Highlight text in documents

txtmarker highlights text in documents. txtmarker takes a list of (name, text) pairs, scans an input document and creates a modified version with highlights embedded.

Current file formats supported:

• pdf

The easiest way to install is via pip and PyPI

pip install txtmarker

Python 3.10+ is supported. Using a Python virtual environment is recommended.

txtmarker can also be installed directly from GitHub to access the latest, unreleased features.

pip install git+https://github.com/neuml/txtmarker

The examples directory has a series of examples and notebooks giving an overview of txtmarker. See the list of notebooks below.

Notebook	Description
Introducing txtmarker	Overview of the functionality provided by txtmarker
Highlighting with Transformers	AI-driven highlighting with Transformers

The following section gives an overview of highlighters and available methods/configuration. See the notebooks above for detailed examples.

Creates a new highlighter instance.

from txtmarker.factory import Factory
highlighter = Factory.create("pdf")

extension: string

Type of highlighter to create (i.e. pdf)

formatter: callable

Formats queries and input text using this method. Helps with cleanup of files with lots of symbols and other content.

chunks: int

Splits queries into multiple chunks. This is designed for very long text matches.

Extracts page text from infile and returns as a generator. This enables analysis on the text exactly as it will appear to the highlighter.

highlighter.pages("input.pdf")

infile: string

Full path to input file

Highlights using provided annotations. Annotated file is stored as outfile.

highlighter.highlight("input.pdf", "output.pdf", [("name", "text to highlight")])

infile: string

Full path to input file

outfile: string

Full path to output file, i.e. the highlighted file

highlights: list of (string, string|regex)

List of highlight elements. Each pair has a name (can be None) and text value. The text can either be a string or a regular expression. When using string matching, make sure to escape regular expressions (i.e. call re.escape).