Description
First of all, thank you very much for echo, I've used it in several projects and really like it so far!
For my current project I have an URL structure like this: /some/:id/and/:more/to/:come for some endpoints and something like /another/:id/ where both of these endpoints use the same interface handler on a struct which then calls some functions on that struct and so on. As the handler is abstract, I couldn't set the url params directly in the handler, so I started implementing a custom binder, which would enable you to define a struct like so:
type Test struct {
ID int64 `param:"id"`
Thing string `param:"more"`
Other string `param:"come"`
// ... and other struct fields.
}The binder would then map the values of the url params in /some/:id/and/:more/to/:come to their respective values in the struct.
When I was about half way through implementing this as a custom binder, I realized 95% of my custom binder was already done with (b *DefaultBinder) bindData(), so I tried to implement this directly into the standard binder and it worked pretty well. I did something like this (at the end of bindData(), right before L75):
paramNames := c.ParamNames()
paramValues := c.ParamValues()
paramVars := make(map[string][]string)
for in, name := range paramNames {
paramVars[name] = append(paramVars[name], paramValues[in])
}
b.bindData(i, paramVars, "param")No need to reinvent the wheel, I thought.
My question now is: Should I send a PR to "really" implement this feature? Is the way I implemented it pure garbage? Or would it be better to implement it separately on my own as a custom binder?
I thought this can be useful for others, so I shared it. This issue is a general one to discuss, as the feature is probably a pretty big one, so I thought it'd be better to create an issue first and discuss a bit instead of sending a pr right away.