improve(genswagger:template):added support for google.protobuf.Timestamp

[EranAvidor]

google.protobuf.Timestamp is constructed from 2 primitives (int32, 64) which results in a wrong swagger file.

[yugui]

github is failing to check if Eran has signed a CLA.
I have asked grpc team about this issue.

@EranAvidor please sign the CLA if you have not yet

[zinuga]

@EranAvidor shows up as not signed CLA yet.

[EranAvidor]

Hi @yugui @zinuga, I just signed the CLA.

[achew22]

Eran,

Thanks for your contribution to the project! Is there any chance I could get you to update one of the example .proto files to include the usage examples for swagger in examples/examplepb? Once you do that you should be able to run make clean examples. That will update the example swagger output in files like this.

Thanks again

[EranAvidor]

sorry for the delay @achew22 , enjoy your weekend.

[achew22]

@wing328, To my novice eyes this looks like it generates a swagger template that will do time. However, the documentation on getting times working is a little sparse so this is based on my generating code. Would you be willing to take a look at the generated swagger output and say if this is the correct approach?

[wing328]

@achew22 is it correct to say that you want me to review this file to see if it's a valid Swagger/OpenAPI spec?

I did a search on the keyword "time" in that file but only found 2 references to "timeout"

[achew22]

@wing328 The lines in question are here. By my of the spec, specifically this section:

Common Name	type	format	Comments
dateTime	string	date-time	As defined by date-time - RFC3339

It looks like you can do one of two things.

i. put the format as the literal token "date-time"
ii. put the format as a string representing RFC3339

Which way is the correct way to interpret that? Now that I'm reading it more closely I suspect that it might be the prior.

[EranAvidor]

Hi @achew22 and @wing328 ,
I came up with this documentation in order to achieve ease of use.
The options above will lead developers to search for usage examples (that can't be directly conclude from the specs or the literal token).
What do you think?

[wing328]

@achew22 if the datetime string conforms to RFC3339, then you can specify the format as date-time

This is the first time I saw people documenting a datetime string in the following format:

          "type": "string",
          "format": "0000-00-00T00:00:00.000000000Z"

and the format is allowed by the spec (https://github.com/OAI/OpenAPI-Specification/blob/master/versions/2.0.md#data-types)

Tools like swagger codegen only recognize date-time but not 0000-00-00T00:00:00.000000000Z.

[achew22]

Thanks @wing328 for taking a look.

@EranAvidor Do you think you could update this to say "date-time" instead of "0000-00-00T00:00:00.0000000Z"?

[EranAvidor]

Hi @achew22 , I just replaced the format as we talked.

[EranAvidor]

rebased with master

[achew22]

This looks good to me. I generated clients in a couple of languages and the generated code has these new fields typed as Date objects.

Can you regenerate the examples to make the tests pass and then upload again?

Thank you!

[EranAvidor]

@achew22 done.

[achew22]

Looks good to me. Merged