530 lines
21 KiB
TypeScript
530 lines
21 KiB
TypeScript
// 从MongoDB的trace表同步数据到ClickHouse的events表
|
||
import { getVariable, setVariable } from "npm:windmill-client@1";
|
||
import { MongoClient, ObjectId } from "https://deno.land/x/mongo@v0.32.0/mod.ts";
|
||
|
||
interface MongoConfig {
|
||
host: string;
|
||
port: string;
|
||
db: string;
|
||
username: string;
|
||
password: string;
|
||
}
|
||
|
||
interface ClickHouseConfig {
|
||
clickhouse_host: string;
|
||
clickhouse_port: number;
|
||
clickhouse_user: string;
|
||
clickhouse_password: string;
|
||
clickhouse_database: string;
|
||
clickhouse_url: string;
|
||
}
|
||
|
||
interface TraceRecord {
|
||
_id: ObjectId;
|
||
slugId: ObjectId;
|
||
label: string | null;
|
||
ip: string;
|
||
type: number;
|
||
platform: string;
|
||
platformOS: string;
|
||
browser: string;
|
||
browserVersion: string;
|
||
url: string;
|
||
createTime: number;
|
||
}
|
||
|
||
interface SyncState {
|
||
last_sync_time: number;
|
||
records_synced: number;
|
||
last_sync_id?: string;
|
||
}
|
||
|
||
export async function main(
|
||
batch_size = 1000,
|
||
max_records = 9999999,
|
||
timeout_minutes = 60,
|
||
skip_clickhouse_check = false,
|
||
force_insert = false,
|
||
database_override = "shorturl_analytics" // 添加数据库名称参数，默认为shorturl_analytics
|
||
) {
|
||
const logWithTimestamp = (message: string) => {
|
||
const now = new Date();
|
||
console.log(`[${now.toISOString()}] ${message}`);
|
||
};
|
||
|
||
logWithTimestamp("开始执行MongoDB到ClickHouse的同步任务");
|
||
logWithTimestamp(`批处理大小: ${batch_size}, 最大记录数: ${max_records}, 超时时间: ${timeout_minutes}分钟`);
|
||
if (skip_clickhouse_check) {
|
||
logWithTimestamp("⚠️ 警告: 已启用跳过ClickHouse检查模式，不会检查记录是否已存在");
|
||
}
|
||
if (force_insert) {
|
||
logWithTimestamp("⚠️ 警告: 已启用强制插入模式，将尝试插入所有记录");
|
||
}
|
||
|
||
// 设置超时
|
||
const startTime = Date.now();
|
||
const timeoutMs = timeout_minutes * 60 * 1000;
|
||
|
||
// 检查是否超时
|
||
const checkTimeout = () => {
|
||
if (Date.now() - startTime > timeoutMs) {
|
||
console.log(`运行时间超过${timeout_minutes}分钟，暂停执行`);
|
||
return true;
|
||
}
|
||
return false;
|
||
};
|
||
|
||
// 获取MongoDB和ClickHouse的连接信息
|
||
let mongoConfig: MongoConfig;
|
||
let clickhouseConfig: ClickHouseConfig;
|
||
|
||
try {
|
||
const rawMongoConfig = await getVariable("f/shorturl_analytics/mongodb");
|
||
console.log("原始MongoDB配置:", JSON.stringify(rawMongoConfig));
|
||
|
||
// 尝试解析配置，如果是字符串形式
|
||
if (typeof rawMongoConfig === "string") {
|
||
try {
|
||
mongoConfig = JSON.parse(rawMongoConfig);
|
||
} catch (e) {
|
||
console.error("MongoDB配置解析失败:", e);
|
||
throw e;
|
||
}
|
||
} else {
|
||
mongoConfig = rawMongoConfig as MongoConfig;
|
||
}
|
||
|
||
const rawClickhouseConfig = await getVariable("f/shorturl_analytics/clickhouse");
|
||
console.log("原始ClickHouse配置:", JSON.stringify(rawClickhouseConfig));
|
||
|
||
// 尝试解析配置，如果是字符串形式
|
||
if (typeof rawClickhouseConfig === "string") {
|
||
try {
|
||
clickhouseConfig = JSON.parse(rawClickhouseConfig);
|
||
