000,25 @@ PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
                         ROUND(SUM(source_total.actualAmount), 2) AS totalAmount
                     from (
                         select distinct
-                            staff.id_card AS objectKey,
+                            project_staff.idCard AS objectKey,
                             pt.purchase_id AS purchaseId,
                             IFNULL(pt.actual_amount, 0) AS actualAmount
                         from ccdi_purchase_transaction pt
-                        inner join ccdi_base_staff staff on CAST(staff.staff_id AS CHAR) = pt.applicant_id
+                        inner join (
+                            <include refid="projectScopedDirectStaffSql"/>
+                        ) project_staff on project_staff.staffId = pt.applicant_id
                         where IFNULL(pt.actual_amount, 0) > 0
 
                         union
 
                         select distinct
-                            staff.id_card AS objectKey,
+                            project_staff.idCard AS objectKey,
                             pt.purchase_id AS purchaseId,
                             IFNULL(pt.actual_amount, 0) AS actualAmount
                         from ccdi_purchase_transaction pt
-                        inner join ccdi_base_staff staff on CAST(staff.staff_id AS CHAR) = pt.purchase_leader_id
+                        inner join (
+                            <include refid="projectScopedDirectStaffSql"/>
+                        ) project_staff on project_staff.staffId = pt.purchase_leader_id
                         where pt.purchase_leader_id is not null
                           and IFNULL(pt.actual_amount, 0) > 0
                     ) source_total

ccdi-project/src/test/java/com/ruoyi/ccdi/project/mapper/CcdiBankTagAnalysisMapperXmlTest.java

@@ -138,6 +138,16 @@ class CcdiBankTagAnalysisMapperXmlTest {
         }
     }
 
+    @Test
+    void purchaseRules_shouldBeScopedToCurrentProjectStaff() throws Exception {
+        String xml = readXml(RESOURCE);
+
+        assertAll(
+            () -> assertPurchaseRuleScopedByProject(xml, "selectLargePurchaseTransactionStatements"),
+            () -> assertPurchaseRuleScopedByProject(xml, "selectSupplierConcentrationObjects")
+        );
+    }
+
     @Test
     void assetRegistrationMismatchRules_shouldUseRealSqlAndAssetTable() throws Exception {
         String xml = readXml(RESOURCE);
@@ -192,4 +202,21 @@ class CcdiBankTagAnalysisMapperXmlTest {
         assertTrue(selectSql.contains("reasonDetail"), () -> selectId + " 缺少 reasonDetail");
         assertTrue(!selectSql.contains("where 1 = 0"), () -> selectId + " 仍是占位 SQL");
     }
+
+    private void assertPurchaseRuleScopedByProject(String xml, String selectId) {
+        String selectSql = extractSelectSql(xml, selectId);
+        String scopeSql = extractSqlFragment(xml, "projectScopedDirectStaffSql");
+        assertTrue(selectSql.contains("projectScopedDirectStaffSql"), () -> selectId + " 未引用项目员工范围片段");
