davinov / weaverbird

A visual data pipeline builder with various backends

Weaverbird

Weaverbird is Toucan Toco's data pipelines toolkit, it contains :

• a pipeline Data Model, currently supporting more than 40 transformation steps
• a friendly User Interface for building those pipelines without writing any code, made with TypeScript, VueJS & VueX
• a set of BackEnds to use those pipelines :
  • the MongoDB Translator that generate Mongo Queries, written in TypeScript
  • the Pandas Executor that compute the result using Pandas dataframes, written in Python

For in depth user & technical documentation, have a look at weaverbird.toucantoco.com
or at the documentation's source files in the docs directory.

Last but not least, you can play with Weaverbird on our online playground !

Badges

UI

Server

Project setup

yarn install

Requirement: node > v11

Compiles target library

yarn build-bundle

This will generate an importable JS VisualQueryBuilder library in the dist directory.

Important note: While we do our best to embrace semantic versioning, we do not guarantee full backward compatibility until version 1.0.0 is realeased.

Run your tests

The basic command to run all tests is:

yarn test:unit

You can also use a watcher so that tests rerun automatically on a change:

yarn test:unit --watchAll

To run a single test file:

yarn test:unit path/to/yourfile.ts

Finally, you can deactivate typescript checks to run tests quicker:

yarn test:quick

This can be useful to accelerate your development cycle temporarily when developing a new feature or fixing a bug. Under the hood, this will use the babel-jest transformer on typescript files instead of ts-jest.

Lints and fixes files

yarn lint

Build the API documentation

yarn build-doc

This will run typedoc on the src/ directory and generate the corresponding documentation in the dist/docs directory.

Build and run the documentation website

The web documentation is powered by Jekyll.

You can find all the sources into the docs folder.

To build and locally launch the documentation you need Ruby and gem before starting, then:

# install ruby
sudo apt install ruby ruby-dev

# install bundler
gem install bundler

# run jekyll and a local server with dependencies :
cd docs
bundle install
bundle exec jekyll serve

Enrich it!

put your .md file into the docs folder. You can add a folder as well to better organization

into your .md file don't forget to declare this at the beginning of the file :

---
title: your title doc name
permalink: /docs/your-page-doc-name/
---

to finish to get your page into the doc navigation you have to add it in `_data/docs.yml``

example :

- title: Technical documentation
  docs:
  - steps
  - stepforms
  - your-page-doc-name

Run the storybook

Storybook uses the bundled lib, so all showcased components must be in the public API.

yarn storybook

This will run storybook, displaying the stories (use cases) of UI components.

Stories are defined in the stories/ directory.

Customize configuration

See Configuration Reference.

Publication

This library is published on npm under the name weaverbird automatically each time a release is created in GitHub.

Create a release (frontend)

• Define new version using semantic versioning

• Create a new local branch release/X.Y.Z from master

  ex: release/0.20.0

• Update the version property in package.json and in sonar-project.properties

• Check differences between last release and current and fill CHANGELOG.md with updates

  • Delete the ##changes title at start of the CHANGELOG.md if provided

  • Add the date and version at start of CHANGELOG.md following this convention

    [X.Y.Z] - YYYY-MM-DD
    

    ex: [0.20.0] - 2020-08-03

  • Add link to the CHANGELOG.md from this version to the previous one at the end of the CHANGELOG.md

    [X.Y.Z]: https://github.com/ToucanToco/weaverbird/compare/voldX.oldY.oldZ...vX.Y.Z
    

    ex: [0.20.0]: https://github.com/ToucanToco/weaverbird/compare/v0.19.2...v0.20.0

• Commit changes with version number

  ex: v0.20.0

• Push branch

• Create a pull request into master from your branch

• When pull request is merged, create a release with the version number in tag version and title (no description needed)

  ex: v0.20.0

• Hit the release "publish release" button (this will automatically create a tag and trigger the package publication )

Create a release (backend)