} catch (e) {
|
||
console.error("ClickHouse配置解析失败:", e);
|
||
throw e;
|
||
}
|
||
} else {
|
||
clickhouseConfig = rawClickhouseConfig as ClickHouseConfig;
|
||
}
|
||
|
||
// 检查并修复数据库配置
|
||
if (!clickhouseConfig.clickhouse_database || clickhouseConfig.clickhouse_database === "undefined") {
|
||
logWithTimestamp(`⚠️ 警告: 数据库名称未定义或为'undefined'，使用提供的默认值: ${database_override}`);
|
||
clickhouseConfig.clickhouse_database = database_override;
|
||
}
|
||
|
||
console.log("MongoDB配置解析为:", JSON.stringify(mongoConfig));
|
||
console.log("ClickHouse配置解析为:", JSON.stringify({
|
||
...clickhouseConfig,
|
||
clickhouse_password: "****" // 隐藏密码
|
||
}));
|
||
|
||
logWithTimestamp(`将使用ClickHouse数据库: ${clickhouseConfig.clickhouse_database}`);
|
||
} catch (error) {
|
||
console.error("获取配置失败:", error);
|
||
throw error;
|
||
}
|
||
|
||
// 构建MongoDB连接URL
|
||
let mongoUrl = "mongodb://";
|
||
if (mongoConfig.username && mongoConfig.password) {
|
||
mongoUrl += `${mongoConfig.username}:${mongoConfig.password}@`;
|
||
}
|
||
mongoUrl += `${mongoConfig.host}:${mongoConfig.port}/${mongoConfig.db}`;
|
||
|
||
console.log(`MongoDB连接URL: ${mongoUrl.replace(/:[^:]*@/, ":****@")}`);
|
||
|
||
// 连接MongoDB
|
||
const client = new MongoClient();
|
||
try {
|
||
await client.connect(mongoUrl);
|
||
console.log("MongoDB连接成功");
|
||
|
||
const db = client.database(mongoConfig.db);
|
||
const traceCollection = db.collection<TraceRecord>("trace");
|
||
|
||
// 构建查询条件，获取所有记录
|
||
const query: Record<string, unknown> = {
|
||
type: 1 // 只同步type为1的记录
|
||
};
|
||
|
||
// 计算总记录数
|
||
const totalRecords = await traceCollection.countDocuments(query);
|
||
console.log(`找到 ${totalRecords} 条记录需要同步`);
|
||
|
||
// 限制此次处理的记录数量
|
||
const recordsToProcess = Math.min(totalRecords, max_records);
|
||
console.log(`本次将处理 ${recordsToProcess} 条记录`);
|
||
|
||
if (totalRecords === 0) {
|
||
console.log("没有记录需要同步，任务完成");
|
||
return {
|
||
success: true,
|
||
records_synced: 0,
|
||
message: "没有记录需要同步"
|
||
};
|
||
}
|
||
|
||
// 检查ClickHouse连接状态
|
||
const checkClickHouseConnection = async (): Promise<boolean> => {
|
||
if (skip_clickhouse_check) {
|
||
logWithTimestamp("已启用跳过ClickHouse检查，不测试连接");
|
||
return true;
|
||
}
|
||
|
||
try {
|
||
logWithTimestamp("测试ClickHouse连接...");
|
||
const clickhouseUrl = `${clickhouseConfig.clickhouse_url}`;
|
||
const response = await fetch(clickhouseUrl, {
|
||
method: "POST",
|
||
headers: {
|
||
"Content-Type": "application/x-www-form-urlencoded",
|
||
"Authorization": `Basic ${btoa(`${clickhouseConfig.clickhouse_user}:${clickhouseConfig.clickhouse_password}`)}`,
|
||
},
|
||
body: `SELECT 1 FROM ${clickhouseConfig.clickhouse_database}.events LIMIT 1`,
|
||
// 设置5秒超时
|
||
signal: AbortSignal.timeout(5000)
|
||
});
|
||
|
||
if (response.ok) {
|
||
logWithTimestamp("ClickHouse连接测试成功");
|
||
return true;
|
||
} else {
|
||
const errorText = await response.text();
|
||
logWithTimestamp(`ClickHouse连接测试失败: ${response.status} ${errorText}`);
|
||
return false;
