Files
gongxue-base/apps/server/src/attendance/attendance-import.service.ts

357 lines
13 KiB
TypeScript

import { BadRequestException, Injectable, Logger } from '@nestjs/common';
import { InjectRepository } from '@nestjs/typeorm';
import { Repository, In } from 'typeorm';
import { Subject, Observable } from 'rxjs';
import {
DingAttendanceRaw,
Student,
StudentDingMapping,
} from '../entities';
import { DingTalkService, DingTalkAttendanceResult } from '../integration/dingtalk.service';
import { AttendanceService } from './attendance.service';
import type { ImportProgressEvent, ImportResult } from './dto/dingtalk-import.dto';
/**
* Service for importing DingTalk attendance data into the system.
*
* ## Stream Architecture
* Uses RxJS Subjects to emit progress events during the import pipeline:
* fetch → parse → deduplicate → save → auto-match
*
* Progress is exposed as an Observable so the controller can relay it via SSE.
* Only ONE import can run at a time (guarded by `isRunning`).
*/
@Injectable()
export class AttendanceImportService {
private readonly logger = new Logger(AttendanceImportService.name);
/** RxJS Subject emitting live progress during import */
private progressSubject = new Subject<ImportProgressEvent>();
private isRunning = false;
/** ID of the user who triggered the current import (for SSE scoping) */
private importingUserId?: number;
constructor(
@InjectRepository(DingAttendanceRaw)
private readonly dingRawRepo: Repository<DingAttendanceRaw>,
@InjectRepository(Student)
private readonly studentRepo: Repository<Student>,
@InjectRepository(StudentDingMapping)
private readonly studentDingMappingRepo: Repository<StudentDingMapping>,
private readonly dingTalkService: DingTalkService,
private readonly attendanceService: AttendanceService,
) {}
/**
* Expose progress as a read-only Observable for SSE.
*/
get progress$(): Observable<ImportProgressEvent> {
return this.progressSubject.asObservable();
}
/**
* Whether an import is currently in progress.
*/
get running(): boolean {
return this.isRunning;
}
/**
* Run the DingTalk attendance import pipeline.
*
* Pipeline stages:
* 1. Fetch attendance results from DingTalk (paginated)
* 2. Parse and validate each record
* 3. Deduplicate by `dingId` (unique in DB)
* 5. Optionally auto-match to students by name
*/
async importFromDingTalk(params: {
startDate: string;
endDate: string;
userIds?: string[];
autoMatch?: boolean;
/** ID of the HTTP user triggering the import (for SSE event scoping) */
userId?: number;
}): Promise<ImportResult> {
if (this.isRunning) {
throw new Error('An import is already in progress');
}
const startedAt = Date.now();
this.isRunning = true;
this.importingUserId = params.userId;
// Safety timeout: auto-reset isRunning after 30 minutes in case of
// an unhandled exception that bypasses the finally block (extremely rare).
const SAFETY_TIMEOUT_MS = 30 * 60 * 1000;
const safetyTimer = setTimeout(() => {
if (this.isRunning) {
this.logger.error('Import safety timeout triggered — force-resetting isRunning');
this.isRunning = false;
}
}, SAFETY_TIMEOUT_MS);
const errors: string[] = [];
let imported = 0;
let skipped = 0;
let matched = 0;
try {
// Stage 1: Fetch
this.emit('fetching', 0, 0, 'Fetching attendance results from DingTalk...');
const rawResults = await this.fetchAllPages(params);
const total = rawResults.length;
this.emit('fetching', total, total, `Fetched ${total} raw attendance records`);
// Stage 2: Parse & deduplicate
this.emit('parsing', 0, total, `Parsing ${total} records...`);
const existingByDingId = await this.getExistingRecordsByDingId(rawResults);
const newRecords = rawResults.filter((r) => !existingByDingId.has(r.checkId));
const duplicateRecords = rawResults.filter((r) => existingByDingId.has(r.checkId));
await this.refreshDuplicatePunchMetadata(duplicateRecords, existingByDingId);
skipped = rawResults.length - newRecords.length;
this.emit('parsing', newRecords.length, total, `${newRecords.length} new records, ${skipped} duplicates skipped`);
if (newRecords.length === 0) {
if (params.autoMatch) matched = await this.autoMatchUnmatched();
this.emit('complete', imported + skipped, total, 'Nothing new to import');
return { success: true, imported, skipped, matched, errors, duration: Date.now() - startedAt };
