imgpdf2text (shell script)

Automated text extraction of a especific PDF made of images

Description

3 step process

1. Extract pdf images (pdfimages)
2. Extract text from images (tesseract OCR)
3. Concat all text files

Caveat

The first step extracts all the images and removes those without text.

This process is tuned for the test pdf and probably will fail in other scenarios.

Caveat explained

The test pdf has multiple images but only some of them have the targeted text.

The list of images is show with:

$ pdfimages -list "$TARGET_IMGPDF"
page   num  type   width height color comp bpc  enc interp  object ID
---------------------------------------------------------------------
   1     0 image    1240  1753  rgb     3   8  jpeg   no        56  0
   1     1 mask      304    76  -       1   1  ccitt  no        57  0
   1     2 mask      368    36  -       1   1  ccitt  no        58  0
   1     3 mask     1968  3092  -       1   1  ccitt  no        59  0
   1     4 mask      152    96  -       1   1  ccitt  no        60  0
   2     5 image    1240  1753  rgb     3   8  jpeg   no         4  0
# ...etc

The target images share the attribute of having 3000+ height, so that "fingerprint" is used.

Target images id adquisition:

TARGET_IMAGE_IDS=$(pdfimages -list "$TARGET_IMGPDF" | grep -E '3[0-9]{3}' | awk '{ print $2; }')

That behaviour could be changed at "function step1()" for other scenarios.

Let me know by means of Github issues of any problems, solutions, proposals. :D

Usage

./imgpdf2text.sh $TARGET_FILE

Outputs creates a file named $TARGET_FILE.txt with the text extracted

Dependencies

• bash (4.3.11(1)-release)
• pdfimages (0.24.5)
• tesseract (3.03)

Install

Linux Mint 17 Qiana (Ubuntu 14.04)

sudo apt-get install poppler-utils
sudo apt-get install tesseract