feat(iit): Implement event-level QC architecture V3.1 with dynamic rule filtering, report deduplication and AI intent enhancement
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
979
backend/src/modules/iit-manager/services/QcReportService.ts
Normal file
979
backend/src/modules/iit-manager/services/QcReportService.ts
Normal file
@@ -0,0 +1,979 @@
|
||||
/**
|
||||
* QcReportService - 质控报告生成服务
|
||||
*
|
||||
* 功能：
|
||||
* - 聚合质控统计数据
|
||||
* - 生成 LLM 友好的 XML 报告
|
||||
* - 缓存报告以提高查询效率
|
||||
*
|
||||
* 设计原则：
|
||||
* - 报告驱动：预计算报告，LLM 只需阅读
|
||||
* - 双模输出：人类可读 + LLM 友好格式
|
||||
* - 智能缓存：报告有效期内直接返回缓存
|
||||
*/
|
||||
|
||||
import { PrismaClient } from '@prisma/client';
|
||||
import { logger } from '../../../common/logging/index.js';
|
||||
|
||||
const prisma = new PrismaClient();
|
||||
|
||||
// ============================================================
|
||||
// 类型定义
|
||||
// ============================================================
|
||||
|
||||
/**
|
||||
* 报告摘要
|
||||
*/
|
||||
export interface ReportSummary {
|
||||
totalRecords: number;
|
||||
completedRecords: number;
|
||||
criticalIssues: number;
|
||||
warningIssues: number;
|
||||
pendingQueries: number;
|
||||
passRate: number;
|
||||
lastQcTime: string | null;
|
||||
}
|
||||
|
||||
/**
|
||||
* 问题项
|
||||
*/
|
||||
export interface ReportIssue {
|
||||
recordId: string;
|
||||
ruleId: string;
|
||||
ruleName: string;
|
||||
severity: 'critical' | 'warning' | 'info';
|
||||
message: string;
|
||||
field?: string;
|
||||
actualValue?: any;
|
||||
expectedValue?: any;
|
||||
evidence?: Record<string, any>;
|
||||
detectedAt: string;
|
||||
}
|
||||
|
||||
/**
|
||||
* 表单统计
|
||||
*/
|
||||
export interface FormStats {
|
||||
formName: string;
|
||||
formLabel: string;
|
||||
totalChecks: number;
|
||||
passed: number;
|
||||
failed: number;
|
||||
passRate: number;
|
||||
}
|
||||
|
||||
/**
|
||||
* 质控报告
|
||||
*/
|
||||
export interface QcReport {
|
||||
projectId: string;
|
||||
reportType: 'daily' | 'weekly' | 'on_demand';
|
||||
generatedAt: string;
|
||||
expiresAt: string | null;
|
||||
summary: ReportSummary;
|
||||
criticalIssues: ReportIssue[];
|
||||
warningIssues: ReportIssue[];
|
||||
formStats: FormStats[];
|
||||
topIssues: TopIssue[]; // V2.1: Top 问题统计
|
||||
groupedIssues: GroupedIssues[]; // V2.1: 按受试者分组
|
||||
llmFriendlyXml: string; // V2.1: LLM 友好格式
|
||||
legacyXml?: string; // V2.1: 兼容旧格式
|
||||
}
|
||||
|
||||
/**
|
||||
* 报告选项
|
||||
*/
|
||||
export interface ReportOptions {
|
||||
forceRefresh?: boolean; // 强制刷新，忽略缓存
|
||||
reportType?: 'daily' | 'weekly' | 'on_demand';
|
||||
expirationHours?: number; // 报告有效期（小时）
|
||||
format?: 'xml' | 'llm-friendly'; // V2.1: 输出格式
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 按受试者分组的问题
|
||||
*/
|
||||
export interface GroupedIssues {
|
||||
recordId: string;
|
||||
issueCount: number;
|
||||
issues: ReportIssue[];
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: Top 问题统计
|
||||
*/
|
||||
export interface TopIssue {
|
||||
ruleName: string;
|
||||
ruleId: string;
|
||||
count: number;
|
||||
affectedRecords: string[];
|
||||
}
|
||||
|
||||
// ============================================================
|
||||
// QcReportService 实现
|
||||
// ============================================================
|
||||
|
||||
class QcReportServiceClass {
|
||||
/**
|
||||
* 获取项目质控报告
|
||||
*
|
||||
* @param projectId 项目 ID
|
||||
* @param options 报告选项
|
||||
* @returns 质控报告
|
||||
*/
|
||||
async getReport(projectId: string, options?: ReportOptions): Promise<QcReport> {
|
||||
const reportType = options?.reportType || 'on_demand';
|
||||
const expirationHours = options?.expirationHours || 24;
|
||||
|
||||
// 1. 检查缓存
|
||||
if (!options?.forceRefresh) {
|
||||
const cached = await this.getCachedReport(projectId, reportType);
|
||||
if (cached) {
|
||||
logger.debug('[QcReportService] Returning cached report', {
|
||||
projectId,
|
||||
reportType,
|
||||
generatedAt: cached.generatedAt,
|
||||
});
|
||||
return cached;
|
||||
}
|
||||
}
|
||||
|
||||
// 2. 生成新报告
|
||||
logger.info('[QcReportService] Generating new report', {
