go-archetype
Dead simple project blueprints for the rest of us.
go-archetype is a tool for creating project blueprints (aka archetypes) in your native language.
If you or your company create new projects and you want all these projects to preserve a similar structure or simply create a good starting point for other develoeprs, use go-archetype to define project archtypes (templates).
Concept
Typically two roles are involved:
- The archetype creator (typically a senior developer/architect)
- The developer creating a new project
The archetype creator
As the archetype creator you create a template (aka blueprint) project. This project is fully functional, valid code written in your native language, be it Golang, JavaScript, Java, Python etc. (go-archetype is written in golang but can be used to generate tempaltes in any language) Then you define a set of transformations. E.g. ask the user for the project name and replace here, here and here.
What you don’t need to do is: Create a bluprint in a meta-language. Other templating tools, such as the python cookiecutter require you to write your blueprint (template) project in a meta-language, e.g. cookiecutter’s language. With go-archetype you don’t have to do that; your write your tempalte in your native language and then define a simple set of transformations for generating templates.
Developer using the archetype
As a developer creating a new project based on the archetype, you run go-archetype
with a few parameters, you answer a few questions (defined by the template creator such as what is your new project’s name, these questions are defined by the archetype creator) and your new code is generated.
Summary
Here are the steps:
- Architect creates an actual working project. This project is used as a blueprint or an archetype. It’s a barebones project, no templating involved.
- Architect defines a set of transformations in transformations.yml. These transformations are used for simple search-and-replace in the blueprint. (and there are more compicated types, see below)
- Developer runs go-archetype to create a new project based on the blueprint.
Example project
Example project using go-archetype: https://github.com/rantav/go-template
Usage
go-archetype transform --transformations=transformations.yml --source=. --destination=/path/to/your/new/project
The transformations.yml file
The transformations file contains two main sections:
- inputs
- and transformations
The inputs section defines user inputs, see description below. The transformations section defines the list of ordered transformations as explained below.
See the file transformations.yml in this very project as an example.
Requesting for user input
It is common to request user inputs in order to apply a set of transformations. For example you might want to request for the project name, description, whether to include this or that functionality.
There are following types of input: text
, yesno
and select
.
text
provides simple, single-line text inputs.yesno
originates a boolean [y/N] question.select
allows to select from a list of predefined options.
Example:
inputs:
- id: ProjectName
text: What is the project name?
type: text
- id: IncludeReadme
text: Would you like to include the readme file?
type: yesno
- id: ProjectType
text: Select project type
type: select
options:
- simple
- advanced
The id
must be unique and is later also used for performing the transformations (see below).
A user may provide the required inputs interactively when promped to when running go-archetype.
Likewise, The user may also provide the input as CLI arguments, which is useful for automation. (see example below)
Interactive question example: (what the user sees while running go-archetype)
? Would you like to include the readme file? (y/N)
And when providing the input as CLI args:
go-archetype transform --transformations=transformations.yml \
--source=. \
--destination=.tmp/go/my-go-project \
-- \
--ProjectName my-go-project \
--IncludeReadme yes \
--ProjectType simple
In this example there are following inputs: ProjectName
, IncludeReadme
and ProjectType
.
To seperate program args from user input we use --
. After the --
the list of user inputs is provided.
Templating user input
After accepting user input you might want to transform and tempalte it.
We use go templates as a templatting languate. For the full overview of the templating language see text/template Following is a list of several examples to get you started.
- User inputs are wrapped in
{{
and}}
- User inputs are prepended with
.
, so for example{{.ProjectName}}
would yield the user’s provided project name. - You may combine user input with constats, for example
{{.ProjectName}} - {{.ProjectDescription}}
, which concatenates the project’s name with a hyphen and then it’s description.
Pipelines and functions are a useful concept. For example you might request the project’s name and then display this project name in uppercase in the readme file and lowercase in source files. To that end we provide a host of template pipelines.
{{ .ProjectName | upper }} # ProjectName is provided by the user and we transform it to all uppercase. with the upper pipeline
{{ wrap 80 .Description }} # Wordwrap the project description by 80 characters
We include out the the box the sprig library, which includes many differnet string manipulation functions, to list a few: trim
, trimAll
, trimSuffix
, trimPrefix
, upper
, lower
, title
, wrap
, plural
, snakecase
, camelcase
, kebabcase
etc.
