This is basically an upgrade of this repository, which in turn is based on this repository and this repository. Things that I have fixed/changed in the Dockerfile:
- starting image: httpd-buster, as python2.7 is no longer supported by newer version of Debian (and httpd)
- debian package: python is now python2.7
- symbolic link for Python, as Brat expects /usr/bin/python
- latest version of Brat
- install.sh should be executable
docker build -t brat .
docker run --name brat-server -p 8080:80 -v "$(pwd)/shared:/usr/local/apache2/htdocs/brat/data/shared" -d brat
The web interface is available at http://127.0.0.1:8080/brat/
.
Using the above launch command, the shared/
folder is shared with the container and will be available as a subfolder in BRAT's workspace.
First, the environment must be configured. Create shared/annotation.conf
with the following content:
[entities]
Foobar
[relations]
[events]
[attributes]
And shared/visual.conf
with:
[labels]
Foobar | Foo | F
[drawing]
Foobar bgColor:red
More information about configuration files is available in the official documentation.
Next, you can add your dataset. Each plain text file (e.g. foo.txt
) must have an associated annotation file (i.e. foo.ann
). If you add new files manually, you must create an (empty) .ann
file for each of them.