This Plugin provides parameter type that can request a REST endpoint and offer a list of values parsed from the response value at build start time. The REST responses may contain Json or XML, which can be parsed with Json-Path and xPath respectively. If the REST endpoint requires authentication, then either Basic or Bearer authentication can be used.
- Introduction
- Changelog
- Features
- Configuration
- Json-Path configuration
- xPath configuration
- Authentication configuration
- Custom HTTP headers
- Pagination
- Parameter Config Validation
- Loading values on the build form
- Value Caching
- Optional Value Ordering
- Definition in Pipeline
- Definition in JobDsl
- REST Multi List Parameter
- Set Global Options via JCasC
- Contributing
- Licence
- Bundled Project Sources
- Development
Release notes can be found in the GitHub Releases
-
Specify any REST/Web endpoint that returns either a Json or XML response with values to select from for a build
-
Authenticate against the endpoint if required (either with
BASICorBEARERauthentication) -
Add custom HTTP headers to REST requests
-
Follow paginated REST APIs (
Linkheader or continuation token) to offer values beyond the first page -
Filter/Pars the values of the web response with Json-Path or xPath (depending on MIME type)
-
Set a per-selected default value (as long as that value is within the received value list)
-
Pick several values in one build with the REST Multi List Parameter
-
Values load in the background after the
Build with Parameterspage opens, each parameter on its own, with a refresh action -
User-friendly error communication, with request details for users who can configure the job
-
Value caching independent of the endpoint’s caching headers, with a bounded fetch time
-
Advanced configuration validation
-
Optional value post processing
-
Filter the already parsed values via a REGEX
-
Apply a sort order to the values
-
If you expect the response of the specified endpoint to produce a Json response and need to utilize Json-Path, then you have to set the MIME type to APPLICATION_JSON (this will also send this MIME as ACCEPT header) and provide the desired Json-Path in the Value Expression field.
The result of such a configuration will be the application of the Json-Path on the response Json to pars the string values later provided in the Build with Parameter dropdown of this parameter.
|
Note
|
Added in v1.5.x thanks to rolaca11 |
The optional 'display value' configuration, hidden within the advanced configuration, allows the use of a REST service that returns full Json objects that get pased into the Pipeline run but to display only an identifying value within the select dropdown.
[
{
"name": "v10.7.7",
"zipball_url": "https://api.github.com/repos/jellyfin/jellyfin/zipball/v10.7.7",
"tarball_url": "https://api.github.com/repos/jellyfin/jellyfin/tarball/v10.7.7",
"commit": {
"sha": "53186c766bddf954483d7b40e385b382eabd0d8e",
"url": "https://api.github.com/repos/jellyfin/jellyfin/commits/53186c766bddf954483d7b40e385b382eabd0d8e"
},
"node_id": "MDM6UmVmMTYxMDEyMDE5OnJlZnMvdGFncy92MTAuNi40"
},
{
"...": "..."
}
]
With a Value Expression such as $.* and a Display Expression such as $.name the values in the dropdown would only show up as the name property of each JSON object (i.e. v10.7.7).
|
Important
|
If you are a user of this functionality please have a look at the pined issue #99 |
If you expect the response of the specified endpoint to produce an XML response and need to utilize xPath, then you have to set the MIME type to APPLICATION_XML (this will also send this MIME as ACCEPT header) and provide the desired xPath in the Value Expression field.
The result of such a configuration will be the application of the xPath on the response XML to pars the string values later provided in the Build with Parameter dropdown of this parameter.
Depending on the authentication requirement for the REST/Web endpoint there are three supported methods:
-
NONEif there is no authentication requirement simply don’t select any Credential ID -
BASICif this authentication type is required, then simply select any Username and Password credential in the Credential ID field -
BEARERif this authentication type is required, then simply select any Secret Text credential in the Credential ID field
|
Important
|
The Authentication header will be build and added based on the type of the selected credential type. |
Custom HTTP headers can be added from the parameter’s advanced configuration. This is useful for APIs that require headers such as X-API-Key, X-Auth-Token, or a custom Authorization format that is not covered by the built-in Basic or Bearer authentication support.
Each custom header entry supports:
-
Header name- for exampleX-API-Key,X-Auth-Token, orAuthorization -
Static value- stored by Jenkins as aSecret -
Secret Text credential- recommended for API keys, tokens, and other sensitive values -
Value prefix- optional text prepended to the static value or credential value, for example `Token `
For sensitive header values, create a Jenkins credential of type Secret text and select it in the custom header row. This keeps tokens out of the job configuration and allows credentials to be rotated centrally.
| Header name | Credential / value | Resulting request header |
|---|---|---|
|
Secret Text credential |
|
|
Secret Text credential |
|
|
Value prefix |
|
Custom headers are applied after the default Accept header and the optional authentication header from Credential ID. If a custom Authorization header is configured, it overrides the default Basic or Bearer authentication header.
