layout | permalink |
---|---|
docs |
/documentation/ |
Oya is a command line tool aiming to help you bootstrap and manage deployable projects.
Install oya and its dependencies:
$ curl https://oya.sh/get | bash
Initialize a project:
$ oya init OyaExample
Add an example task to the bottom of the generated Oyafile
:
$ cat Oyafile
Project: project
build: |
echo "Hello, world"
A task is simply a bash-compatible script. Its name is any valid camel-case identifier starting with a lowercase letter.
Identifiers starting with caps are reserved by Oya.
List available tasks:
$ oya tasks
Run the task:
$ oya run build
Hello, world
If you're familiar with Makefiles, you may have noticed some similarity here. The main difference is because we're using standard YAML files, is the pipe character after task name. An added bonus you don't have to use tabs. :>
If all Oya offered was poor-man's Makefiles, you'd better find something better to do. Fortunately, Oya has much more to offer so keep reading.
- Oyafile - is an YAML file containing Oya config and task definitions.
- Oya task - a named bash-compatible script you can run using
oya run <task name>
. - Oya project - is a directory and any number of subdirectories containing
Oyafiles
; the top-levelOyafile
must contain aProject:
directive. - Oya pack - an installable Oya project containing reusable tasks you can easily use in other projects.
To install the latest version of Oya run the following command:
$ curl https://oya.sh/get | bash
You can also specify which version should be installed
$ curl https://oya.sh/get | bash -s v0.0.7 $ oya --version
To get started using Oya in an existing project you need to initialize it by running the following command in its top-level directory:
$ oya init OyaExample
All the command does is generate a file named Oyafile
that looks like this.
$ cat Oyafile
Project: OyaExample
Oya task is a named bash script defined in an Oyafile
. Let's pretend our
project is a Golang HTTP server and we need tasks for building and running the
server. Edit the generated Oyafile
so it looks like this:
$ cat Oyafile
Project: OyaExample
build: |
go build .
start: |
go run .
Notice the pipe characters after task names. This is YAML and the pipe is required for multi-line script definitions.
Here's how you can list available tasks:
$ oya tasks
# in ./Oyafile
oya run build
oya run start
To make it work, let's create a simple server implementation:
$ cat app.go
package main
import (
"flag"
"fmt"
"log"
"net/http"
)
func handler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "Hello, world!")
}
func main() {
host := flag.String("host", "0.0.0.0", "host name to bind to")
port := flag.Int("port", 8080, "port number")
flag.Parse()
http.HandleFunc("/", handler)
bind := fmt.Sprintf("%s:%d", *host, *port)
fmt.Printf("Starting web server on %s\n", bind)
log.Fatal(http.ListenAndServe(bind, nil))
}
Well, to really make it work, you also need the Go language tools installed.
Let's start the server
$ oya run start
Ok, but does our server work?
$ curl localhost:8000
Hello, world!
Success!
Alongside the Oyafile
you can create any number of YAML files with an .oya
extension containing predefined values you can use in your tasks (and in
generated boilerplate as you'll find out later).
Let's put the default port number and host name into an '.oya' file. Create file
named values.oya
with the following contents:
$ cat values.oya
port: 4000
host: localhost
You can use any names for your values as long as they start with a lowercase letter. By convention, the names are camel-case.
Let's now modify our task definitions so we pass port and host name explicitly when starting the server:
$ cat Oyafile
Project: OyaExample
build: |
go build .
start: |
go run . --port ${Oya[port]} --host ${Oya[host]}
The
${Oya[...]}
syntax is how you access bash associative arrays. Oya comes with its own shell implementation aiming to be compatible with Bash 4.
After restarting the server (using oya run start
) it's reachable on a
different port:
$ curl localhost:4000
Hello, world!
This is a YAML file you can use maps, arrays and nest values. Let's modify values.oya
slightly:
$ cat values.oya
port: 4000
host: localhost
app:
version: v0.1.0
This is how you can use it in the start
task:
$ cat Oyafile
[...]
start: |
echo "Starting server version ${Oya[app.version]}"
go run . --port ${Oya[port]} --host ${Oya[host]}
TODO: Override port/host via flags.
You can also store confidential data right in your projects. Oya uses SOPS to store them in an encrypted format.
Imagine you want to protect our oh so very secret HTTP endpoint using a password you need to supply as a parameter, example:
$ curl localhost:4000?password=badpassword
Unauthorized
First, configure SOPS for encryption method, check https://github.com/mozilla/sops/blob/master/README.rst#usage.
