Variables in pipelines

Codefresh provides a set of predefined variables automatically in each build, that you can use to parameterize the way your pipeline works. You can also define your own variables. Some common examples of predefined variables include:

• CF_BRANCH is the Git branch that was used for this pipeline.
• CF_REVISION is the Git hash that was used for this pipeline.
• CF_BUILD_URL is the url of the pipeline build.

Using Codefresh variables in your pipelines

There are two ways to use a Codefresh variable in your pipelines:

1. Expose as UNIX environment variables This is the default method, where all variables in all freestyle steps are exposed as UNIX environment variables: $MY_VARIABLE_EXAMPLE.
2. Directly in YAML properties Use variables can be used in YAML properties with the syntax ${{MY_VARIABLE_EXAMPLE}}.

TIP:
If you are unsure about which form you need to use, feel free to use ${{MY_VARIABLE_EXAMPLE}} everywhere. This is the Codefresh specific form and should function in all sections of codefresh.yml.

Example: Print out the branch as an environment variable
In this example, we use simple echo commands, but any program or script that reads environment variables can also read them in the same manner.

YAML

MyOwnStep:
  title: Variable example
  image: alpine
  commands: 
    - echo $CF_BUILD_ID 
    - echo $CF_BRANCH_TAG_NORMALIZED 

Example: Use directly in YAML properties

YAML

MyAppDockerImage:
  title: Building Docker Image
  type: build
  image_name: my-own-app
  tag: ${{CF_BRANCH_TAG_NORMALIZED}}

Example: Concatenating variables

YAML

MyAppDockerImage:
  title: Building Docker Image
  type: build
  image_name: my-own-app
  tag: ${{CF_BRANCH_TAG_NORMALIZED}}-${{CF_SHORT_REVISION}}

The concatenation above creates Docker images with tags such as:

master-df6a04c
develop-ba1cd68
feature-vb145dh

NOTE:
This syntax is specific to Codefresh, and is only available within the Codefresh YAML file itself. If you want to write scripts or programs that use the Codefresh variables, you need to make them aware of the environment variable form.

System variables

System variables are automatically injected to any freestyle step as environment variables.

NOTE:
It is important to understand that all Git-related variables such CF_BRANCH, CF_COMMIT_MESSAGE, CF_REVISION etc. come directly from the Git provider you use, and therefore have the same limitations of that provider.

For example, GitLab sends less information in pull request events than normal pushes, and Bitbucket sends only the short hash of a commit in pull request events. We suggest you read the documentation of your Git provider first to understand what information is available for every Git event.

Gerrit has change-Id and Changes that you can map to CF_PULL_REQUEST variables, such as CF_PULL_REQUEST_ID and more.

Context-related Variables

Context-related variables are created dynamically during the workflow execution and according to the used steps.

A very common pattern in Codefresh pipelines, is to create a Docker image in one step, and then run a command on its container in the next step (e.g. run unit tests):

YAML

version: '1.0'
steps:
  MyAppDockerImage:
    title: Building Docker Image
    type: build
    image_name: my-own-app
  MyUnitTests:
    title: Running Unit tests
    image: ${{MyAppDockerImage}}
    commands: 
      - ./my-unit-tests.sh

In the example above you can see the MyAppDockerImage variable that denotes a Docker image created dynamically within this single pipeline. In the second step we use it as a Docker context in order to run unit tests. See also the unit testing example app.

Step variables

Every step in a Codefresh pipeline also exposes several built-in variables. You can access them via the global steps parent variable.

• Each step creates a variable based on the name of the step. You can then use the members of each variable for status conditions such as: steps.MyUnitTests.result == 'error' for a step called MyUnitTests.
• To access variables that have a non-standard (i.e. only alphanumeric and _ characters) names, use the Variable() function.

Step Member variables

Variables that are created by steps can have members. The members depend on the step type. For example if you have a build step named myBuildStep you can get the ID of the docker image that gets created with echo ${{steps.myBuildStep.imageId}}

GitHub release variables

GitHub allows you to create releases for marking specific Git tags for general availability.

You can set a trigger for GitHub releases. When a GitHub release happens, the following variables are also available:

GitHub Pull Request variables

When a pull request is closed in GitHub, the following variables are also available

Gerrit changeId & change message variables

Gerrit has no explicit concept of pull requests, as in other version control systems, to map trigger event payloads to builds. Instead, Gerrit uses Changes which serves a similar purpose and functionality as pull requests. You can achieve the same functionality in Codefresh with our CF_PULL_REQUEST group of environment variables.

User-defined variables

User variables can be defined at 6 levels:

1. Manually within a step using the export command or in any subsequent step with the cf_export command
2. Freestyle step definition (using the environment field)
3. Specific build Execution (after clicking the “Build” button open the “Build Variables” section, or use the CLI)
4. Pipeline Definition (under “Environment variables” section in the pipeline view)
5. Shared Configuration (defined under your account settings, and used using the “Import from shared configuration” button under the “Environment Variables” section in the pipeline view)
6. Variables defined on the Project level (Under the variables tab on any project view)

The options are listed in order of priority (from the most important to the least important), so in case of multiple variables defined at different locations with the same name, the order of overriding will be as listed here.