Test Configuration sends the custom headers exactly as a build would, including rows you have added or changed but not saved yet. The REST Endpoint field check does not send custom headers. If the endpoint rejects that check with 401 or 403, it shows a warning instead of an error; use Test Configuration to verify the headers.
Many APIs return their values in pages, for example GitHub (at most 100 items per page) or Nexus 3. Without pagination, the parameter only offers the values of the first page. Check Follow paginated responses in the parameter’s advanced configuration and pick a strategy:
| Strategy | How the next page is found | Used by |
|---|---|---|
|
The URL of the |
GitHub, GitLab, Gitea, Harbor |
|
A token read from the Json body with the |
Nexus 3, Bitbucket Server ( |
Pagination ends when a page has no next link, or when the token expression finds nothing, null or an empty string.
The Value Expression and Display Expression are applied to each page on its own, so write them for a single page (for Nexus 3, $.items[*].version, not an expression over all pages). The entries of all pages are joined in page order, and the Value Filter Pattern and Apply Sort Order are then applied once to the combined list. A page without matches is fine. Only when all pages together yield nothing is it reported as "no values" (subject to Allow Empty Value).
- Page limit
-
Max Pageslimits the number of requests per fetch, counting the first one: 10 by default, at most 100 (larger values are treated as 100). When the limit is reached while the last page still points to a next page, the values fetched so far are used and a warning is logged. All pages are requested one after another when the parameter’s values load, so each page adds to the time until they appear, and all of them together are bounded by the global fetch timeout. The combined values of all pages are cached as one list for theCache Max Age(see Value Caching). - Same-host rule
-
Every page is requested with the same credentials and custom headers as the first one. To keep them from leaking, pagination only follows URLs with the same scheme, host and port as the
REST Endpoint. A next page on another host fails the fetch withNext page is not on the same host as the Rest Endpoint: <origin>, and no request is sent to it. - Failures and loops
-
A failure on any page fails the whole list. For page 2 and later, the error names the page, for example
Encountered Http Server Error: 503 (page 4). A next page that was already requested, or a repeated token, ends pagination with a warning and keeps the values fetched so far.
Test Configuration uses the form’s pagination settings, including unsaved changes, and reports the pages fetched, for example Test Successful! 57 Values from 3 pages, first: v1.0. When the page limit was reached, the result is a warning ending in (page limit reached, more values may exist).
Link header)
RESTList(
name: 'TAG',
description: '',
restEndpoint: 'https://api.github.com/repos/jenkinsci/jenkins/tags?per_page=100',
credentialId: '',
mimeType: 'APPLICATION_JSON',
valueExpression: '$.*.name',
pagination: linkHeader(maxPages: 5)
)
RESTList(
name: 'VERSION',
description: '',
restEndpoint: 'https://nexus.example.com/service/rest/v1/components?repository=releases',
credentialId: 'nexus-user',
mimeType: 'APPLICATION_JSON',
valueExpression: '$.items[*].version',
pagination: continuationToken(tokenExpression: '$.continuationToken', queryParameter: 'continuationToken')
)
Omitting pagination keeps a single request per fetch. The same pagination property is available on the REST Multi List Parameter.
|
Note
|
Added in v1.1.0 |
The configuration page of this plugin features supportive form validation (which was already present in a basic form in v1.0.x). With v1.1.x+ the validation got expanded to be more helpful and prevent the constant switch from the configuration to a build and back, just to validate the configuration. The most notable expansion in validation was the REST Endpoint field validation, which now works similar to the validation found for the SCM fields (Endpoint and credential). Additionally, a Test Configuration button got added, which once pressed, attempts to query the endpoint and list the number of parsed values as well as the first value in the list.
The Build with Parameters page opens without waiting for any endpoint. Each REST List Parameter and REST Multi List Parameter shows a loading indicator and requests its values on its own, so a slow endpoint only delays its own parameter. Once the values arrive, the parameter shows them in its dropdown or multi-select, with the usual default preselection.
-
While any parameter is still loading, the form is not submitted; clicking
BuildshowsValues are still loadingbeneath the parameters still loading. A parameter whose load failed does not block the form: the submitted value is validated as usual. -
A failed load shows its error beneath the parameter with a
Retryaction. Users who can configure the job also see the endpoint URL (without query string and user info), the page that failed, the HTTP status or exception type and the duration; response bodies and secrets are never shown. -
The refresh button next to each parameter loads its values again, bypassing all caches. The selected entry stays selected if it is still offered, and custom values picked with validation disabled are kept, in the dropdown as in a multi-select.