For our example we can use a sample PGP key:
$ export SOPS_PGP_FP="317D 6971 DD80 4501 A6B8 65B9 0F1F D46E 2E8C 7202"
Oya secrets commands:
$ oya secrets --help
...
edit Edit secrets file
encrypt Encrypt secrets file
view View secrets
...
Let's first slightly modify our HTTP server so it checks if the provided
password matches one in an environment variable. Change app.go
so it looks
like this:
$ cat app.go
package main
import (
"flag"
"fmt"
"log"
"net/http"
"os"
)
func handler(w http.ResponseWriter, r *http.Request) {
requiredPassword := os.Getenv("PASSWORD")
password, ok := r.URL.Query()["password"]
if !ok || len(password) != 1 || password[0] != requiredPassword {
http.Error(w, "Unauthorized", http.StatusUnauthorized)
return
}
fmt.Fprintf(w, "Hello, world!")
}
func main() {
host := flag.String("host", "0.0.0.0", "host name to bind to")
port := flag.Int("port", 8080, "port number")
flag.Parse()
http.HandleFunc("/", handler)
bind := fmt.Sprintf("%s:%d", *host, *port)
fmt.Printf("Starting web server on %s\n", bind)
log.Fatal(http.ListenAndServe(bind, nil))
}
Long story short, the server now requires the PASSWORD
environment variable to
be present. Let's modify our Oyafile
so it sets that variable:
$ cat Oyafile
Project: OyaExample
build: |
go build .
start: |
PASSWORD=${password} go run . --port ${Oya[port]} --host ${Oya[host]}
Because we don't want to store the password in the plain, we'll encrypt it.
First you need to create secrets.oya
file and encrypt it:
$ oya secrets encrypt secrets.oya
$ cat secrets.oya
password: hokuspokus
EOT
$ oya secrets encrypt secrets.oya
There's nothing special about the name of the file. You can encrypt any YAML file with .oya extension. In larger projects you could keep your secrets in several encrypted .oya files, grouping secrets by function.
Now our precious secret is safe!
$ cat secrets.oya
{
"data": "ENC[AES256_GCM,data:XXXX=,tag:XXXX==,type:str]",
"sops": {
...
"pgp": [...],
...
}
}
Only SOPS metadata is out in the plain, the password itself is encrypted.
Restart the HTTP server and test it:
$ curl localhost:4000?password=badpassword
Unauthorized
$ curl localhost:4000?password=hokuspokus
Hello, world!
To view or edit an encrypted file later:
$ oya secrets view secrets.oya
password: hokuspokus
$ oya secrets edit secrets.oya
You can use your favorite editor by setting the
EDITOR
environment variable. The default is vim but you should be able to make it work even with GUI editors.
Pack is an installable Oya project containing reusable tasks you can easily use in other projects.
Oya installs pack in your home
~/.oya
directory by default but you can change the location by setting theOYA_HOME
environment variable.
In this tutorial, let's use the docker
pack to generate a Dockerfile for the
application:
$ oya import github.com/tooploox/oya-packs/docker
Import automatically resolves dependencies using the newest available version of
the pack, pinning it in the Require:
section:
$ cat Oyafile
Project: OyaExample
Require:
github.com/tooploox/oya-packs/docker: v0.0.6
[...]
It makes the pack's tasks available under an alias. In case of this pack, it's
docker
:
$ oya tasks
# in ./Oyafile
oya run build
oya run docker.build
oya run docker.generate
oya run docker.run
oya run docker.stop
oya run docker.version
oya run start
You can change the alias by editing the alias in the Import
section:
$ cat Oyafile
Project: OyaExample
[...]
Import:
docker: github.com/tooploox/oya-packs/docker
[...]
Let's now generate a Dockerfile for our server:
$ oya run docker.generate
This is what the Dockerfile looks like:
$ cat Dockefile
FROM golang
COPY . /go/src/app
WORKDIR /go/src/app
RUN go get
RUN go build -o app
CMD [ "app" ]
You can build the image and start the server in a container:
$ oya run docker.build
[...]
$ oya run docker.run
Starting web server on 0.0.0.0:8080
Oya can also render files and even entire directories from templates. Oya uses Plush templating engine.
In an earlier section you were asked to copy & paste a simple web server. For the sake of illustration imagine that you want to make creating HTTP web servers easier by generating the boilerplate from a template.
Let's create app.go
in templates/
directory:
$ cat templates/app.go
package main
import (
"flag"
"fmt"
"log"
"net/http"
)
func handler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "Hello, world!")
