Parsing urls with missing protocol

[lusy]

Hi, I've just used URI.js for the first time and I was excited:

var urlParser = require('URIjs');

urlParser("http://example.com:8080/pathname/lala").protocol() --> http
urlParser("http://example.com:8080/pathname/lala").hostname() --> example.com
urlParser("http://example.com:8080/pathname/lala").port() --> 8080
urlParser("http://example.com:8080/pathname/lala").pathname() --> /pathname/lala

However, if I try it on a url where the protocol part is missing, everything gets messed up:

var urlParser = require('URIjs');

urlParser("example.com:8080/pathname/lala").protocol() --> example.com
urlParser("example.com:8080/pathname/lala").hostname() --> 
urlParser("example.com:8080/pathname/lala").port() --> 
urlParser("example.com:8080/pathname/lala").pathname() --> 8080/pathname/lala

Could we do support for parsing urls where the protocol part is missing? I think they come up pretty often.. Should I pull-request that? Or is it intetionally left out? Thanks!

[ooxi]

That will not be really possible since example.com:8080/pathname/lala is no valid URI. There is support for URIs without the scheme part in URI.js but in your example that would be ://example.com:8080/pathname/lala.

There have been discussions before of adding heuristics for such behaviour but it's generally very messy to operate on invalid input :(

[rodneyrehm]

This behavior has caused confusion in the past and I fully expect this to raise more issues in the future. Welcome to the weird world of string semantics…

According to RFC 3986 the string example.com:8080/pathname/lala is to be considered an URN with the scheme example.com and the path 8080/pathname/lala, which URI.js is getting right. To treat this as an HTTP URL, either prepend the proper scheme, or use the scheme-relative notation presented by @ooxi