|
||
}
|
||
} catch (err) {
|
||
const error = err as Error;
|
||
logWithTimestamp(`ClickHouse连接测试失败: ${error.message}`);
|
||
return false;
|
||
}
|
||
};
|
||
|
||
// 检查记录是否已经存在于ClickHouse中
|
||
const checkExistingRecords = async (records: TraceRecord[]): Promise<TraceRecord[]> => {
|
||
if (records.length === 0) return [];
|
||
|
||
// 如果跳过ClickHouse检查或强制插入，则直接返回所有记录
|
||
if (skip_clickhouse_check || force_insert) {
|
||
logWithTimestamp(`已跳过ClickHouse重复检查，准备处理所有 ${records.length} 条记录`);
|
||
return records;
|
||
}
|
||
|
||
logWithTimestamp(`正在检查 ${records.length} 条记录是否已存在于ClickHouse中...`);
|
||
|
||
try {
|
||
// 验证数据库名称
|
||
if (!clickhouseConfig.clickhouse_database || clickhouseConfig.clickhouse_database === "undefined") {
|
||
throw new Error("数据库名称未定义或无效，请检查配置");
|
||
}
|
||
|
||
// 提取所有记录的ID
|
||
const recordIds = records.map(record => record.slugId.toString()); // 使用slugId作为link_id查询
|
||
logWithTimestamp(`待检查的记录ID: ${recordIds.join(', ')}`);
|
||
|
||
// 构建查询SQL，检查记录是否已存在，确保添加FORMAT JSON来获取正确的JSON格式响应
|
||
const query = `
|
||
SELECT link_id, visitor_id
|
||
FROM ${clickhouseConfig.clickhouse_database}.events
|
||
WHERE link_id IN ('${recordIds.join("','")}')
|
||
FORMAT JSON
|
||
`;
|
||
|
||
logWithTimestamp(`执行ClickHouse查询: ${query.replace(/\n\s*/g, ' ')}`);
|
||
|
||
// 发送请求到ClickHouse，添加10秒超时
|
||
const clickhouseUrl = `${clickhouseConfig.clickhouse_url}`;
|
||
const response = await fetch(clickhouseUrl, {
|
||
method: "POST",
|
||
headers: {
|
||
"Content-Type": "application/x-www-form-urlencoded",
|
||
"Authorization": `Basic ${btoa(`${clickhouseConfig.clickhouse_user}:${clickhouseConfig.clickhouse_password}`)}`
|
||
},
|
||
body: query,
|
||
signal: AbortSignal.timeout(10000)
|
||
});
|
||
|
||
if (!response.ok) {
|
||
const errorText = await response.text();
|
||
throw new Error(`ClickHouse查询错误: ${response.status} ${errorText}`);
|
||
}
|
||
|
||
// 获取响应文本以便记录
|
||
const responseText = await response.text();
|
||
logWithTimestamp(`ClickHouse查询响应: ${responseText.slice(0, 200)}${responseText.length > 200 ? '...' : ''}`);
|
||
|
||
if (!responseText.trim()) {
|
||
logWithTimestamp("ClickHouse返回空响应，假定没有记录存在");
|
||
return records; // 如果响应为空，假设没有记录
|
||
}
|
||
|
||
// 解析结果
|
||
let result;
|
||
try {
|
||
result = JSON.parse(responseText);
|
||
} catch (err) {
|
||
logWithTimestamp(`ClickHouse响应不是有效的JSON: ${responseText}`);
|
||
throw new Error(`解析ClickHouse响应失败: ${(err as Error).message}`);
|
||
}
|
||
|
||
// 确保result有正确的结构
|
||
if (!result.data) {
|
||
logWithTimestamp(`ClickHouse响应缺少data字段: ${JSON.stringify(result)}`);
|
||
return records; // 如果没有data字段，假设没有记录
|
||
}
|
||
|
||
// 提取已存在的记录ID
|
||
const existingIds = new Set(result.data.map((row: { link_id: string }) => row.link_id));
|
||
|
||
logWithTimestamp(`检测到 ${existingIds.size} 条记录已存在于ClickHouse中`);
|
||
if (existingIds.size > 0) {
|
||
logWithTimestamp(`已存在的记录ID: ${Array.from(existingIds).join(', ')}`);
|
||
}
|
||
|
||
// 过滤出不存在的记录
|
||