}
func main() {
host := flag.String("host", "0.0.0.0", "host name to bind to")
port := flag.Int("port", 8080, "port number")
flag.Parse()
http.HandleFunc("/", handler)
bind := fmt.Sprintf("%s:%d", *host, *port)
fmt.Printf("Starting web server on %s\n", bind)
log.Fatal(http.ListenAndServe(bind, nil))
}
EOT
This is how you can render it to the current directory (the command will
override app.go
so ****be careful!**):
$ oya render templates/app.go
Ok, all is good and fine but how to make the ^ code available when creating new projects? Easy, turn it into a pack!
Technically, all you need to do to turn the project we created into an Oya pack
is push it to Github and tag it with a version number by adding & pushing a git
tag in the right format (e.g. v0.1.0
) along with a few small changes.
Rather than doing that, let's create it step-by-step in a fresh new git repo.
First, let's add a task you'll use to generate boilerplate to the original Oyafile so it looks like this:
$ cat Oyafile
Project: project
build: |
go build .
start: |
go run . --port ${Oya[port]} --host ${Oya[host]}
generate: |
oya render ${Oya[BasePath]}/templates/app.go
The new generate
task will generate files into the current directory based on
the contents of the templates directory.
BasePath is the base directory of the path so the
oya render
command knows where to take templates from.
Because the script needs port
and host
, let's also create values.oya
containing pack defaults:
$ cat values.oya
port: 8080
host: localhost
To share your pack all you need is push the project to a Github repository and tag it with a version number, roughly:
$ git push origin
$ git tag v0.1.0
$ git push --tags
That's it!
Currently, only Github is supported as a way of sharing packs but if you want to help with adding support for Bitbucket and others, do get in touch!
So how do you use the pack? Easy. First, create a new empty project and initialize it:
$ oya init OyaExample
Then import the pack. Here, I'm assuming it's under github.com/tooploox/oya-gohttp:
$ oya import github.com/tooploox/oya-gohttp
That's it! Let's see what tasks we have available:
$ oya tasks
# in ./Oyafile
oya run oya-gohttp.build
oya run oya-gohttp.generate
oya run oya-gohttp.start
Let's generate the server source:
$ oya run oya-gohttp.generate
This will generate app.go
file in the current directory:
$ cat app.go
package main
import (
[...]
Let's start the server:
$ oya run oya-gohttp.start
TODO: Overriding port.
TODO: Describe multiple packs in a single repo (i.e. how to version and use them).
So far we only talked about one Oyafile in the project's top-level directory. But you can put Oyafiles in subdirectories.
This is especially useful in monorepos so we'll use that as an example. Imagine we have a web application consisting of a REST API back-end server and front-end SPA application.
For the sake of illustration let's create a mono-repository containing both back-end and front-end in separate directories.
$ tree
.
├── Oyafile
├── backend
│ ├── Oyafile
│ ├── server.go
└── frontend
├── Oyafile
└── main.ts
2 directories, 3 files
In addition to the top-level Oyafile
both front-end and back-end have their
own Oyafiles
, let's have a look at each.
The top level Oyafile contains a Project:
directive to mark the top-level
project directory and a task named build
preparing the output directory.
$ cat Oyafile
Project: myproject
build: |
echo "Preparing build/ folder"
rm -rf build/ && mkdir -p build/public
In this example, let's assume that back-end is an HTTP API written in Go. Here
is how the back-end Oyafile
looks like:
$ cat backend/Oyafile
build: |
echo "Compiling server"
go build -o ../build/server .
Notice that it contains just one task, named build
for compiling the back-end
server.
Front-end is a TypeScript SPA application. The build
task compiles TypeScript
source to JavaScript:
$ cat frontend/Oyafile
build: |
echo "Compiling front-end"
tsc main.ts --outFile ../build/public/main.js
Now let’s list the available tasks. To do it for the whole project including
subdirectories we need to use -r or --recurse
flag.
$ oya tasks -r
# in ./Oyafile
oya run build
# in ./backend/Oyafile
oya run build
# in ./frontend/Oyafile
oya run build
As you can see we have three build
tasks one per Oyafile. We can now run them
all.
$ oya run -r build
Preparing build/ folder
Compiling server
Compiling front-end
The result is the compiled server as well as the JavaScript it serves.
TODO: For a working example, clone ...
- Install go 1.11 (goenv is recommended, example:
goenv install 1.11.4
). - Checkout oya outside GOHOME.
- Install godog:
go get -u github.com/cucumber/godog/cmd/godog
. - Run acceptance tests:
godog
.
For all test to pass you need to import the PGP key used to encrypt secrets. $ gpg --import testutil/pgp/private.rsa
- Run tests:
go test ./...
. - Run Oya:
go run oya.go
.