Noodlesoft Forums

What is the best way to evaluate a large number of pdf files and see which have already undergone OCR?

I use hazel to process pdf files and OCR them but i do not want to run them thru OCR twice.

Thank you.

There's no standard way to detecting that but search the forums as I believe someone has come up with a script to check whether any fonts are used, which is usually a good indication.

I use Spotlight or Finder to search "contents" for "a" on the basis that the vast majority of documents contains "a" and if pdf has been ocr'd, then Finder contents will find an "a" as text. Is this thinking correct ?
By default, Spotlight searches insides files for text. But using Finder, I search for Kind pdf first, then + to add Contents search for "a" .

I have a Hazel folder - called ScanThis - to scan using Pdfpen and colour file green and purple - so I know which files have text layer. Katie Floyd has a script for this.

Possibly. Not sure how reliable that is. The script others are using look for a font directive which usually indicates text content.

SimonBrowneNZ wrote:I use Spotlight or Finder to search "contents" for "a" on the basis that the vast majority of documents contains "a" and if pdf has been ocr'd, then Finder contents will find an "a" as text. Is this thinking correct ?
By default, Spotlight searches insides files for text. But using Finder, I search for Kind pdf first, then + to add Contents search for "a" .

I have a Hazel folder - called ScanThis - to scan using Pdfpen and colour file green and purple - so I know which files have text layer. Katie Floyd has a script for this.

Could you please post the code for how you perfoem the above check? Thanks!

I use a program called ‘pdftotext’ from https://www.bluem.net/files/pdftotext.dmg

It is abandonware (no longer being developed) but it works for now, and if it stops working in the future, I’ll deal with it then.

The use is very simple:

Code: Select all

pdftotext file.pdf file.txt


then check to see if ‘file.txt’ is just a newline. If it is, that means it wasn’t able to find any text in ‘file.pdf’.

I’ve yet to find a more reliable test.

You may also try Grooper to tackle this particular problem: https://www.bisok.com/grooper-data-capt ... -pass-ocr/ . I have used it for this job several times and have been pretty happy with the results. Good luck!

Steelersfan, thanks, that works!

Hi, here is my Hazel rule

Code: Select all

If all of the following conditions are met

Kind is PDF
Contents do not contain match a


Code: Select all

Do the following to the matched file or folder:
Run AppleScript embedded script
Toggle extension visible
Rename with pattern: name extension
Continue matching rules


embedded script:

Code: Select all

tell application "PDFpenPro"
   open theFile as alias
   tell document 1
      ocr
      repeat while performing ocr
         delay 1
      end repeat
      delay 1
      close with saving
   end tell
   quit
end tell


gacorrea wrote:

SimonBrowneNZ wrote:I use Spotlight or Finder to search "contents" for "a" on the basis that the vast majority of documents contains "a" and if pdf has been ocr'd, then Finder contents will find an "a" as text. Is this thinking correct ?
By default, Spotlight searches insides files for text. But using Finder, I search for Kind pdf first, then + to add Contents search for "a" .

I have a Hazel folder - called ScanThis - to scan using Pdfpen and colour file green and purple - so I know which files have text layer. Katie Floyd has a script for this.

Could you please post the code for how you perfoem the above check? Thanks!

I have had success with the following shell script to determine if the PDF has been OCR'd it is pretty fast...

Code: Select all

#! /bin/bash
if ! grep Font "$1"
then
     exit 0
else
     exit 1
fi