-
With
Enable Value Validationunchecked, aREST List Parameteralso accepts a value that is not among the entries: type it into the dropdown’s search field and pickUse "<text>", and it is submitted verbatim. -
Loading the values requires the Build permission on the job. In a Pipeline
inputstep, users without Build on the job cannot load the values. -
The values are shown by JavaScript, as Select2 already required.
- Fetch timeout
-
One fetch of a parameter’s values, including all its pages, may take at most the global
Fetch Timeout(default 60 seconds). When it is reached, the parameter showsFetching values timed out after <n> secondsinstead of values.
|
Note
|
Added in v1.2.0; since the values load in the background, the processed values are cached independently of the endpoint’s caching headers. |
The ability to use value caching allows for great performance improvements in cases of repetitive building and querying the same REST endpoints. Additionally, it drastically decreases the amount of endpoint queries, which is especially useful for rate limited endpoints. Per default caching is disabled for new and old REST List Parameter, but can be activated at any time by setting how many minutes the values should stay cached for.
With a Cache Max Age greater than 0, the final values of a successful fetch (all pages, after the expressions, the filter and the order) are kept in memory for that many minutes. This works whatever caching headers the endpoint sends, including Cache-Control: no-store. The build form, the validation of a submitted value and Rebuild all use these values:
-
Values are cached per job, parameter name and value source (endpoint, credentials, MIME type, expressions, filter, order, custom headers and pagination), so jobs with different credentials never share values, and changing the configuration makes earlier values unused.
-
Failed fetches are not cached, and do not remove values cached earlier.
-
A submitted value found in fresh cached values is accepted without contacting the endpoint. Otherwise the values are fetched once more, bypassing all caches, before the value is accepted or rejected, so a value published after the values were cached (for example a release triggered by a Pipeline) is accepted.
-
The cache holds the values of up to 1000 parameters, evicting the least recently used ones first (system property
io.jenkins.plugins.restlistparam.ValueCache.maxEntries), is never written to disk, and is empty after Jenkins restarts.
Cached values are replaced with fresh ones by the refresh button on the build form, and administrators can clear all of them with Clear Value Cache in the global configuration, without saving it. Below this cache, endpoint responses are still kept in the HTTP response cache on disk, bounded by the global cache size, as long as the endpoint’s own caching headers allow it.
Additionally, there are global configuration values now, that give the user control over the global cache size (default 50 MiB), as well as the default value for the cache max age (default 0 minutes) and the fetch timeout (default 60 seconds). These values can be overwritten within the UI or via the ConfigurationAsCode plugin and applying a config with this added.
|
Note
|
Added in v1.2.0 |
This option in the "Advanced" section of the parameter configuration allows the user to apply a sort order on the values received from the endpoint specified. The default value of this option is NONE, so no additional sporting gets applied on the values received. The other two options offer tha ability to apply either an ascending or descending sort order to the received values.
pipeline {
agent any
parameters {
RESTList(
name: 'REST_VALUE',
description: '',
restEndpoint: 'http://test.lan',
credentialId: '',
mimeType: 'APPLICATION_JSON',
valueExpression: '$.*',
cacheTime: 10, // optional
defaultValue: '', // optional
filter: '.*', // optional
valueOrder: 'ASC', // optional
customHeaders: [ // optional
[name: 'X-API-Key', credentialId: 'example-api-key'],
[name: 'Authorization', credentialId: 'example-auth-token', valuePrefix: 'CustomFormat ']
],
pagination: linkHeader(maxPages: 5) // optional, see Pagination
)
}
stages {
stage('Test') {
steps {
println env.REST_VALUE
}
}
}
}
pipelineJob('DemoJob') {
parameters {
RESTList {
name('REST_VALUE')
description('TEST')
restEndpoint('http://test.lan')
credentialId('credID')
mimeType('APPLICATION_JSON')
valueExpression('$.*')
cacheTime(10) // optional
defaultValue('') // optional
filter('.*') // optional
valueOrder('NONE') // optional
}
}
definition {
cps {
script("""
pipeline {
agent any
stages {
stage('Test') {
steps {
println env.REST_VALUE
}
}
}
}
""")
sandbox()
}
}
}
The REST Multi List Parameter lets the user pick several entries from the same kind of REST response in one build. It offers the same configuration as the REST List Parameter (endpoint, credentials, MIME type, value and display expressions, filter, order, cache and custom headers) and fetches its entries the same way. The build form shows a searchable multi-select.
Its value is an ordered list of strings. Entries are kept in the order they are listed in, and duplicates are removed.