+        assertTrue(scopeSql.contains("ccdi_bank_statement"), () -> selectId + " 缺少项目流水范围约束");
+        assertTrue(scopeSql.contains("#{projectId}"), () -> selectId + " 缺少 projectId 过滤条件");
+    }
+
+    private String extractSqlFragment(String xml, String sqlId) {
+        Pattern pattern = Pattern.compile(
+            "<sql\\s+id=\"" + sqlId + "\"[\\s\\S]*?</sql>"
+        );
+        Matcher matcher = pattern.matcher(xml);
+        assertTrue(matcher.find(), () -> "未找到 sql 片段: " + sqlId);
+        return matcher.group();
+    }
 }

docs/plans/backend/2026-03-23-credit-parsing-mock-server-backend-implementation.md

@@ -0,0 +1,447 @@
+# Credit Parsing Mock Server Backend Implementation Plan
+
+> **For agentic workers:** REQUIRED: Use superpowers:subagent-driven-development (if subagents available) or superpowers:executing-plans to implement this plan. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** 在现有 `lsfx-mock-server` 中新增征信解析 Mock 能力，提供 `POST /xfeature-mngs/conversation/htmlEval` 接口，支持稳定随机生成征信 `payload`、最小错误码模拟与健康检查能力。
+
+**Architecture:** 复用现有 FastAPI 应用与测试基础设施，不新建独立服务。征信解析能力拆成 `schema 配置 + payload 生成 service + 调试 service + 独立 router` 四层，接口层只负责 `form-data` 接收与响应组装，字段生成和错误处理分别下沉到独立 service 中，避免与现有 LSFX Mock 逻辑混杂。
+
+**Tech Stack:** Python 3, FastAPI, pytest, python-multipart, JSON, Bash
+
+---
+
+## File Structure
+
+- `lsfx-mock-server/config/credit_feature_schema.json`: 固化 Excel 中 30 个征信指标字段的主题域、字段名、类型与枚举范围。
+- `lsfx-mock-server/config/credit_response_examples.json`: 固化成功与失败响应模板，避免响应结构散落在路由中。
+- `lsfx-mock-server/services/credit_payload_service.py`: 负责稳定随机种子生成、schema 加载和 `payload` 构造。
+- `lsfx-mock-server/services/credit_debug_service.py`: 负责参数校验、调试标记识别与错误响应封装。
+- `lsfx-mock-server/routers/credit_api.py`: 负责新增征信解析接口和征信健康检查接口。
+- `lsfx-mock-server/main.py`: 注册征信解析 router，并在应用描述中补充接口说明。
+- `lsfx-mock-server/README.md`: 补充征信解析 Mock 的启动方式、接口路径和调试用法。
+- `lsfx-mock-server/tests/test_credit_payload_service.py`: 锁定稳定随机与字段类型语义。
+- `lsfx-mock-server/tests/test_credit_api.py`: 锁定征信解析接口成功、缺参、错误码与健康检查行为。
+- `lsfx-mock-server/tests/test_startup.py`: 锁定新 router 已被主应用注册。
+- `docs/reports/implementation/2026-03-23-credit-parsing-mock-server-backend-record.md`: 记录本次后端实施内容。
+- `docs/tests/records/2026-03-23-credit-parsing-mock-server-backend-verification.md`: 记录本次后端验证命令与结果。
+
+### Task 1: 建立征信字段 schema 和稳定随机 payload 生成能力
+
+**Files:**
+- Create: `lsfx-mock-server/config/credit_feature_schema.json`
+- Create: `lsfx-mock-server/config/credit_response_examples.json`
+- Create: `lsfx-mock-server/services/credit_payload_service.py`
+- Create: `lsfx-mock-server/tests/test_credit_payload_service.py`
+- Reference: `docs/design/2026-03-23-credit-parsing-mock-server-design.md`
+- Reference: `assets/征信解析/征信解析接口payload.xlsx`
+
+- [ ] **Step 1: Write the failing test**
+
+在 `lsfx-mock-server/tests/test_credit_payload_service.py` 中先补两条失败用例，锁定“同一输入返回稳定随机相同结果”和“schema 中的状态字段严格落在枚举范围内”：
+
+```python
+from services.credit_payload_service import CreditPayloadService
+
+
+def test_generate_payload_should_be_stable_for_same_input():
