knowledge-clustering

Command-line tool to help with the use of the knowledge LaTeX package. A tutorial on how to use both knowledge and knowledge-clustering can be found here.

Principle

The goal of knowledge-clustering is to help the user write a LaTeX document with the knowledge package. It has three features:

• Clustering: provide suggestions to the user of what notions should be grouped together.
• Add quotes: find where you might have missed some quotes in your document.
• Anchor points: find where you might have missed anchor points in your document.

The clustering algorithm is meant to be used while writing your document, while the last two tools should be used when your document is (nearly) ready to be published, to check if everything is right.

Installation

To install (or upgrade) knowledge-clustering, you need to have Python 3.9 (or a more recent version), and then run

python3 -m pip install --upgrade knowledge-clustering

and then

knowledge init

To check if you have the latest version of knowledge-clustering, you can run

knowledge --version

Clustering notions

Syntax

Usage: knowledge cluster [OPTIONS]

  Defines, as a comment and in the knowledge files, all the knowledges
  occuring in the file.

Options:
  -k, --knowledge FILE        File containing the knowledges that are already
                              defined. Multiple files are allowed; new
                              knowledges will be written in the last one. If
                              the option is not specified, all .kl file in the
                              current directory (and subdirectory,
                              recursively) will be taken. If there are
                              multiple files, exactly one of them must end
                              with `default.kl`.
  -d, --diagnose FILE         Diagnose file produced by LaTeX. If the option
                              is not specified, the unique .diagnose file in
                              the current directory (and subdirectory,
                              recursively) is taken instead.
  -l, --lang [en|fr]          Language of your TeX document.
  -S, --scope / --no-scope    Print the scopes defined in the knowledge file
                              and print the possible meaning of those scope
                              inferred by knowledge-clustering.
  -P, --print / --no-print    Print all new knowledges.
  -N, --no-update / --update  Don't look on PyPI if a newer version of
                              knowledge-clustering is available.
  -c, --config-file TEXT      Specify the configuration file. By default the
                              configuration file in the folder
                              /Users/rmorvan/knowledge-
                              clustering/knowledge_clustering/data
                              corresponding to your language is used.
  --help                      Show this message and exit.

Example

Example files can be found in the examples/ folder.

While writing some document, you have defined some knowledges in a file called preservation.kl (distinct from your main LaTeX). You continued writing your LaTeX document (not provided in the examples/ folder) for some time, and used some knowledges that were undefined. When compiling, LaTeX and the knowledge package gives you a warning and writes in a .diagnose file some information explaining what went wrong. This .diagnose file contains a section called "Undefined knowledges" containing all knowledges used in your main LaTeX file but not defined in preservation.kl. We reproduced this section in the preservation.diagnose file.

Normally, you would add every undefined knowledge, one after the other, in your preservation.kl. This is quite burdensome and can largely be automated. This is precisely what knowledge-clustering does: after running

knowledge cluster -k preservation.kl -d preservation.diagnose

your file preservation.diagnose is left unchanged but preservation.kl is updated with comments.

The cluster command is optional: you can also write knowledge -k preservation.kl -d preservation.diagnose.

Now you simply have to check that the recommendations of knowledge-clustering are correct, and uncomment those lines.

Autofinder

If the current directory (and its recursive subdirectories) contains a unique .diagnose file and a unique .kl file, you can simply write knowledge cluster (or knowledge): the files will be automatically found.

Multiple knowledge files

If you have multiple knowledge files, you can use the -k option multiple times. For instance, you could write:

knowledge cluster -k 1.kl -k 2.kl -d ordinal.diagnose

Synonyms of knowledges defined in 1.kl (resp. 2.kl) will be defined, as comments, in 1.kl (resp. 2.kl). New knowledges will always be added, as comments, to the last file, which is 2.kl in the example.

