Human to machine

This program translates human language to Meteor Kitchen input .json file.

This is just initial commit, more details will be added soon.

Usage

human2machine input_file.txt [output_file.json]

Meteor Kitchen automatically executes human2machine, converts .txt file to .json and builds application:

meteor-kitchen input_file.txt ./output_dir

(you need latest version of meteor kitchen)

Example input files

Example 1 - CRUD

I want site with three pages: home, customers and about.
In home page I want jumbotron with title: "This application is written in English language!", text: "Human to describe app, machine to write code!", button url: "customers".
Please create one collection: customers.
In customers collection I want three fields: name, address and e-mail.
In customers page I want CRUD for customers collection.
In about page I want text: "This application is written in English language using Meteor Kitchen, source code generator for Meteor".

Live application is here.

Example 2 - IoT

I want site with one page.
Create one collection: sensors.
In sensors collection I want fields: topic and message.
Connect sensors collection to mosquitto server: "test.mosquitto.org" and subscribe to topic: revspace/sensors/#.
In home page I want table showing sensors collection.

Live application is here.

Syntax

It's important to finish each sentence with a dot. Newlines are not important and are ignored by parser.
To define a list you can say 4 pages or four pages. After type name you can list element names after colon and separate items with comma: three fields: name, address and e-mail. If element names are not listed, they will be automatically named as type + item_index ("field0", "field1", "field2"). Exception is page names: "page0" will be named "home" and other pages: "page1", "page2"...
To add component or text into page, say in page_name page (word between in and page is interpreted as page name). If you make a typo and page name is unknown - complete sentence is ignored.
To define collection fields say in collection_name collection (word between in and collection is interpreted as collection name). If you make a typo and collection name is unknown - complete sentence is ignored.
In examples above, most of words are added to make text readable to human and are ignored by parser:

Example 1 with important words only:

Three pages: home, customers and about.
In home page jumbotron title: "This application is written in human language!",  text: "Human to describe app, machine to write code!", button url: "customers".
One collection: customers.
In customers collection three fields: name, address and e-mail.
In customers page CRUD for customers collection.
In about page text: "This application is written in human language using Meteor Kitchen, code generator for Meteor".

Example 2 with important words only:

One page.
One collection: sensors.
In sensors collection fields: topic and message.
Connect sensors collection mosquitto server: "test.mosquitto.org", topic: revspace/sensors/#.
In home page table showing sensors collection.

(to be continued)

Datacraft-GaaS / human2machine