@naturalcycles/db-lib
Version:
Lowest Common Denominator API to supported Databases
61 lines (60 loc) • 2.65 kB
JavaScript
import { localTime } from '@naturalcycles/js-lib/datetime/localTime.js';
import { ErrorMode } from '@naturalcycles/js-lib/error/errorMode.js';
import { pMap } from '@naturalcycles/js-lib/promise/pMap.js';
import { _passthroughMapper } from '@naturalcycles/js-lib/types';
import { boldWhite, dimWhite, grey, yellow } from '@naturalcycles/nodejs-lib/colors';
import { NDJsonStats } from '@naturalcycles/nodejs-lib/stream';
import { DBQuery } from '../query/dbQuery.js';
/**
* Pipeline from input stream(s) to CommonDB .saveBatch().
* Input stream can be a stream from CommonDB.streamQuery()
* Allows to define a mapper and a predicate to map/filter objects between input and output.
* Handles backpressure.
*/
export async function dbPipelineCopy(opt) {
const { chunkSize = 100, dbInput, dbOutput, concurrency = 16, limit = 0, sinceUpdated, mapperPerTable = {}, saveOptionsPerTable = {}, transformMapOptions, errorMode = ErrorMode.SUPPRESS, } = opt;
let { tables } = opt;
const sinceUpdatedStr = sinceUpdated ? ' since ' + grey(localTime(sinceUpdated).toPretty()) : '';
console.log(`>> ${dimWhite('dbPipelineCopy')} started...${sinceUpdatedStr}`);
tables ||= await dbInput.getTables();
console.log(`${yellow(tables.length)} ${boldWhite('table(s)')}:\n` + tables.join('\n'));
const statsPerTable = {};
await pMap(tables, async (table) => {
let q = DBQuery.create(table).limit(limit);
if (sinceUpdated) {
q = q.filter('updated', '>=', sinceUpdated);
}
const saveOptions = saveOptionsPerTable[table] || {};
const mapper = mapperPerTable[table] || _passthroughMapper;
const started = Date.now();
let rows = 0;
await dbInput
.streamQuery(q)
.logProgress({
logEvery: 1000,
...opt,
metric: table,
})
.map(mapper, {
errorMode,
...transformMapOptions,
metric: table,
})
.flattenIfNeeded()
.tapSync(() => rows++)
.chunk(chunkSize)
.forEach(async (dbms) => {
await dbOutput.saveBatch(table, dbms, saveOptions);
});
const stats = NDJsonStats.create({
tookMillis: Date.now() - started,
rows,
sizeBytes: 0, // n/a
});
console.log(`>> ${grey(table)}\n` + stats.toPretty());
statsPerTable[table] = stats;
}, { concurrency, errorMode });
const statsTotal = NDJsonStats.createCombined(Object.values(statsPerTable));
console.log(statsTotal.toPretty('total'));
return statsTotal;
}