|
||||
projectId,
|
||||
reportType,
|
||||
});
|
||||
|
||||
const report = await this.generateReport(projectId, reportType, expirationHours);
|
||||
|
||||
// 3. 缓存报告
|
||||
await this.cacheReport(report);
|
||||
|
||||
return report;
|
||||
}
|
||||
|
||||
/**
|
||||
* 获取缓存的报告
|
||||
*/
|
||||
private async getCachedReport(
|
||||
projectId: string,
|
||||
reportType: string
|
||||
): Promise<QcReport | null> {
|
||||
try {
|
||||
const cached = await prisma.iitQcReport.findFirst({
|
||||
where: {
|
||||
projectId,
|
||||
reportType,
|
||||
OR: [
|
||||
{ expiresAt: null },
|
||||
{ expiresAt: { gt: new Date() } },
|
||||
],
|
||||
},
|
||||
orderBy: { generatedAt: 'desc' },
|
||||
});
|
||||
|
||||
if (!cached) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const issuesData = cached.issues as any || {};
|
||||
|
||||
return {
|
||||
projectId: cached.projectId,
|
||||
reportType: cached.reportType as 'daily' | 'weekly' | 'on_demand',
|
||||
generatedAt: cached.generatedAt.toISOString(),
|
||||
expiresAt: cached.expiresAt?.toISOString() || null,
|
||||
summary: cached.summary as unknown as ReportSummary,
|
||||
criticalIssues: (issuesData.critical || []) as ReportIssue[],
|
||||
warningIssues: (issuesData.warning || []) as ReportIssue[],
|
||||
formStats: (issuesData.formStats || []) as FormStats[],
|
||||
topIssues: (issuesData.topIssues || []) as TopIssue[],
|
||||
groupedIssues: (issuesData.groupedIssues || []) as GroupedIssues[],
|
||||
llmFriendlyXml: cached.llmReport,
|
||||
legacyXml: issuesData.legacyXml,
|
||||
};
|
||||
} catch (error: any) {
|
||||
logger.warn('[QcReportService] Failed to get cached report', {
|
||||
projectId,
|
||||
error: error.message,
|
||||
});
|
||||
return null;
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 生成新报告
|
||||
*
|
||||
* V2.1 优化：支持双格式输出
|
||||
*/
|
||||
private async generateReport(
|
||||
projectId: string,
|
||||
reportType: 'daily' | 'weekly' | 'on_demand',
|
||||
expirationHours: number
|
||||
): Promise<QcReport> {
|
||||
const startTime = Date.now();
|
||||
|
||||
// 1. 获取项目信息
|
||||
const project = await prisma.iitProject.findUnique({
|
||||
where: { id: projectId },
|
||||
select: {
|
||||
id: true,
|
||||
name: true,
|
||||
fieldMappings: true,
|
||||
},
|
||||
});
|
||||
|
||||
if (!project) {
|
||||
throw new Error(`项目不存在: ${projectId}`);
|
||||
}
|
||||
|
||||
// 2. 聚合质控统计
|
||||
const summary = await this.aggregateStats(projectId);
|
||||
|
||||
// 3. 获取问题列表
|
||||
const { criticalIssues, warningIssues } = await this.getIssues(projectId);
|
||||
|
||||
// 4. 获取表单统计
|
||||
const formStats = await this.getFormStats(projectId);
|
||||
|
||||
// 5. V2.1: 计算 Top Issues 和分组
|
||||
const allIssues = [...criticalIssues, ...warningIssues];
|
||||
const topIssues = this.calculateTopIssues(allIssues);
|
||||
const groupedIssues = this.groupIssuesByRecord(criticalIssues);
|
||||
|
||||
// 6. V2.1: 生成双格式 XML 报告
|
||||
const llmFriendlyXml = this.buildLlmXmlReport(
|
||||
projectId,
|
||||
project.name || projectId,
|
||||
summary,
|
||||
criticalIssues,
|
||||
warningIssues,
|
||||
formStats
|
||||
);
|
||||
|
||||
const legacyXml = this.buildLegacyXmlReport(
|
||||
projectId,
|
||||
project.name || projectId,
|
||||
summary,
|
||||
criticalIssues,
|
||||
warningIssues,
|
||||
formStats
|
||||
);
|
||||
|
||||
const generatedAt = new Date();
|
||||
const expiresAt = new Date(generatedAt.getTime() + expirationHours * 60 * 60 * 1000);
|
||||
|
||||
const duration = Date.now() - startTime;
|
||||
logger.info('[QcReportService] Report generated', {
|
||||
projectId,
|
||||
reportType,
|
||||
duration: `${duration}ms`,
|
||||
criticalCount: criticalIssues.length,
|
||||
warningCount: warningIssues.length,
|
||||
topIssuesCount: topIssues.length,
|
||||
groupedRecordCount: groupedIssues.length,
|
||||
});
|
||||
|
||||
return {
|
||||
projectId,
|
||||
reportType,
|
||||
generatedAt: generatedAt.toISOString(),
|
||||
expiresAt: expiresAt.toISOString(),
|
||||
summary,
|
||||
criticalIssues,
|
||||
warningIssues,
|
||||
formStats,
|
||||
topIssues,
|
||||
groupedIssues,
|
||||
llmFriendlyXml,
