pisa25-cdata-wrap.groovy

pisa25-cdata-wrap.groovy

/**
 * Usage : Put this script in <ScriptsDir>/project_changed folder. Create a folder if it doesn't exists.
 *
 * @authors 	Yu Tang, Manuel Souto Pico
 * @version 	0.5.6 (dynamic) -- match only XML files for PISA
 * @date		2023.05.25
 */


import static org.omegat.core.events.IProjectEventListener.PROJECT_CHANGE_TYPE.*
// import groovy.xml.XmlSlurper
// import org.apache.commons.lang.StringEscapeUtils
// import groovy.xml.MarkupBuilder

// check that this is a PISA project
prop = project.getProjectProperties()
proj_name = prop.projectName
// The container is the first entity in the project name (before the first underscore)
container = (proj_name =~ /^[^_]+/)[0]

if (container.contains("pisa")) {
  console.println("This script runs on PISA25 XML files, let's continue!")
} else {
  console.println("This project is not PISA, let's stop here. Good bye!")
  return
}

// prepare
String dir
def replacePair

def skipTraverse(eventType) {
    if (!eventType.metaClass.hasProperty(eventType, 'skipTraverse')) {
        eventType.metaClass.skipTraverse = false
    }
    eventType.skipTraverse
}

// tag = "(/?(?:[bui]|br/?|h1|h2|h3|h4|em|strong|ul|ol|li|span|div|p|sup|table|tbody|d|th|tr))"

switch (eventType) {
    case LOAD:
        // Skip traverse
        if (skipTraverse(LOAD)) {
            LOAD.skipTraverse = false // reset the flag
            return
        }
        dir = project.projectProperties.sourceRoot
        replacePair = []
        break
    case COMPILE:
        dir = project.projectProperties.targetRoot
        
        replacePair = [
            [find: /(?s)<text>([\s\S]+?)<\/text>/, replacement: /<text><![CDATA[$1]]><\/text>/]
        ]
        break
    default:
        return null // No output
}

String ENCODING = 'UTF-8'
File rootDir = new File(dir)
int modifiedFiles = 0

// options as map
def options = [
    type       : groovy.io.FileType.FILES,
    nameFilter : ~/.*\.xml/
]

// replacer as closure
def replacer = {file ->
    String text = file.getText ENCODING
    String replaced = text
    replacePair.each { replaced = replaced.replaceAll(it.find, it.replacement) }
    if (text != replaced) {
        file.setText replaced, ENCODING
        console.println "modified: $file"
        modifiedFiles++
    }
}

def reloadProjectOnetime = {
    LOAD.skipTraverse = true    // avoid potentially infinity reloading loop
    javax.swing.SwingUtilities.invokeLater({
        org.omegat.gui.main.ProjectUICommands.projectReload()
    } as Runnable)
}

// do replace
rootDir.traverse options, replacer

if (modifiedFiles > 0 && eventType == LOAD) {
    console.println "$modifiedFiles file(s) modified."
    reloadProjectOnetime()
}