README.md

jsons

Usage

Get an iterator with jsons.parser(str),
where str is a string containing one json value.

Get a pretty json string with jsons.pretty(iter)
or a minified json string with jsons.minify(iter).

The iterator will return 0, 2 or 4 values:

• type: one of: nil, "object", "array", "string", "number", "boolean", "nil";
• where: current position in str;
• if type is not "object" or "array":
• • value: value of the above Lua type;
• • valuestr: string, the value copied verbatim from str.

An "object" or "array" type opens one and a nil closes either.
Use a for loop to iterate over their contents.

Example input:

[123, 456, ["foo", {"qux": "baz", "test": 789}, "bar"], null, false]

Example output (empty line represents no return value at all):

"array"  , 1
"number" , 2 , 123   , "123"
"number" , 7 , 456   , "456"
"array"  , 12
"string" , 13, "foo" , '"foo"'
"object" , 20
"string" , 21, "qux" , '"qux"'
"string" , 28, "baz" , '"baz"'
"string" , 35, "test", '"test"'
"number" , 43, 789   , "789"

"string" , 49, "bar" , '"bar"'

"nil"    , 57, nil   , "null"
"boolean", 63, false , "false"

68
cannot resume a dead coroutine

Example function that validates and returns an array of numbers:

local function arr(str)
	local parser = require("jsons").parser(str)
	assert(parser() == "array", "Not an array")
	local out = {}
	for type, value, value in parser do
		assert(type == "number", "Array contains a non-number")
		table.insert(out, value)
	end
	return out
end

for k,v in ipairs(arr("[1, 2, 3, 4, 5]")) do print(k, v) end
for k,v in ipairs(arr("[1, 2, [], 4, 5]")) do print(k, v) end

Notes

The iterator throws syntax errors when it encounters them.
Thus, avoid causing side effects until the entire document has been parsed.

If you don't like any of the values that the iterator returns, then just make a function that calls the iterator and returns only the ones you want:

local parser = jsons.parser(str)
local function parser2()
	local type, value, value = parser()
	return type, value
end

This will return only the type and the value, but not the position or the verbatim value.

It does not validate the UTF-8 within strings.

Motivation

This library was created for a pretty-printing/beautification command for the text editor I use. The point is to be non-destructive - a DOM parser might rearrange values in objects and mistreat nulls.

Use for anything else but the above at your own risk, because it has never been used for anything else.