|
||||
legacyXml,
|
||||
};
|
||||
}
|
||||
|
||||
/**
|
||||
* 聚合质控统计
|
||||
*
|
||||
* V3.1: 修复记录数统计和 issues 格式兼容性
|
||||
*/
|
||||
private async aggregateStats(projectId: string): Promise<ReportSummary> {
|
||||
// 获取记录汇总（用于 completedRecords 统计）
|
||||
const recordSummaries = await prisma.iitRecordSummary.findMany({
|
||||
where: { projectId },
|
||||
});
|
||||
|
||||
const completedRecords = recordSummaries.filter(r =>
|
||||
r.completionRate && (r.completionRate as number) >= 100
|
||||
).length;
|
||||
|
||||
// V3.1: 获取每个 record+event 的最新质控日志（避免重复）
|
||||
const latestQcLogs = await prisma.$queryRaw<Array<{
|
||||
record_id: string;
|
||||
event_id: string | null;
|
||||
form_name: string | null;
|
||||
status: string;
|
||||
issues: any;
|
||||
created_at: Date;
|
||||
}>>`
|
||||
SELECT DISTINCT ON (record_id, COALESCE(event_id, ''))
|
||||
record_id, event_id, form_name, status, issues, created_at
|
||||
FROM iit_schema.qc_logs
|
||||
WHERE project_id = ${projectId}
|
||||
ORDER BY record_id, COALESCE(event_id, ''), created_at DESC
|
||||
`;
|
||||
|
||||
// V3.1: 从质控日志获取独立 record_id 数量
|
||||
const uniqueRecordIds = new Set(latestQcLogs.map(log => log.record_id));
|
||||
const totalRecords = uniqueRecordIds.size;
|
||||
|
||||
// V3.1: 统计问题数量（按 recordId + ruleId 去重）
|
||||
const seenCritical = new Set<string>();
|
||||
const seenWarning = new Set<string>();
|
||||
let pendingQueries = 0;
|
||||
|
||||
for (const log of latestQcLogs) {
|
||||
// V3.1: 兼容两种 issues 格式
|
||||
const rawIssues = log.issues as any;
|
||||
let issues: any[] = [];
|
||||
|
||||
if (Array.isArray(rawIssues)) {
|
||||
issues = rawIssues;
|
||||
} else if (rawIssues && Array.isArray(rawIssues.items)) {
|
||||
issues = rawIssues.items;
|
||||
}
|
||||
|
||||
for (const issue of issues) {
|
||||
const ruleId = issue.ruleId || issue.ruleName || 'unknown';
|
||||
const key = `${log.record_id}:${ruleId}`;
|
||||
const severity = issue.severity || issue.level;
|
||||
|
||||
if (severity === 'critical' || severity === 'RED' || severity === 'error') {
|
||||
seenCritical.add(key);
|
||||
} else if (severity === 'warning' || severity === 'YELLOW') {
|
||||
seenWarning.add(key);
|
||||
}
|
||||
}
|
||||
|
||||
if (log.status === 'UNCERTAIN' || log.status === 'PENDING') {
|
||||
pendingQueries++;
|
||||
}
|
||||
}
|
||||
|
||||
const criticalIssues = seenCritical.size;
|
||||
const warningIssues = seenWarning.size;
|
||||
|
||||
// 计算通过率
|
||||
const passedRecords = latestQcLogs.filter(log =>
|
||||
log.status === 'PASS' || log.status === 'GREEN'
|
||||
).length;
|
||||
const passRate = totalRecords > 0
|
||||
? Math.round((passedRecords / totalRecords) * 100 * 10) / 10
|
||||
: 0;
|
||||
|
||||
// 获取最后质控时间
|
||||
const lastQcLog = await prisma.iitQcLog.findFirst({
|
||||
where: { projectId },
|
||||
orderBy: { createdAt: 'desc' },
|
||||
select: { createdAt: true },
|
||||
});
|
||||
|
||||
return {
|
||||
totalRecords,
|
||||
completedRecords,
|
||||
criticalIssues,
|
||||
warningIssues,
|
||||
pendingQueries,
|
||||
passRate,
|
||||
lastQcTime: lastQcLog?.createdAt?.toISOString() || null,
|
||||
};
|
||||
}
|
||||
|
||||
/**
|
||||
* 获取问题列表
|
||||
*/
|
||||
private async getIssues(projectId: string): Promise<{
|
||||
criticalIssues: ReportIssue[];
|
||||
warningIssues: ReportIssue[];
|
||||
}> {
|
||||
// V3.1: 获取每个 record+event 的最新质控日志
|
||||
const latestQcLogs = await prisma.$queryRaw<Array<{
|
||||
record_id: string;
|
||||
event_id: string | null;
|
||||
form_name: string | null;
|
||||
status: string;
|
||||
issues: any;
|
||||
created_at: Date;
|
||||
}>>`
|
||||
SELECT DISTINCT ON (record_id, COALESCE(event_id, ''))
|
||||
record_id, event_id, form_name, status, issues, created_at
|
||||
FROM iit_schema.qc_logs
|
||||
WHERE project_id = ${projectId}
|
||||
ORDER BY record_id, COALESCE(event_id, ''), created_at DESC
|
||||
`;
|
||||
|
||||
const criticalIssues: ReportIssue[] = [];
|
||||
const warningIssues: ReportIssue[] = [];
|
||||
|
||||
for (const log of latestQcLogs) {
|
||||
// V2.1: 兼容两种 issues 格式
|
||||
// 新格式: { items: [...], summary: {...} }
|
||||
// 旧格式: [...]
