Partial expand

[Andrey Popp]

It would be nice to have partial expand feature which works by binding supplied values to variables and produces another template:

URLTemplate.new("{a}/{b}").partial_expand(a: 1)
=> <URLTemplate @pattern="1/{b}">

[Hannes Voigt-Georg]

Yep, I agree. I left it out at first because I didn't see the immediate value, but now I some pointful usecases.

[Hannes Voigt-Georg]

Okay, I've implemented something in the 0.6.0 release. Please test and report if something doesn't work like you expect it.

[Andrey Popp]

This is cool, though I expect

URITemplate.new('{+path}').expand_partial(path: 'x').to_s == 'x{+path}'

and

URITemplate.new('{?x,y}').expand_partial(path: 'x').to_s == '?x=1{&y}'

and so on...

[Hannes Voigt-Georg]

Hm, how should this case be handled:

URITemplate.new('{?x,y}').expand_partial(y: 'z')

{?x}&y=z is not guaranteed to issue a leading questionmark.

[Andrey Popp]

Do you think components inside {?...} should be expanded in order they are defined? I didn't find anything about it in RFC6570. I think that:

URITemplate.new('{?x,y}').expand_partial(y: 'z') # => URITemplate('?y=z{&x}')

would be reasonable.

[Hannes Voigt-Georg]

I guess you're right. Just because the example code uses the order of the variables doesn't mean it's mandatory.

[Hannes Voigt-Georg]

Hi, I've written some more code and some more specs. I've also grant you push access in case you want to add some more specs or change the behavior.

[Hannes Voigt-Georg]

Hi @andreypopp

Have you tested the changed behavior? Can you give me some feedback?

[Arne Brasseur]

Hi, I notice the behavior of URITemplate has changed between 0.6.0 and 0.7.0

URITemplate.new('/foo/{x}/{y}').expand_partial(x: 123).to_s
# 0.6.0 => "/foo/123/{y}"
# 0.7.0 => "/foo/123{x}/{y}"

is there an API to get the old behaviour? I'm using URITemplate for Yaks and this is blocking me from upgrading.