Transformers
There are two types of transformers: the include and replace transformers.
The include transformer
The include transformer allows inclusion or exclusion of whole files as well as parts of files in the generated project. A simple example: you ask then user whether to include a README file and based on the answer you include or exlude this entire file.
Including or exluding whole files
All files that pass the global ignore filter are included by default. Unless one or more of the include rules with an empty region_marker
applies to them, in which case the files are included only if the condition evaluates to true.
Condition is a go template condition as can be used in an {{if}}
expression. As such a simple boolean is expressed as simply .var
where var
is typically a user input.
For example with the following user input:
inputs:
- id: IncludeReadme
text: Would you like to include the readme file?
type: yesno
It is possbible to define the following include transformer:
transformations:
- name: include the readme file
type: include
region_marker: # When there's no marker, the entire file(s) is included
condition: .IncludeReadme
files: ["README.md"]
Note that when there’s no region_marker
that simply means that the entire file is included/excluded based on the user’s input.
More sophisticated expresions could also be utilized, such as boolean algebra, using and .x .y
, or .x .y
etc (x
and y
are user inputs). For a complete reference, see go templates.
Don’t forget the dot .
Keep in mind that go templates require a dot (.
) to prepend a value. So when utilizing user input, for example such as IncludeReadme
be sure to prepend the dot, e.g. .IncludeReadme
whenever used in conditions or replacements.
One caveat to that is that for simplicity go-archetype allows dot-less conditions when they are very simple, e.g. only variable
, such as IncludeReadme
. So the following conditions are actually equivalent:
condition: .IncludeReadme
and
condition: IncludeReadme # No dot here
This is done in order to simplify the simple single-operand conditions. However, with more complex conditions be sure to prepend the dot. For example the following condition is valid where x and y are user inputs:
condition: and .x .y
But the following is not valid:
condition: and x y # Not valid. x and y need to be prepended by a dot .
Including or exluding parts of files
Sometimes it’s useful to conditionally include or exlude parts of files and not the entire files. To do this we utilize special region markers.
Example:
import (
"net"
"net/http"
"golang.org/x/sync/errgroup"
// BEGIN __INCLUDE_GRPC__
channelz "github.com/rantav/go-grpc-channelz"
channelzservice "google.golang.org/grpc/channelz/service"
// END __INCLUDE_GRPC__
)
The includes channelz
and channelzservice
are only required in cases where the user selected to add gRPC functionality to the project.
The corresponding transformations.yml file sections are:
inputs:
- id: include_grpc
text: Should gRPC functionality be included?
type: yesno
transformations:
- name: "include grpc - parts of files"
type: include
region_marker: __INCLUDE_GRPC__
condition: .include_grpc
files: ["Makefile", "**/*.go", "deployments/*"] # optional, just as an example
To summarize, in cases where you’d like to include or exlude just parts of files (and not the entire file), you use special region markers inside the source code file. These region markers are simply comments in Go or in Java, PHP etc.
The replace transformer
This transformer performs a search and replace functionality. For example, you might ask the user for the project name and then replace the generic template project name with the user’s provided name
Example:
inputs:
- id: name
text: What is the project name? (e.g. my-awesome-go-project)
type: text
transformations:
- name: project name
type: replace # The type of the transformer is **replace**
pattern: go-template # The text pattern to search and replace
replacement: "{{ .name }}" # The text to replace. You may use go tempalates and perform arbitrary replacements.
files: ["*.go", "**/*.go", "**/*.sh", ".gitignore", "README.md"]
The rename transformer
This transformer renames files. For example, you might want to rename all the files with the pattern <go-template>
with the user’s provided project name.
Example:
Let’s imagine, you have a project template with the following structure:
go-template
├── cmd
│ └── go-template
│ ├── go-template.go
│ └── main.go
├── Dockerfile
└── README.md
And the following transformations.yml
file:
inputs:
- id: name
text: What is the project name? (e.g. my-awesome-go-project)
type: text
transformations:
- name: rename files to the project name
type: rename # The type of the transformer is **rename**
pattern: go-template # The text pattern to search and replace
replacement: "{{ .name }}" # The text to replace. You may use go tempalates and perform arbitrary replacements.
files: ["**"]
Then, after transformations you will get a new project with the renamed files like these:
my-awesome-go-project
├── cmd
│ └── my-awesome-go-project
│ ├── my-awesome-go-project.go
│ └── main.go
├── Dockerfile
└── README.md
Recepies
A list of useful recepies.
