pagexml-tools

Utility functions for reading PageXML files

installing

using poetry

poetry add pagexml-tools

using pip

pip install pagexml-tools

Using

PageXML-tools contains functions for parsing and for a range of analysis tasks.

Parsing PageXML files and the Physical Document model

There is a tutorial that demonstrates the physical document model API

PageXML-tools contains basic functionality for parsing a PageXML file that returns a document model representing the content of the file. The HTR/OCR process that generates PageXML, recognises text in an image of a physical document.

from pagexml.parser import parse_pagexml_file

pagexml_file = "path/to/pagexml_file.xml"

page_doc = parse_pagexml_file(pagexml_file)

# a page document has an ID
print(page_doc.id)

# print descriptive statistics
print(page_doc.stats)

# iterative over text regions and lines
for tr in page_doc.text_regions:
    # a text_region has an ID and a bounding box derived from its coordinates
    print(tr.id, tr.coords.box)
    # a text_region can have sub-text_regions and lines
    for line in tr.lines:
        # a line has an ID, coordinates and text
        print(line.id, line.coords.box, line.text)

In addition to the basic parsing and handling of PageXML output, there is functionality to support a range of tasks:

• reading sets of PageXML files from a archive (tar, zip) file (tutorial),
• searching in text (keyword in context, keywords or fuzzy search)
• reading and working with tables (table processing)
• classifying physical document types in a large set of PageXML documents (tutorial),
• checking the quality of the HTR/OCR process (tutorial),
• comparing subsets (tutorial),
• identifying document sections in sequences of PageXML documents (tutorial),
• turning text lines into running text (tutorial),
• supporting different reading orders (tutorial),
• reinterpreting and restructuring text regions and lines (tutorial),
• turning physical structure into logical structure,

Using PageXML-tools in R:

#Python
# run in CMD (Command) first (first time only): 
py -m pip install pagexml-tools
# Install and start Python 3.11 in background 

# R
# run in different window in R first time only! 
install.packages("reticulate")
# run in different window in R first: 
library(reticulate)
# run in different window in R next: 
py_require(c("pagexml-tools")) 

# Import PageXML file parser:

from pagexml.parser import parse_pagexml_file

Thanks to Milan van Lange for this R example.

USAGE | CONTRIBUTING | LICENSE