const newRecords = records.filter(record => !existingIds.has(record.slugId.toString())); // 使用slugId匹配link_id
|
||
logWithTimestamp(`过滤后剩余 ${newRecords.length} 条新记录需要插入`);
|
||
|
||
return newRecords;
|
||
} catch (err) {
|
||
const error = err as Error;
|
||
logWithTimestamp(`ClickHouse查询出错: ${error.message}`);
|
||
if (skip_clickhouse_check) {
|
||
logWithTimestamp("已启用跳过ClickHouse检查，将继续处理所有记录");
|
||
return records;
|
||
} else {
|
||
throw error; // 如果没有启用跳过检查，则抛出错误
|
||
}
|
||
}
|
||
};
|
||
|
||
// 在处理记录前先检查ClickHouse连接
|
||
const clickhouseConnected = await checkClickHouseConnection();
|
||
if (!clickhouseConnected && !skip_clickhouse_check) {
|
||
logWithTimestamp("⚠️ ClickHouse连接测试失败，请启用skip_clickhouse_check=true参数来跳过连接检查");
|
||
throw new Error("ClickHouse连接失败，无法继续同步");
|
||
}
|
||
|
||
// 处理记录的函数
|
||
const processRecords = async (records: TraceRecord[]) => {
|
||
if (records.length === 0) return 0;
|
||
|
||
logWithTimestamp(`开始处理批次数据，共 ${records.length} 条记录...`);
|
||
|
||
// 检查记录是否已存在
|
||
let newRecords;
|
||
try {
|
||
newRecords = await checkExistingRecords(records);
|
||
} catch (err) {
|
||
const error = err as Error;
|
||
logWithTimestamp(`检查记录是否存在时出错: ${error.message}`);
|
||
if (!skip_clickhouse_check && !force_insert) {
|
||
throw error;
|
||
}
|
||
// 如果跳过检查或强制插入，则使用所有记录
|
||
logWithTimestamp("将使用所有记录进行处理");
|
||
newRecords = records;
|
||
}
|
||
|
||
if (newRecords.length === 0) {
|
||
logWithTimestamp("所有记录都已存在，跳过处理");
|
||
return 0;
|
||
}
|
||
|
||
logWithTimestamp(`准备处理 ${newRecords.length} 条新记录...`);
|
||
|
||
// 准备ClickHouse插入数据
|
||
const clickhouseData = newRecords.map(record => {
|
||
const eventTime = new Date(record.createTime);
|
||
// 转换MongoDB记录为ClickHouse格式，匹配ClickHouse表结构
|
||
return {
|
||
// UUID将由ClickHouse自动生成 (event_id)
|
||
event_time: eventTime.toISOString().replace('T', ' ').replace('Z', ''),
|
||
event_type: record.type === 1 ? "visit" : "custom",
|
||
event_attributes: `{"mongo_id":"${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_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: "",
|
||
user_name: "",
|
||
user_email: "",
|
||
user_attributes: "{}",
|
||
team_id: "",
|
||
team_name: "",
|
||
team_attributes: "{}",
|
||
project_id: "",
|
||
project_name: "",
|
||
project_attributes: "{}",
|
||
qr_code_id: "",
|
||
qr_code_name: "",
|
||
qr_code_attributes: "{}",
|
||
visitor_id: record._id.toString(), // 使用MongoDB ID作为访客ID
|
||
session_id: record._id.toString() + "-" + record.createTime, // 创建一个唯一会话ID
|
||
ip_address: record.ip,
|
||
country: "", // 这些字段在MongoDB中不存在，使用默认值
|
||
city: "",
|
||
device_type: record.platform || "unknown",
|
||
browser: record.browser || "",
|
||
os: record.platformOS || "",
|
||
user_agent: record.browser + " " + record.browserVersion,
|
||
referrer: record.url || "",
|
||
utm_source: "",
|
||
utm_medium: "",
|
||
utm_campaign: "",
|
||
utm_term: "",
|
||
utm_content: "",
|
||
time_spent_sec: 0,
|
||
is_bounce: true,
|
||
is_qr_scan: false,
|
||
conversion_type: "visit",
|
||
conversion_value: 0,
