Feature Complete:

- MongoDB Check before Initialising
- Output to file or S3
- Output to correct format for JSON Schema
This commit is contained in:
2026-01-07 19:51:07 +00:00
parent 03ac6431ff
commit 068cca32b4
7 changed files with 2128 additions and 35 deletions

View File

@@ -1,12 +1,18 @@
import { log } from './logger.js'
import { Readable } from 'node:stream'
import { createWriteStream } from 'node:fs'
import { finished } from 'node:stream/promises'
import { processAndStore } from './sss.js'
import { getLatestPackageName, getRequestStream } from './sources/gitea.js'
import { processPisStream } from './process.js'
import { isPackageProcessed } from './database.js'
async function main() {
const SERVICE_NAME = process.env.SERVICE_NAME;
if (!SERVICE_NAME) {
log('ERROR', "SERVICE_NAME env variable must be set");
process.exit(1);
}
try {
const packageInfo = await getLatestPackageName();
log('INFO', `Latest PIS Package: ${packageInfo.name}`);
@@ -16,12 +22,19 @@ async function main() {
process.exit(9);
}
if (await isPackageProcessed(SERVICE_NAME, packageInfo.name)) {
log('INFO', `Database matches latest release. Exiting`);
process.exit(0);
}
const inputStream: Readable = await getRequestStream(packageInfo.assets[0].browser_download_url);
const objectGenerator = processPisStream(inputStream);
const outputPath = './review_output.ndjson';
log('DEBUG', `Processing stream to: ${outputPath}`);
const filename = `${packageInfo.name.replace(/\s+/g, '_')}_pis_data_ndjson`;
await writeToTemporaryFile(processPisStream(inputStream), outputPath);
log('DEBUG', `Processing stream to: ${filename}`);
await processAndStore(objectGenerator, filename);
log('DEBUG', 'Done');
} catch (err) {
log('ERROR', 'Fatal error in pipeline: ', err);
@@ -29,18 +42,4 @@ async function main() {
}
}
async function writeToTemporaryFile(generator: AsyncGenerator<any>, path: string) {
const writer = createWriteStream(path);
for await (const record of generator) {
const line = JSON.stringify(record) + '\n';
if (!writer.write(line)) {
await new Promise((resolve) => writer.once('drain', resolve));
}
}
writer.end();
return finished(writer);
}
main();