GUIDE深度 · Deep dive
Text Layer or OCR? Check First
PDFs come in two broad types: text-layer (selectable/searchable) and scanned images. Extract the text layer first for speed/accuracy; OCR only scanned pages. Text layers can still be garbled/incomplete, so spot-check exports.
Selectable/copyable text = text layer.
Only the whole page image selects = scan.
Search misses can also mean a scan.
Numbers/amounts/dates/IDs.
Negations/proper nouns.
FAQ
Do I still OCR a text-layer PDF?
Usually not; OCR can turn correct text into errors.
Can OCR-ed scans be searched?
The exported TXT can; the original PDF needs to be saved as searchable.