|
||
req_full_path: record.url || ""
|
||
};
|
||
});
|
||
|
||
// 生成ClickHouse插入SQL
|
||
const insertSQL = `
|
||
INSERT INTO ${clickhouseConfig.clickhouse_database}.events
|
||
(event_time, event_type, event_attributes, link_id, link_slug, link_label, link_title,
|
||
link_original_url, link_attributes, link_created_at, link_expires_at, link_tags,
|
||
user_id, user_name, user_email, user_attributes, team_id, team_name, team_attributes,
|
||
project_id, project_name, project_attributes, qr_code_id, qr_code_name, qr_code_attributes,
|
||
visitor_id, session_id, ip_address, country, city, device_type, browser, os, user_agent,
|
||
referrer, utm_source, utm_medium, utm_campaign, utm_term, utm_content, time_spent_sec,
|
||
is_bounce, is_qr_scan, conversion_type, conversion_value, req_full_path)
|
||
VALUES ${clickhouseData.map(record => {
|
||
// 确保所有字符串值都是字符串类型，并安全处理替换
|
||
const safeReplace = (val: unknown): string => {
|
||
// 确保值是字符串，如果是null或undefined则使用空字符串
|
||
const str = val === null || val === undefined ? "" : String(val);
|
||
// 安全替换单引号
|
||
return str.replace(/'/g, "''");
|
||
};
|
||
|
||
return `('${record.event_time}', '${safeReplace(record.event_type)}', '${safeReplace(record.event_attributes)}',
|
||
'${record.link_id}', '${safeReplace(record.link_slug)}', '${safeReplace(record.link_label)}', '${safeReplace(record.link_title)}',
|
||
'${safeReplace(record.link_original_url)}', '${safeReplace(record.link_attributes)}', '${record.link_created_at}',
|
||
${record.link_expires_at === null ? 'NULL' : `'${record.link_expires_at}'`}, '${safeReplace(record.link_tags)}',
|
||
'${safeReplace(record.user_id)}', '${safeReplace(record.user_name)}', '${safeReplace(record.user_email)}',
|
||
'${safeReplace(record.user_attributes)}', '${safeReplace(record.team_id)}', '${safeReplace(record.team_name)}',
|
||
'${safeReplace(record.team_attributes)}', '${safeReplace(record.project_id)}', '${safeReplace(record.project_name)}',
|
||
'${safeReplace(record.project_attributes)}', '${safeReplace(record.qr_code_id)}', '${safeReplace(record.qr_code_name)}',
|
||
'${safeReplace(record.qr_code_attributes)}', '${safeReplace(record.visitor_id)}', '${safeReplace(record.session_id)}',
|
||
'${safeReplace(record.ip_address)}', '${safeReplace(record.country)}', '${safeReplace(record.city)}',
|
||
'${safeReplace(record.device_type)}', '${safeReplace(record.browser)}', '${safeReplace(record.os)}',
|
||
'${safeReplace(record.user_agent)}', '${safeReplace(record.referrer)}', '${safeReplace(record.utm_source)}',
|
||
'${safeReplace(record.utm_medium)}', '${safeReplace(record.utm_campaign)}', '${safeReplace(record.utm_term)}',
|
||
'${safeReplace(record.utm_content)}', ${record.time_spent_sec}, ${record.is_bounce}, ${record.is_qr_scan},
|
||
'${safeReplace(record.conversion_type)}', ${record.conversion_value}, '${safeReplace(record.req_full_path)}')`;
|
||
}).join(", ")}
|
||
`;
|
||
|
||
if (insertSQL.length === 0) {
|
||
console.log("没有新记录需要插入");
|
||
return 0;
|
||
}
|
