public AllrisScraper(
String baseUrl,
String allrisUrl,
boolean scrape,
String scrapeStartDate,
String cacheFolder,
int timeout,
String outputFormat,
String outputFolder,
String outputUrlBase) {
this.baseUrl = formatUrl(baseUrl);
this.allrisUrl = formatUrl(allrisUrl);
this.scrape = scrape;
this.scrapeStartDate = scrapeStartDate;
this.cacheFolder = cacheFolder;
this.timeout = timeout;
this.outputFormat = outputFormat;
this.outputFolder = outputFolder;
this.outputUrlBase = formatUrl(outputUrlBase);
List<Drucksache> drucksachen = new ArrayList<Drucksache>();
System.out.println("Hole Drucksachen.");
DrucksachenScraper druckScraper = new DrucksachenScraper(this.baseUrl, this.cacheFolder);
if (scrape) {
druckScraper.init(
this.baseUrl + "/" + this.allrisUrl + "/vo040.asp?selfaction=ws&template=xyz");
List<String> failedUris = druckScraper.scrapeDrucksachen(true);
if (failedUris.size() > 0) {
System.out.println("Failed to parse " + failedUris.size() + " URLS:");
for (String failed : failedUris) {
System.out.println(failed);
}
}
drucksachen = druckScraper.getDrucksachen();
} else {
System.out.println("Lese Drucksachen aus " + this.cacheFolder);
drucksachen = druckScraper.readDrucksachenFromFolder();
}
List<Fraktion> fraktionen = new ArrayList<Fraktion>();
System.out.println("Scrape Fraktionen.");
FraktionScraper fruckScraper = new FraktionScraper(this.baseUrl);
fraktionen =
fruckScraper.parseFraktionen(
this.baseUrl + "/" + this.allrisUrl + "/fr010.asp", this.timeout);
System.out.println("Writer erstellt Datenmodell.");
RDFWriter writer = new RDFWriter(drucksachen, fraktionen, outputUrlBase);
Model rdfModel = writer.getRdfModel();
System.out.println("Schreibe Datenmodell in " + this.outputFolder);
writer.print(this.outputFormat, this.outputFolder, rdfModel);
}
public static void main(String[] args) throws IOException {
String resultFile = "owl-results.rdf";
if (args.length >= 1) {
resultFile = args[0];
}
WebOntTestHarness harness = new WebOntTestHarness();
harness.runTests();
// harness.runTest("http://www.w3.org/2002/03owlt/AnnotationProperty/Manifest004#test");
// harness.runTest("http://www.w3.org/2002/03owlt/AnnotationProperty/Manifest003#test");
// harness.runTest("http://www.w3.org/2002/03owlt/Thing/Manifest001#test");
// harness.runTest("http://www.w3.org/2002/03owlt/Thing/Manifest002#test");
// harness.runTest("http://www.w3.org/2002/03owlt/Thing/Manifest003#test");
// harness.runTest("http://www.w3.org/2002/03owlt/Thing/Manifest004#test");
// harness.runTest("http://www.w3.org/2002/03owlt/Thing/Manifest005#test");
RDFWriter writer = harness.testResults.getWriter("RDF/XML-ABBREV");
OutputStream stream = new FileOutputStream(resultFile);
writer.setProperty("showXmlDeclaration", "true");
harness.testResults.setNsPrefix("", "http://www.w3.org/1999/xhtml");
writer.write(harness.testResults, stream, BASE_RESULTS_URI);
}
/**
* LDP-Style to serialize a resource.
*
* @param writer the writer to serialize to
* @param subject the resource to serialize
* @param iteration the Iteration containing the data
* @throws RDFHandlerException
* @throws RepositoryException
*/
public static void exportIteration(
RDFWriter writer, URI subject, CloseableIteration<Statement, RepositoryException> iteration)
throws RDFHandlerException, RepositoryException {
writer.startRDF();
writer.handleNamespace(LDP.PREFIX, LDP.NAMESPACE);
writer.handleNamespace(RDF.PREFIX, RDF.NAMESPACE);
writer.handleNamespace(XSD.PREFIX, XSD.NAMESPACE);
writer.handleNamespace(DCTERMS.PREFIX, DCTERMS.NAMESPACE);
writer.handleNamespace("parent", subject.getNamespace());
writer.handleNamespace("child", subject.stringValue().replaceFirst("/*$", "/"));
writer.handleNamespace("this", subject.stringValue().replaceFirst("/*$", "#"));
while (iteration.hasNext()) {
writer.handleStatement(iteration.next());
}
writer.endRDF();
}