|
||||
const rawIssues = log.issues as any;
|
||||
let issues: any[] = [];
|
||||
|
||||
if (Array.isArray(rawIssues)) {
|
||||
// 旧格式：直接是数组
|
||||
issues = rawIssues;
|
||||
} else if (rawIssues && Array.isArray(rawIssues.items)) {
|
||||
// 新格式：对象包含 items 数组
|
||||
issues = rawIssues.items;
|
||||
} else {
|
||||
continue;
|
||||
}
|
||||
|
||||
for (const issue of issues) {
|
||||
// V2.1: 构建自包含的 LLM 友好消息
|
||||
const llmMessage = this.buildSelfContainedMessage(issue);
|
||||
|
||||
const reportIssue: ReportIssue = {
|
||||
recordId: log.record_id,
|
||||
ruleId: issue.ruleId || issue.ruleName || 'unknown',
|
||||
ruleName: issue.ruleName || issue.message || 'Unknown Rule',
|
||||
severity: this.normalizeSeverity(issue.severity || issue.level),
|
||||
message: llmMessage, // V2.1: 使用自包含消息
|
||||
field: issue.field,
|
||||
actualValue: issue.actualValue,
|
||||
expectedValue: issue.expectedValue || this.extractExpectedFromMessage(issue.message),
|
||||
evidence: issue.evidence,
|
||||
detectedAt: log.created_at.toISOString(),
|
||||
};
|
||||
|
||||
if (reportIssue.severity === 'critical') {
|
||||
criticalIssues.push(reportIssue);
|
||||
} else if (reportIssue.severity === 'warning') {
|
||||
warningIssues.push(reportIssue);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// V3.1: 按 recordId + ruleId 去重，保留最新的问题（避免多事件重复）
|
||||
const deduplicateCritical = this.deduplicateIssues(criticalIssues);
|
||||
const deduplicateWarning = this.deduplicateIssues(warningIssues);
|
||||
|
||||
return { criticalIssues: deduplicateCritical, warningIssues: deduplicateWarning };
|
||||
}
|
||||
|
||||
/**
|
||||
* V3.1: 按 recordId + ruleId 去重问题
|
||||
*
|
||||
* 同一记录的同一规则只保留一条（最新的，因为数据已按时间倒序排列）
|
||||
*/
|
||||
private deduplicateIssues(issues: ReportIssue[]): ReportIssue[] {
|
||||
const seen = new Map<string, ReportIssue>();
|
||||
|
||||
for (const issue of issues) {
|
||||
const key = `${issue.recordId}:${issue.ruleId}`;
|
||||
if (!seen.has(key)) {
|
||||
seen.set(key, issue);
|
||||
}
|
||||
// 如果已存在，跳过（因为按时间倒序，第一个就是最新的）
|
||||
}
|
||||
|
||||
return Array.from(seen.values());
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 构建自包含的 LLM 友好消息
|
||||
*/
|
||||
private buildSelfContainedMessage(issue: any): string {
|
||||
const ruleName = issue.ruleName || issue.message || 'Unknown';
|
||||
const actualValue = issue.actualValue;
|
||||
const expectedValue = issue.expectedValue || this.extractExpectedFromMessage(issue.message);
|
||||
|
||||
// 如果已经有自包含格式，直接返回
|
||||
if (issue.llmMessage) {
|
||||
return issue.llmMessage;
|
||||
}
|
||||
|
||||
// 构建自包含格式
|
||||
const actualDisplay = actualValue !== undefined && actualValue !== null && actualValue !== ''
|
||||
? `**${actualValue}**`
|
||||
: '**空**';
|
||||
|
||||
if (expectedValue) {
|
||||
return `**${ruleName}**: 当前值 ${actualDisplay} (标准: ${expectedValue})`;
|
||||
}
|
||||
|
||||
return `**${ruleName}**: 当前值 ${actualDisplay}`;
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 从原始消息中提取期望值
|
||||
* 例如："年龄不在 25-35 岁范围内" -> "25-35 岁"
|
||||
*/
|
||||
private extractExpectedFromMessage(message: string): string | undefined {
|
||||
if (!message) return undefined;
|
||||
|
||||
// 尝试提取数字范围：如 "25-35"
|
||||
const rangeMatch = message.match(/(\d+)\s*[-~至到]\s*(\d+)/);
|
||||
if (rangeMatch) {
|
||||
return `${rangeMatch[1]}-${rangeMatch[2]}`;
|
||||
}
|
||||
|
||||
// 尝试提取日期范围
|
||||
const dateRangeMatch = message.match(/(\d{4}-\d{2}-\d{2})\s*至\s*(\d{4}-\d{2}-\d{2})/);
|
||||
if (dateRangeMatch) {
|
||||
return `${dateRangeMatch[1]} 至 ${dateRangeMatch[2]}`;
|
||||
}
|
||||
|
||||
return undefined;
|
||||
}
|
||||
|
||||
/**
|
||||
* 标准化严重程度
|
||||
*/
|
||||
