For work I and colleagues have for the last nine months been writing a lot of different small shell scripts with curl that interacts with the Go API.

Most of them are quick n' dirty scripts that aren't very robust, and because they're written in the mindset of the now we just end up copy pasting it everywhere. As well as figuring out what API endpoints are available, how they work and so on.

The goal of this project is to make these and similar tasks super simple, without having to write a so-so reliable bash script. And for the most common things just one invocation.

This is still early in the development and a bit rough around the edges. Any bug reports, feature suggestions, etc are greatly appreciated. :)

I'm planning to add support for all the API endpoints that make sense from a CLI perspective. And also to handle some of the scenarios where we ended up writing shell scripts.

Installation

Since this is a Python package available on PyPi you can install it like any other Python package.

# on modern systems with Python you can install with pip
$ pip install gocd-cli
# on older systems you can install using easy_install
$ easy_install gocd-cli

Usage The commands should be mostly self-documenting in how they are defined, which is made available through the help command.

$ gocd
usage: gocd <command> <subcommand> [<posarg1>, ...] [--kwarg1=value, ...]
Commands:
   pipeline
      check: Check whether a pipeline has run successfully
      check-all: Checks all pipelines to be green/non-stalled
      list: Lists all pipelines with their current status
      pause: Pauses the named pipeline
      retrigger-failed: Retrigger a pipeline/stage that has failed
      trigger: Triggers the named pipeline
      unlock: Unlocks the named pipeline if it's currently locked
      unpause: Unpauses the named pipeline

$ gocd help pipeline retrigger-failed
retrigger-failed <name> [--counter] [--stage] [--retrigger]

Retrigger a pipeline/stage that has failed

Flags:
   counter: the pipeline counter to check. Default: latest
   stage: if given the pipeline will only be retriggered if
     this stage failed
   retrigger: possible values (pipeline, stage) default pipeline.
     When pipeline and there's a failed stage retriggers the pipeline.
     When stage and there's a failure retriggers only that stage.
$ gocd pipeline retrigger-failed Integration --stage external-points --retrigger stage

This script has been prepared to be run two situations:

1. From your local machine
2. From inside of Go

Because of this the configuration is handled by a config file and it can be overridden by environment variables.

The current options are:

The configuration file is stored in ~/.gocd/gocd-cli.cfg and is an ini file. Example:

[gocd]
server = http://localhost:8153/
user = admin
password = badger

The environment variables are prefixed with GOCD_ and always ALL CAPS. Example:

GOCD_SERVER=http://loaclhost:8153/
GOCD_USER=admin
GOCD_PASSWORD=badger

Encrypted configuration keys

From version 0.9 there's support for encrypted configuration keys. There's a builtin module for Rot13, or Caesar cipher, as well as a standalone module using blowfish called gocd-cli.encryption.blowfish.

This feature was added to handle a very specific use case, where the password to the Go server was not allowed to be stored in plaintext. But it was okay if the decryption key was stored on the same machine. The builtin implementation is to be seen as a reference implementation and not to be used. But then again, if you just need it to not be plaintext…

Usage

To then encrypt the current plaintext password do:

$ gocd settings encrypt --key password
encryption_module = gocd_cli.encryption.caesar
password_encrypted = fhcre frperg

Copy these two values into your ~/.gocd/gocd-cli.cfg file and remove the old password and next time it'll use the encrypted password instead.

To decrypt:

$ gocd settings decrypt --key password
encryption_module = gocd_cli.encryption.caesar
password = super secret

This project uses namespaced packages which means that you as a plugin/command author will extend the official namespace with your commands.

There are several advantages to this:

• The CLI can dynamically be updated with new commands, just install a Python package to get it integrated
• Internal/private commands can easily be used side-by-side with public commands, no need to maintain a fork for your personal commands
• Low entry to making your own commands

The way the cli searches for commands is quite straightforward:

• The first argument is the package the command belongs to
• The second argument is the class to call
• Any unnamed parameters are passed in the same order as on the cli
• Any --parameters gets the dashes stripped and sent as keyword arguments

To make it work this way there's a pattern to keep to. For each package the __init__.py file will have to provide all the subcommands in the __all__ variable. Each command is a class and it's the name of those classes that are in the __all__ variable. There is an example gocd-cli.commands.echo which only does the bare minimum to show how all this works.

The subcommands will on the command line be divided by dashes, meaning that RetriggerFailed will become retrigger-failed on the command line.

$ gocd <command> <subcommand> posarg1 --kwarg1
# or how it's referred to in code
$ gocd <package> <command class> posarg1 --kwarg1
# or when used
$ gocd pipeline retrigger-failed Simple-with-lock --stage=firstStage \
    --retrigger=stage

Calling help for a command or subcommand will list all available commands, for more information about each command ask for help on each in turn.