usability: Parsing binary information from names on golang.org/dl is difficult

[kevinburke]

From the Gophers Slack channel, someone is attempting to parse the filenames here: https://storage.googleapis.com/golang into their component parts.

Examples:

• go1.3.darwin-386-osx10.6.tar.gz
• go1.3.3.windows-amd64.zip
• go1.3beta1.darwin-386-osx10.6.pkg
• go1.3rc1.freebsd-amd64.tar.gz

Parsing these is a little less than trivial:

• The version string may have one or two periods, and may include the words "beta" or "rc"

• The machine may have two parts ("freebsd-386") or three ("darwin-amd64-osx10.8").

• Multiple different possible endings - pkg, tar.gz, msi, zip.

It might be good to write a canonical version-string parser, or point people to an easier-to-parse information source, if one exists.

[OneOfOne]

When all else fails, regexp will save us all, mark my words!

https://play.golang.org/p/krFqYD5HUy

[bradfitz]

Is this about getting JSON out of golang.org/x/tools/godoc/dl/dl.go?

[kevinburke]

I'm not sure - @weingart was the initial requester.

[gopherbot]

Timed out in state WaitingForInfo. Closing.

(I am just a bot, though. Please speak up if this is a mistake or you have the requested information.)

[josharian]

This seems rather relevant to the one-line install discussion. Re-opening to give @spf13 a chance to weigh in.

[josharian]

Thanks, @gopherbot, but I want this open.

[bradfitz]

Hmm, @gopherbot shouldn't be fighting you. Most of the gopherbot tasks try hard to not re-do actions if it looks like somebody did the opposite. Maybe that task has a bug?

[weingart]

Didn't realize that I was paged here. I was attempting to write a fairly simple download/install thing that could be used to "track" stable/beta/rc style candidates. While working on this, I also noticed that a new binary with a name/key of "getgo/installer_*" appeared... however, at the time I checked, the source for this was not available (looks to be part of a private github repo). I was optimistic that it would do what I wished... however, even the "dry_run" feature ended up splatting things all over my home directory/etc.

Documenting the filename format, possibly renaming a few of the files to be more consistent, or even publishing a go package that parses the filename format or gives you a way to list based on some parameters and get the URL for a download, would be awesome.

[weingart]

There's likely easier ways to do this, but this is what I ended up with, which seems sub-optimal. Various regexes may improve the verbosity and/or readability. I've not explored that direction much yet.

https://gist.github.com/weingart/7567dcee16796e7242356f3547a7d5b5

[agnivade]

@weingart - We have https://golang.org/dl/?mode=json which returns json that you can parse. On top of it, this CL adds support for returning Archive and Unstable downloads too.

I guess that should suffice your requirements. Please let us know if you need anything else.