||
|
||
// 发送请求到ClickHouse，添加20秒超时
|
||
const clickhouseUrl = `${clickhouseConfig.clickhouse_url}`;
|
||
try {
|
||
logWithTimestamp("发送插入请求到ClickHouse...");
|
||
const response = await fetch(clickhouseUrl, {
|
||
method: "POST",
|
||
headers: {
|
||
"Content-Type": "application/x-www-form-urlencoded",
|
||
"Authorization": `Basic ${btoa(`${clickhouseConfig.clickhouse_user}:${clickhouseConfig.clickhouse_password}`)}`
|
||
},
|
||
body: insertSQL,
|
||
signal: AbortSignal.timeout(20000)
|
||
});
|
||
|
||
if (!response.ok) {
|
||
const errorText = await response.text();
|
||
throw new Error(`ClickHouse插入错误: ${response.status} ${errorText}`);
|
||
}
|
||
|
||
logWithTimestamp(`成功插入 ${newRecords.length} 条记录到ClickHouse`);
|
||
return newRecords.length;
|
||
} catch (err) {
|
||
const error = err as Error;
|
||
logWithTimestamp(`向ClickHouse插入数据失败: ${error.message}`);
|
||
throw error;
|
||
}
|
||
};
|
||
|
||
// 批量处理记录
|
||
let processedRecords = 0;
|
||
let totalBatchRecords = 0;
|
||
|
||
for (let page = 0; processedRecords < recordsToProcess; page++) {
|
||
// 检查超时
|
||
if (checkTimeout()) {
|
||
logWithTimestamp(`已处理 ${processedRecords}/${recordsToProcess} 条记录，因超时暂停执行`);
|
||
break;
|
||
}
|
||
|
||
// 每批次都输出进度
|
||
logWithTimestamp(`开始处理第 ${page+1} 批次，已完成 ${processedRecords}/${recordsToProcess} 条记录 (${Math.round(processedRecords/recordsToProcess*100)}%)`);
|
||
|
||
logWithTimestamp(`正在从MongoDB获取第 ${page+1} 批次数据...`);
|
||
const records = await traceCollection.find(
|
||
query,
|
||
{
|
||
allowDiskUse: true,
|
||
sort: { createTime: 1 },
|
||
skip: page * batch_size,
|
||
limit: batch_size
|
||
}
|
||
).toArray();
|
||
|
||
if (records.length === 0) {
|
||
logWithTimestamp("没有找到更多数据，同步结束");
|
||
break;
|
||
}
|
||
|
||
// 找到数据，开始处理
|
||
logWithTimestamp(`获取到 ${records.length} 条记录，开始处理...`);
|
||
// 输出当前批次的部分数据信息
|
||
if (records.length > 0) {
|
||
logWithTimestamp(`批次 ${page+1} 第一条记录: ID=${records[0]._id}, 时间=${new Date(records[0].createTime).toISOString()}`);
|
||
if (records.length > 1) {
|
||
logWithTimestamp(`批次 ${page+1} 最后一条记录: ID=${records[records.length-1]._id}, 时间=${new Date(records[records.length-1].createTime).toISOString()}`);
|
||
}
|
||
}
|
||
|
||
const batchSize = await processRecords(records);
|
||
processedRecords += records.length;
|
||
totalBatchRecords += batchSize;
|
||
|
||
logWithTimestamp(`第 ${page+1} 批次处理完成。已处理 ${processedRecords}/${recordsToProcess} 条记录，实际插入 ${totalBatchRecords} 条 (${Math.round(processedRecords/recordsToProcess*100)}%)`);
|
||
}
|
||
|
||
return {
|
||
success: true,
|
||
records_processed: processedRecords,
|
||
records_synced: totalBatchRecords,
|
||
message: "数据同步完成"
|
||
};
|
||
} catch (err) {
|
||
console.error("同步过程中发生错误:", err);
|
||
return {
|
||
success: false,
|
||
error: err instanceof Error ? err.message : String(err),
|
||
stack: err instanceof Error ? err.stack : undefined
|
||
};
|
||
} finally {
|
||
// 关闭MongoDB连接
|
||
await client.close();
|
||
console.log("MongoDB连接已关闭");
|
||
}
|
||
}
|