private normalizeSeverity(severity: string): 'critical' | 'warning' | 'info' {
|
||||
const lower = (severity || '').toLowerCase();
|
||||
if (lower === 'critical' || lower === 'red' || lower === 'error') {
|
||||
return 'critical';
|
||||
} else if (lower === 'warning' || lower === 'yellow') {
|
||||
return 'warning';
|
||||
}
|
||||
return 'info';
|
||||
}
|
||||
|
||||
/**
|
||||
* 获取表单统计
|
||||
*/
|
||||
private async getFormStats(projectId: string): Promise<FormStats[]> {
|
||||
// 获取表单标签映射
|
||||
const project = await prisma.iitProject.findUnique({
|
||||
where: { id: projectId },
|
||||
select: { fieldMappings: true },
|
||||
});
|
||||
const formLabels: Record<string, string> =
|
||||
((project?.fieldMappings as any)?.formLabels) || {};
|
||||
|
||||
// 按表单统计
|
||||
const formStatsRaw = await prisma.$queryRaw<Array<{
|
||||
form_name: string;
|
||||
total: bigint;
|
||||
passed: bigint;
|
||||
failed: bigint;
|
||||
}>>`
|
||||
SELECT
|
||||
form_name,
|
||||
COUNT(*) as total,
|
||||
COUNT(*) FILTER (WHERE status = 'PASS' OR status = 'GREEN') as passed,
|
||||
COUNT(*) FILTER (WHERE status = 'FAIL' OR status = 'RED') as failed
|
||||
FROM (
|
||||
SELECT DISTINCT ON (record_id, form_name)
|
||||
record_id, form_name, status
|
||||
FROM iit_schema.qc_logs
|
||||
WHERE project_id = ${projectId} AND form_name IS NOT NULL
|
||||
ORDER BY record_id, form_name, created_at DESC
|
||||
) latest_logs
|
||||
GROUP BY form_name
|
||||
ORDER BY form_name
|
||||
`;
|
||||
|
||||
return formStatsRaw.map(row => {
|
||||
const total = Number(row.total);
|
||||
const passed = Number(row.passed);
|
||||
const failed = Number(row.failed);
|
||||
return {
|
||||
formName: row.form_name,
|
||||
formLabel: formLabels[row.form_name] || this.formatFormName(row.form_name),
|
||||
totalChecks: total,
|
||||
passed,
|
||||
failed,
|
||||
passRate: total > 0 ? Math.round((passed / total) * 100 * 10) / 10 : 0,
|
||||
};
|
||||
});
|
||||
}
|
||||
|
||||
/**
|
||||
* 格式化表单名称
|
||||
*/
|
||||
private formatFormName(formName: string): string {
|
||||
return formName
|
||||
.replace(/_/g, ' ')
|
||||
.replace(/\b\w/g, c => c.toUpperCase());
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 按受试者分组问题
|
||||
*/
|
||||
private groupIssuesByRecord(issues: ReportIssue[]): GroupedIssues[] {
|
||||
const grouped = new Map<string, ReportIssue[]>();
|
||||
|
||||
for (const issue of issues) {
|
||||
const existing = grouped.get(issue.recordId) || [];
|
||||
existing.push(issue);
|
||||
grouped.set(issue.recordId, existing);
|
||||
}
|
||||
|
||||
return Array.from(grouped.entries())
|
||||
.map(([recordId, issues]) => ({
|
||||
recordId,
|
||||
issueCount: issues.length,
|
||||
issues,
|
||||
}))
|
||||
.sort((a, b) => a.recordId.localeCompare(b.recordId, undefined, { numeric: true }));
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 计算 Top Issues 统计
|
||||
*/
|
||||
private calculateTopIssues(issues: ReportIssue[], limit: number = 5): TopIssue[] {
|
||||
const ruleStats = new Map<string, { ruleName: string; ruleId: string; records: Set<string> }>();
|
||||
|
||||
for (const issue of issues) {
|
||||
const key = issue.ruleId || issue.ruleName;
|
||||
const existing = ruleStats.get(key) || {
|
||||
ruleName: issue.ruleName,
|
||||
ruleId: issue.ruleId,
|
||||
records: new Set<string>(),
|
||||
};
|
||||
existing.records.add(issue.recordId);
|
||||
ruleStats.set(key, existing);
|
||||
}
|
||||
|
||||
return Array.from(ruleStats.values())
|
||||
.map(stat => ({
|
||||
ruleName: stat.ruleName,
|
||||
ruleId: stat.ruleId,
|
||||
count: stat.records.size,
|
||||
affectedRecords: Array.from(stat.records),
|
||||
}))
|
||||
.sort((a, b) => b.count - a.count)
