3gpp-mcp-server/src/download-dataset.ts

MCP Server for querying 3GPP telecom protocol specifications

import { TSpecLLMDownloader, DatasetConfig } from './utils/dataset-downloader';
import * as path from 'path';

async function main() {
  const config: DatasetConfig = {
    repositoryUrl: 'https://huggingface.co/datasets/rasoul-nikbakht/TSpec-LLM',
    localPath: path.join(__dirname, '..', 'data', 'TSpec-LLM'),
    formats: ['markdown', 'docx'],
  };

  const downloader = new TSpecLLMDownloader(config);

  try {
    console.log('Starting TSpec-LLM dataset download...');
    await downloader.downloadDataset();

    console.log('Analyzing dataset structure...');
    const structure = await downloader.analyzeStructure();

    console.log('\n=== Dataset Analysis ===');
    console.log(`Total files: ${structure.totalFiles}`);
    console.log(`Markdown files: ${structure.formats.markdown}`);
    console.log(`DOCX files: ${structure.formats.docx}`);
    console.log('\nReleases found:');
    Object.entries(structure.releases).forEach(([release, count]) => {
      console.log(`  ${release}: ${count} files`);
    });

    console.log('\nSample files:');
    structure.sampleFiles.forEach(file => {
      console.log(`  ${file.path} (${Math.round(file.size / 1024)}KB, ${file.format})`);
    });

  } catch (error) {
    console.error('Error:', error);
    process.exit(1);
  }
}

main();