hocr_utils

hocr_utils is a package to transform, plot and simplify the use of hOCR files.

Installation

Dependencies

hocr-utils requires:

• Python (>= |3.7|)

Optional Dependencies

The functions to plot, transform pdf into hOCR require the following additional dependencies:

• pytesseract
• pdf2image
• opencv-python

Additionaly tesseract language pack need to be install for non-english ocr.

Example: install french language package on ubuntu with:

apt-get install tesseract-ocr-fra

User Installation

The easiest way to install scikit-learn is using pip:

pip install -U hocr_utils

Usecases

Transform PIL Images to hOCR

Requires pytesseract dependency and the requested tesseract language pack.

from hocr_utils import utils
from PIL import Image

image = Image.open('./data/sample.png')
hocr = utils.images_to_hocr([image])

Transform pdf to hOCR

Requires pytesseract, pdf2image dependencies as well as the requested tesseract language pack.

from hocr_utils import utils

hocr = utils.pdf_to_hocr('./data/sample.pdf')

Transform hOCR to list of dictionary

from hocr_utils import utils

hocr_dict = utils.hocr_to_dict(hocr)

This can then be transformed into pandas dataFrame using pd.dataFrame.from_records(hocr_dict)

By default there will be one entry per line in the hOCR, However it is possible to group the list by ['paragraph', 'line', 'word'] using by argument.

Get a single page from hOCR

from hocr_utils import utils

hocr_1 = utils.get_page(hocr, 1)