Add comments to PDF files automagically with regular expressions [closed]

Question

I've been grading academic papers for a couple of years now and I've started to see numerous patterns in spelling and grammer mistakes. Also, I've noticed that less experienced academics tend to use certain constructs that immediately raise "smells" to more experienced researchers.

I would like to automagically recognize and annotate these in PDF files. Is anyone aware of a script that I could use to automagically annotate and comment PDF files? Perhaps it's dead simple, but I feel like I'm one of the first ones to ask this question.

Programming is no problem.

Answer 1

To solve this task, you need 3 things:

1. A good text extracting tool to get the contents from the PDFs (you're basically asking for this).
2. The knowledge about what keywords you want to use in order to create appropriate textual notes/comments and trigger a PDF annotation automatism (you say you have this).
3. A method to insert your comments into the PDF, preferably on the correct pages, or even on the exactly correct spot on the page (you're asking for this).

Text extraction

PDFlib's TET (text extraction toolkit) lets you extract text from any PDF. It's the most powerful of available PDF text extraction tools out there that allows you access via commandline and scripting. It can handle such weirdies (from the p.o.v. of text extraction) as ligatures as well as different text encodings. More important, it can tell you the exact page number and coordinates on the PDF page for any character or text string it extracted.

Inserting PDF annnotations

After you parsed the text, and your logic decided which comment to add for which page, you can use PDFlib or Ghostscript to add comments ("annotations") to the original PDF.

I'm not delivering a tutorial about how to use PDFlib in order to add annotations to existing PDFs here. But I will leak some insider knowledge about how Ghostscript can do it:

Using Ghostscript for adding annotations to PDFs

To add an annotation with Ghostscript to an existing PDF, first create a text file called my-pdfmarks.txt (or whatever name you prefer). Now type into that textfile the content of your annotation, using the following syntax:

 [ /Title (Annotation experiments by -pipitas-)
   /Author (pipitas)
   /Subject (I'm trying to add annotations to existing PDFs with the help of Ghostscript...)
   /Keywords (comma, separated, keywords, spelling mistakes, grammar mistakes, raising "smells")
   /ModDate (D:20101219192842)
   /CreationDate (D:20101219092842)
   /Creator (pipitas' brainz)
   /Producer (Ghostscript under the direction of pipitas)
   /DOCINFO pdfmark

 [ /Contents (Smell: This statement was bloody well rebutted by decades of academic research...)
   /Rect [10 10 50 50]
   /Subtype /Text
   /Name Note
   /SrcPg 2
   /Open true
   /ModDate (D:20101220193344)
   /Title (A Comment on Page 2)
   /Color [.5 .5 0]
   /ANN pdfmark

Then, run Ghostscript command like the following. I'm assuming Windows now -- for Linux/Unix/MacOSX use gs instead of gswin32c.exe for the executable, and use \ instead of ^ for the line continuation marks:

gs ^
  -o original-annotated.pdf ^
  -sDEVICE=pdfwrite ^
  -dPDFSETTINGS=/prepress ^
   original.pdf ^
   my-pdfmarks.txt

Voila! Your output PDF now has an annotation on page 2.

Now you probably didn't understand what exactly you were doing:

• The first part of the my-pdfmarks.txt file manipulates the PDF's meta data. Just delete it if you don't want this.
• The second part adds an annotation ('/Subtype /Text' and '/Name /Note') on Page 2 ('/SrcPg 2') of the output PDF at the lower left corner, 10 points away from each page border ('/Rect [10 10 50 50]'), using a greenish DeviceRGB color ('/Color [0.5 0.5 0]'), and opening it by default ('/Open true') when accessing the page.

Tweakable parameter values (after each keyword) in the my-annotations.txt file are all BUT the following:

1. "/DOCINFO pdfmark"
2. "/Subtype /Text"
3. "/Name /Note"
4. "/ANN pdfmark"

For example, to make the annotation appear in pure red, use /Color [1 0 0].

In order to fully understand the pdfmark syntax (and add more tweaks to your procedure), you'll need to google for Adobe's pdfmark Reference Manual and read that.

Since you said 'programming is no problem' you now have all the building blocks to automate this with any scripting language of your choice.

Answer 2

If I were you I would start with the PDF Library SDK which supports the things you're looking for:

• Extract content
• Add comments to documents

One drawback is that you have to apply for it and Adobe may refuse your request.

EDIT:

PDFedit seems promising. It's an open source GUI application that allows you to modify PDF manually or by scripting.