| Age | Commit message (Collapse) | Author |
|---|
|
After parsing a URI, we don't want the square brackets
to remain in the host element, because those were only
there to delimit it in the context of a URI string.
|
|
Add unit tests for stand-alone modules that were formerly
part of Uri and so were previously tested along with Uri.
|
|
|
|
Path normalization is hideously broken for now.
|
|
Such a URI should be considered equivalent to a path of "/"
because in both cases the path is an absolute path.
|
|
Fix assumption that the path of an absolute URI is
considered to have a relative path if the path
is empty.
|
|
For normalization "step 2C", if the output path was
empty, we don't want to pop the end of it off.
|
|
* Code the neat example in section 6.2.2 of the RFC.
* Add equality/inequality operators for Uri.
|
|
|
|
|
|
|
|
|
|
|
|
Path may also have colon, so make sure we don't scan
into the path element if there is one.
|
|
* Detect bad characters in host names.
* Incorporate splitting host and port into the state
machine that is parsing/decoding the host.
NOTE:
IPv6address is not checked for bad characters yet.
More research is needed to learn exactly what are
the various ways to write an IPv6 address.
|
|
A colon may be in the authority, if present, so limit
the search for scheme delimiter so we aren't scanning
the authority part, when parsing the scheme.
|
|
|
|
|
|
|
|
* Add IsRelativeReference.
* Add IsRelativePath.
* Add Query.
* Add Fragment.
* Add UserInfo.
* Fix parsing of URIs that have no scheme.
|
|
|
|
* Parts of a path are called "segments", not "steps",
in the RFC.
* The RFC specifies that path separators are always
forward slashes, so don't support other separators.
|
|
* Can now parse URIs from strings.
* This supports scheme, host, and path.
* Path separator defaults to "/" but may be customized.
|
|
|