You can also use the autofinder in this case, using knowledge cluster or knowledge: if multiple .kl files are present in the current directory (and its recursive subdirectories), exactly one of them must end with default.kl—this is where new knowledges will be put.

Adding quotes

/!\ This feature is somewhat experimental.

Usage: knowledge addquotes [OPTIONS]

  Finds knowledges defined in the knowledge files that appear in tex file
  without quote symbols. Proposes to add quotes around them.

Options:
  -t, --tex FILE              Your TeX file.  [required]
  -k, --knowledge FILE        File containing the knowledges that are already
                              defined. Multiple files are allowed; new
                              knowledges will be written in the last one. If
                              the option is not specified, all .kl file in the
                              current directory (and subdirectory,
                              recursively) will be taken. If there are
                              multiple files, exactly one of them must end
                              with `default.kl`.
  -p, --print INTEGER         When finding a match, number of lines (preceding
                              the match) that are printed in the prompt to the
                              user.
  -N, --no-update / --update
  --help                      Show this message and exit.

After running

knowledge addquotes -t mydocument.tex -k knowledges1.kl -k knowledges2.kl

your prompt will propose to add quotes around defined knowledges, and to define synonyms of knowledges that occur in your TeX file. For instance, if "algorithm" is a defined knowledge and "algorithms" occurs in your TeX file, then it will propose to you to define "algorithms" as a synonym of the knowledge "algorithm", and to add a pair of quotes around the string "algorithms" that occurs in your TeX file.

Whenever the algorithm finds a match for a knowledge, it will print the line of the document where it found the match, and emphasize the string corresponding to the knowledge. If you want to print more than one line, you can use the -p (or --print) option to print more than one line.

Finding missing anchor points

Usage: knowledge anchor [OPTIONS]

  Prints warning when a knowledge is introduced but is not preceded by an
  anchor point.

Options:
  -t, --tex FILE              Your TeX file.  [required]
  -s, --space INTEGER         Number of characters tolerated between an anchor
                              point and the introduction of a knowledge.
                              (Default value: 200)
  -N, --no-update / --update
  --help                      Show this message and exit.

When one runs

knowledge anchor -t mydocument.tex

the tool will print the lines of the document containing the introduction of a knowledge that is not preceded by an anchor point. The tolerance on how far away the anchor point can be from the introduction of a knowledge can be changed with the -s (or --space) option. The default value is 150 characters (corresponding to 2-3 lines in a TeX document).

Devel using virtualenv

Using venv and the --editable option from pip allows for an easy setup of a development environment that will match a future user install without the hassle.

For bash and Zsh users

python3 -m venv kl.venv
source ./kl.venv/bin/activate
python3 -m pip install --editable .

For fish users

python3 -m venv kl.venv
source ./kl.venv/bin/activate.fish
python3 -m pip install --editable .

FAQ

• knowledge: command not found after installing knowledge-clustering

  Make sure you have Python>=3.9.

• When running knowledge, I obtain a long message error indicating "Resource punkt not found."

  Run knowledge init.

• My shell doesn't autocomplete the command knowledge.

  Depending on whether you use zsh or bash write

  eval "`pip completion --<shellname>`"
  

  (where <shellname> is either zsh or bash) in your .zshrc (or .bashrc) file and then, either launch a new terminal or run source ~/.zshrc (or source ~/.bashrc).

• Error: Got unexpected extra argument when using multiple knowledge files.

  You should use the option -k before every knowledge file, like in

  knowledge cluster -k 1.kl -k 2.kl -d blabla.diagnose 
  

• I've updated knowledge-clustering but I still don't have the last version (which can be checked using knowledge --version): This can happen if you have multiple versions of python (and multiple versions of knowledge-clustering).

  Type where python3, and uninstall knowledge-clustering from everywhere (using <path>/python3 -m pip uninstall knowledge-clustering). Try to then reinstall knowledge-clustering by running python3 -m pip install --upgrade knowledge-clustering.