Skip to content

Latest commit

 

History

History

connect-publish

README.md

rstudio/actions/connect-publish

This action may be use for publishing to RStudio Connect.

Important Note: The only currently supported type of publishing is "from manifest". This type of publishing requires that a manifest.json has already been generated by either the R or Python rsconnect package, which may be done within an earlier GitHub Actions step or performed locally and stored in git.

Using this Action

R users may find it easiest to set up scaffolding for GitHub Actions using the usethis package.

You can import and use this GitHub Action in your project using

usethis::use_github_action(url = "https://raw.githubusercontent.com/rstudio/actions/main/examples/connect-publish.yaml")

Once imported, you can adjust the settings below to match your use case.

Inputs

url

Required RStudio Connect URL of the instance to which content will be published. The value of the URL may contain an API key genereted in RStudio Connect (equivalent to the api-key input) in the "user" or "password" part.

dir

Directory or directories containing the content to publish, optionally with :<app-identifier> suffixes, separated by newlines. Without a :<app-identifier> suffix, the value of the directory path will be used to match any existing published content. (Default ".")

NOTE The :<app-identifier> may be one of the following:

  • GUID, e.g. ffffffff-aaaa-ffff-aaaa-ffffffffffff, which is used when the string is a valid GUID
  • name, e.g. name-without-slashes, which is tried when the string does not contain at least one slash
  • vanity URL (path), e.g. /name-with/a-slash, which is tried when the string does contain at least one slash

working-directory

Directory relative to which the 'dir' argument(s) will be resolved. (Default ".").

api-key

An API key genereted in RStudio Connect. This value may be included in url as the "user" or "password" part. (No default)

access-type

Access type to apply to published content. (No default)

Valid values are:

  • all publicly accessible
  • logged_in accessible only when logged into RStudio Connect
  • acl accessible to specific users and groups (managed within RStudio Connect)

When not specified, the RStudio Connect server configuration default applies. Access types disallowed by RStudio Connect server configuration will result in the publish being canceled.

namespace

Namespace prefix for published paths. (No default)

This value is used in the construction of the "vanity URL" path, e.g. this configuration:

  - uses: rstudio/actions/connect-publish@main
    with:
      url: https://${{ secrets.RSTUDIO_CONNECT_API_KEY }}@connect.example.org
      namespace: experimental-beta
      dir: ./very-shiny-app/:/shiny/app/path/

will result in a published URL of

https://connect.example.org/experimental-beta/shiny/app/path/

NOTE Because the namespace is used as the first segment of a slash-delimited path, there are restrictions on what is allowed by RStudio Connect. For example, a namespace of connect is not allowed and will result in an error at publish time.

show-logs

Show all publishing logs. (Default false)

force

Force publish even if up to date. (Default false)

NOTE: The determination of a given publish being up to date is based on the bundle generated from the manifest matching the current bundle according to RStudio Connect. If available, the bundle SHA1 digest is compared, else the bundle size is compared.

require-vanity-path

Require setting the vanity path. (Default false)

A given RStudio Connect server may be configured such that an application's vanity path may not be set by users with the publisher role. The default behavior will attempt to set the vanity path and any failure to do so will be logged in the debug logs. By setting require-vanity-path: true, any failure to set the vanity path will bubble up as an error and cancel publishing.

update-env

Update content environment variables. (Default true)

The environment variables configured for published content may be configured via the running process environment and via a specific set of files relative to the content directory.

setting from process environment

Any environment variable in the process environment that begins with the prefix CONNECT_ENV_SET_ will be set without the prefix, e.g.:

# ...
steps:
  - name: Publish some things
    uses: rstudio/actions/connect-publish@main
    env:
      CONNECT_ENV_SET_MYVAR: available at runtime
    with:
      url: https://${{ secrets.RSTUDIO_CONNECT_API_KEY }}@connect.example.org
      access-type: logged_in
      dir: |
        ./flask-api/

will result in the following variable being configured on the published content:

  • MYVAR='available at runtime'

setting from environment file

The first of the following files that is found will be read:

  • .rstudio-connect.env
  • rstudio-connect.env
  • rsconnect/env
  • rsconnect-python/env

The content of the file is expected to be shell-like with variables defined, and may contain comments, e.g.:

# the following variable is available at runtime
SECRET_INGREDIENT='tartlet'

# referencing the variable name without an assignment will read the
# variable from the process environment
MYVAR

# variables with leading "export" or "set" work, too
export FRIENDS="great"
set PASTA='usually awesome'

Outputs

results

JSON-serialized array of objects, each containing a dir, url, id, and success key indicating per-directory publish success. In case of an unhandled error, the value will be an empty array.

Example usage

steps:
  - uses: actions/checkout@v2

  - name: Publish some things
    uses: rstudio/actions/connect-publish@main
    with:
      url: https://${{ secrets.RSTUDIO_CONNECT_API_KEY }}@connect.example.org
      access-type: logged_in
      dir: |
        ./very-shiny-app/:/shiny/app/path/
        ./useful-rmarkdown-report/:useful-report-for-cats
        ./slick-flask-api/
        ./insightful-jupyter-notebook/:528f3dc0-3bd0-41e6-a191-aa264c451416