Deployment

Quick start

The most straight forward way is to run make run-dev-env

In case only API is required:

make run-dev-api

Requirements

If running everything on Docker stack, only docker and docker-compose are required. If you prefer to run directly, each part is described bellow.

Building blocks

This project consists of serveral independant parts which are combined together using docker-compose, which relies on Docker containers. Each individual part has it's Dockerfile which can be used as a reference launching directly on machine.

GF

Dockerfile

It launches an instance of Grammatical Framework with a custom REST API to be able to generate text and get result by single HTTP request.

Grammatical Framework can be downloaded and installed directly (Downloads Page) and core/gf/server.py can be simply launched via python server.py or gunicorn server(Works on Python 2.7 and Python 3+).

If pgf module is missing, that means runtime libraries are not installed. Runtime Libraries can be found here, you always need to compile C runtime library and only then compile python (same goes for other runtimes, C is required)

API

Dockerfile

It launches REST API as well as GraphQL.

GraphQL gives access to: - Document Plans - Dictionary items - Query uploaded CSV files

REST API gives access to: - Request to generate text. Read generation result - Upload CSV data file

It can be launched via: make run-dev-api, or docker-compose -f docker-compose.yml up

If you want to be able to run it locally, you firstly need to compile it. 1. Install Clojure 2. Do clojure -A:uberjar. It will create uberjar (fatjar) which contains all of the dependencies, code etc. 3. Copy target/api-1.0.0-SNAPSHOT-standalone.jar to server it will be running on

To run it, simply do java -jar api-1.0.0-SNAPSHOT-standalone.jar (Assuming server has Java JVM installed).

Here are environment variables with their default values that will need to be changed accordingly:

GF_ENDPOINT: "http://gf:8000"
DB_IMPLEMENTATION: "datomic"
DB_URI: "datomic:free://datomic-transactor:4334/acc-text?password=datomic"
DATA_FILES: "/opt/data-files"
DICT_PATH: "/opt/dictionary"
DOCUMENT_PLANS: "/opt/document-plans"
ENABLED_LANGUAGES: "eng"
SHOW_FLAGS: "true"

Front-End

Dockerfile

FrontEnd runs on nodeJS, thus node and npm should be installed to be able to run it directly.

It can be launched via make run-front-end-dev, or if you prefer npm, npm install && npm run inside front-end directory.

Because UI is essentially a "glue" which binds every part together, it needs to know where's all other parts are located.

It is done using environment variables.

Example.:

ACC_TEXT_API_URL: http://0.0.0.0:3001
ACC_TEXT_GRAPHQL_URL: http://0.0.0.0:3001/_graphql