☰

Scribd Enhancer All-in-One (v2.7.3)

Scribd Enhancer with OCR, export, print, parallel scraping, and clean UI. Auto language detection, smart filtering, output splitting — full feature set restored and refined. By Eliminater74.

  1. // ==UserScript==
  2. // @name         Scribd Enhancer All-in-One (v2.7.3)
  3. // @namespace    https://greatest.deepsurf.us/users/Eliminater74
  4. // @version      2.7.3
  5. // @description  Scribd Enhancer with OCR, export, print, parallel scraping, and clean UI. Auto language detection, smart filtering, output splitting — full feature set restored and refined. By Eliminater74.
  6. // @author       Eliminater74
  7. // @license      MIT
  8. // @match        *://*.scribd.com/*
  9. // @grant        none
  10. // @icon         https://s-f.scribdassets.com/favicon.ico
  11. // ==/UserScript==
  12.  
  13. (function () {
  14.   'use strict';
  15.  
  16.   const SETTINGS_KEY = 'scribdEnhancerSettings';
  17.   const defaultSettings = {
  18.     unblur: true,
  19.     autoScrape: false,
  20.     darkMode: false,
  21.     showPreview: true,
  22.     enableOCR: true,
  23.     ocrLang: 'auto',
  24.     splitEvery: 0
  25.   };
  26.   const settings = { ...defaultSettings, ...JSON.parse(localStorage.getItem(SETTINGS_KEY) || '{}') };
  27.   const saveSettings = () => localStorage.setItem(SETTINGS_KEY, JSON.stringify(settings));
  28.  
  29.   const tesseractScript = document.createElement('script');
  30.   tesseractScript.src = 'https://cdn.jsdelivr.net/npm/tesseract.js@4.0.2/dist/tesseract.min.js';
  31.   document.head.appendChild(tesseractScript);
  32.  
  33.   const style = document.createElement('style');
  34.   style.textContent = `
  35.     #se-ui {
  36.       position: fixed; bottom: 20px; right: 20px; background: #222; color: #fff;
  37.       border-radius: 10px; padding: 10px; z-index: 9999; width: 320px;
  38.       font-family: sans-serif; font-size: 13px; box-shadow: 0 0 10px #000;
  39.     }
  40.     #se-ui label, #se-ui select, #se-ui button {
  41.       display: block; width: 100%; margin: 4px 0;
  42.     }
  43.     #se-ui input[type="checkbox"] { margin-right: 6px; }
  44.     #se-ui button {
  45.       background: #444; color: white; border: none; border-radius: 6px; padding: 6px;
  46.     }
  47.     #se-preview {
  48.       position: fixed; top: 10px; right: 20px; bottom: 140px; width: 360px;
  49.       background: #f4f4f4; color: #000; overflow: auto; padding: 10px;
  50.       font-family: monospace; font-size: 12px; white-space: pre-wrap;
  51.       border: 1px solid #999; z-index: 9998; border-radius: 10px;
  52.     }
  53.     .dark-mode #se-preview {
  54.       background: #222; color: #eee; border-color: #555;
  55.     }
  56.     .dark-mode * {
  57.       background-color: transparent !important;
  58.       color: #e0e0e0 !important;
  59.       border-color: #444 !important;
  60.     }
  61.   `;
  62.   document.head.appendChild(style);
  63.  
  64.   function applyDarkMode() {
  65.     document.documentElement.classList.toggle('dark-mode', settings.darkMode);
  66.     document.body.classList.toggle('dark-mode', settings.darkMode);
  67.   }
  68.  
  69.   function unblurContent() {
  70.     if (!settings.unblur) return;
  71.     const cleanup = () => {
  72.       document.querySelectorAll('.blurred_page, .promo_div, [unselectable="on"]').forEach(el => el.remove());
  73.       document.querySelectorAll('*').forEach(el => {
  74.         const cs = getComputedStyle(el);
  75.         if (cs.color === 'transparent') el.style.color = '#111';
  76.         if (cs.textShadow?.includes('white')) el.style.textShadow = 'none';
  77.       });
  78.     };
  79.     cleanup();
  80.     new MutationObserver(cleanup).observe(document.body, { childList: true, subtree: true });
  81.   }
  82.  
  83.   function cleanOCRText(text) {
  84.     return text.split('\n').map(t => t.trim()).filter(line =>
  85.       line.length >= 3 && /[a-zA-Z]/.test(line) && !/^[^a-zA-Z0-9]{3,}$/.test(line)
  86.     ).join('\n');
  87.   }
  88.  