The symbols are RESTMultiList, RestMultiList and RESTMultiListParam. They take the same properties as RESTList.
pipeline {
agent any
parameters {
RESTMultiList(
name: 'TARGETS',
description: 'Where to deploy',
restEndpoint: 'https://inventory.example.com/api/targets',
credentialId: '',
mimeType: 'APPLICATION_JSON',
valueExpression: '$.*.name',
defaultValue: '["staging"]', // optional, see below
allowEmptyValue: false, // optional
enableValidation: true // optional
)
}
stages {
stage('Deploy') {
steps {
script {
params.TARGETS.each { target ->
echo "Deploying to ${target}"
}
}
}
}
}
}
In Pipeline, params.TARGETS is a List of strings. The same list can be used for fan-out, e.g. parallel params.TARGETS.collectEntries { t -> [(t): { echo "deploy ${t}" }] }.
In Job DSL the parameter is declared as RESTMultiList { ... }, with the same properties as RESTList.
The environment variable (and its upper-cased alias) holds the list as a compact JSON array of JSON strings, for example ["staging","prod"]. An empty list is []. Every element is written as a JSON string, even when its text is JSON itself: an element {"name":"v1"} becomes ["{\"name\":\"v1\"}"].
In a shell step, iterate over the elements with jq:
echo "$TARGETS" | jq -r '.[]' | while read -r target; do
echo "Deploying to $target"
done
The CLI, the remote API, the Pipeline build step with a string parameter and re-runs pass the value as one string. It is read as follows:
-
A JSON array whose elements are all JSON strings becomes that list:
["a","c"]isa,c. -
The empty string and
[]are the empty list. -
Any other string is a list with that one string:
aisa, anda,cis the single elementa,c.
Strings are never split on commas or any other delimiter.
A single element that is itself a JSON array of strings must be wrapped in an array: ["[\"x\"]"] is the single element ["x"].
build job: 'deploy', parameters: [string(name: 'TARGETS', value: '["staging","prod"]')]
On the remote API, pass the list as one URL-encoded JSON array:
curl -X POST "$JENKINS_URL/job/deploy/buildWithParameters" \
--user "$USER:$TOKEN" \
--data-urlencode 'TARGETS=["staging","prod"]'
Repeating the parameter name (?TARGETS=staging&TARGETS=prod) is not supported: Jenkins rejects the request and no build starts.
With Enable Value Validation checked (the default), every element must be the value of an entry. The list is accepted from the cached values when they are fresh and contain every element; otherwise the entries are fetched once when the build is submitted and every element is checked against them. If that fetch fails, any non-empty list is rejected.
With it unchecked, the user can also add entries that are not in the list: typing into the multi-select offers the typed text as Use "<text>", and picking it adds the text verbatim as its own element. The endpoint is then not contacted to validate the submission, which helps while it is unavailable and for values it does not (yet) return.
An empty list is accepted only when Allow Empty Value is checked. Empty elements are always rejected. A rejected value stops the build with Illegal value for parameter <name>: <detail>, where the detail is the first offending element, "" for an empty element, or [] for a disallowed empty list.
The default value refers to entries by their display value and is read with the same string rule: ["Alpha","Gamma"] preselects two entries, and Beta preselects one. Default values that match no entry are ignored when validation is enabled; when it is disabled, they are preselected as custom entries. When a build is re-run, the previous selection becomes the new form’s default.
The value of this parameter is a list, not a string. Plugins that assume a parameter value is a string may therefore handle it differently than a REST List Parameter:
-
Parameterized Trigger’s "Current build parameters" copies the parameter value object unchanged, so the list is expected to reach the triggered build as is.
-
Rebuilder (checked with version 332): "Rebuild" opens the multi-select with the previous selection preselected, and the rebuilt build receives the same list.
I welcome all contributions and pull requests! If you have a larger feature in mind please open an issue, so we can discuss the implementation before you start.
|
Note
|
I prefer GitHub Issues over Jira Issues, but I check both regularly. |
For further contributing info please have a look at the JenkinsCI contribution guidelines.
|
Note
|
The vendored Select2-Bootstrap4-Theme was removed; the dropdowns are styled with the plugin’s own Select2 theme, built on the Jenkins design tokens. |
No third-party sources are bundled with this plugin.
-
Java 8 or newer
-
Maven 3 or newer
-
(optional) a test Jenkins instance to deploy the plugin SNAPSHOT to for testing
$ # build, test and package to hpi (hpi can be deployed to test Jenkins)
$ mvn -B package --file pom.xml
$ # build, test, package and launch test Jenkins
$ mvn -B hpi:run --file pom.xml
|
Note
|
This Plugin uses SemVer to version its releases |
To creat a new release follow the instruction found for the Incremental tools to create a release whilst incrementing the correct position of the SemVer.











