net.sf.saxon.s9api.SaxonApiException Java Examples

@Test
public void testDomSource() throws SaxonApiException, SAXException, IOException {
    final DocumentBuilder builder = TestObjectFactory.createProcessor().newDocumentBuilder();

    final BuildingContentHandler handler = builder.newBuildingContentHandler();
    handler.startDocument();
    handler.startElement("http://some.ns", "mynode", "mynode", new AttributesImpl());
    final Document dom = NodeOverNodeInfo.wrap(handler.getDocumentNode().getUnderlyingNode()).getOwnerDocument();
    final Input domInput = InputFactory.read(new DOMSource(dom), "MD5", "id".getBytes());
    assertThat(domInput).isNotNull();
    final Source source = domInput.getSource();
    assertThat(source).isNotNull();
    final Result<XdmNode, XMLSyntaxError> parsed = Helper.parseDocument(domInput);
    assertThat(parsed.isValid()).isTrue();

    // read twice
    assertThat(Helper.parseDocument(domInput).getObject()).isNotNull();
}
 

@Override
protected boolean doProcess2(Record inputRecord, InputStream stream) throws SaxonApiException, XMLStreamException {
  incrementNumRecords();      
  for (Fragment fragment : fragments) {
    Record outputRecord = inputRecord.copy();
    removeAttachments(outputRecord);   
    XdmNode document = parseXmlDocument(stream);
    LOG.trace("XSLT input document: {}", document);
    XsltTransformer evaluator = fragment.transformer;
    evaluator.setInitialContextNode(document);
    XMLStreamWriter morphlineWriter = new MorphlineXMLStreamWriter(getChild(), outputRecord);
    evaluator.setDestination(new XMLStreamWriterDestination(morphlineWriter));
    evaluator.transform(); //  run the query and push into child via RecordXMLStreamWriter
  }      
  return true;
}
 

private Result<Boolean, XMLSyntaxError> validate(final Bag results, final Scenario scenario) {
    log.debug("Validating document using scenario {}", scenario.getConfiguration().getName());
    final CollectingErrorEventHandler errorHandler = new CollectingErrorEventHandler();
    try ( final SourceProvider validateInput = resolveSource(results) ) {

        final Validator validator = this.factory.createValidator(scenario.getSchema());
        validator.setErrorHandler(errorHandler);
        validator.validate(validateInput.getSource());
        return new Result<>(!errorHandler.hasErrors(), errorHandler.getErrors());
    } catch (final SAXException | SaxonApiException | IOException e) {
        final String msg = String.format("Error processing schema validation for scenario %s", scenario.getConfiguration().getName());
        log.error(msg, e);
        results.addProcessingError(msg);
        return new Result<>(Boolean.FALSE);
    }
}
 

private String evaluateQuery(String query, Document dataDoc)
           throws CodeletExecutionException {

    try {
        String result = SaxonUtil.evaluateQuery(query, dataDoc);
        if (result != null)
            return result;
        else
            throw new CodeletExecutionException("No data produced for query: '" +
                                                 query + "'.");
    }
    catch (SaxonApiException sapie) {
        throw new CodeletExecutionException(
                "Invalid query: '" + query + "'.\n" +
                "Message from parser: [" + sapie.getMessage() + "]");
    }
    catch (Exception e) {
        throw new CodeletExecutionException(
                "Exception in query evaluation: '" + query + "'.");
    }
}
 

/**
 * Parsed und überprüft ein übergebenes Dokument darauf ob es well-formed ist. Dies stellt den ersten
 * Verarbeitungsschritt des Prüf-Tools dar. Diese Funktion verzichtet explizit auf die Validierung gegenüber einem
 * Schema.
 * 
 * @param content ein Dokument
 * @return Ergebnis des Parsings inklusive etwaiger Fehler
 */
public Result<XdmNode, XMLSyntaxError> parseDocument(final Input content) {
    if (content == null) {
        throw new IllegalArgumentException("Input may not be null");
    }
    Result<XdmNode, XMLSyntaxError> result;
    try {
        final DocumentBuilder builder = this.processor.newDocumentBuilder();
        builder.setLineNumbering(true);
        final XdmNode doc = builder.build(content.getSource());
        result = new Result<>(doc, Collections.emptyList());
    } catch (final SaxonApiException | IOException e) {
        log.debug("Exception while parsing {}", content.getName(), e);
        final XMLSyntaxError error = new XMLSyntaxError();
        error.setSeverityCode(XMLSyntaxErrorSeverity.SEVERITY_FATAL_ERROR);
        error.setMessage(String.format("IOException while reading resource %s: %s", content.getName(), e.getMessage()));
        result = new Result<>(Collections.singleton(error));
    }

