# CLAUDE.md - system-admin

This file provides guidance to Claude Code when working in the system-admin module.

## Purpose

Main Spring Boot application module. Entry point: `com.weather.AdminApplication`. Depends on `system-common` and `system-dynamic-datasource`.

## Module Structure

```
com.weather/
├── AdminApplication.java          # @SpringBootApplication, @EnableAsync
├── common/                        # Module-level shared code
│   ├── annotation/                # @DataFilter, @LogOperation
│   ├── aspect/                    # DataFilterAspect, LogOperationAspect
│   ├── config/                    # JacksonConfig, MybatisPlusConfig, SwaggerConfig, FileServerConfig
│   ├── exception/                 # CustomExceptionHandler
│   ├── handler/                   # FieldMetaObjectHandler (MyBatis-Plus auto-fill)
│   ├── interceptor/               # DataFilterInterceptor, DataScope
│   ├── utils/                     # ExcelUtils, TimeUtils
│   └── validator/group/           # Cloud storage validator groups
└── modules/
    ├── job/                       # Quartz scheduling
    ├── log/                       # Audit logs (operation, login, error)
    ├── oss/                       # Cloud file storage (Alibaba/Qiniu/Tencent)
    ├── region/                    # Geographic region tree
    ├── security/                  # Shiro auth, OAuth2 filter, login
    ├── sys/                       # System management (users, roles, menus, depts, dicts, params, alerts)
    └── weather/                   # Weather domain
        ├── dailydata/             # Daily weather data CRUD, import, caching
        ├── filescan/              # WatchService file scanning pipeline
        └── station/               # Weather station management
```

## Service Layer Pattern

All business modules follow this structure under `modules/<domain>/`:

```
controller/   → @RestController, delegates to service interface
service/      → Interface extends CrudService<Entity, Dto>
service/impl/ → Extends CrudServiceImpl<Dao, Entity, Dto>, implements getWrapper()
dao/          → Interface extends BaseDao / BaseMapper
entity/       → Extends BaseEntity, maps to DB table
dto/          → Request/response DTOs
excel/        → @ExcelProperty-annotated beans for EasyExcel import/export
```

Entity-to-DTO conversion uses `ConvertUtils.sourceToTarget()`.

## Security / Auth

- **Framework**: Apache Shiro with Jakarta-compatible artifacts (classifier: `jakarta`)
- **Token flow**: Client sends `token` header/param -> `Oauth2Filter` extracts it -> `Oauth2Realm` validates against `sys_user_token` table -> returns `UserDetail` (user + dept scope + permissions)
- **Public paths** (no auth): `/login`, `/captcha`, `/druid/**`, `/doc.html`, `/swagger/**`, `/v3/api-docs/**`, `/favicon.ico`
- **Password hashing**: Custom `BCryptPasswordEncoder` with `PasswordUtils` (configurable rounds via sys_params)
- **Async context**: `UserContextHolder` (ThreadLocal) preserves user context across async task boundaries (used in file import, scheduled tasks). `SecurityUser.getUser()` falls back to this when Shiro Subject is unavailable.

## Data Scoping (Dept-Based Row-Level Security)

`DataFilterInterceptor` is a MyBatis-Plus `InnerInterceptor`. When a service method is annotated with `@DataFilter`, the `DataFilterAspect` injects a `DataScope` object containing a SQL filter clause into the query parameters. The interceptor then appends this WHERE clause to the SQL, restricting results to the user's authorized departments.

Example service usage:
```java
@DataFilter
public PageData<Dto> page(Map<String, Object> params) {
    // SQL automatically filtered by dept scope
}
```

## File Scanning Pipeline

`FileWatchServiceManager` (`modules/weather/filescan/`) monitors `{FILE_SCAN_ROOT_PATH}/receive/` using `java.nio.file.WatchService`:

1. **Directory layout**: `receive/<deptName>/`, `display/<deptName>/`, `archive/<deptName>/`
2. **Detection**: WatchService detects `ENTRY_CREATE` and `ENTRY_MODIFY` events
3. **Wait**: `waitForFileReady()` polls file size stability + file lock to ensure write completion
4. **Dedup**: MD5 hash check against `weather_file_scan_record` table
5. **Parse**: `FileNameParser` extracts region, category, period from filename
6. **Store**: Record inserted; file copied to `display/`; old version moved to `archive/`
7. **Root files**: Files in `receive/` root (no dept subdir) are classified as "model forecast" with `deptId = null`

`FileScanStartupRunner` (ApplicationRunner) triggers full directory scan + WatchService registration on startup.