}
// Stage 3: Batch save
this.emit('saving', 0, newRecords.length, `Saving ${newRecords.length} records...`);
const batchSize = 100;
for (let i = 0; i < newRecords.length; i += batchSize) {
const batch = newRecords.slice(i, i + batchSize);
const entities = await Promise.all(batch.map((record) => this.mapToEntity(record)));
try {
await this.dingRawRepo.save(entities, { chunk: 50 });
imported += entities.length;
this.emit('saving', imported, newRecords.length, `Saved ${imported}/${newRecords.length}`);
} catch (err: unknown) {
const msg = err instanceof Error ? err.message : String(err);
errors.push(`Batch save error at offset ${i}: ${msg}`);
this.logger.error(`Batch save error: ${msg}`);
}
}
// Stage 4: Auto-match (optional)
if (params.autoMatch && imported > 0) {
this.emit('matching', 0, imported, 'Auto-matching records to students...');
matched = await this.autoMatchUnmatched();
this.emit('matching', matched, imported, `Matched ${matched} records to students`);
}
const duration = Date.now() - startedAt;
this.emit('complete', imported, rawResults.length, `Import complete: ${imported} new, ${skipped} skipped, ${matched} matched (${duration}ms)`);
this.logger.log(`DingTalk attendance import done: ${imported} imported, ${skipped} skipped, ${matched} matched`);
return { success: true, imported, skipped, matched, errors, duration };
} catch (err: unknown) {
const msg = err instanceof Error ? err.message : String(err);
errors.push(msg);
this.emit('error', imported, 0, `Import failed: ${msg}`, msg);
this.logger.error(`DingTalk attendance import failed: ${msg}`);
return { success: false, imported, skipped, matched, errors, duration: Date.now() - startedAt };
} finally {
clearTimeout(safetyTimer);
this.isRunning = false;
this.importingUserId = undefined;
}
}
/**
* DingTalk requires userIds, accepts at most 50 users per request, and
* allows a maximum inclusive date range of 7 calendar days.
*/
private async fetchAllPages(params: {
startDate: string;
endDate: string;
userIds?: string[];
}): Promise<DingTalkAttendanceResult[]> {
const userIds = [...new Set((params.userIds ?? []).filter(Boolean))];
if (userIds.length === 0) {
throw new BadRequestException('拉取钉钉考勤必须指定人员范围');
}
if (params.startDate > params.endDate) {
throw new BadRequestException('开始日期不能晚于结束日期');
}
const allResults: DingTalkAttendanceResult[] = [];
const userBatches = this.chunk(userIds, 50);
const dateRanges = this.splitDateRanges(params.startDate, params.endDate, 7);
const totalRequests = userBatches.length * dateRanges.length;
let completedRequests = 0;
for (const range of dateRanges) {
for (const users of userBatches) {
const batch = await this.dingTalkService.fetchAttendanceResults({
startDate: range.startDate,
endDate: range.endDate,
userIds: users,
});
allResults.push(...batch);
completedRequests++;
this.emit(
'fetching',
completedRequests,
totalRequests,
`已完成 ${completedRequests}/${totalRequests} 批,获取 ${allResults.length} 条记录`,
);
}
}
return allResults;
}
private chunk<T>(items: T[], size: number): T[][] {
const result: T[][] = [];
for (let index = 0; index < items.length; index += size) {
result.push(items.slice(index, index + size));
}
return result;
}
private splitDateRanges(
startDate: string,
endDate: string,
maxDays: number,
): Array<{ startDate: string; endDate: string }> {
const ranges: Array<{ startDate: string; endDate: string }> = [];
let cursor = this.parseDate(startDate);
const end = this.parseDate(endDate);
while (cursor.getTime() <= end.getTime()) {
const rangeEnd = new Date(cursor);
rangeEnd.setUTCDate(rangeEnd.getUTCDate() + maxDays - 1);
if (rangeEnd.getTime() > end.getTime()) rangeEnd.setTime(end.getTime());
ranges.push({
startDate: this.formatDate(cursor),
endDate: this.formatDate(rangeEnd),
});
cursor = new Date(rangeEnd);
cursor.setUTCDate(cursor.getUTCDate() + 1);
}
return ranges;
}
private parseDate(value: string): Date {
const date = new Date(`${value}T00:00:00.000Z`);
if (Number.isNaN(date.getTime())) {
throw new BadRequestException(`无效日期: ${value}`);
}
return date;
}
private formatDate(value: Date): string {
return value.toISOString().slice(0, 10);
}
/**
* Query which dingIds already exist to skip duplicates.
