DocIndex is the batch process used to feed DocMag, a front-end to Elasticsearch that allows server-side document searching to be simple.
DocIndex can be run directly on an OS, however it is recommended to be run within a Docker container. The container is composed within the DocMag docker-compose.yml.
Usually you won't want to build and run docidx locally, instead it is best to run the docker container published at: https://hub.docker.com/r/deckerego/docidx/
Since docidx relies heavily on computer vision and image processing, bindings to native libraries are heavily used. Packaged Java distributions with native libraries are a giant pain in the butt - hence leveraging Docker containers to ship things by default. If you just want to get docidx up and running Docker will be the easiest way to go, but if you would like to tweak the code and run it locally you will need to jump through some hoops to install the native libs.
docidx uses bindings for OpenCV and Tesseract native libraries. The OpenCV libraries are especially version-sensitive. To install the native Tessearact libriaries in MacOS you can use Homebrew, as in:
brew install tesseract
Unfortunately OpenCV 3.2 does not build properly under Homebrew. For MacOS, OpenCV needs to be built from source. This can be done with:
wget https://github.com/opencv/opencv/archive/3.2.0.tar.gz
tar xzf 3.2.0.tar.gz
mkdir opencv-3.2.0/build
cd opencv-3.2.0/build
cmake .. -DBUILD_opencv_java=ON
make
make install
Linux distributions often ship with Tesseract and OpenCV 3.2, such as with Ubuntu (Bionic):
apt-get install tesseract-ocr libopencv3.2-jni
After the native libraries are installed, building and testing can be performed locally with Maven and Spring Boot:
mvn -DargLine="-Djava.library.path=/usr/local/share/OpenCV/java/" install
If you would also like to spin up a local Elasticsearch and Kibana instance for
testing, you can deploy both with Docker configs in the tests/ directory:
cd tests
docker-compose up -d
To search within your documents, use DocMag available at https://github.com/deckerego/docmag
You could also query Elasticsearch directly using the API or Kibana's dev tools. A query sent over the API might be:
GET /docidx/_search
{"query": { "simple_query_string" :
{ "query": "water bill" }
}}