Serban Cezar
Serban Cezar

Reputation: 535

Groovy multiline string interpolation whitespace

I am trying to generate some generic Groovy code for Jenkins but I seem to have trouble with multi line strings and extra white space. I've tried everything I could find by Googling but I can't seem to get it working.

My issue isn't related to simple multi line strings. I managed to trim white space by using the stripIndent() and stripMargin() methods for simple cases. My issue is caused by having interpolated methods inside my strings.

Groovy info: Groovy Version: 3.0.2 JVM: 13.0.2 Vendor: Oracle Corporation OS: Mac OS X

String method2(String tier, String jobName) {
    return """
            Map downstreamJobs = [:]
            stage ("${jobName}-${tier}-\${region}_${jobName}") {
                test
            }
        """.stripIndent().stripMargin()
}

static String simpleLog() {
    return """
            script {
               def user = env.BUILD_USER_ID
            }
          """.stripIndent().stripMargin()
}

static String method1() {
    return """\
            import jenkins.model.Jenkins
            currentBuild.displayName = "name"

            ${simpleLog()}
        """.stripIndent().stripMargin()
}

String generateFullDeploymentPipelineCode() {
    return """Text here
            ${method1()}
            ${method2("test1", "test2")}
            """.stripIndent().stripMargin()
}

println(generateFullDeploymentPipelineCode())

This is what it prints(or writes to disk):

Text here
                      import jenkins.model.Jenkins
          currentBuild.displayName = "name"

script {
   def user = env.BUILD_USER_ID
}



Map downstreamJobs = [:]
stage ("test2-test1-${region}_test2") {
    test
}

Why the extra space around the import lines? I know the indentation method is supposed to trim all white space according to the least number of leading spaces, so that's why we use backslash (example here https://stackoverflow.com/a/19882917/7569335).

That works for simple strings, but it breaks down once use start using interpolation. Not with regular variables, just when you interpolate an entire method.

Upvotes: 4

Views: 5754

Answers (2)

daggett
daggett

Reputation: 28564

as variant - use just stripMargin() and only once on a final string

String method2(String tier, String jobName) {
    return """\
            |Map downstreamJobs = [:]
            |stage ("${jobName}-${tier}-\${region}_${jobName}") {
            |    test
            |}
        """
}

static String simpleLog() {
    return """\
            |script {
            |   def user = env.BUILD_USER_ID
            |}
          """
}

static String method1() {
    return """\
            |import jenkins.model.Jenkins
            |currentBuild.displayName = "name"

            ${simpleLog()}
        """
}

String generateFullDeploymentPipelineCode() {
    return """\
            |Text here
            ${method1()}
            ${method2("test1", "test2")}
            """.stripIndent().stripMargin()
}

println(generateFullDeploymentPipelineCode())

result:

Text here
import jenkins.model.Jenkins
currentBuild.displayName = "name"

script {
   def user = env.BUILD_USER_ID
}

Map downstreamJobs = [:]
stage ("test2-test1-${region}_test2") {
    test
}

another variant with trim() and stripIndent()

def method2(String tier, String jobName) {
    return """
            Map downstreamJobs = [:]
            stage ("${jobName}-${tier}-\${region}_${jobName}") {
                test
            }
        """.trim()
}

def simpleLog() {
    return """
            script {
               def user = env.BUILD_USER_ID
            }
          """.trim()
}

def method1() {
    return """
            import jenkins.model.Jenkins
            currentBuild.displayName = "name"
            ${simpleLog()}
        """.trim()
}

def generateFullDeploymentPipelineCode() {
    return """\
            Text here
            ${method1()}
            ${method2("test1", "test2")}
            """.stripIndent()
}

println(generateFullDeploymentPipelineCode())

Upvotes: 2

zett42
zett42

Reputation: 27756

When you insert a string through interpolation you only indent the first line of it. The following lines of the inserted string will be indented differently, which messes everything up.

Using some lesser-known members of GString (namely .strings[] and .values[]), we can align the indentation of all lines of each interpolated value.

String method2(String tier, String jobName) {
    indented """
        Map downstreamJobs = [:]
        stage ("${jobName}-${tier}-\${region}_${jobName}") {
            test
        }
    """
}

String simpleLog() {
    indented """\
        script {
           def user = env.BUILD_USER_ID
        }
    """
}

String method1() {
    indented """\
        import jenkins.model.Jenkins
        currentBuild.displayName = "name"

        ${simpleLog()}
    """
}

String generateFullDeploymentPipelineCode() {
    indented """\
        Text here
        ${method1()}
        ${method2("test1", "test2")}
    """
}

println generateFullDeploymentPipelineCode()

//---------- Move the following code into its own script ----------

// Function to adjust the indentation of interpolated values so that all lines
// of a value match the indentation of the first line.
// Finally stripIndent() will be called before returning the string.

String indented( GString templ ) {

    // Iterate over the interpolated values of the GString template.
    templ.values.eachWithIndex{ value, i ->

        // Get the string preceding the current value. Always defined, even
        // when the value is at the beginning of the template.
        def beforeValue = templ.strings[ i ]

        // RegEx to match any indent substring before the value.
        // Special case for the first string, which doesn't necessarily contain '\n'. 
        def regexIndent = i == 0
                          ? /(?:^|\n)([ \t]+)$/
                          : /\n([ \t]+)$/

        def matchIndent = ( beforeValue =~ regexIndent )
        if( matchIndent ) {
            def indent = matchIndent[ 0 ][ 1 ]
            def lines = value.readLines()
            def linesNew = [ lines.head() ]  // The 1st line is already indented.
            // Insert the indentation from the 1st line into all subsequent lines.
            linesNew += lines.tail().collect{ indent + it }
            // Finally replace the value with the reformatted lines.
            templ.values[ i ] = linesNew.join('\n')
        }
    }

    return templ.stripIndent()
}

// Fallback in case the input string is not a GString (when it doesn't contain expressions)
String indented( String templ ) {
    return templ.stripIndent()  
}

Live Demo at codingground

Output:

Text here
import jenkins.model.Jenkins
currentBuild.displayName = "name"

script {
   def user = env.BUILD_USER_ID
}

Map downstreamJobs = [:]
stage ("test2-test1-${region}_test2") {
    test
}

Conclusion:

Using the indented function, a clean Groovy syntax for generating code from GString templates has been achieved.

This was quite a learning experience. I first tried to do it completely different using the evaluate function, which turned out to be too complicated and not so flexible. Then I randomly browsed through some posts from mrhaki blog (always a good read!) until I discovered "Groovy Goodness: Get to Know More About a GString". This was the key to implementing this solution.

Upvotes: 1

Related Questions