+    service = CreditPayloadService("config/credit_feature_schema.json")
+
+    payload1 = service.generate_payload(
+        model="LXCUSTALL",
+        h_type="PERSON",
+        filename="credit-report-a.html",
+    )
+    payload2 = service.generate_payload(
+        model="LXCUSTALL",
+        h_type="PERSON",
+        filename="credit-report-a.html",
+    )
+
+    assert payload1 == payload2
+    assert set(payload1.keys()) == {"lx_header", "lx_debt", "lx_publictype"}
+    assert len(payload1["lx_debt"]) == 21
+    assert len(payload1["lx_publictype"]) == 6
+
+
+def test_generate_payload_should_use_schema_type_rules():
+    service = CreditPayloadService("config/credit_feature_schema.json")
+
+    payload = service.generate_payload(
+        model="LXCUSTALL",
+        h_type="ENTERPRISE",
+        filename="credit-report-b.html",
+    )
+
+    assert payload["lx_debt"]["uncle_bank_house_state"] in {"正常", "逾期", "不良"}
+    assert payload["lx_header"]["report_time"].count("-") == 2
+    assert payload["lx_publictype"]["civil_cnt"].isdigit()
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_credit_payload_service.py -v
+```
+
+Expected:
+
+- `FAIL`
+- 原因是 `CreditPayloadService`、schema 配置文件和响应模板尚不存在
+
+- [ ] **Step 3: Write minimal implementation**
+
+按最小路径落地：
+
+1. 在 `lsfx-mock-server/config/credit_feature_schema.json` 中把 Excel 30 个字段完整整理为 JSON，至少包含：
+
+```json
+[
+  { "domain": "lx_header", "field": "query_cert_no", "type": "string" },
+  { "domain": "lx_header", "field": "query_cust_name", "type": "string" },
+  { "domain": "lx_header", "field": "report_time", "type": "string" },
+  { "domain": "lx_debt", "field": "uncle_bank_house_state", "type": "status", "options": ["正常", "逾期", "不良"] }
+]
+```
+
+2. 在 `lsfx-mock-server/config/credit_response_examples.json` 中固化成功/失败模板：
+
+```json
+{
+  "success": { "message": "成功", "payload": {}, "status_code": "0" },
+  "errors": {
+    "ERR_99999": { "message": "关键参数缺失，参数名: XX", "payload": null, "status_code": "ERR_99999" }
+  }
+}
+```
+
+3. 在 `lsfx-mock-server/services/credit_payload_service.py` 中实现最小服务：
+
+```python
+class CreditPayloadService:
+    def __init__(self, schema_path: str):
+        self.schema_path = schema_path
+        self.schema = self._load_schema()
+
+    def generate_payload(self, model: str, h_type: str, filename: str) -> dict:
+        rng = random.Random(self._build_seed(model, h_type, filename))
+        ...
+```
+
+4. 生成规则只保留设计文档确认的 4 类：
+   - `string`
+   - `amount`
+   - `count`
+   - `status`
+
+- [ ] **Step 4: Run test to verify it passes**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_credit_payload_service.py -v
+```
+
+Expected:
+
+- `PASS`
+- 同一输入生成稳定一致结果，不同字段类型输出格式符合约定
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lsfx-mock-server/config/credit_feature_schema.json lsfx-mock-server/config/credit_response_examples.json lsfx-mock-server/services/credit_payload_service.py lsfx-mock-server/tests/test_credit_payload_service.py
+git commit -m "新增征信解析字段配置与生成服务"
+```
+
+### Task 2: 实现参数校验、错误码模拟和征信解析接口
