Harvest

What Is It

• Ad-hoc query engine for relational databases
  • Uses Django to build SQL and execute the query
  • Uses a simple JSON-based DSL for declaring the query
  • Uses metadata index for query evaluation and preparation
  • Does not require knowing how data model components are related
• Modern HTML5 Web client for building and interacting with queries completely driven by a robust REST API
• Save queries for future reference
• Share queries for increased knowledge dissemination
• Export data into CSV, Excel, and JSON formats
  • Export an R or SAS bundles containing the data and script for streamlined data analysis
• Persisted metadata index of the relational data model
  • Relies on pre-defined Django models to build metadata index
• User-centric annotations on metadata for improved domain specificity
• Lightweight "concept" model for improving how fields in a data model are presented for query or viewing purposes
  • Abstraction used for presentation which is a translation between raw data model and end user.
  • Descriptors
  • Custom query interface
  • Custom output

Install

$ pip install harvest

Harvest CLI

Dependencies

This version of Harvest requires Python 2.6 or 2.7.

Start A New Project

$ harvest init [--verbose] [--no-env] [--no-input] project_name

This command performs the following steps:

• Create a new virtualenv environment (name project_name-env)
• Installs Django
• Creates a starter project using the built-in Harvest template
• Installs the base dependencies
• Syncs and migrates a SQLite database, this requires you to answer a couple prompts (unless --no-input is passed)
• Collects the static CSS and JavaScript files (mainly due to Cilantro)
• Prints out a message to perform a couple commands in your shell

Arguments

project_name - The name of the project which must be a valid Python identifier since it will be an importable Python package. This means it can only contain alphanumeric characters and underscores and cannot start with a number, such as myproject, my_project, and project1, but not 1project, my-project or -myproject.

Options

--verbose - Pass to get all output printed to stdout. Multiple flags can be passed to increase the verbosity, e.g. -vv.

--no-env - Pass to prevent creating a virtualenv. If set, it is assumed the virtualenv is active prior to running this command to ensure dependencies are installed in the correct site-packages directory.

--no-input - Pass to prevent being prompted during the setup. This currently includes the prompt for setting up a superuser during the database sync. This is primarily useful for performing scripted builds.

--template - Specify a template to base your Harvest application on. By default harvest init will base its build off of https://github.com/cbmi/harvest-template/archive/HEAD.zip. By passing a URL to this option harvest init will attempt to bootstrap the project based on the endpoint specified. Additionally, if your provided template contains a Fabric fabfile containing a harvest_bootstrap task the init command will offload all bootstrapping tasks beyond creating the virtualenv and installing of dependencies to the harvest_bootstrap task. This could be useful in situations where further assumptions can be made about a new Harvest deployment (i.e. containerization, use of a specific DB, specific Django models, etc.).

--venv-wrap - If you are using virtualenvwrapper to handle your python virtual environments you can set this flag to create a virtualenv in accordance with the conventions of that utility -- The name of your environment will correspond to your project name and will be created in the directory specified by the WORKON_HOME environment variable.

Post-Setup

After creating a new Harvest project, the next step is to define a few Django models. Run python bin/manage.py avocado check to see what needs to be further setup as well as optional settings and dependencies that can be installed.

Update Harvest

This command updates itself to the lastest stable version from PyPi.

$ harvest update

Install Demo

This command installs one of the Harvest demos.

$ harvest init-demo [--verbose] [--no-env] demo_name

Arguments

demo_name - The name of an available demo which is currently only openmrs.

Options

--verbose - Pass to get all output printed to stdout. Multiple flags can be passed to increase the verbosity, e.g. -vv.

--no-env - Pass to prevent creating a virtualenv. If set, it is assumed the virtualenv is active prior to running this command to ensure dependencies are installed in the correct site-packages directory.