radio-explorer/server/scripts/import-underground.js

// One-shot importer: resolves a curated list of underground / experimental /
// DJ-led stations against Radio-Browser, then writes
// data/seed/stations-underground.json. Re-running is safe: the seeder merges
// by Radio-Browser UUID.
//
// Usage:  node server/scripts/import-underground.js
//
// All entries are real DJ-broadcast / community / college / free-form stations
// (no playlist bots), spread across punk, house, jazz, eclectic, underground.

import fs from 'node:fs';
import path from 'node:path';
import { fileURLToPath } from 'node:url';

const __dirname = path.dirname(fileURLToPath(import.meta.url));
const ROOT = path.resolve(__dirname, '..', '..');
const OUT = path.join(ROOT, 'data', 'seed', 'stations-underground.json');
const RB = 'https://de1.api.radio-browser.info';
const UA = 'OnlineRadioExplorer/0.1 (+import-underground)';

// Each entry: [searchName, countryCodeOrNull, expectedNameSubstring]
// expectedNameSubstring is a lowercase substring used to filter out wrong-station
// hits when the same name is shared (e.g. two "Skylab Radio"s in different countries).
const PICKS = [
    // === UK / IE underground & DJ-led ===
    ['Foundation FM', 'GB', 'foundation fm'],
    ['Aaja Radio | Channel 1', 'GB', 'aaja'],
    ['Aaja Radio | Channel 2', 'GB', 'aaja'],
    ['Bloop London Radio', 'GB', 'bloop london'],
    ['Reform Radio', 'GB', 'reform radio'],
    ['1BTN', 'GB', '1btn'],
    ['Sub.fm', 'GB', 'sub.fm'],
    ['Radio Wigwam', 'GB', 'wigwam'],
    ['Skylab Radio', 'GB', 'skylab'],

    // === NL underground ===
    ['Echobox', 'NL', 'echobox'],
    ['Operator Radio', 'NL', 'operator'],

    // === DE / AT / CH underground ===
    ['byte.fm', 'DE', 'byte'],
    ['Radio 80000', 'DE', '80000'],
    ['FluxFM', 'DE', 'fluxfm'],
    ['FluxFM - Techno Underground', 'DE', 'techno'],
    ['Radio Eins', 'DE', 'radio eins'],
    ['Radio Helsinki 98,5 Mhz', 'FI', 'helsinki'],
    ['Radio Helsinki', 'AT', 'helsinki'],
    ['RTS Couleur 3', 'CH', 'couleur 3'],

    // === GR / HU / FR / EE underground ===
    ['Movement', 'GR', 'movement.radio 1'],
    ['Movement', 'GR', 'movement.radio 2'],
    ['Tilos Rádió', 'HU', 'tilos'],
    ['Radio Campus Paris', 'FR', 'radio campus paris'],

    // === US college & free-form (the punk-friendly ones) ===
    ['WMBR 88.1', 'US', 'wmbr'],
    ['WHRB 95.3', 'US', 'whrb'],
    ['KALX 90.7FM Berkeley', 'US', 'kalx'],
    ['WREK 91.1', 'US', 'wrek'],
    ['WXYC 89.3', 'US', 'wxyc'],
    ['KZSC 88.1', 'US', 'kzsc'],
    ['KDVS Davis', 'US', 'kdvs'],
    ['WPRB 103.3 FM', 'US', 'wprb'],
    ['WZBC', 'US', 'wzbc'],
    ['KFJC', null, 'kfjc'],
    ['KXLU 88.9FM', 'US', 'kxlu'],
    ['KCSB', 'US', 'kcsb'],
    ['WLUW', 'US', 'wluw'],
    ['WUSB 90.1', 'US', 'wusb'],
    ['BFF.fm', 'US', 'bff.fm'],

    // === Punk / shoegaze / noise leaning ===
    ['DKFM Shoegaze Radio', 'CA', 'dkfm'],
    ['idobi Anthm', 'US', 'anthm'],
    ['idobi Howl', 'US', 'howl'],
    ['Radio Free Brooklyn', 'US', 'radio free brooklyn'],

