JobData/.planning/phases/05-data-pipeline/05-02-PLAN.md

---
phase: 5
plan: 2
wave: 2
title: "公司招聘信息 channel 区分（DATA-04）+ 推送 API 确认（DATA-02）"
depends_on:
  - "05-01-PLAN.md"
files_modified:
  - app/services/ingest/configs/boss.py
  - app/services/ingest/configs/qcwy.py
  - app/services/ingest/configs/zhilian.py
  - app/schemas/ingest.py
  - app/services/company_jobs_sync.py
autonomous: true
requirements:
  - DATA-02
  - DATA-04
---

# Phase 5 Plan 02: company_job channel 区分（DATA-04 + DATA-02 确认）

## Objective

### DATA-02（确认已完成）

`app/api/v1/job/job.py` 已有三个推送接收端点：
- `POST /data/store`（单条）
- `POST /data/batch`（批量同步）
- `POST /data/batch-async`（批量异步）

全部走 `IngestService`，`channel` 和 `platform` 由调用方在请求体中指定。
**spiderJobs 只需调用这些 API，DATA-02 不需要后端改动。**

### DATA-04（需要修复）

`company_jobs_sync.py` 目前调用 `router.store_batch(source, "mini", "job", jobs)`，
公司职位和搜索职位混用同一个 `channel="mini"`，无法区分来源。

**修复：** 
1. `ChannelType` 枚举加 `COMPANY = "company"` 值
2. 各平台 `ingest/configs/*.py` 注册 `channel="company"` 的配置
3. `company_jobs_sync.py` 调用改为 `channel="company"`

## Must Haves

- [ ] `app/schemas/ingest.py` 中 `ChannelType` 枚举加 `COMPANY = "company"`
- [ ] `ingest/configs/boss.py` 注册 `channel="company", data_type="job"` 配置（共用 boss_job 表）
- [ ] `ingest/configs/qcwy.py` 同样注册 `channel="company"` 配置
- [ ] `ingest/configs/zhilian.py` 同样注册 `channel="company"` 配置
- [ ] `company_jobs_sync.py` 中 `store_batch(source, "mini", "job", jobs)` → `store_batch(source, "company", "job", jobs)`
- [ ] `pipenv run python -c "from app.services.ingest import IngestService; print(IngestService.get_registry_info())"` 显示 channel=company 条目
- [ ] 全量回归 `pytest tests/` 无失败

---

## Wave 2（依赖 Plan 01 完成）

### Task 2.1: 更新 ChannelType 枚举

<read_first>
- `app/schemas/ingest.py`（需读取确认当前 ChannelType 定义）
</read_first>

<action>
在 `ChannelType` 枚举中添加：
```python
COMPANY = "company"
```
</action>

---

### Task 2.2: 各平台 configs 注册 channel="company" 配置

<read_first>
- `app/services/ingest/configs/boss.py`（当前有 mini+job、mini+company 两条注册）
- `app/services/ingest/configs/qcwy.py`
- `app/services/ingest/configs/zhilian.py`
</read_first>

<action>
在三个 configs 文件中各添加一条新注册：

**boss.py 追加：**
```python
register(PlatformConfig(
    platform="boss", channel="company", data_type="job",
    table="boss_job",  # 共用同一张表，通过 channel 列区分来源
    dedup_fields=(DedupFieldSpec(column="job_id", extractor=_extract_job_id),),
))
```

**qcwy.py 追加：**
```python
register(PlatformConfig(
    platform="qcwy", channel="company", data_type="job",
    table="qcwy_job",
    dedup_fields=(
        DedupFieldSpec(column="job_id", extractor=_extract_job_id),
        DedupFieldSpec(column="update_date_time", extractor=_extract_update_dt),
    ),
))
```

**zhilian.py 追加：**（需读取 zhilian.py 确认 extractor 函数名）
```python
register(PlatformConfig(
    platform="zhilian", channel="company", data_type="job",
    table="zhilian_job",
    dedup_fields=(DedupFieldSpec(column="job_number", extractor=_extract_job_number),),
))
```
</action>

---

### Task 2.3: 修改 company_jobs_sync.py

<action>
将 `sync_company_jobs()` 中：
```python
# 修改前
store_result = await router.store_batch(source, "mini", "job", jobs)

# 修改后
store_result = await router.store_batch(source, "company", "job", jobs)
```
</action>

<acceptance_criteria>
- `grep "company.*job" app/services/ingest/configs/boss.py` 有输出
- `grep '"company"' app/services/company_jobs_sync.py` 有输出
- `grep '"mini"' app/services/company_jobs_sync.py` 无输出（全部替换）
</acceptance_criteria>

---

## Verification

```bash
# 1. 确认 channel=company 注册存在
pipenv run python -c "
from app.services.ingest.configs import boss, qcwy, zhilian  # 触发 register()
from app.services.ingest.registry import list_configs
company_configs = [c for c in list_configs() if c.channel == 'company']
print(f'channel=company 配置数: {len(company_configs)}')
for c in company_configs:
    print(f'  {c.platform} / {c.channel} / {c.data_type} → {c.table}')
assert len(company_configs) >= 3, '❌ 缺少注册配置'
print('✅ 注册配置完整')
"

# 2. 全量回归
pipenv run python -m pytest tests/ -v --tb=short
```