|
||||
.slice(0, limit);
|
||||
}
|
||||
|
||||
/**
|
||||
* 构建 LLM 友好的 XML 报告
|
||||
*
|
||||
* V2.1 优化：
|
||||
* - 按受试者分组
|
||||
* - 添加 Top Issues 统计
|
||||
* - 自包含的 message 格式
|
||||
*/
|
||||
private buildLlmXmlReport(
|
||||
projectId: string,
|
||||
projectName: string,
|
||||
summary: ReportSummary,
|
||||
criticalIssues: ReportIssue[],
|
||||
warningIssues: ReportIssue[],
|
||||
formStats: FormStats[]
|
||||
): string {
|
||||
const now = new Date().toISOString();
|
||||
|
||||
// V2.1: 计算 Top Issues
|
||||
const allIssues = [...criticalIssues, ...warningIssues];
|
||||
const topIssues = this.calculateTopIssues(allIssues);
|
||||
|
||||
// V2.1: 按受试者分组
|
||||
const groupedCritical = this.groupIssuesByRecord(criticalIssues);
|
||||
const failedRecordCount = groupedCritical.length;
|
||||
|
||||
let xml = `<?xml version="1.0" encoding="UTF-8"?>
|
||||
<qc_context project_id="${projectId}" project_name="${this.escapeXml(projectName)}" generated="${now}">
|
||||
|
||||
<!-- 1. 宏观统计 (Aggregate) -->
|
||||
<summary>
|
||||
- 状态: ${failedRecordCount}/${summary.totalRecords} 记录存在严重违规 (${summary.totalRecords > 0 ? Math.round((failedRecordCount / summary.totalRecords) * 100) : 0}% Fail)
|
||||
- 通过率: ${summary.passRate}%
|
||||
- 严重问题: ${summary.criticalIssues} | 警告: ${summary.warningIssues}
|
||||
${topIssues.length > 0 ? ` - Top ${Math.min(3, topIssues.length)} 问题:
|
||||
${topIssues.slice(0, 3).map((t, i) => ` ${i + 1}. ${t.ruleName} (${t.count}人)`).join('\n')}` : ''}
|
||||
</summary>
|
||||
|
||||
`;
|
||||
|
||||
// V3.1: 严重问题详情（按受试者分组，显示所有问题）
|
||||
if (groupedCritical.length > 0) {
|
||||
xml += ` <!-- 2. 严重问题详情 (按受试者分组) -->\n`;
|
||||
xml += ` <critical_issues record_count="${groupedCritical.length}" issue_count="${criticalIssues.length}">\n\n`;
|
||||
|
||||
for (const group of groupedCritical) {
|
||||
xml += ` <record id="${group.recordId}">\n`;
|
||||
xml += ` **严重违规 (${group.issueCount}项)**:\n`;
|
||||
|
||||
// V3.1: 显示所有问题，不再限制
|
||||
for (let i = 0; i < group.issues.length; i++) {
|
||||
const issue = group.issues[i];
|
||||
const llmLine = this.buildLlmIssueLine(issue, i + 1);
|
||||
xml += ` ${llmLine}\n`;
|
||||
}
|
||||
|
||||
xml += ` </record>\n\n`;
|
||||
}
|
||||
|
||||
xml += ` </critical_issues>\n\n`;
|
||||
} else {
|
||||
xml += ` <critical_issues record_count="0" issue_count="0" />\n\n`;
|
||||
}
|
||||
|
||||
// V3.1: 警告问题（显示所有）
|
||||
const groupedWarning = this.groupIssuesByRecord(warningIssues);
|
||||
if (groupedWarning.length > 0) {
|
||||
xml += ` <!-- 3. 警告问题 -->\n`;
|
||||
xml += ` <warning_issues record_count="${groupedWarning.length}" issue_count="${warningIssues.length}">\n`;
|
||||
|
||||
for (const group of groupedWarning) {
|
||||
xml += ` <record id="${group.recordId}">\n`;
|
||||
xml += ` **警告 (${group.issueCount}项)**:\n`;
|
||||
for (let i = 0; i < group.issues.length; i++) {
|
||||
const issue = group.issues[i];
|
||||
const llmLine = this.buildLlmIssueLine(issue, i + 1);
|
||||
xml += ` ${llmLine}\n`;
|
||||
}
|
||||
xml += ` </record>\n`;
|
||||
}
|
||||
|
||||
xml += ` </warning_issues>\n\n`;
|
||||
}
|
||||
|
||||
xml += `</qc_context>`;
|
||||
|
||||
return xml;
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 构建 LLM 友好的单行问题描述
|
||||
*
|
||||
* 格式: [规则ID] **问题类型**: 当前 **值** (标准: xxx)
|
||||
*/
|
||||
private buildLlmIssueLine(issue: ReportIssue, index: number): string {
|
||||
const ruleId = issue.ruleId !== 'unknown' ? `[${issue.ruleId}]` : '';
|
||||
|
||||
// 尝试使用 llmMessage（如果 HardRuleEngine 已经生成）
|
||||
if (issue.message && issue.message.includes('**')) {
|
||||
// 已经是自包含格式
|
||||
return `${index}. ${ruleId} ${issue.message}`;
|
||||
}
|
||||