    return result;
}
 

protected Element evaluateTreeQuery(String query, Document document)
        throws YQueryException {

    try {
        logger.debug("Evaluating XQuery: " + query);
        return SaxonUtil.evaluateTreeQuery(query, document);
    } catch (SaxonApiException e) {
        YQueryException qe = new YQueryException(
                "Something Wrong with Process Specification:\n" +
                        "The engine failed to parse an invalid query.\n" +
                        "Please check task:\n\t" +
                        "id[ " + getID() + " ]\n\t" +
                        "query: \n\t" + query + ".\n" +
                        "Message from parser: [" + e.getMessage() + "]");
        qe.setStackTrace(e.getStackTrace());
        throw qe;
    }
}
 

protected void checkXQuery(String xQuery, String param, YVerificationHandler handler) {
    if (!StringUtil.isNullOrEmpty(xQuery)) {
        if (ExternalDataGatewayFactory.isExternalDataMappingExpression(xQuery)) {
            checkExternalMapping(xQuery, handler);
        } else {
            try {
                SaxonUtil.compileXQuery(xQuery);
            } catch (SaxonApiException e) {
                handler.error(this, this + " [id= " + this.getID() +
                        "] the XQuery could not be successfully" +
                        " parsed [" + e.getMessage() + "]");
            }
        }
    } else handler.error(this, this + " [id= " + this.getID() +
            "] the XQuery for param [" + param +
            "] cannot be equal to null or the empty string.");
}
 

public String execute(Node node, String expression)
		throws SaxonApiException {

	Processor proc = new Processor(false);
	XPathCompiler xpath = proc.newXPathCompiler();
	DocumentBuilder builder = proc.newDocumentBuilder();

	String fileName = getClass().getResource(
			map.getLogicalSource().getIdentifier()).getFile();

	XdmNode doc = builder.build(new File(fileName));
	String expre = replace(node, expression);
	expression = expression.replaceAll(
			"\\{" + expression.split("\\{")[1].split("\\}")[0] + "\\}", "'"
					+ expre + "'");

	XPathSelector selector = xpath.compile(expression).load();
	selector.setContextItem(doc);

	// Evaluate the expression.
	Object result = selector.evaluate();

	return result.toString();

}
 

/**
 * Lädt ein XSL von der angegebenen URI
 *
 * @param uri die URI der XSL Definition
 * @return ein XSLT Executable
 */
public XsltExecutable loadXsltScript(final URI uri) {
    log.info("Loading XSLT script from  {}", uri);
    final XsltCompiler xsltCompiler = getProcessor().newXsltCompiler();
    final CollectingErrorEventHandler listener = new CollectingErrorEventHandler();
    try {
        xsltCompiler.setErrorListener(listener);
        if (getResolver() != null) {
            // otherwise use default resolver
            xsltCompiler.setURIResolver(getResolver());
        }

        return xsltCompiler.compile(resolveInRepository(uri));
    } catch (final SaxonApiException e) {
        listener.getErrors().forEach(event -> event.log(log));
        throw new IllegalStateException("Can not compile xslt executable for uri " + uri, e);
    } finally {
        if (!listener.hasErrors() && listener.hasEvents()) {
            log.warn("Received warnings or errors while loading a xslt script {}", uri);
            listener.getErrors().forEach(e -> e.log(log));
        }
    }
}
 

private boolean evaluateSplitQuery(String query, YIdentifier tokenToSend)
        throws YQueryException {

    // check for timer predicates first
    if (isTimerPredicate(query)) {
        return evaluateTimerPredicate(query, tokenToSend);
    }

    // next check for plugin evaluator expressions and have the evaluators replace
    // them with simple values
    query = PredicateEvaluatorCache.process(getDecompositionPrototype(), query,
            tokenToSend);

    // now we have a standard query - check if it evaluates to true
    String xquery = "boolean(" + query + ")";
    try {
        logger.debug("Evaluating XQuery: " + xquery);
        String result = SaxonUtil.evaluateQuery(xquery, _net.getInternalDataDocument());

        if (result != null) {
            if (result.equalsIgnoreCase("true")) {
                logger.debug("XQuery evaluated TRUE.");
                return true;
            } else if (result.equalsIgnoreCase("false")) {
                logger.debug("XQuery evaluated FALSE.");
                return false;
            }
        }

        // either result is null or result is not a boolean string
        logger.error("Evaluated XQuery did not return a singular boolean result.");
        throw new YQueryException("Evaluated XQuery did not return a singular " +
                "boolean result. Evaluated: '" + xquery + "'");
    } catch (SaxonApiException e) {
        logger.error("Invalid XQuery expression (" + xquery + ").", e);
        throw new YQueryException("Invalid XQuery expression (" + xquery + ").");
    }
}
 

