Run CodeNarc over the code blocks on the groovy user-guide webpages

scanner.groovy

@Grab(group='org.codenarc', module='CodeNarc', version='0.15')
@Grab(group='net.sourceforge.nekohtml', module='nekohtml', version='1.9.15')
import org.codenarc.AnalysisContext
import org.codenarc.CodeNarc
import org.codenarc.CodeNarcRunner
import org.codenarc.analyzer.SourceAnalyzer
import org.codenarc.analyzer.StringSourceAnalyzer
import org.codenarc.report.TextReportWriter
import org.codenarc.ruleset.PropertiesFileRuleSetConfigurer
import org.codenarc.results.Results
import org.codenarc.ruleregistry.RuleRegistryInitializer

// URL of the groovy site that you want to analyse
def url = 'http://groovy.codehaus.org/Convert+SQL+Result+To+XML'

// Load it in via XmlSlurper
def page = new XmlSlurper(new org.cyberneko.html.parsers.SAXParser()).parse( url )

// And fetch out all the <PRE> blocks with class='code-java'
def data = page.depthFirst().grep{ it.name() == 'PRE' && it.@class.toString() == 'code-java' }*.text().flatten()

// Join all the code blocks into a single String
String pageCode = new StringWriter().with { w -> data.each { w << "$it\n" } ; w }.toString()

// Then, run the analysis
new Narc().execute( pageCode )

///////////////////////////////////////////////////////////
// Utility classes that I required to get CodeNarc running
// for code in Strings instead of Files. There is probably
// an easier way than this that I missed, but this works of
// a fashion

class NarcWriter extends TextReportWriter {
  void writeReport(Writer writer, AnalysisContext analysisContext, Results results) {
    initializeResourceBundle()
    def printWriter = new PrintWriter( writer )
    results.violations.each {
      println """$it.rule
\tLine   : $it.lineNumber
\tSrc    : $it.sourceLine
\tMessage: $it.message"""
    }
    println "Got ${results.violations.size()} violations"
  }
}

class NarcRunner extends CodeNarcRunner {
  Results execute() {
    new RuleRegistryInitializer().initializeRuleRegistry()
    def ruleSet = createRuleSet()
    // Hack to remove rules that don't make sense in this context
    def disabledRules = [
      'Println',                 // println is fine in script examples
      'JavaIoPackageAccess',     // use of File is fine
    ]
    ruleSet.rules = ruleSet.rules.findAll { !( it.name in disabledRules ) }
    def results = sourceAnalyzer.analyze(ruleSet)
    def analysisContext = new AnalysisContext(ruleSet:ruleSet, sourceDirectories:sourceAnalyzer.sourceDirectories)
    reportWriters.each { reportWriter ->
      reportWriter.writeReport(analysisContext, results)
    }
    results
  }
}

class Narc extends CodeNarc {
  protected void execute( src ) {
    // Load the latest rules from sourceforge
    setDefaultsIfNecessary()
    new NarcRunner().with { nr ->
      nr.ruleSetFiles = 'http://codenarc.svn.sourceforge.net/viewvc/codenarc/trunk/src/site/resources/StarterRuleSet-AllRules.groovy.txt?revision=813'
      nr.reportWriters = [ new NarcWriter() ]
      nr.sourceAnalyzer = new StringSourceAnalyzer( src )
      nr.execute()
    }
  }
}