+
+**Files:**
+- Create: `lsfx-mock-server/services/credit_debug_service.py`
+- Create: `lsfx-mock-server/routers/credit_api.py`
+- Create: `lsfx-mock-server/tests/test_credit_api.py`
+- Modify: `lsfx-mock-server/tests/conftest.py`
+- Reference: `docs/design/2026-03-23-credit-parsing-mock-server-design.md`
+
+- [ ] **Step 1: Write the failing test**
+
+在 `lsfx-mock-server/tests/test_credit_api.py` 中先补 4 类失败/成功用例，确保接口不走 FastAPI 默认 422，而是走说明书约定的业务响应：
+
+```python
+def test_html_eval_should_return_credit_payload(client):
+    response = client.post(
+        "/xfeature-mngs/conversation/htmlEval",
+        data={"model": "LXCUSTALL", "hType": "PERSON"},
+        files={"file": ("credit.html", b"<html></html>", "text/html")},
+    )
+
+    assert response.status_code == 200
+    data = response.json()
+    assert data["status_code"] == "0"
+    assert data["message"] == "成功"
+    assert "lx_header" in data["payload"]
+
+
+def test_html_eval_should_return_err_99999_for_missing_model(client):
+    response = client.post(
+        "/xfeature-mngs/conversation/htmlEval",
+        data={"hType": "PERSON"},
+        files={"file": ("credit.html", b"<html></html>", "text/html")},
+    )
+
+    assert response.status_code == 200
+    assert response.json()["status_code"] == "ERR_99999"
+
+
+def test_html_eval_should_return_err_10003_for_invalid_h_type(client):
+    response = client.post(
+        "/xfeature-mngs/conversation/htmlEval",
+        data={"model": "LXCUSTALL", "hType": "JSON"},
+        files={"file": ("credit.html", b"<html></html>", "text/html")},
+    )
+
+    assert response.status_code == 200
+    assert response.json()["status_code"] == "ERR_10003"
+
+
+def test_html_eval_should_support_debug_error_marker(client):
+    response = client.post(
+        "/xfeature-mngs/conversation/htmlEval",
+        data={"model": "error_ERR_10001", "hType": "PERSON"},
+        files={"file": ("credit.html", b"<html></html>", "text/html")},
+    )
+
+    assert response.status_code == 200
+    assert response.json()["status_code"] == "ERR_10001"
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_credit_api.py -v
+```
+
+Expected:
+
+- `FAIL`
+- 原因是征信解析 router 和调试 service 尚不存在
+
+- [ ] **Step 3: Write minimal implementation**
+
+按设计文档实现最小闭环：
+
+1. 在 `lsfx-mock-server/services/credit_debug_service.py` 中定义错误码映射和业务校验入口：
+
+```python
+class CreditDebugService:
+    ERROR_MESSAGES = {
+        "ERR_99999": "关键参数缺失，参数名: XX",
+        "ERR_10001": "无效的证件号码",
+        "ERR_10002": "无效的主题域",
+        "ERR_10003": "报文类型无效，仅支持JSON/XML",
+        "ERR_10004": "无效机构号或行社号",
+        "ERR_10005": "无权限访问",
+        "ERR_10006": "尽调报告生成异常：异步事件发送失败",
+    }
+```
+
+2. 在 `lsfx-mock-server/routers/credit_api.py` 中让参数全部使用可空 `Form(None)` / `File(None)`，再手工校验，避免缺参时被 FastAPI 直接拦成 422：
+
+```python
+@router.post("/xfeature-mngs/conversation/htmlEval")
+async def html_eval(
+    model: Optional[str] = Form(None),
+    hType: Optional[str] = Form(None),
+    file: Optional[UploadFile] = File(None),
+):
+    ...
+```
