fix URL.resolve and PathInfo.normalize

[itboy87]

Changes in URL.resolve

• Added new test cases in URLTest.testResolve.
• Removed old query on the copy of the path.
• Added protocol for a relative URL.
  • assertEquals("https://example2.com/one", URL.resolve("https://example.com/", "//example2.com/one"))
• Now returns the base URL if the access URL is empty.
  • assertEquals("https://example.com/one", URL.resolve("https://example.com/one", ""))
• Introduced a new extension function String.normalizeUrl.
  • Avoids normalizing query parameters or URL fragments.
  • Keeps end slashes after normalization.

Changes in PathInfo.normalize

• Added new test cases in URLTest.testNormalize.
• Added a parameter for removeEndSlash.
  • This parameter is introduced to match the standard of the web, but in filesystem paths, it's not required, and the default value is true.
  • assertEquals("g", "g/".pathInfo.normalize())
  • assertEquals("g/", "g/".pathInfo.normalize(removeEndSlash = false))

[itboy87]

Additionally, as per RFC3986 Secion-3.1 , URL schemes should be case-insensitive, producing only lowercase scheme names for consistency. Should I apply this change in this pull request as well?

[itboy87]

@soywiz I have one more scenario to handle: if the base URL is empty or not absolute, how should it be resolved? I think we should return the access URL if it's absolute; otherwise, return an empty string or null.
In java.net.URL("") or java.net.URL("test"), a java.net.MalformedURLException exception is thrown.

[soywiz]

Thanks @itboy87! I have merged it now as it cover and fixes already some scenarios.

Additionally, as per RFC3986 Secion-3.1 , URL schemes should be case-insensitive, producing only lowercase scheme names for consistency. Should I apply this change in this pull request as well?

Makes sense. Feel free to make a separate PR for that.

@soywiz I have one more scenario to handle: if the base URL is empty or not absolute, how should it be resolved? I think we should return the access URL if it's absolute; otherwise, return an empty string or null.
In java.net.URL("") or java.net.URL("test"), a java.net.MalformedURLException exception is thrown.

I would have URL.resolveOrNull returning null for invalid scenarios, and then make resolve throw the exception; that would be consistent with the Kotlin API. What would you think?

In this case too, let's make those changes in a separate PR.

[soywiz]

I have created two issues for covering your suggestions:

• URL schemes should be case-insensitive, producing only lowercase scheme names for consistency #2092
• if the base URL is empty or not absolute, how should it be resolved? I think we should return the access URL if it's absolute; otherwise, return an empty string or null. #2093

Feel free to make PRs to handle those. And thanks!

[itboy87]

I have created two issues for covering your suggestions:

• URL schemes should be case-insensitive, producing only lowercase scheme names for consistency #2092
• if the base URL is empty or not absolute, how should it be resolved? I think we should return the access URL if it's absolute; otherwise, return an empty string or null. #2093

Feel free to make PRs to handle those. And thanks!

ok Great.