feup-infolab / latex-index-tools

LaTeX index generation tools for dissertations, theses, books, or reports.

latex-index-tools

LaTeX index generation tools for dissertations, theses, books, or reports.

Instalation

Requirements:

• python 3.x (tested using python 3.7.6)
• pip (tested using pip 20.0.2)
• detex

Install dependencies using:

pip install -r requirements.txt

Dependencies:

• pyahocorasick 1.1.0 (we use iter_long() from this branch: git+https://github.com/WojciechMula/pyahocorasick.git@iter_long)
• python-rake 1.4.5
• yake 0.4.3
• nltk 3.4.5

Usage

While nothing replaces the manual selection of concepts to add to your back-of-the-book index, we can make this job easier by relying on automatic tools. LaTeX Index Tools enables you to automatically extract the most salient concepts, via keyword extraction, providing an easy way to edit them and add additional matching expressions for the same index entry:

./extract_concepts.py -i thesis.tex \
  -o reviewed_concepts.csv -ml 2 -n 1000 -m rake -s

This will use detex to extract the text from the LaTeX document, considering the content from any \input and \include. The extracted keywords will be saved to a CSV file with three columns: concept, match and weight. The concept is the same as the index entry, while match is the text after which the index entry should be added --- thus, multiple rows containing the same concept, but different match entries can be added. In this example, the minimum length of concepts is 2, and we extract 1000 concepts using RAKE. The -s option is used to provide an interactive command line review of concepts. It can be removed for a fully automatic extraction process, or when manually editting the resulting CSV file is preferred.

After the CSV file containing concepts and matches is prepared, we can automatically add the index entries to our LaTeX files:

./index_concepts.py -i thesis.tex -c reviewed_concepts.csv \
  -o /tmp/output_dir -e arsclassica-settings.tex \
  -e listings-settings.tex -e hyphenations.tex -e "FrontBackMatter/*.tex" \
  -m stop -m dist

All LaTeX files within the thesis.tex directory are considered. We do not parse the main LaTeX to find dependencies. Instead, we simply consider all *.tex files, recursively, excluding those that do not matter using the -e option. No chances will be done to your files. Instead, a new copy will be created in the directory pointed out by -o. We also provide heuristics to avoid overcrowding the back-of-the-book index:

• stop, which relies on the THESIS_STOPWORDS constant within index_concepts.py, which you can edit to setup your own stopword-like concepts to be ignored (use the distribution of concepts provided as output of this script to and a cutoff of about 100, at most, to select these stopwords;
• dist will ensure a given distance (set in the DISTANCE constant withint index_concepts.py) in number of lines between concept mentions (this is reset between chapters).