用于在OCR PDF中搜索空格/断行词的正则表达式（goo d ni g ht）

Question

I need searching lots of OCR PDFs. I realized the words and sentences are perfect visually, but if I copy an paste the content, there are spaces which shouldn't be there!

I can see in the text: good night

If I copy and paste somewhere: goo d ni g ht

I would appreciate advices to handle this situation through a Regex expression considering:

a) The simple example for short words as \\bgood night\\b for goo d ni g ht

b) When there is line break in the sentence. I mean, the Regex expression isn't able to search from one line to another in the PDF even the paragraph is the same. In looking for \\bthe sun set and the night comes\\b , but the PDF content is like that when pasted:

line 1: t he sun set an d th e

line 2: nig ht co m es

Many thanks, Cadu

Answer 1

This random occurence of spaces in the middle of words can happen in PDF. The reason behind it is the complex format that PDF actually is. You see, a PDF document is actually a container of instructions for rendering the text in a viewer.

Imagine instructions like:

• go to position 50, 50.
• draw the character 'G'
• go to position 56, 50.
• draw the character 'O'
• etc

Whenever you select something in a viewer (for instance Adobe), the program has to figure out what content overlaps with your selection (already this is not an easy problem). If it's text, it then needs to decide where to add spaces and line-breaks. Different viewers (or software) might use different metrics for this. A typical one for instance is "insert a space if two characters are further apart than the width of the space character in the same font"

The point is, getting text out of a PDF document is always kind of guesswork. And if you add the fact that it's an OCR PDF, you are adding a further layer of difficulties.