  89.   function detectLanguage(text) {
  90.     const map = { spa: /ñ|á|í|ó|ú/, fra: /é|è|ê|ç/, deu: /ä|ö|ü|ß/, ron: /ș|ț|ă|î|â/ };
  91.     for (const [lang, regex] of Object.entries(map)) {
  92.       if (regex.test(text)) return lang;
  93.     }
  94.     return 'eng';
  95.   }
  96.  
  97.   async function preprocessImage(src) {
  98.     return new Promise(resolve => {
  99.       const img = new Image();
  100.       img.crossOrigin = 'anonymous';
  101.       img.onload = () => {
  102.         if (img.naturalWidth < 100 || img.naturalHeight < 100 || /logo|icon|watermark/i.test(src)) return resolve(null);
  103.         const canvas = document.createElement('canvas');
  104.         canvas.width = img.width; canvas.height = img.height;
  105.         const ctx = canvas.getContext('2d');
  106.         ctx.drawImage(img, 0, 0);
  107.         const imageData = ctx.getImageData(0, 0, canvas.width, canvas.height);
  108.         for (let i = 0; i < imageData.data.length; i += 4) {
  109.           const avg = (imageData.data[i] + imageData.data[i+1] + imageData.data[i+2]) / 3;
  110.           imageData.data[i] = imageData.data[i+1] = imageData.data[i+2] = avg;
  111.         }
  112.         ctx.putImageData(imageData, 0, 0);
  113.         resolve(canvas.toDataURL('image/png'));
  114.       };
  115.       img.src = src;
  116.     });
  117.   }
  118.  
  119.   function createPreview() {
  120.     const preview = document.createElement('div');
  121.     preview.id = 'se-preview';
  122.     if (settings.showPreview) {
  123.       preview.textContent = '[Preview Initialized]\n';
  124.       document.body.appendChild(preview);
  125.     }
  126.     return preview;
  127.   }
  128.  
  129.   function exportOutput(content, ext) {
  130.     const split = settings.splitEvery;
  131.     const lines = content.split(/(?=\[Page \d+])/);
  132.     if (!split || split < 1) {
  133.       const blob = new Blob([content], { type: `text/${ext}` });
  134.       const a = document.createElement('a');
  135.       a.href = URL.createObjectURL(blob);
  136.       a.download = `scribd_output.${ext}`;
  137.       a.click();
  138.     } else {
  139.       for (let i = 0; i < lines.length; i += split) {
  140.         const chunk = lines.slice(i, i + split).join('\n');
  141.         const blob = new Blob([chunk], { type: `text/${ext}` });
  142.         const a = document.createElement('a');
  143.         a.href = URL.createObjectURL(blob);
  144.         a.download = `scribd_part${Math.floor(i / split) + 1}.${ext}`;
  145.         a.click();
  146.       }
  147.     }
  148.   }
  149.  
  150.   function printToPDF(content) {
  151.     const win = window.open('', 'PrintView');
  152.     win.document.write(`<html><head><title>Scribd Print</title></head><body><pre>${content}</pre></body></html>`);
  153.     win.document.close();
  154.     win.focus();
  155.     setTimeout(() => win.print(), 600);
  156.   }
  157.  
  158.   async function scrapePages(pages, preview) {
  159.     const concurrency = 4;
  160.     let index = 0;
  161.     const firstText = [];
  162.  
  163.     async function scrape(page, i) {
  164.       page.scrollIntoView();
  165.       await new Promise(r => setTimeout(r, 300));
  166.  
  167.       let found = false;
  168.       const text = page.innerText.trim();
  169.       if (text) {
  170.         preview.textContent += `[Page ${i + 1}] ✅\n${text}\n\n`;
  171.         firstText.push(text);
  172.         found = true;
  173.       }
  174.  
  175.       if (settings.enableOCR && window.Tesseract) {
  176.         const imgs = page.querySelectorAll('img');
  177.         for (let img of imgs) {
  178.           const src = img.src || '';
  179.           const processed = await preprocessImage(src);
  180.           if (!processed) continue;
  181.           const lang = settings.ocrLang === 'auto' ? detectLanguage(firstText.join(' ')) : settings.ocrLang;
  182.           const result = await window.Tesseract.recognize(processed, lang);
  183.           const ocrText = cleanOCRText(result.data.text || '');
  184.           if (ocrText) {
  185.             preview.textContent += `[OCR] ${ocrText}\n\n`;
  186.             found = true;
  187.           }
  188.         }
  189.       }
  190.  