|
||||
// 回退：手动构建
|
||||
const actualDisplay = issue.actualValue !== undefined && issue.actualValue !== null && issue.actualValue !== ''
|
||||
? `**${issue.actualValue}**`
|
||||
: '**空**';
|
||||
|
||||
const expectedDisplay = issue.expectedValue
|
||||
? ` (标准: ${issue.expectedValue})`
|
||||
: '';
|
||||
|
||||
return `${index}. ${ruleId} **${issue.ruleName}**: 当前值 ${actualDisplay}${expectedDisplay}`;
|
||||
}
|
||||
|
||||
/**
|
||||
* 构建原始 XML 报告（兼容旧格式）
|
||||
*/
|
||||
private buildLegacyXmlReport(
|
||||
projectId: string,
|
||||
projectName: string,
|
||||
summary: ReportSummary,
|
||||
criticalIssues: ReportIssue[],
|
||||
warningIssues: ReportIssue[],
|
||||
formStats: FormStats[]
|
||||
): string {
|
||||
const now = new Date().toISOString();
|
||||
|
||||
let xml = `<?xml version="1.0" encoding="UTF-8"?>
|
||||
<qc_report project_id="${projectId}" project_name="${this.escapeXml(projectName)}" generated="${now}">
|
||||
|
||||
<summary>
|
||||
<total_records>${summary.totalRecords}</total_records>
|
||||
<completed_records>${summary.completedRecords}</completed_records>
|
||||
<critical_issues>${summary.criticalIssues}</critical_issues>
|
||||
<warning_issues>${summary.warningIssues}</warning_issues>
|
||||
<pending_queries>${summary.pendingQueries}</pending_queries>
|
||||
<pass_rate>${summary.passRate}%</pass_rate>
|
||||
<last_qc_time>${summary.lastQcTime || 'N/A'}</last_qc_time>
|
||||
</summary>
|
||||
|
||||
`;
|
||||
|
||||
// V3.1: 严重问题列表（显示所有）
|
||||
if (criticalIssues.length > 0) {
|
||||
xml += ` <critical_issues count="${criticalIssues.length}">\n`;
|
||||
for (const issue of criticalIssues) {
|
||||
xml += this.buildIssueXml(issue, ' ');
|
||||
}
|
||||
xml += ` </critical_issues>\n\n`;
|
||||
} else {
|
||||
xml += ` <critical_issues count="0" />\n\n`;
|
||||
}
|
||||
|
||||
// V3.1: 警告问题列表（显示所有）
|
||||
if (warningIssues.length > 0) {
|
||||
xml += ` <warning_issues count="${warningIssues.length}">\n`;
|
||||
for (const issue of warningIssues) {
|
||||
xml += this.buildIssueXml(issue, ' ');
|
||||
}
|
||||
xml += ` </warning_issues>\n\n`;
|
||||
} else {
|
||||
xml += ` <warning_issues count="0" />\n\n`;
|
||||
}
|
||||
|
||||
// 表单统计
|
||||
if (formStats.length > 0) {
|
||||
xml += ` <form_statistics>\n`;
|
||||
for (const form of formStats) {
|
||||
xml += ` <form name="${this.escapeXml(form.formName)}" label="${this.escapeXml(form.formLabel)}">\n`;
|
||||
xml += ` <total_checks>${form.totalChecks}</total_checks>\n`;
|
||||
xml += ` <passed>${form.passed}</passed>\n`;
|
||||
xml += ` <failed>${form.failed}</failed>\n`;
|
||||
xml += ` <pass_rate>${form.passRate}%</pass_rate>\n`;
|
||||
xml += ` </form>\n`;
|
||||
}
|
||||
xml += ` </form_statistics>\n\n`;
|
||||
}
|
||||
|
||||
xml += `</qc_report>`;
|
||||
|
||||
return xml;
|
||||
}
|
||||
|
||||
/**
|
||||
* 构建单个问题的 XML
|
||||
*/
|
||||
private buildIssueXml(issue: ReportIssue, indent: string): string {
|
||||
let xml = `${indent}<issue record="${issue.recordId}" rule="${this.escapeXml(issue.ruleId)}" severity="${issue.severity}">\n`;
|
||||
xml += `${indent} <rule_name>${this.escapeXml(issue.ruleName)}</rule_name>\n`;
|
||||
xml += `${indent} <message>${this.escapeXml(issue.message)}</message>\n`;
|
||||
|
||||
if (issue.field) {
|
||||
xml += `${indent} <field>${this.escapeXml(String(issue.field))}</field>\n`;
|
||||
}
|
||||
if (issue.actualValue !== undefined) {
|
||||
xml += `${indent} <actual_value>${this.escapeXml(String(issue.actualValue))}</actual_value>\n`;
|
||||
}
|
||||
if (issue.expectedValue !== undefined) {
|
||||
xml += `${indent} <expected_value>${this.escapeXml(String(issue.expectedValue))}</expected_value>\n`;