For example if a pipeline variable is defined both in project level and as an execution parameter of a specific build, then the final result will be the value defined as a build parameter and the project level variable will not take effect.

Exporting environment variables from a freestyle step

Steps defined within steps are scoped to the step they were created in (even if you used the export command). To use variables across steps, we provide a shared file through which you can import and export variables. There are two ways to add variables to the shared file:

• Using the cf_export command
• Writing directly to the file

Exporting variables with cf_export

Within every freestyle step, the cf_export command allows you to export variables across steps by writing to the shared variables file.

The variables exported through cf_export overrides those at the pipeline-level.

You can either:

• Explicitly state a VAR=VAL pair
• State the name of an existing exported environment variable (like EXISTING_VAR).

version: '1.0'
steps:
  freestyle-step-1:
    description: Freestyle step..
    title: Free styling
    image: alpine:latest
    commands:
      # Normal export will only work in a single step
      - export EXISTING_VAR=www.example.com

      # CF export will now work in all other subsequent steps
      - cf_export VAR1=alpine:latest VAR2=VALUE2 EXISTING_VAR

  freestyle-step-2:
    description: Freestyle step..
    title: Free styling 2
    image: ${{VAR1}}
    commands:
      - echo $VAR2
      - echo http://$EXISTING_VAR/index.php

cf_export syntax

cf_export has the same syntax as the bash export command.
This means that when you use it you don’t need any dollar signs for the variable that is created/assigned.

cf_export $MY_VAR # Don't do this
cf_export MY_VAR # Correct syntax

Masking variables within cf_export

Mask variables within cf_export by defining the --mask flag.
Values of masked variables in cf_export commands are replaced with asterisks in the build logs. This helps to ensure that sensitive information is not exposed also in the variables list, in addition to the logs.

Here is an example with standard and masked versions of the same variable in cf_export commands.

version: '1.0'
steps:
  freestyle-step:
    description: Freestyle step..
    title: Free styling
    image: alpine:latest
    commands:
      - export EXISTING_VAR=some-value

      - cf_export VAR1=alpine:latest VAR2=VALUE2 EXISTING_VAR --mask
      

Export variables to all steps with cf_export

By default, cf_export works only on subsequent steps.
To export a variable both in the current and to all the remaining steps, do the following:

export MY_VAR='example' # Makes MY_VAR available in this step only
cf_export MY_VAR='example' # Makes MY_VAR available also to all steps after this one

There is nothing really magic about cf_export. It is a normal script. You can see its contents on your own by entering the command cat /codefresh/volume/cf_export on any Codefresh freestyle step inside a pipeline.

For more information on its limitations, see the troubleshooting page.

Directly writing to the file

For more advanced use cases, you can write directly to the shared variable file that Codefresh reads to understand which variables need to be available to all steps.

This file has a simple format where each line is a variable and its value in the form of VARIABLE=VALUE. The cf_export command mentioned in the previous section is just a shorthand for writing on this file.

The variables file is available inside freestyle steps in the following path: ${{CF_VOLUME_PATH}}/env_vars_to_export

version: '1.0'
steps:
  freestyle-step-1:
    description: Freestyle step..
    title: Free styling
    image: alpine:latest
    commands:
      - echo VAR1=192.168.0.1 >> ${{CF_VOLUME_PATH}}/env_vars_to_export
      - echo hey=alpine:3.9 >> ${{CF_VOLUME_PATH}}/env_vars_to_export
          
  freestyle-step-2:
    description: Freestyle step..
    title: Free styling 2
    image: ${{hey}}
    commands:
      - echo http://$VAR1/index.php
      

Use this technique if you have complex expressions that have issues with the cf_export command.

Masking variables in logs

Codefresh has the built-in capability to automatically mask variables in logs if they are encrypted. The values of encrypted variables will be replaced with asterisks in build logs.

The variables can be defined in any of the usual ways Codefresh offers such as shared configuration or within the pipeline:

Notice that this feature is currently available only in Enterprise accounts.

Encrypt variables for pipeline build runs

When you run a pipeline manually, you can either add variables on the fly or change the value of existing variables for the specific run of the build. You can also encrypt the build-specific variables, new or existing, at the same time.

1. Do one of the following:
   • From the Pipelines page, select the pipeline and click Run on the right.
   • From the Builds page, select the pipeline to run manually, and then click Run.
2. Expand Build Variables.
3. To add a variable, click Add Variable.
   1. Enter the Key and Value for the new variable.
   2. To encrypt, click Encrypt, and confirm.

Escape special characters in variables

When passing special characters through environmental variables, use \ as an escape character.
For example, to pass a cassandra connection string, you might do something like Points\=hostname\;Port\=16376\;Username\=user\;Password\=password.

This will safely escape ; and =.

Related articles

Steps in pipelines
Conditional execution of steps