This extension allows Antora to retrieve content from maven coordinates in addition to git repositories. If parts of the documentation are generated or post-processed it's usually more convenient to package and publish the docs to a maven repository instead of making the build commit generated files into a git repo. Also, it's probably more common for Java hackers to refer to published artifacts than to git repos.
This extension recommends maven artifacts to be versioned according to SemVer
The extension has been tested with Antora 3.0.1
up to 3.1.9
.
Publish an artifact (zip
, jar
and tgz
supported) to a maven repository. Make sure the archive content adheres to
the antora folder structure (somewhere, since start paths
are supported, too).
Make sure it's available for antora (i.e. installed globally or along the playbook):
npm i @kiwigrid/antora-maven-content
Then add the extension to the playbook:
antora:
extensions:
- require: "@kiwigrid/antora-maven-content"
mavenSettings: true # defaults to false, true resolves to '$HOME/.m2/settings.xml' or '$M2_HOME/conf/settings.xml', a string is taken as is
repositories: # optional
- baseUrl: https://www.example.com # required for a repo
fetchOptions: # optional
headers:
"Authorization": "Basic <base64 encoded user:password>"
sources:
- groupId: "com.example" # required
artifactId: "antora-module" # required
version: "1.x.x" # defaults to '*'
versionScheme: "SemVer" # defaults to 'SemVer'
limit: 4 # defaults to 1
limitBy: minor # defaults to 'major', one of 'major', 'minor', 'patch', 'any'
includeSnapshots: true # defaults to false, true has no effect if includePrerelease is false as SNAPSHOTS are SemVer pre releases
includePrerelease: true # defaults to true
classifier: "" # defaults to 'docs'
extension: "tgz" # defaults to 'zip'
startPath: ~ # defaults to null
startPaths: "docs/*" # defaults to null
edit_url: "https://git.example.com/repos/myRepo/browse/{path}" # defaults to false
# ...
With above example configuration the extension is going to download all available versions for com.example:antora-module
and picks the 3 highest versions which:
-
match the SemVer Range
1.x.x
-
do not equal an already picked version when reduced to the
minor
version -
for example:
Available Versions Decision 2.0.0 ✖ does not match 1.x.x 1.2.1 ✅ 1.2.0 ✖ equals 1.2.1 when reduced to minor (1.2) 1.1.0 ✅ 1.0.2 ✅ 1.0.1 ✖ equals 1.0.2 when reduced to minor (1.0) 1.0.0 ✖ equals 1.0.2 when reduced to minor (1.0) 0.9.2 ✖ does not match 1.x.x 0.9.1 ✖ does not match 1.x.x 0.9.0 ✖ does not match 1.x.x
So the configuration basically picks the most recent patch versions of the four highest minor releases of major version 1. For each picked version a corresponding playbook content source entry is created which:
- points to a local transient cached on-demand git repository the artifact has been extracted to
- is configured with the same start path(s)
- is configured with an edit url
Note the Antora component version is totally unrelated to the maven artifact version and will be read by Antora from the component descriptor file antora.yml
contained in the artifact archive.
Scheme | Structure | version format | notes |
---|---|---|---|
SemVer |
<major>.<minor>.<patch>+<metadata>-<prerelease> |
any valid SemVer Range | recommended |
OSGI |
<major>.<minor>.<micro>.<qualifier> |
any valid OSGI range | micro is exposed as patch , there is no order between qualifiers |
Lexicographically |
any | any valid regular expression | minor and patch are always 0 , the complete version is the major part |
If mavenSettings
is given a maven settings.xml is parsed for repositories and authentication data. The value of the
option can be true
to use $HOME/.m2/settings.xml
or $M2_HOME/conf/settings.xml
or a string pointing to a settings
file. Only repositories of profiles which are active by default are extracted, mirrors are properly resolved.