|
||||
}
|
||||
if (issue.evidence && Object.keys(issue.evidence).length > 0) {
|
||||
xml += `${indent} <evidence>\n`;
|
||||
for (const [key, value] of Object.entries(issue.evidence)) {
|
||||
xml += `${indent} <${this.escapeXml(key)}>${this.escapeXml(String(value))}</${this.escapeXml(key)}>\n`;
|
||||
}
|
||||
xml += `${indent} </evidence>\n`;
|
||||
}
|
||||
|
||||
xml += `${indent} <detected_at>${issue.detectedAt}</detected_at>\n`;
|
||||
xml += `${indent}</issue>\n`;
|
||||
|
||||
return xml;
|
||||
}
|
||||
|
||||
/**
|
||||
* XML 转义
|
||||
*/
|
||||
private escapeXml(str: string): string {
|
||||
return str
|
||||
.replace(/&/g, '&')
|
||||
.replace(/</g, '<')
|
||||
.replace(/>/g, '>')
|
||||
.replace(/"/g, '"')
|
||||
.replace(/'/g, ''');
|
||||
}
|
||||
|
||||
/**
|
||||
* 缓存报告
|
||||
*/
|
||||
private async cacheReport(report: QcReport): Promise<void> {
|
||||
try {
|
||||
await prisma.iitQcReport.create({
|
||||
data: {
|
||||
projectId: report.projectId,
|
||||
reportType: report.reportType,
|
||||
summary: report.summary as any,
|
||||
issues: {
|
||||
critical: report.criticalIssues,
|
||||
warning: report.warningIssues,
|
||||
formStats: report.formStats,
|
||||
topIssues: report.topIssues, // V2.1
|
||||
groupedIssues: report.groupedIssues, // V2.1
|
||||
legacyXml: report.legacyXml, // V2.1
|
||||
} as any,
|
||||
llmReport: report.llmFriendlyXml,
|
||||
generatedAt: new Date(report.generatedAt),
|
||||
expiresAt: report.expiresAt ? new Date(report.expiresAt) : null,
|
||||
},
|
||||
});
|
||||
|
||||
logger.debug('[QcReportService] Report cached', {
|
||||
projectId: report.projectId,
|
||||
reportType: report.reportType,
|
||||
});
|
||||
} catch (error: any) {
|
||||
logger.warn('[QcReportService] Failed to cache report', {
|
||||
projectId: report.projectId,
|
||||
error: error.message,
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 获取 LLM 友好的报告（用于问答）
|
||||
*
|
||||
* @param projectId 项目 ID
|
||||
* @param format 格式：'llm-friendly' (默认) 或 'xml' (兼容格式)
|
||||
* @returns XML 报告
|
||||
*/
|
||||
async getLlmReport(projectId: string, format: 'llm-friendly' | 'xml' = 'llm-friendly'): Promise<string> {
|
||||
const report = await this.getReport(projectId);
|
||||
|
||||
if (format === 'xml') {
|
||||
return report.legacyXml || report.llmFriendlyXml;
|
||||
}
|
||||
|
||||
return report.llmFriendlyXml;
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 获取 Top Issues 统计
|
||||
*/
|
||||
async getTopIssues(projectId: string, limit: number = 5): Promise<TopIssue[]> {
|
||||
const report = await this.getReport(projectId);
|
||||
return report.topIssues.slice(0, limit);
|
||||
}
|
||||
|
||||
/**
|
||||
* V2.1: 获取按受试者分组的问题
|
||||
*/
|
||||
async getGroupedIssues(projectId: string): Promise<GroupedIssues[]> {
|
||||
const report = await this.getReport(projectId);
|
||||
return report.groupedIssues;
|
||||
}
|
||||
|
||||
/**
|
||||
* 强制刷新报告
|
||||
*
|
||||
* @param projectId 项目 ID
|
||||
* @returns 新生成的报告
|
||||
*/
|
||||
async refreshReport(projectId: string): Promise<QcReport> {
|
||||
return this.getReport(projectId, { forceRefresh: true });
|
||||
}
|
||||
|
||||
/**
|
||||
* 清理过期报告
|
||||
*/
|
||||
async cleanupExpiredReports(): Promise<number> {
|
||||
const result = await prisma.iitQcReport.deleteMany({
|
||||
where: {
|
||||
expiresAt: {
|
||||
lt: new Date(),
|
||||
},
|
||||
},
|
||||
});
|
||||
|
||||
logger.info('[QcReportService] Cleaned up expired reports', {
|
||||
count: result.count,
|
||||
});
|
||||
|
||||
return result.count;
|
||||
}
|
||||
}
|
||||
|
||||
// 单例导出
|
||||
export const QcReportService = new QcReportServiceClass();
|
||||
Reference in New Issue
Block a user