  191.       if (!found) preview.textContent += `[Page ${i + 1}] ❌ No content\n\n`;
  192.     }
  193.  
  194.     const tasks = Array(concurrency).fill(null).map(async () => {
  195.       while (index < pages.length) {
  196.         const i = index++;
  197.         await scrape(pages[i], i);
  198.       }
  199.     });
  200.     await Promise.all(tasks);
  201.     alert(`✅ Scraped ${pages.length} pages.`);
  202.   }
  203.  
  204.   function createUI(preview) {
  205.     const ui = document.createElement('div');
  206.     ui.id = 'se-ui';
  207.     ui.innerHTML = `
  208.       <label><input type="checkbox" id="opt-unblur"> Unblur</label>
  209.       <label><input type="checkbox" id="opt-autoscrape"> Auto Scrape</label>
  210.       <label><input type="checkbox" id="opt-dark"> Dark Mode</label>
  211.       <label><input type="checkbox" id="opt-preview"> Show Preview</label>
  212.       <label>OCR:
  213.         <select id="opt-lang">
  214.           <option value="auto">Auto</option>
  215.           <option value="eng">English</option>
  216.           <option value="spa">Spanish</option>
  217.           <option value="fra">French</option>
  218.           <option value="deu">German</option>
  219.         </select>
  220.       </label>
  221.       <label>Split Every:
  222.         <select id="opt-split">
  223.           <option value="0">Off</option>
  224.           <option value="100">100</option>
  225.           <option value="250">250</option>
  226.           <option value="500">500</option>
  227.         </select>
  228.       </label>
  229.       <button id="btn-scrape">📖 Scrape Pages</button>
  230.       <button id="btn-export">💾 Export TXT</button>
  231.       <button id="btn-html">🧾 Export HTML</button>
  232.       <button id="btn-print">🖨️ Print to PDF</button>
  233.     `;
  234.     document.body.appendChild(ui);
  235.  
  236.     ui.querySelector('#opt-unblur').checked = settings.unblur;
  237.     ui.querySelector('#opt-autoscrape').checked = settings.autoScrape;
  238.     ui.querySelector('#opt-dark').checked = settings.darkMode;
  239.     ui.querySelector('#opt-preview').checked = settings.showPreview;
  240.     ui.querySelector('#opt-lang').value = settings.ocrLang;
  241.     ui.querySelector('#opt-split').value = settings.splitEvery;
  242.  
  243.     ui.querySelectorAll('input, select').forEach(input => {
  244.       input.onchange = () => {
  245.         settings.unblur = ui.querySelector('#opt-unblur').checked;
  246.         settings.autoScrape = ui.querySelector('#opt-autoscrape').checked;
  247.         settings.darkMode = ui.querySelector('#opt-dark').checked;
  248.         settings.showPreview = ui.querySelector('#opt-preview').checked;
  249.         settings.ocrLang = ui.querySelector('#opt-lang').value;
  250.         settings.splitEvery = parseInt(ui.querySelector('#opt-split').value);
  251.         saveSettings();
  252.         applyDarkMode();
  253.       };
  254.     });
  255.  
  256.     ui.querySelector('#btn-scrape').onclick = () => {
  257.       const pages = [...document.querySelectorAll(
  258.         '.page, .reader_column, [id^="page_container"], .outer_page, .abs_page, .scribd_page, .text_layer'
  259.       )];
  260.       if (!pages.length) return alert('❌ No pages found.');
  261.       scrapePages(pages, preview);
  262.     };
  263.     ui.querySelector('#btn-export').onclick = () => exportOutput(preview.textContent, 'txt');
  264.     ui.querySelector('#btn-html').onclick = () => exportOutput(`<html><body><pre>${preview.textContent}</pre></body></html>`, 'html');
  265.     ui.querySelector('#btn-print').onclick = () => printToPDF(preview.textContent);
  266.   }
  267.  
  268.   window.addEventListener('load', () => {
  269.     applyDarkMode();
  270.     unblurContent();
  271.     const preview = createPreview();
  272.     createUI(preview);
  273.     if (settings.autoScrape) document.querySelector('#btn-scrape').click();
  274.   });
  275. })();