## Weather Daily Data

Core table: `weather_daily_data` (entity: `WeatherDailyDataEntity`). Fields: stationId, observeDate, avgTemp, maxTemp/minTemp (with time), rainfall (20-20 and 08-08), relativeHumidity, atmospheres, wind (avg/max/extreme speed + direction + time), deptId.

### Import Flow

`WeatherDataImportManager` handles Excel uploads:
1. Save MultipartFile to temp file
2. Submit async task (`CompletableFuture`), returns taskId immediately
3. First pass: `EasyExcel.read()` with count listener to get total rows
4. Second pass: `WeatherDataListener` reads rows in batches, calls `insertBatch()`
5. `insertBatch()` resolves station-to-dept mapping, fills audit fields, uses `insertBatchMultiRow()` (custom MySQL multi-row INSERT)
6. Clears weather summarize Redis cache on completion
7. Progress tracked in-memory (`ConcurrentHashMap<String, ImportProgress>`)

### Summarize Caching

`WeatherSummarizeCacheTask` (Quartz job) precomputes "same month-day across years" summary data per station into Redis. Cache key: `weather:summarize:{month}:{day}`. On cache hit, `getCachedSummarize()` filters by station + year range. On miss, falls back to `selectSummarizeByMonthDay()` DB query.

## Alert / Notification System

- **Storage**: `SysAlertEntity` in `sys_alert` table
- **Real-time push**: `SseAlertService` manages active SSE connections (`CopyOnWriteArraySet<SseEmitter>`), broadcasts on Spring events:
  - `AlertCreatedEvent` -> SSE event `alert`
  - `AlertWithdrawnEvent` -> SSE event `alert-withdrawn`
  - `AlertDeletedEvent` -> SSE event `alert-deleted`
- **External sources**: `AlertSourcePollingTask` (Quartz) polls external services; `AlertSourceCollector` gathers results

## Job Scheduling (Quartz)

Tables: `schedule_job`, `schedule_job_log`, plus standard `qrtz_*` tables.

- `ScheduleConfig`: Quartz `SchedulerFactoryBean` configuration
- `ScheduleJob` entity: bean class, cron expression, params, status (PAUSE/NORMAL)
- `ScheduleUtils`: Create/update/delete/pause/resume Quartz triggers
- `JobCommandLineRunner`: On startup, restores all NORMAL-status jobs from DB
- `ITask` interface: `run(String params)` method -- all job classes implement this
- Concurrency control via `@DisallowConcurrentExecution`

## Database

Schema: `system-admin/db/weather_data_system.sql`

### Core Tables

| Table | Purpose |
|-------|---------|
| `sys_user`, `sys_role`, `sys_menu`, `sys_dept` | RBAC |
| `sys_role_user`, `sys_role_menu`, `sys_role_data_scope` | RBAC associations |
| `sys_user_token` | Shiro auth tokens |
| `sys_dict_type`, `sys_dict_data` | Dictionary system |
| `sys_params` | Key-value system parameters |
| `sys_alert` | Alert/notification records |
| `weather_station` | Weather station registry |
| `weather_daily_data` | Daily meteorological observations |
| `weather_file_scan_record` | File tracking (receive -> display -> archive) |
| `sys_region` | Geographic region tree (province/city/county) |
| `schedule_job`, `schedule_job_log` | Quartz job definitions and execution history |
| `qrtz_*` | Quartz internal scheduler tables |
| `sys_log_operation`, `sys_log_login`, `sys_log_error` | Audit logs |
| `sys_oss` | Cloud storage object records |