feat(scripts): strip translator credits and promo lines from fetched and existing EPUBs

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-25 17:18:05 -03:00
parent fba9a35075
commit 750c2824ce
3 changed files with 82 additions and 1 deletions
@@ -0,0 +1,38 @@
+// Translator credits and promo lines that sites paste into every chapter
+// ("Translator: …", "Discord: https://…", "Join the Discord…", ko-fi/Patreon).
+// A paragraph is dropped only when it is short and the whole thing is promo, so
+// story text that happens to mention a word like "discord" stays.
+// The illustrator's reader.js carries the same rule for in-app reading.
+const MAX_PROMO_LENGTH = 200;
+const URL = /https?:\/\/\S+|\b(dsc|discord)\.gg\/\S*/gi;
+const MAX_LINK_LABEL = 60; // "Discord: <url>", "Ko-Fi: <url>", but not prose with a link
+const PROMO = [
+  /^(discord|patreon|ko-?fi)\b\s*:?/i,
+  /\b(join|support)\b.{0,40}\b(discord|patreon|ko-?fi)\b/i,
+  /\bko-?fi\b|\bpatreon\b/i,
+  /^(translators?|translated by|translation|editors?|edited by|editing|proofreaders?|proofread by|tl|tln|pr|ed)(\s*\/\s*\w+)*\s*[:-]/i,
+  /\bread (the )?(latest|advanced?|more) chapters?\b/i,
+];
+
+export const isPromo = text => {
+  const t = text.replace(/\s+/g, ' ').trim();
+  if (!t || t.length > MAX_PROMO_LENGTH) return false;
+  const label = t.replace(URL, '').trim();
+  if (label.length < t.length && label.length <= MAX_LINK_LABEL) return true;
+  return PROMO.some(r => r.test(t));
+};
+
+/** Remove promo paragraphs from a cheerio document; returns how many. */
+export const stripPromo = $ => {
+  let removed = 0;
+  $('p, div, h3, h4, h5, h6, li').each((_, el) => {
+    const node = $(el);
+    // Innermost blocks only, so a wrapper holding the whole chapter is never dropped.
+    if (node.find('p, div, li').length) return;
+    if (isPromo(node.text())) {
+      node.remove();
+      removed++;
+    }
+  });
+  return removed;
+};