Whole-page Genius dumps extract to just the lyrics

A full select-all paste (nav, bio, Q&A, credits, translations) now
reduces to the block between the first [Section] header and the
Embed/About footer, with "You might also like" recommendation inserts
dropped through to the next section. Lyrics-only pastes — with or
without [tags] — pass through untouched.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-06-07 12:48:36 -04:00
parent 5cb419e385
commit 00106f584a
+33
View File
@@ -958,10 +958,43 @@ document.getElementById('exportBtn').addEventListener('click', async ()=>{
(dont -> don + t), and lyrics sites ship cruft lines. Pasted
text gets normalized and de-crufted on the way in.
============================================================ */
const SECTION_RE = /^\[[^\]]{1,48}\]$/;
function extractLyricBlock(t){
// a whole-page Genius dump: lyrics live between the first [Section]
// header and the Embed/About/Comments footer
const lines = t.split('\n');
const first = lines.findIndex(l=>SECTION_RE.test(l.trim()));
if(first < 0) return null;
let end = -1;
for(let i = first; i < lines.length; i++){
const s = lines[i].trim();
if(/^\d*\s*Embed$/.test(s) || /^About$/.test(s) || /^Comments$/.test(s)){ end = i; break; }
}
const head = lines.slice(0, first).join('\n');
const pageish = /Contributors|Translations|Lyrics/i.test(head);
if(end < 0 && !pageish) return null; // just lyrics with [tags] — leave alone
let block = lines.slice(first, end < 0 ? lines.length : end);
// recommendation inserts run from "You might also like" to the next [Section]
const out = [];
let skipping = false;
for(const l of block){
const s = l.trim();
if(/^You might also like/i.test(s)){ skipping = true; continue; }
if(skipping){
if(SECTION_RE.test(s)) skipping = false;
else continue;
}
out.push(l);
}
while(out.length && /^\d*$/.test(out[out.length - 1].trim())) out.pop();
return out.join('\n').trim() + '\n';
}
function cleanPaste(t){
t = t.replace(/[\u2018\u2019\u02BC]/g, "'")
.replace(/[\u201C\u201D]/g, '"')
.replace(/\u00A0/g, ' ');
const extracted = extractLyricBlock(t);
if(extracted) return extracted;
const out = [];
t.split('\n').forEach((l, i)=>{
const s = l.trim();