+
+3. 正常流程调用 `CreditPayloadService.generate_payload()`。
+4. 失败流程统一返回 `message + payload(null) + status_code`。
+5. 在 `tests/conftest.py` 中按需补共享 fixture，例如测试上传的默认 HTML 文件内容。
+
+- [ ] **Step 4: Run test to verify it passes**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_credit_api.py -v
+```
+
+Expected:
+
+- `PASS`
+- 接口成功、缺参、非法 `hType`、调试错误码都返回预期业务结构
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lsfx-mock-server/services/credit_debug_service.py lsfx-mock-server/routers/credit_api.py lsfx-mock-server/tests/test_credit_api.py lsfx-mock-server/tests/conftest.py
+git commit -m "新增征信解析接口与错误模拟"
+```
+
+### Task 3: 注册新 router 并补启动文档
+
+**Files:**
+- Modify: `lsfx-mock-server/main.py`
+- Modify: `lsfx-mock-server/README.md`
+- Modify: `lsfx-mock-server/tests/test_startup.py`
+- Reference: `lsfx-mock-server/dev.py`
+
+- [ ] **Step 1: Write the failing test**
+
+先在 `lsfx-mock-server/tests/test_startup.py` 中补应用注册断言，锁定主应用已经包含征信解析接口和健康检查接口：
+
+```python
+from main import app
+
+
+def test_app_should_register_credit_mock_routes():
+    paths = {route.path for route in app.routes}
+
+    assert "/xfeature-mngs/conversation/htmlEval" in paths
+    assert "/credit/health" in paths
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_startup.py -v
+```
+
+Expected:
+
+- `FAIL`
+- 原因是 `main.py` 还未注册征信解析 router
+
+- [ ] **Step 3: Write minimal implementation**
+
+1. 在 `lsfx-mock-server/main.py` 中引入并注册 `credit_api.router`：
+
+```python
+from routers import api, credit_api
+
+app.include_router(api.router, tags=["流水分析接口"])
+app.include_router(credit_api.router, tags=["征信解析接口"])
+```
+
+2. 在应用描述中补充征信解析能力说明。
+3. 在 `lsfx-mock-server/README.md` 中补充：
+   - 新接口路径
+   - `curl` 或 `requests` 调用示例
+   - `error_ERR_10001` 调试方式
+   - `GET /credit/health` 用法
+
+- [ ] **Step 4: Run test to verify it passes**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_startup.py tests/test_credit_api.py -v
+```
+
+Expected:
+
+- `PASS`
+- 主应用已注册征信解析 router，README 与运行方式保持一致
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add lsfx-mock-server/main.py lsfx-mock-server/README.md lsfx-mock-server/tests/test_startup.py
+git commit -m "注册征信解析Mock路由并补充说明"
+```
+
+### Task 4: 端到端验证并沉淀实施记录
+
+**Files:**
+- Create: `docs/reports/implementation/2026-03-23-credit-parsing-mock-server-backend-record.md`
+- Create: `docs/tests/records/2026-03-23-credit-parsing-mock-server-backend-verification.md`
+- Reference: `docs/design/2026-03-23-credit-parsing-mock-server-design.md`
+- Reference: `docs/plans/backend/2026-03-23-credit-parsing-mock-server-backend-implementation.md`
+
+- [ ] **Step 1: Run targeted automated tests**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 -m pytest tests/test_credit_payload_service.py tests/test_credit_api.py tests/test_startup.py -v
+```
+
+Expected:
+
+- `PASS`
+- 征信字段生成、接口返回和路由注册全部通过
+
+- [ ] **Step 2: Run manual startup and smoke verification**
+
+Run:
+
+```bash
+cd lsfx-mock-server
+python3 main.py > /tmp/credit-mock-server.log 2>&1 &
+echo $! > /tmp/credit-mock-server.pid
+```
+
+再执行：
+
+```bash
