Java URI.resolve

Question

I'm trying to resolve two URIs, but it's not as straightforward as I'd like it to be.

URI a = new URI("http://www.foo.com");
URI b = new URI("bar.html");

The trouble is that a.resolve(b).toString() is now "http://www.foo.combar.html". How can I get away with that?

Answer 1

Example with the different posibilities:

    URI uri = new URI( "http://www.example.org");
    System.out.println( "*** "+uri+" ***" );
    System.out.println( uri.resolve( "bar") );
    System.out.println( uri.resolve( "/bar") );
    System.out.println( uri.resolve( "bar/") );
    System.out.println( uri.resolve( "/bar/") );
    System.out.println();
    
    uri = new URI( "http://www.example.org/");
    System.out.println( "*** "+uri+" ***" );
    System.out.println( uri.resolve( "bar") );
    System.out.println( uri.resolve( "/bar") );
    System.out.println( uri.resolve( "bar/") );
    System.out.println( uri.resolve( "/bar/") );
    System.out.println();

    uri = new URI( "http://www.example.org/foo1/foo2");
    System.out.println( "*** "+uri+" ***" );
    System.out.println( uri.resolve( "bar") );
    System.out.println( uri.resolve( "/bar") );
    System.out.println( uri.resolve( "bar/") );
    System.out.println( uri.resolve( "/bar/") );
    System.out.println();
    
    uri = new URI( "http://www.example.org/foo1/foo2/");
    System.out.println( "*** "+uri+" ***" );
    System.out.println( uri.resolve( "bar") );
    System.out.println( uri.resolve( "/bar") );
    System.out.println( uri.resolve( "bar/") );
    System.out.println( uri.resolve( "/bar/") );

produces as output:

*** http://www.example.org ***
http://www.example.orgbar
http://www.example.org/bar
http://www.example.orgbar/
http://www.example.org/bar/

*** http://www.example.org/ ***
http://www.example.org/bar
http://www.example.org/bar
http://www.example.org/bar/
http://www.example.org/bar/

*** http://www.example.org/foo1/foo2 ***
http://www.example.org/foo1/bar
http://www.example.org/bar
http://www.example.org/foo1/bar/
http://www.example.org/bar/

*** http://www.example.org/foo1/foo2/ ***
http://www.example.org/foo1/foo2/bar
http://www.example.org/bar
http://www.example.org/foo1/foo2/bar/
http://www.example.org/bar/

In conclusion:

• cases 1-4: If the original URI has not path, resolve uses it argument as new path. That can produce a wrong domain if the resolve argument doesn't start by slash.
• cases 5-8: If the original URI is a domain plus slash, resolve uses its argument as new path. No problem with double slashes.
• cases 10, 12, 14 and 16: If the original URI has a path and resolve argument starts by slash, resolve replaces URI path by its argument. Initial path is totally discarded.
• cases 9, 11: If the original URI has a path not ended by slash and resolve argument doesn't start by slash, resolve discards the last element of the initial uri path and adds its argument.
• cases 13, 15: If the original URI has a path ended by slash and resolve argument doesn't start by slash, resolve adds its argument.

Answer 2

URI.resolve behaves like if you are on a HTML page like http://example.org/path/to/menu.html and click a link with href="page1.html": It cuts off the last segment (here menu.html) and puts page1.html in its place.

(http://example.org/path/to/menu.html, page1.html) → http://example.org/path/to/page1.html

This works also, if the object you call resolve on is a directory, denoted by ending in a slash:

(http://example.org/path/to/, page1.html) → http://example.org/path/to/page1.html

If it does not end in a slash, the outcome is not what you might expect:

(http://example.org/path/to, page1.html) → http://example.org/path/page1.html (missing "to")

If you know that the first argument of the URIs to concatenate is a directory, but you don’t know in which format you get it (with or without trailing slash), this might help you:

static URI asDirectory(URI uri) {
    String uriString = uri.toString();
    return !uriString.endsWith("/") ? URI.create(uriString.concat("/")) : uri;
}

Answer 3

URI should contain the final separator('/') as well to resolve the way you want:

URI a = new URI("http://www.foo.com/");

Answer 4

Ok, appears from URL deffinition scheme://domain:port/path?query_string#fragment_id there should be 3 slashes before path (two after scheme and one directly before path)
2 situation can occure:

there are 3 slashes in your URI => everything is OK

there are less then 3 slashes in your URI => you need to add slash at the end of URI

there is my snappet of code:

String url = "http://www.foo.com";
String endSlash="";
int indexOfSlash = 0;
for(int i = 0;i<3;i++){
   int nextIndex = url.indexOf('/',indexOfSlash);
   if(!(nextIndex>0)){
      if(i>1){
         endSlash="/";
      }else{
         throw new MalformedURLException("Bad given url format, mising :// after schema");
      }
   }else{
      indexOfSlash = ++nextIndex;
   }
}
URL rightUrl = new URL(url+endSlash);

Answer 5

Sounds like you probably want to use URL rather than URI (which is more general and needs to deal with a less strict syntax.)

URI a = new URI("http://www.foo.com");
URI b = new URI("bar.html");
URI c = a.resolve(b);
c.toString()     -> "http://www.foo.combar.html"
c.getAuthority() -> "www.foo.com"
c.getPath()      -> "bar.html"

URI's toString() doesn't behave as you might expect, but given its general nature it may be that it should be forgiven.

Sadly URI's toURL() method doesn't behave quite as I would have hoped to give you what you want.

URL u = c.toURL();
u.toString()     -> "http://www.foo.combar.html"
u.getAuthority() -> "www.foo.combar.html"  --- Oh dear :(

So best just to start straight out with a URL to get what you want:

URL x = new URL("http://www.foo.com");
URL y = new URL(x, "bar.html");
y.toString() -> "http://www.foo.com/bar.html"