    // === AU community indie ===
    ['Triple R 102.7', 'AU', 'triple r'],
    ['PBS 106.7FM', 'AU', 'pbs 106.7'],
    ['FBi Radio', 'AU', 'fbi radio'],
    ['2SER 107.3 FM', 'AU', '2ser'],
    ['4zzz', 'AU', '4zzz'],
    ['RTRFM', 'AU', 'rtrfm']
];

function detectFormat(codec, url) {
    const c = (codec || '').toLowerCase();
    if (c.includes('mp3')) return 'mp3';
    if (c.includes('aac')) return 'aac';
    if (c.includes('ogg') || c.includes('vorbis') || c.includes('opus')) return 'ogg';
    if (url?.endsWith('.m3u8')) return 'hls';
    if (url?.endsWith('.m3u')) return 'm3u';
    if (url?.endsWith('.pls')) return 'pls';
    return 'unknown';
}

function slugify(name) {
    return name.toLowerCase()
        .replace(/[^a-z0-9]+/g, '-')
        .replace(/^-+|-+$/g, '')
        .slice(0, 80);
}

async function rbSearch(name, country) {
    const params = new URLSearchParams({
        name, limit: '8', hidebroken: 'true', order: 'clickcount', reverse: 'true'
    });
    if (country) params.set('countrycode', country);
    const res = await fetch(`${RB}/json/stations/search?${params}`, { headers: { 'User-Agent': UA } });
    if (!res.ok) throw new Error(`RB ${res.status}`);
    return res.json();
}

function pickBest(list, target, expectSub) {
    if (!list.length) return null;
    const sub = (expectSub || '').toLowerCase();
    const t = target.toLowerCase().trim();
    // 1. exact name match
    const exact = list.find((s) => (s.name || '').toLowerCase().trim() === t);
    if (exact) return exact;
    // 2. expected substring match
    if (sub) {
        const subMatch = list.find((s) => (s.name || '').toLowerCase().includes(sub));
        if (subMatch) return subMatch;
    }
    // 3. fall back to top hit
    return list[0];
}

function toEntry(s) {
    const stream = {
        url: s.url_resolved || s.url,
        format: detectFormat(s.codec, s.url_resolved || s.url),
        bitrate: s.bitrate || null,
        label: s.codec ? `${s.codec} ${s.bitrate || ''}`.trim() : null,
        priority: 0
    };
    return {
        uuid: s.stationuuid,
        slug: `rb-${s.stationuuid.slice(0, 8)}-${slugify(s.name).slice(0, 40)}`,
        name: s.name,
        category: 'underground',
        country: s.countrycode || null,
        homepage: s.homepage || null,
        genres: (s.tags || '').split(',').map((t) => t.trim()).filter(Boolean).slice(0, 6),
        description: null,
        image_url: s.favicon || null,
        source: 'radiobrowser',
        source_ref: s.stationuuid,
        streams: [stream]
    };
}

async function main() {
    const out = [];
    const seenUuids = new Set();
    let okCount = 0;
    let missCount = 0;
    let dupCount = 0;

    for (const [name, country, expectSub] of PICKS) {
        try {
            const hits = await rbSearch(name, country);
            const pick = pickBest(hits, name, expectSub);
            if (!pick) { console.warn('  miss:', name); missCount++; continue; }
            if (seenUuids.has(pick.stationuuid)) {
                console.warn('  dup :', name, '->', pick.name);
                dupCount++;
                continue;
            }
            seenUuids.add(pick.stationuuid);
            out.push(toEntry(pick));
            console.log('  ok  :', name.padEnd(32), '->', pick.name, `(${pick.codec || '?'} ${pick.bitrate || ''})`);
            okCount++;
        } catch (err) {
            console.warn('  err :', name, err.message);
            missCount++;
        }
        await new Promise((r) => setTimeout(r, 80));
    }

    fs.writeFileSync(OUT, JSON.stringify(out, null, 2) + '\n', 'utf8');
    console.log(`\nwrote ${out.length} entries to ${path.relative(ROOT, OUT)}`);
    console.log(`  ok=${okCount}  miss=${missCount}  dup=${dupCount}`);
}

main().catch((e) => { console.error(e); process.exit(1); });