*/
private async getExistingRecordsByDingId(
results: DingTalkAttendanceResult[],
): Promise<Map<string, DingAttendanceRaw>> {
const dingIds = results.map((r) => r.checkId).filter(Boolean);
if (dingIds.length === 0) return new Map();
const existing = await this.dingRawRepo.find({
where: { dingId: In(dingIds) },
});
return new Map(existing.map((entity) => [entity.dingId, entity]));
}
private async refreshDuplicatePunchMetadata(
results: DingTalkAttendanceResult[],
existingByDingId: Map<string, DingAttendanceRaw>,
): Promise<void> {
const changed: DingAttendanceRaw[] = [];
for (const result of results) {
const entity = existingByDingId.get(result.checkId);
if (!entity) continue;
const punchSource = result.sourceType || entity.punchSource || null;
const punchDeviceName = result.deviceName || entity.punchDeviceName || null;
const punchDeviceId = result.deviceId || entity.punchDeviceId || null;
if (
entity.punchSource === punchSource &&
entity.punchDeviceName === punchDeviceName &&
entity.punchDeviceId === punchDeviceId
) {
continue;
}
entity.punchSource = punchSource;
entity.punchDeviceName = punchDeviceName;
entity.punchDeviceId = punchDeviceId;
entity.rawData = JSON.stringify(result);
changed.push(entity);
}
if (changed.length > 0) {
await this.dingRawRepo.save(changed, { chunk: 50 });
}
}
/**
* Map a DingTalk API result to a DingAttendanceRaw entity.
*/
private async mapToEntity(r: DingTalkAttendanceResult): Promise<DingAttendanceRaw> {
const entity = new DingAttendanceRaw();
entity.dingUserId = r.userId;
entity.userName = r.userName || await this.resolveStudentName(r.userId);
entity.attendanceDate = r.workDate;
entity.dingId = r.checkId;
entity.attendanceType = r.checkType || 'OnDuty';
entity.timeResult = r.timeResult;
entity.locationResult = r.locationResult || '';
entity.punchSource = r.sourceType || null;
entity.punchDeviceName = r.deviceName || null;
entity.punchDeviceId = r.deviceId || null;
// Parse check-in/out times
if (r.actualCheckTime) {
const dt = new Date(r.actualCheckTime);
if (!isNaN(dt.getTime())) {
if (r.checkType === 'OnDuty') {
entity.checkInTime = dt;
} else {
entity.checkOutTime = dt;
}
}
}
entity.matchStatus = 'unmatched';
entity.rawData = JSON.stringify(r);
return entity;
}
private async resolveStudentName(dingUserId: string): Promise<string> {
const mapping = await this.studentDingMappingRepo.findOne({
where: { dingUserId },
});
if (!mapping) return '';
const student = await this.studentRepo.findOne({ where: { id: mapping.studentId } });
return student?.name || '';
}
/**
* Auto-match unmatched records to students via the dingUserId → userId mapping chain.
*
* Delegates to {@link AttendanceService.autoMatchDingRecords} to avoid duplicate logic.
*/
private async autoMatchUnmatched(): Promise<number> {
const result = await this.attendanceService.autoMatchDingRecords();
return result.matched;
}
/**
* Emit a progress event to the stream.
*/
private emit(
phase: ImportProgressEvent['phase'],
current: number,
total: number,
message: string,
error?: string,
): void {
this.progressSubject.next({ phase, current, total, message, error, userId: this.importingUserId });
}
}