A Robust URL Parser and Builder for Lua

This small Lua library provides a few functions to parse URL with querystring and build new URL easily.

    url = require "net.url"

URL parser

The library converts an URL to a table of the elements as described in RFC : scheme, host, path, etc.

    u = url.parse("http://www.example.com/test/?start=10")
    print(u.scheme)
    -- http
    print(u.host)
    -- www.example.com
    print(u.path)
    -- /test/

URL normalization

    u = url.parse("http://www.FOO.com:80///foo/../foo/./bar"):normalize()
    print(u)
    -- http://www.foo.com/foo/bar

URL resolver

URL resolution follows the examples provided in the RFC 2396.

    u = url.parse("http://a/b/c/d;p?q"):resolve("../../g")
    print(u)
    -- http://a/g

Path builder

Path segments can be added using the __div metatable or u.addSegment().

    u = url.parse('http://example.com')
    u / 'bands' / 'AC/DC'
    print(u)
    -- http://example.com/bands/AC%2FDC

Module Options

• separator is used to specify which separator is used between query parameters. It is & by default.

• cumulative_parameters is false by default. If true, query parameters with the same name will be stored in a table.

• legal_in_path is a table of characters that will not be url encoded in path components.

• legal_in_query is a table of characters that will not be url encoded in query values. Query parameters on the other hand only support a small set of legal characters (-_.).

• query_plus_is_space is true by default, so a plus sign in a query value will be converted to %20 (space), not %2B (plus).

If one wants to have the + sign as is in path segments, one can add it to the list of legal characters in path. For example:

    url = require "net.url"
    url.options.legal_in_path["+"] = true;

Querystring parser

The library supports brackets in querystrings, like PHP. It means you can use brackets to build multi-dimensional tables. The parsed querystring has a tostring() helper. As usual with Lua, if no index is specified, it starts from index 1.

    query = url.parseQuery("first=abc&a[]=123&a[]=false&b[]=str&c[]=3.5&a[]=last")
    print(query)
    -- a[1]=123&a[2]=false&a[3]=last&b[1]=str&c[1]=3.5&first=abc
    print(query.a[1])
    -- 123

Querystring builder

    u = url.parse("http://www.example.com")
    u.query.foo = "bar"
    print(u)
    -- http://www.example.com/?foo=bar
    
    u:setQuery{ json = true, skip = 100 }
    print(u)
    -- http://www.example.com/?json=true&skip=100

Differences with luasocket/url.lua

• Luasocket/url.lua can't parse http://www.example.com?url=net correctly because there are no path.

• Luasocket/url.lua can't clean and normalize url, for example by removing default port, extra zero in port, empty authority, uppercase scheme, domain name.

• Luasocket/url.lua doesn't parse the query string parameters.

• Luasocket/url.lua is less compliant with RFC 2396 and will resolve http://a/b/c/d;p?q and : ../../../g to http://ag instead of http://a/g ../../../../g to http://a../g instead of http://a/g g;x=1/../y to http://a/b/c/g;x=1/../y instead of http://a/b/c/y /./g to http://a/./g instead of http://a/g g;x=1/./y to http://a/b/c/g;x=1/./y instead of http://a/b/c/g;x=1/y