Add req_full_path to Event interface, implement activities API for event retrieval, and enhance sync script with short link details

This commit is contained in:
2025-04-17 22:23:38 +08:00
parent 53e1611670
commit 2cb45781c7
3 changed files with 190 additions and 11 deletions

View File

@@ -33,6 +33,23 @@ interface TraceRecord {
createTime: number;
}
// 添加 ShortRecord 接口定义
interface ShortRecord {
_id: ObjectId;
slug: string; // 短链接的slug部分
origin: string; // 原始URL
domain?: string; // 域名
createTime: number; // 创建时间戳
user?: string; // 创建用户
title?: string; // 标题
description?: string; // 描述
tags?: string[]; // 标签
active?: boolean; // 是否活跃
expiresAt?: number; // 过期时间戳
teamId?: string; // 团队ID
projectId?: string; // 项目ID
}
interface SyncState {
last_sync_time: number;
records_synced: number;
@@ -181,6 +198,8 @@ export async function main(
const db = client.database(mongoConfig.db);
const traceCollection = db.collection<TraceRecord>("trace");
// 添加对short集合的引用
const shortCollection = db.collection<ShortRecord>("short");
// 构建查询条件,根据上次同步状态获取新记录
const query: Record<string, unknown> = {
@@ -380,9 +399,23 @@ export async function main(
logWithTimestamp(`准备处理 ${newRecords.length} 条新记录...`);
// 获取链接信息 - 新增代码
const slugIds = newRecords.map(record => record.slugId);
logWithTimestamp(`正在查询 ${slugIds.length} 条短链接信息...`);
const shortLinks = await shortCollection.find({
_id: { $in: slugIds }
}).toArray();
// 创建映射用于快速查找 - 新增代码
const shortLinksMap = new Map(shortLinks.map(link => [link._id.toString(), link]));
logWithTimestamp(`获取到 ${shortLinks.length} 条短链接信息`);
// 准备ClickHouse插入数据
const clickhouseData = newRecords.map(record => {
const eventTime = new Date(record.createTime);
// 获取对应的短链接信息 - 新增代码
const shortLink = shortLinksMap.get(record.slugId.toString());
// 转换MongoDB记录为ClickHouse格式匹配ClickHouse表结构
return {
// UUID将由ClickHouse自动生成 (event_id)
@@ -390,22 +423,26 @@ export async function main(
event_type: record.type === 1 ? "visit" : "custom",
event_attributes: `{"mongo_id":"${record._id.toString()}"}`,
link_id: record.slugId.toString(),
link_slug: "", // 这些字段可能需要从其他表获取
link_slug: shortLink?.slug || "", // 新增: 从short获取slug
link_label: record.label || "",
link_title: "",
link_original_url: "",
link_attributes: "{}",
link_created_at: eventTime.toISOString().replace('T', ' ').replace('Z', ''), // 暂用访问时间代替,可能需要从其他表获取
link_expires_at: null,
link_tags: "[]",
user_id: "",
link_title: shortLink?.title || "", // 新增: 从short获取标题
link_original_url: shortLink?.origin || "", // 新增: 从short获取原始URL
link_attributes: JSON.stringify({ domain: shortLink?.domain || null }), // 新增: 从short获取域名信息
link_created_at: shortLink?.createTime
? new Date(shortLink.createTime).toISOString().replace('T', ' ').replace('Z', '')
: eventTime.toISOString().replace('T', ' ').replace('Z', ''), // 新增: 使用真实的创建时间
link_expires_at: shortLink?.expiresAt
? new Date(shortLink.expiresAt).toISOString().replace('T', ' ').replace('Z', '')
: null, // 新增: 使用真实的过期时间
link_tags: shortLink?.tags ? JSON.stringify(shortLink.tags) : "[]", // 新增: 从short获取标签
user_id: shortLink?.user || "", // 新增: 从short获取用户ID
user_name: "",
user_email: "",
user_attributes: "{}",
team_id: "",
team_id: shortLink?.teamId || "", // 新增: 从short获取团队ID
team_name: "",
team_attributes: "{}",
project_id: "",
project_id: shortLink?.projectId || "", // 新增: 从short获取项目ID
project_name: "",
project_attributes: "{}",
qr_code_id: "",