This library allows you to auto-generate C4 component diagrams out from the Golang code.
If you want to get directly into usage of the library check my blog post with a step-by-step implementation guide.
You may also find a couple of examples implemented in the cmd
directory. In order to run any of those examples, please run the shell script attached.
The library provides a set of tools (Scraper and View) that allow you to scrape and render given Golang structures into a C4 component diagram in *.plantuml format.
Scraper component reflects given structure in accordance with structure interfaces, predefined rules and configuration.
You may pass the scraped structure into a View definition which you can then render into a plantUML diagram code.
Scraper identifies components to scrape in one of the following cases:
- type that is being examined implements an interface
model.HasInfo
. - type that is being examined applies to one of the rules registered in the scraper.
Structure model.Info
is a basic structure that defines a component included in the scraped structure of your code.
type Info struct {
Kind string // kind of scraped component
Name string // component name
Description string // component description
Technology string // technology used within the component
Tags []string // tags are used to match view styles to component
}
Scraper may be instantiated in one of two ways:
- from the code
- from the YAML file
In order to instantiate the scraper you need to provide scraper configuration which contains a slice of prefixes of packages that you want to reflect. Types that do not match any of the given prefixes will not be traversed.
config := scraper.NewConfiguration(
"github.com/org/pkg",
)
s := scraper.NewScraper(config)
Having a scraper instantiated, you can register a set of rules that will allow the scraper to identify the components to include in the output structure.
Each rule consists of:
- Set of package regexp's - only types in a package matching at least one of the package regexp's will be processed
- Name regexp - only type of name matching regexp will be processed
- Apply function - function that produces
model.Info
describing the component included in the scraped structure.
r, err := scraper.NewRule().
WithPkgRegexps("github.com/org/pkg/foo/.*").
WithNameRegexp("^.*Client$").
WithApplyFunc(
func(name string, _ ...string) model.Info {
return model.ComponentInfo(name, "foo client", "gRPC", "TAG")
}).
Build()
err = s.RegisterRule(r)
The apply function has two arguments: name and groups matched from the name regular expression.
See the example:
r, err := scraper.NewRule().
WithPkgRegexps("github.com/org/pkg/foo/.*").
WithNameRegexp(`^(\w*)\.(\w*)Client$`).
WithApplyFunc(
func(_ string, groups ...string) model.Info {
// Do some groups sanity checks first, then:
n := fmt.Sprintf("Client of external %s service", groups[1])
return model.ComponentInfo(n, "foo client", "gRPC", "TAG")
}).
Build()
err = s.RegisterRule(r)
Alternatively, you can instantiate the scraper form YAML configuration file:
// go-structurizr.yml
configuration:
pkgs:
- "github.com/org/pkg"
rules:
- name_regexp: "^.*Client$"
pkg_regexps:
- "github.com/org/pkg/foo/.*"
component:
description: "foo client"
technology: "gRPC"
tags:
- TAG
Regex groups may also be used within yaml rule definition. Here you can find an example:
rules:
- name_regexp: "(\\w*)\\.(\\w*)Client$"
pkg_regexps:
- "github.com/org/pkg/foo/.*"
component:
name: "Client of external {1} service"
description: "foo client"
technology: "gRPC"
tags:
- TAG
s, err := scraper.NewScraperFromConfigFile("./go-structurizr.yml")
Eventually, having the scraper instantiated and configured you can use it to scrape any structure you want. Scraper returns a struct model.Structure
.
structure := s.Scrape(app)
Similarly, to the scraper, view may be instantiated in one of two ways:
- from the code
- from the YAML file
In order to render scraped structure, you will need to instantiate and configure a view. View consists of:
- title
- component styles - styles are applied to the components by matching first of component tags with style ids
- additional styling (i.e. line color)
- component tags - if specified, view will contain only components tagged with one of the view tags. When no tag is defined, all components will be included in the rendered view.
- root component tags - if specified, view will contain only those components which have connection (direct or in-direct) to at least one of components with root tag.
In order to instantiate default view, use the view builder:
v := view.NewView().Build()
In case you need to customize it, use available builder methods:
v := view.NewView().
WithTitle("Title")
WithComponentStyle(
view.NewComponentStyle("TAG").
WithBackgroundColor(color.White).
WithFontColor(color.Black).
WithBorderColor(color.Black).
WithShape("database").
Build(),
).
WithComponentTag("TAG").
WithRootComponentTag("ROOT").
Build()
Alternatively, you can instantiate the view form YAML configuration file:
// go-structurizr.yml
view:
title: "Title"
line_color: 000000ff
styles:
- id: TAG
background_color: ffffffff
font_color: 000000ff
border_color: 000000ff
shape: database
root_component_tags:
- ROOT
component_tags:
- TAG
v, err := view.NewViewFromConfigFile("./go-structurizr.yml")
As the view is initialized, you can now render the structure into planUML diagram.
outFile, _ := os.Create("c4.plantuml")
defer func() {
_ = outFile.Close()
}()
err = v.RenderStructureTo(structure, outFile)
In order to see detailed scraping or view rendering logs, set LOG_LEVEL
env variable with debug
of DEBUG
.
The best results and experience in using the library will be ensured by enforcing the following practices:
- Having a solid and well-organized application context following clean-architecture principles will make your diagrams simple and easy to read. Also, this will allow you to create a short list of component types and styles.
- Following consistent naming conventions will help you in creating simple and straight-forward scraper rules.
https://pkg.go.dev/github.com/krzysztofreczek/go-structurizr