+curl -s http://127.0.0.1:8000/credit/health
+
+curl -s -X POST http://127.0.0.1:8000/xfeature-mngs/conversation/htmlEval \
+  -F model=LXCUSTALL \
+  -F hType=PERSON \
+  -F file=@/tmp/sample-credit.html
+```
+
+Expected:
+
+- 健康检查返回 `healthy`
+- 成功接口返回 `status_code: "0"`，且 `payload` 包含三大主题域
+
+- [ ] **Step 3: Stop the started process**
+
+Run:
+
+```bash
+kill "$(cat /tmp/credit-mock-server.pid)"
+rm -f /tmp/credit-mock-server.pid
+```
+
+Expected:
+
+- 本次验证启动的 Mock 进程已停止，不留下端口占用
+
+- [ ] **Step 4: Write implementation and verification records**
+
+在 `docs/reports/implementation/2026-03-23-credit-parsing-mock-server-backend-record.md` 中记录：
+
+- 新增接口、配置文件、service 和测试文件
+- 稳定随机策略
+- 说明书歧义处理方式
+
+在 `docs/tests/records/2026-03-23-credit-parsing-mock-server-backend-verification.md` 中记录：
+
+- 自动化测试命令与结果
+- 手工 `curl` 验证命令与结果
+- 启停服务过程
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add docs/reports/implementation/2026-03-23-credit-parsing-mock-server-backend-record.md docs/tests/records/2026-03-23-credit-parsing-mock-server-backend-verification.md
+git commit -m "补充征信解析Mock后端实施与验证记录"
+```

docs/plans/frontend/2026-03-23-credit-parsing-mock-server-frontend-implementation.md

@@ -0,0 +1,124 @@
+# Credit Parsing Mock Server Frontend Implementation Plan
+
+> **For agentic workers:** REQUIRED: Use superpowers:subagent-driven-development (if subagents available) or superpowers:executing-plans to implement this plan. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** 在不新增前端页面和业务补丁的前提下，明确本次征信解析 Mock Server 建设对前端的影响边界，并沉淀联调核验与零改动结论。
+
+**Architecture:** 本次需求主体是 `lsfx-mock-server` 的后端 Mock 能力建设，不要求 `ruoyi-ui` 新增页面、路由、接口封装或交互。前端计划采用“零源码改动 + 契约核验 + 记录沉淀”的最短路径；若核验发现必须新增前端消费字段，则应停止执行并回到设计阶段，而不是在本计划中临时扩展功能。
+
+**Tech Stack:** Vue 2, Axios request wrapper, rg, Markdown docs
+
+---
+
+## File Structure
+
+- `ruoyi-ui/src/api/`: 只用于核验当前仓库是否已有征信解析相关调用，不预期修改。
+- `ruoyi-ui/src/views/`: 只用于核验是否存在需要接入征信解析 Mock 的页面触点，不预期修改。
+- `ruoyi-ui/src/utils/request.js`: 只用于确认现有请求封装是否足以承接未来联调，不预期修改。
+- `docs/reports/implementation/2026-03-23-credit-parsing-mock-server-frontend-record.md`: 记录本次前端零代码改动的范围与依据。
+- `docs/tests/records/2026-03-23-credit-parsing-mock-server-frontend-verification.md`: 记录前端契约核验命令、结果和结论。
+
+### Task 1: 核验当前前端不存在征信解析 Mock 接入改造需求
+
+**Files:**
+- Reference: `ruoyi-ui/src/api/`
+- Reference: `ruoyi-ui/src/views/`
+- Reference: `ruoyi-ui/src/utils/request.js`
+- Reference: `docs/design/2026-03-23-credit-parsing-mock-server-design.md`
+
+- [ ] **Step 1: Check the existing frontend touchpoints**
+
+检查当前前端是否已经存在以下任一触点：
+
+- 征信 HTML 上传入口
+- 征信解析结果展示页
+- 指向 `htmlEval`、`xfeature`、`征信解析` 的接口封装
+
+如无上述触点，则本次计划默认保持前端零代码改动。
+
+- [ ] **Step 2: Verify with search commands**
+
+Run:
+
+```bash
+cd ruoyi-ui
+rg -n "htmlEval|xfeature|征信|credit" src
+```
+
+Expected:
+
+- 若无输出或仅命中无关文字，说明当前前端仓库没有现成的征信解析接入点
+- 本次 Mock Server 不需要同步新增前端代码
+
+- [ ] **Step 3: Confirm no contract adaptation is needed**
+
+对照设计文档确认：
+
+- Mock 接口为独立服务能力，不替换现有前端接口返回结构
+- 不要求前端新增必填请求字段
+- 不要求前端渲染新增结果列或交互入口
+
+若上述任一不成立，则停止执行并回到设计阶段。
+
