python-hocr

HOCR conversion to csv based on https://github.com/concordusapps/python-hocr.

Installation

This has been lightly tested on python 2.7 and 3.5, but is still pretty rough. You'll need to install it on your own, though that shouldn't be too hard.

1. Make a new virtualenv for the code to live in by typing something like: $ virtualenv python-hocr_env You can call it whatever you like, I'm just using python-hocr_env as an example.

2. Activate that virtual env, either with the workon command (if virtualenvwrapper is installed) or directly by typing something like: $ source python-hocr_env/bin/activate

3. Get the raw code with $ git clone https://github.com/jsfenfen/python-hocr.git

4. Install the requirements by using pip install -r requirements_dev.txt

convert_hocr.py

Convert hOCR files to .csv or .json format. Assumes that each hOCR file has only one closing </html> tag; some hOCR outputs mangle html by giving each pages an opening and closing html tag (as opposed to just an opening and closing ocr_page tag). This script will only convert the content that appears before the first html closing tag in such files; it's recommended that you pre-process such files ahead of time.

$ python convert_hocr.py  --help
usage: hocr2csv [-h] [--pages PAGES [PAGES ...]] [--format {csv,json}]
                infile outfile

positional arguments:
  infile
  outfile

optional arguments:
  -h, --help            show this help message and exit
  --pages PAGES [PAGES ...]
  --format {csv,json}

examples:

python convert_hocr.py infile.html --pages=1-4 infile.csv

page ranges are inclusive.

python convert_hocr.py infile.html --format=json infile.json

License

Unless otherwise noted, all files contained within this project are liensed under the MIT opensource license. See the included file LICENSE or visit opensource.org for more information.