cleaning up db

This commit is contained in:
2025-03-12 14:08:38 +08:00
parent ae7ea3b567
commit 285a0c780a
11 changed files with 1011 additions and 70 deletions

View File

@@ -0,0 +1,212 @@
// 检查ClickHouse数据库结构的脚本
const { createClient } = require('@clickhouse/client');
const dotenv = require('dotenv');
const path = require('path');
const fs = require('fs');
// 加载环境变量
dotenv.config({ path: path.resolve(__dirname, '../../.env') });
// 定义输出目录
const DB_REPORTS_DIR = path.resolve(__dirname, '../db-reports');
// 获取ClickHouse配置
const clickhouseHost = process.env.CLICKHOUSE_HOST || 'localhost';
const clickhousePort = process.env.CLICKHOUSE_PORT || '8123';
const clickhouseUser = process.env.CLICKHOUSE_USER || 'default';
const clickhousePassword = process.env.CLICKHOUSE_PASSWORD || '';
const clickhouseDatabase = process.env.CLICKHOUSE_DATABASE || 'default';
console.log('ClickHouse配置:');
console.log(` - 主机: ${clickhouseHost}`);
console.log(` - 端口: ${clickhousePort}`);
console.log(` - 用户: ${clickhouseUser}`);
console.log(` - 数据库: ${clickhouseDatabase}`);
// 创建ClickHouse客户端 - 使用0.2.10版本的API
const client = createClient({
url: `http://${clickhouseHost}:${clickhousePort}`,
username: clickhouseUser,
password: clickhousePassword,
database: clickhouseDatabase
});
// 获取所有表
async function getAllTables() {
console.log('\n获取所有表...');
try {
const query = `
SELECT name
FROM system.tables
WHERE database = '${clickhouseDatabase}'
`;
const resultSet = await client.query({
query,
format: 'JSONEachRow'
});
const tables = await resultSet.json();
if (!tables || tables.length === 0) {
console.log(`数据库 ${clickhouseDatabase} 中没有找到任何表`);
return null;
}
console.log(`数据库 ${clickhouseDatabase} 中找到以下表:`);
tables.forEach(table => {
console.log(` - ${table.name}`);
});
return tables.map(table => table.name);
} catch (error) {
console.error('获取所有表时出错:', error);
return null;
}
}
// 获取表结构
async function getTableSchema(tableName) {
console.log(`\n获取表 ${tableName} 的结构...`);
try {
const query = `
DESCRIBE TABLE ${clickhouseDatabase}.${tableName}
`;
const resultSet = await client.query({
query,
format: 'JSONEachRow'
});
const columns = await resultSet.json();
if (!columns || columns.length === 0) {
console.log(`${tableName} 不存在或没有列`);
return null;
}
console.log(`${tableName} 的列:`);
columns.forEach(column => {
console.log(` - ${column.name} (${column.type}, ${column.default_type === '' ? '无默认值' : `默认值: ${column.default_expression}`})`);
});
return columns;
} catch (error) {
console.error(`获取表 ${tableName} 结构时出错:`, error);
return null;
}
}
// 获取表数据示例
async function getTableDataSample(tableName, limit = 5) {
console.log(`\n获取表 ${tableName} 的数据示例 (最多 ${limit} 行)...`);
try {
const query = `
SELECT *
FROM ${clickhouseDatabase}.${tableName}
LIMIT ${limit}
`;
const resultSet = await client.query({
query,
format: 'JSONEachRow'
});
const rows = await resultSet.json();
if (!rows || rows.length === 0) {
console.log(`${tableName} 中没有数据`);
return null;
}
console.log(`${tableName} 的数据示例:`);
rows.forEach((row, index) => {
console.log(`${index + 1}:`);
Object.entries(row).forEach(([key, value]) => {
console.log(` ${key}: ${value}`);
});
});
return rows;
} catch (error) {
console.error(`获取表 ${tableName} 数据示例时出错:`, error);
return null;
}
}
// 主函数
async function main() {
let outputBuffer = '';
const originalConsoleLog = console.log;
// 重定向console.log到buffer和控制台
console.log = function() {
// 调用原始的console.log
originalConsoleLog.apply(console, arguments);
// 写入到buffer
outputBuffer += Array.from(arguments).join(' ') + '\n';
};
try {
// 获取所有表
const tables = await getAllTables();
if (!tables) {
console.error('无法获取表列表');
process.exit(1);
}
console.log('\n所有ClickHouse表:');
console.log(tables.join(', '));
// 获取每个表的结构,但不获取数据示例
for (const tableName of tables) {
await getTableSchema(tableName);
// 移除数据示例检查
// await getTableDataSample(tableName);
}
console.log('\nClickHouse数据库结构检查完成');
// 保存输出到指定目录
const timestamp = new Date().toISOString().replace(/[:.]/g, '-');
// 确保目录存在
if (!fs.existsSync(DB_REPORTS_DIR)) {
fs.mkdirSync(DB_REPORTS_DIR, { recursive: true });
}
const outputPath = path.join(DB_REPORTS_DIR, `clickhouse-schema-${timestamp}.log`);
fs.writeFileSync(outputPath, outputBuffer);
originalConsoleLog(`结果已保存到: ${outputPath}`);
} catch (error) {
console.error('检查ClickHouse数据库结构时出错:', error);
} finally {
// 恢复原始的console.log
console.log = originalConsoleLog;
// 关闭客户端连接
await client.close();
}
}
// 导出函数
module.exports = {
getAllTables,
getTableSchema,
getTableDataSample,
main
};
// 如果直接运行此脚本则执行main函数
if (require.main === module) {
main().catch(error => {
console.error('运行脚本时出错:', error);
process.exit(1);
});
}