embeddings-js
Version:
A NodeJS RAG framework to easily work with LLMs and custom datasets
47 lines (46 loc) • 1.81 kB
JavaScript
import md5 from 'md5';
import usetube from 'usetube';
import createDebugMessages from 'debug';
import { BaseLoader } from '../interfaces/base-loader.js';
import { YoutubeChannelLoader } from './youtube-channel-loader.js';
export class YoutubeSearchLoader extends BaseLoader {
constructor({ searchString }) {
super(`YoutubeSearchLoader${md5(searchString)}`);
Object.defineProperty(this, "debug", {
enumerable: true,
configurable: true,
writable: true,
value: createDebugMessages('embedjs:loader:YoutubeSearchLoader')
});
Object.defineProperty(this, "searchString", {
enumerable: true,
configurable: true,
writable: true,
value: void 0
});
this.searchString = searchString;
}
async *getChunks() {
try {
const { channels } = await usetube.searchChannel(this.searchString);
this.debug(`Search for channels with search string '${this.searchString}' found ${channels.length} entries`);
const channelIds = channels.map((c) => c.channel_id);
for (const channelId of channelIds) {
const youtubeLoader = new YoutubeChannelLoader({ channelId });
for await (const chunk of youtubeLoader.getChunks()) {
yield {
...chunk,
metadata: {
...chunk.metadata,
type: 'YoutubeSearchLoader',
originalSource: this.searchString,
},
};
}
}
}
catch (e) {
this.debug('Could not search for string', this.searchString, e);
}
}
}