Proofreading with Crowbook

Crowbook includes some proofreading features, that can be enabled if you set one of the

output.proofread.html
output.proofread.html_dir
output.proofread.pdf

output files (or include proofread.pdf in the list of formats to render to output). This allows you to generate different files for publishing and proofreading (you probably don't want to publish a version that highlights your grammar errors or your repetitions).

Current proofreading features are:

repetition detection;
grammar check.

Enabling proofreading

Since proofreading can take quite a lot of time, particularly for a long book, it is disabled by default. You'll have to run

$ crowbook --proofread my.book

$ crowbook -p my.book

to generate proofreading copies. Alternatively, if you want it to be activated each time you run crowbook on this book (which is not recommanded for long books, particularly if you want to perform a grammar check), you can set

proofread: true

in the book configuration file.

Repetition detection

Repetition detection is enabled with:

proofread.repetitions: true

It uses Caribon library to detect the repetition in your text. Since the notion of a repetition is relatively arbitrary, it is possible to adapt the settings. Default are:

# The maximum distance between two identical words to
# consider them a repetition 
proofread.repetitions.max_distance: 25
# The minimal number of occurences to consider it a repetition
proofread.repetitions.threshold: 2.0
# Ignore proper nouns (words starting by a capital,
# not at a beginning of a sentence)
proofread.repetitions.ignore_proper: true

# Activate fuzzy string matching
proofread.repetitions.fuzzy: true
# The maximal ratio of difference to consider
# that two words are identical
# (E.g., with 0.2, "Rust" and "Lust" won't be
# considered as the same word, but they will be with 0.5)
proofread.repetitions.fuzzy.threshold: 0.2

For more information, see Caribon's documentation.

Currently, repetitions are not displayed in PDF proofreading output.

Grammar checking

With Languagetool

Crowbook can use LanguageTool to detect grammar errors in your text. It is, however, a bit more complex to activate.

First, you'll have to activate this feature in your book configuration file:

# Activate language tool support
proofread.languagetool: true
# (Optional) Sets the port number to connect to (default below)
proofread.languagetool.port: 8081

You'll then have to download the stand-alone version of LanguageTool. It includes a server mode, which you'll have to launch:

$ java -cp languagetool-server.jar org.languagetool.server.HTTPServer --port 8081

You can also use the LanguageTool GUI (languagetool.jar) and start the server from the menu "Text Checking -> Options". This also allows you to configure LanguageTool more precisely by activating or deactivating rules.

You can then run Crowbook, and it will highlight grammar errors in HTML or PDF proofreading output files.

Note: running a grammar check on a long book (like a novel) can take up to a few minutes.

With Grammalecte

Grammalecte is a grammar checker specialized for the french language. If the language of your book is french, you can use it in a similar fashion to languagetool:

# Activate grammalecte support
proofread.grammalecte: true
# (Optional) Sets the port number to connect to (default below)
proofread.grammalecte.port: 8080

You'll also need to run the Grammalecte server. First download the CLI and server version, then:

$ python3 server.py

You can then run Crowbook with --proofread to check the grammar of your book. It is possible to run both LanguageTool and Grammalecte on the same book (though might take a while for a long book...).

4.1 KiB Raw Permalink Blame History