• Create a new local branch chore/bump-server-version-x-x-x

• Edit server/pyproject.toml & increment the version in [tool.poetry] section

• Push branch

• Create a pull request into master from your branch

• Once the PR is approved & merged in master publish the release in Pypi with make build & make upload

Usage as library

Without any module bundler

<!-- Import styles -->
<link rel="stylesheet" href="weaverbird/dist/weaverbird.umd.min.js" />

<!-- Import scripts -->
<script src="vue.js"></script>
<script src="weaverbird/dist/weaverbird.umd.min.js"></script>

With an ES module bundler (typically webpack or rollup)

import { Pipeline } from 'weaverbird';

By default, the CommonJS module is imported. If you prefer the ES module version, import dist/weaverbird.esm.js.

Styles

If your module bundler can also import CSS (e.g. via styles-loader):

import 'weaverbird/dist/weaverbird.css';

If you prefer to use Sass, you may import directly the scss:

@import '~weaverbird/src/styles/main';

This example makes use of the ~ syntax from webpack's sass-loader to resolve the imported modules.

API

Modules

See the documentation generated in dist/docs directory

Styles

TODO: document here sass variables that can be overriden

Playground

The /playground directory hosts a demo application with a small server that showcases how to integrate the exported components and API. To run it, just run:

yarn playground

which is basically a shortcut for the following steps:

# build the visual query builder bundle
yarn build-bundle --watch
# run the server and enjoy!
node playground/server.js

Once the server is started, you should be able to open the http://localhost:3000 in your favorite browser and enjoy!

Mongo back-end

The default back-end for the playground is a small server passing queries to MongoDB.

The server.js script reads the playground/playground.config.json config file to know which database should be queried or which http port should be used. If you want to customize these values, either edit this json file or override each available option on the commandline, e.g.

node playground/server.js --dburi mongodb://localhost:27018

You can also customize options through environment variables with the following naming pattern VQB_PLAYGROUND_{OPTION}, e.g.

VQB_PLAYGROUND_DBURI=mongodb://localhost:27018 node playground/server.js

You can use the default test dataset by loading the playground/default-dataset.csv file. To do that, use the following command line:

node playground/server.js --reset

If you want to use a custom CSV file, use the defaultDataset command line option:

node playground/server.js --defaultDataset my-dataset.csv --reset

If you don't have mongodb installed, you can use the --automongo flag from the command line. It will use mongodb-prebuilt to download (the first time) and run mongo 4.0.13 and then listen on the port guessed from the --dburi flag.

mongodb-prebuilt uses mongodb-download internally to download mongodb binaries. Unfortunately, the URLs on https://fastdl.mongodb.org used for these binaries have changed and mongodb-download doesn't seem up-to-date, cf. mongodb-js/mongodb-download#36 and mongodb-js/mongodb-prebuilt#59 To bypass this issue, you can manually specify the download URL by setting the MONGODB_DL_URI environment variable. For instance, you can use the following command line: MONGODB_DL_URI=https://fastdl.mongodb.org/linux/mongodb-linux-x86_64-debian92-4.0.13.tgz node playground/server.js --automongo or MONGODB_DL_URI=https://fastdl.mongodb.org/osx/mongodb-osx-ssl-x86_64-4.0.13.tgz node playground/server.js --automongo

Pandas back-end

An alternative back-end for the playground is a small server running in python, executing pipelines with pandas.

Run the sever in server/playground.py: cd server; FLASK_APP=playground FLASK_ENV=development flask run. Go to http://localhost:3000?backend=pandas to see it in action.

To ease these process, the front-end and the pandas back-end can be run in a container. See the Dockerfile. It's also published on Docker Hub, and can be run directly with: docker run --rm -p 3000:3000 toucantoco/weaverbird-playground:master And then visit http://localhost:3000?backend=pandas Note: add -v /path/to/your/folder/with/csv:/weaverbird/playground/datastore to use your own set of csv instead of the default ones