jfpoilpret / yeti-json Goto Github PK

This module defines functions that can be used from Yeti to parse JSON input string in a typesafe way.
To use these functions, you just need to load the "json" module:

    {readJson, readJson'} = load json;
    
Please also ensure that you put the following jars in the classpath:
	- json_simple-1.1.jar
	- json-1.0-SNAPSHOT.jar

If you want to use Yeti REPL then you can type the following command line:

    $ java -cp json_simple-1.1.jar;json-1.0-SNAPSHOT.jar;yeti.jar yeti.lang.compiler.yeti

Usage examples follow:

    > x1 = readJson {a = "", b = 0, c = false} '{"a": "abc", "b": 100, "c": true}';

The first argument acts as a "type template" to which the second argument must be mappable after JSON parsing. An exception will occur if the template type doesn't match the type of the parsed input.

Mapping principles follow:

    JSON type     Yeti type    Comments
    ---------     ---------    --------
    object        struct       all fields from the yeti target struct must exist 
                               in JSON object
    list          list         if yeti target list is not empty, then item type 
                               must match with JSON list items type
    string        string	
    number        number
    boolean       boolean
    
    anything      Some 'a     'a must be convertible to type "anything"
    null          Some 'a     in this case, Some 'a is replaced with None () in the 
                              result
    null          list<'a>    in this case, the result is the empty list []

The mapping is recursive, meaning that if JSON input defines list of lists or structs within structs, then the template should be the same.

Other usage examples with lists:

    > x2 = readJson [] '[1, 2, 3]';
    > x2 = readJson [0] '[1, 2, 3]';

The second function takes a list of variants that allows you to specify several different templates that the JSON input may map to; you would use this function as in the following example:

    > x3 = readJson' [A {a = "", b = 0}, B {b = "", c = false}] '{"a": "abc", "b": 123}';

This allows you to use pattern matching when the JSON input that you get may be of different types.