Dockerfile for LibreGrammar

This repository contains a Dockerfile to create a Docker image for LibreGrammar, a LanguageTool fork maintained by TiagoSantos81.

The main repository can be found on GitLab. There is also a mirror, mainly used for following the original project available on GitHub.

I wrote this image since I'm looking for a Job, so I can't afford to pay LanguageTool premium and LibreGrammar activates most of the rules.

Setup

Prebuilt images

There is an image build in the gitlab registry. This image is automatically built and updated, using GitLab CI/CD, each time a new tag is pushed to the repo. You can just pull it using:

docker pull registry.gitlab.com/py_crash/docker-libregrammar

This would pul the image with the latest tag. If you want an specific image you can browse the registry

Setup using the Dockerfile

This approach could be used when you plan to make changes to the Dockerfile.

git clone https://github.com/py-crash/docker-libregrammar.git -b libregrammar --config core.autocrlf=input
cd libregrammar
docker build -t libregrammar .
docker run --rm -it -p 8081:8081 libregrammar

Configuration

Java heap size

LibreGrammar will be started with a minimal heap size (-Xms) of 256m and a maximum (-Xmx) of 512m. You can overwrite these defaults by setting the environment variables Java_Xms and Java_Xmx.

An example startup configuration:

docker run --rm -it -p 8081:8081 -e Java_Xms=512m -e Java_Xmx=2g libregrammar

LibreGrammar HTTPServerConfig

You are able to use the HTTPServerConfig configuration options by prefixing the fields with langtool_ and setting them as environment variables.

An example startup configuration:

docker run --rm -it -p 8081:8081 -e langtool_pipelinePrewarming=true -e Java_Xms=1g -e Java_Xmx=2g libregrammar

Using n-gram datasets

LibreGrammar can make use of large n-gram data sets to detect errors with words that are often confused, like their and there.

Source: https://dev.languagetool.org/finding-errors-using-n-gram-data

Download the n-gram dataset(s) to your local machine and mount the local n-gram data directory to the /ngrams directory in the Docker container using the -v configuration and set the languageModel configuration to the /ngrams folder.

An example startup configuration:

docker run --rm -it -p 8081:8081 -e langtool_languageModel=/ngrams -v local/path/to/ngrams:/ngrams libregrammar

Improving the spell checker

You can improve the spell checker without touching the dictionary. For single words (no spaces), you can add your words to one of these files:

• spelling.txt: words that the spell checker will ignore and use to generate corrections if someone types a similar word
• ignore.txt: words that the spell checker will ignore but not use to generate corrections
• prohibited.txt: words that should be considered incorrect even though the spell checker would accept them

Source: https://dev.languagetool.org/hunspell-support

The following Dockerfile contains an example on how to add words to spelling.txt. It assumes you have your own list of words in en_spelling_additions.txt next to the Dockerfile. It assumes you already built the LibreGrammar image.

FROM registry.gitlab.com/py_crash/docker-libregrammar

# Improving the spell checker
# http://wiki.languagetool.org/hunspell-support
USER root
COPY en_spelling_additions.txt en_spelling_additions.txt
RUN  (echo; cat en_spelling_additions.txt) >> org/languagetool/resource/en/hunspell/spelling.txt
USER libregrammar

You can build & run the custom Dockerfile with the following two commands:

docker build -t libregrammar-custom .
docker run --rm -it -p 8081:8081 libregrammar-custom

You can add words to other languages by changing the en language tag in the target path. Note that for some languages, e.g., for nl the spelling.txt file is not in the hunspell folder: org/languagetool/resource/nl/spelling/spelling.txt.

Docker Compose

This image can also be used with Docker Compose. An example docker-compose.yml would be:

version: "3"

services:
  libregrammar:
    build: ./docker-libregrammar # For building it yourself
    image: registry.gitlab.com/py_crash/docker-libregrammar # For using the prebuilt image
    container_name: libregrammar
    ports:
        - 8081:8081  # Using default port from the image
    environment:
        - langtool_languageModel=/ngrams  # OPTIONAL: Using ngrams data
        - Java_Xms=512m  # OPTIONAL: Setting a minimal Java heap size of 512 mib
        - Java_Xmx=1g  # OPTIONAL: Setting a maximum Java heap size of 1 Gib
    volumes:
        - /path/to/ngrams/data:/ngrams

This assumes you have cloned the repo into a folder called docker-libregrammar in the same path as your docker-compose.yml

Podman

This image can be also be build and run using rootless podman. In fact, I use podman myself on my computer. Just replace docker for podman and it should work:

$ podman build -t libregrammar ./
$ podman run -d --rm -it -p 8081:8081 -e langtool_languageModel=/ngrams -e Java_Xms=1g -e Java_Xmx=3g -v /path/to/ngrams:/ngrams localhost/libregrammar

Usage

By default this image is configured to listen on port 8081 (the default port for LibreGrammar).

An example cURL request:

curl --data "language=en-US&text=a simple test" http://localhost:8081/v2/check

Please refer to the official LanguageTool documentation and to the Libregrammmar Repo for further usage instructions.