Always ignore
To always exclude some parts of the output regardless of user input, do as follows:
transformers:
- name: do not include template code in the final output
type: include
region_marker: __DO_NOT_INCLUDE__
condition: false
files: ["**"]
And then in your source file(s):
# BEGIN __DO_NOT_INCLUDE__
... Code that should neven be included in the final output, such as
... templating specific scripts, makefiles etc
# END __DO_NOT_INCLUDE__
Conditional replace
Sometimes you need to conditionally replace a pattern. The condition may depend on user input. For example you may ask the user whether they’d like to includ gRPC functionality in the project or not and based on that render a different makefile.
Since the underlying rendering engine uses go templates it is possible to utilize the following condition:
{{ if .include_grpc }}build: build-grpc{{ else }}build:{{end}}
This condition will render build:
if .include_grpc
is false and build: build-grpc
if .include_grpc
is true. include_grpc
is a user input in this case.
The complete example would look as follows then:
inputs:
- id: include_grpc
text: Should gRPC functionality be included?
type: yesno
transformations:
- name: build with grpc or not
type: replace
pattern: "build: build-grpc"
replacement: "{{ if .include_grpc }}build: build-grpc{{ else }}build:{{end}}"
files: ["Makefile"]
Change to CamelCase
If you have a string that you’d like to change to CamelCase, for example my-project
, first convert to snake_case and then to CamelCase. This is unfortunately a limitation of sprig.
Example:
replacement: "{{ .name | snakecase | camelcase }}"
Changing to camelCase where the first letter is lowercased is even more ticky, but here goes:
replacement: "{{ .name | snakecase | camelcase | swapcase | title | swapcase }}"
Yeah it works…
Inclusion By Selection
Select allows to select one of the options and include section depends on selected choice.
- name: simple
type: include
region_marker: __ProjectType_SIMPLE__
condition: "eq .ProjectType \"simple\""
files: ["main.go"]
- name: medium
type: include
region_marker: __ProjectType_MEDIUM__
condition: "eq .ProjectType \"medium\""
files: ["main.go"]
- name: advanced
type: include
region_marker: __ProjectType_ADVANCED__
condition: "eq .ProjectType \"advanced\""
files: ["main.go"]
Order of execution
Transformatinos are executed by the order they appear inside the transformations.yml file. The output of the first transformation is then piped into the input of the second transformation and so forth. That means that the order is important such that if you’re pattern needs to match certain text, you need to make sure that no previous transformation had changed this text. That’s why it’s wise to start with the more specific replacements and then move on to the more generic replacements.
Example:
transformations:
- name: project long description
type: replace
pattern: Use go-archetype to transform project archetypes into existing live projects
replacement: "{{ wrap 80 .ProjectDescription }}"
files: ["cmd/root.go"]
- name: project name
type: replace
pattern: go-archetype
replacement: "{{ .ProjectName }}"
files: ["*.go", "**/*.go"]
project long description
should be placed before project name
. If it weren’t so then after applying ProjetName replacement on all occurences of the string "go-archetype"
then the sentence "Use go-archetype to transform project archetypes into existing live projects"
would have become "Use my-project-name to transform project archetypes into existing live projects"
and then the replacement would not have been matched.
Before and After
The before
and after
hooks allow you to run arbitrary shell command just before running all transformations or just after then.
They are provided with useful context that can be used in the actual command, which includes:
source
(Used as{{ .source }}
)destination
(Used as{{ .destination }}
)- As well as all user inputs
Example:
after:
operations:
- sh:
# You can define commands with the basic syntax.
# If so, each command will be separated by the <newline> symbol. Each line will be executed
# separately.
- cd {{.destination}} && echo $PWD
# Either, you can use the extended syntax.
# cmd defines command line to be executed.
- cmd: echo Done archetyping from {{ .source }} to {{ .destination }} of project {{ .ProjectName }}
# if multiline is set to false, then it runs exactly as basic syntax command
multiline: false
# Or, you can set multiline to true. Then, the command will be executed as-is.
# To make this happen, use YAML multiline operator "|".
# See also: https://yaml-multiline.info/
- cmd: |
if [ 1 == 1 ]; then
echo "OK!"
fi
multiline: true
Operations and debugging
The view detailed logs, run with LOG_LEVEL=debug