{
 "title": "Print-to-PDF versus screenshot capture: what each route keeps",
 "measured": "2026-08-01",
 "document": "English Wikipedia article 'PDF', same URL, same day, both routes",
 "method": {
  "print": "Chrome headless --print-to-pdf, Letter, no header/footer, 15s virtual time budget",
  "screenshot": "Full Page PDF Snap 2.16.0, single-page output, then Tesseract 5 at 150 dpi",
  "ground_truth": "Wikipedia REST API plain text, same day"
 },
 "results": {
  "print": {
   "pages": 26,
   "page_breaks": 25,
   "breaks_mid_sentence": 9,
   "words": 9620,
   "recall_percent": 94.8,
   "embedded_images": 2,
   "file_mb": 1.1,
   "text_layer": true
  },
  "screenshot": {
   "pages": 1,
   "page_breaks": 0,
   "breaks_mid_sentence": 0,
   "words": 8316,
   "recall_percent": 92.7,
   "file_mb": 6.7,
   "text_layer": false
  }
 },
 "reading": "Print wins on text fidelity and file size; it has a real text layer and needs no OCR. Screenshot wins on continuity: one uninterrupted sheet against 26 pages, of which 9 breaks cut a sentence in half. Which matters depends on whether you need the words or the page.",
 "limitations": "One document, single runs. Wikipedia prints comparatively well; pages with heavy CSS layout diverge more.",
 "license": "CC BY 4.0",
 "article": "https://provinglab.dev/measurements/print-to-pdf-vs-screenshot/"
}