+- [ ] **Step 4: Record the no-op conclusion**
+
+在后续实施记录中明确写明：
+
+- 本次需求不涉及 `ruoyi-ui` 源码改动
+- 不为了“看起来完整”而额外创建演示页或调试页
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add docs/reports/implementation/2026-03-23-credit-parsing-mock-server-frontend-record.md docs/tests/records/2026-03-23-credit-parsing-mock-server-frontend-verification.md
+git commit -m "记录征信解析Mock前端零改动结论"
+```
+
+### Task 2: 沉淀前端联调与验证记录
+
+**Files:**
+- Create: `docs/reports/implementation/2026-03-23-credit-parsing-mock-server-frontend-record.md`
+- Create: `docs/tests/records/2026-03-23-credit-parsing-mock-server-frontend-verification.md`
+
+- [ ] **Step 1: Write implementation record**
+
+在 `docs/reports/implementation/2026-03-23-credit-parsing-mock-server-frontend-record.md` 中记录：
+
+- 本次需求核心在 Mock 服务后端
+- 当前前端仓库不存在征信解析接入点
+- 因此本轮实施不修改 `ruoyi-ui` 任何源码
+
+- [ ] **Step 2: Write verification record**
+
+在 `docs/tests/records/2026-03-23-credit-parsing-mock-server-frontend-verification.md` 中记录：
+
+- 执行过的 `rg` 核验命令
+- 查验的目录范围
+- “无需前端改动”的判断依据
+
+- [ ] **Step 3: Verify frontend diff stays empty**
+
+Run:
+
+```bash
+git diff --name-only -- ruoyi-ui
+```
+
+Expected:
+
+- 无输出
+- 证明本次前端计划执行保持零源码改动
+
+- [ ] **Step 4: Confirm no frontend build is required**
+
+在验证记录中明确写明：
+
+- 因为 `ruoyi-ui` 无源码改动，本次不执行 `npm run build:prod`
+- 若后续新增实际接入页面，再补充构建和联调测试
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add docs/reports/implementation/2026-03-23-credit-parsing-mock-server-frontend-record.md docs/tests/records/2026-03-23-credit-parsing-mock-server-frontend-verification.md
+git commit -m "补充征信解析Mock前端核验记录"
+```

docs/reports/implementation/2026-03-23-project-51-warning-count-fix-implementation.md

@@ -0,0 +1,28 @@
+# Project 51 预警人数异常修复记录
+
+## 本次改动
+
+- 修复 `ccdi-project` 模块中采购类规则未按项目人员范围过滤的问题。
+- 在 `ccdi-project/src/main/resources/mapper/ccdi/project/CcdiBankTagAnalysisMapper.xml` 新增 `projectScopedDirectStaffSql` 片段，统一收敛“当前项目流水中已识别员工”的范围。
+- 将 `selectLargePurchaseTransactionStatements` 改为仅统计当前项目员工对应的采购记录。
+- 将 `selectSupplierConcentrationObjects` 改为仅统计当前项目员工对应的采购记录，避免把其他项目或全局采购数据映射进当前项目预警结果。
+- 在 `ccdi-project/src/test/java/com/ruoyi/ccdi/project/mapper/CcdiBankTagAnalysisMapperXmlTest.java` 新增回归测试，约束采购规则必须引用项目员工范围片段。
+
+## 根因说明
+
+- `project_id = 51` 的项目总人数来自 `ccdi_project.target_count`，当前值为 `1`。
+- 预警人数来自 `ccdi_project_overview_employee_result` 聚合结果，排查发现其中除了上传文件对应员工 `韩桂英` 外，还混入了 `罗洋`。
+- 进一步核对 `ccdi_bank_statement_tag_result` 后确认，多出的 `罗洋` 来自 `SUPPLIER_CONCENTRATION` 规则生成的对象命中结果。
+- 原始 SQL 直接扫描 `ccdi_purchase_transaction` 全表，并通过员工工号映射到 `ccdi_base_staff`，没有任何 `projectId` 或项目人员范围限制，导致全局采购数据串入当前项目。
+
+## 变更文件
+
+- `ccdi-project/src/main/resources/mapper/ccdi/project/CcdiBankTagAnalysisMapper.xml`
+- `ccdi-project/src/test/java/com/ruoyi/ccdi/project/mapper/CcdiBankTagAnalysisMapperXmlTest.java`
+
+## 验证结果
+
+- 通过数据库核对确认：`project_id = 51` 当前仅上传成功 1 个文件，但 `ccdi_project_overview_employee_result` 中存在 2 名员工，异常现象可复现。
+- 执行 `mvn -pl ccdi-project -Dtest=CcdiBankTagAnalysisMapperXmlTest test`
+- 结果：`BUILD SUCCESS`
+- 新增回归测试已覆盖采购规则项目范围约束，防止后续再次把全局采购员工带入单个项目的预警统计。