@Test
public void testProcessingError() throws IOException, SaxonApiException {
    final CheckAction.Bag bag = createBag(InputFactory.read(Simple.SIMPLE_VALID.toURL()), true);

    final Scenario scenario = bag.getScenarioSelectionResult().getObject();
    final XsltExecutable exec = mock(XsltExecutable.class);
    final XsltTransformer transformer = mock(XsltTransformer.class);
    doThrow(new SaxonApiException("invalid")).when(transformer).transform();
    when(exec.load()).thenReturn(transformer);
    final ResourceType resourceType = new ResourceType();
    resourceType.setName("invalid internal");
    scenario.setSchematronValidations(Collections.singletonList(new Transformation(exec, resourceType)));
    this.action.check(bag);
    assertThat(bag.getReportInput().getProcessingError().getError()).isNotEmpty();
}
 

@Test
public void testEvilStylesheets() throws IOException {
    final Processor p = TestObjectFactory.createProcessor();
    for (int i = 1; i <= 5; i++) {
        try {
            final URL resource = SaxonSecurityTest.class.getResource(String.format("/evil/evil%s.xsl", i));
            final XsltCompiler compiler = p.newXsltCompiler();
            final RelativeUriResolver resolver = new RelativeUriResolver(Simple.REPOSITORY_URI);
            compiler.setURIResolver(resolver);
            final XsltExecutable executable = compiler.compile(new StreamSource(resource.openStream()));
            final XsltTransformer transformer = executable.load();
            final Source document = InputFactory.read("<root/>".getBytes(), "dummy").getSource();
            // transformer.getUnderlyingController().setUnparsedTextURIResolver(resolver);
            transformer.setURIResolver(resolver);
            transformer.setSource(document);
            final XdmDestination result = new XdmDestination();
            transformer.setDestination(result);
            transformer.transform();

            // wenn der Punkt erreicht wird, sollte wenigstens, das Element evil nicht mit 'bösen' Inhalten gefüllt sein!
            if (StringUtils.isNotBlank(result.getXdmNode().getStringValue())) {
                fail(String.format("Saxon configuration should prevent expansion within %s", resource));
            }

        } catch (final SaxonApiException | RuntimeException e) {
            log.info("Expected exception detected {}", e.getMessage(), e);
        }
    }
}
 

/**
 * Lädt ein XML-Dokument von der gegebenen URL
 * 
 * @param url die url die geladen werden soll
 * @return ein result objekt mit Dokument
 */
public static XdmNode load(final URL url) {
    try ( final InputStream input = url.openStream() ) {
        return TestObjectFactory.createProcessor().newDocumentBuilder().build(new StreamSource(input));
    } catch (final SaxonApiException | IOException e) {
        throw new IllegalStateException("Fehler beim Laden der XML-Datei", e);

    }

}
 

public static String serialize(final XdmNode node) {
    try ( final StringWriter writer = new StringWriter() ) {
        final Processor processor = Helper.getTestProcessor();
        final Serializer serializer = processor.newSerializer(writer);
        serializer.serializeNode(node);
        return writer.toString();
    } catch (final SaxonApiException | IOException e) {
        throw new IllegalStateException("Can not serialize document", e);
    }
}
 

@Override
public void initialize(InputSplit inSplit, TaskAttemptContext context)
        throws IOException, InterruptedException {
    Path file = ((FileSplit)inSplit).getPath();
    FileSystem fs = file.getFileSystem(context.getConfiguration());
    FSDataInputStream fileIn = fs.open(file);
    DocumentBuilder docBuilder = builderLocal.get();
    try {
        Document document = docBuilder.parse(fileIn);
        net.sf.saxon.s9api.DocumentBuilder db = saxonBuilderLocal.get();
        XdmNode xdmDoc = db.wrap(document);
        XPathCompiler xpath = proc.newXPathCompiler();
        xpath.declareNamespace("wp", 
                "http://www.mediawiki.org/xml/export-0.4/");
        XPathSelector selector = xpath.compile(PATH_EXPRESSION).load();
        selector.setContextItem(xdmDoc);
        items = new ArrayList<XdmItem>();
        for (XdmItem item : selector) {
            items.add(item);
        }
    } catch (SAXException ex) {
        ex.printStackTrace();
        throw new IOException(ex);
    } catch (SaxonApiException e) {
        e.printStackTrace();
    } finally {
        if (fileIn != null) {
            fileIn.close();
        }
    }
}
 

/**
 *  PARSING METHODS
 */


private String evaluateXQuery(String expr, Element data) throws RdrConditionException {
    try {
        if (expr.startsWith("{")) expr = deQuote(expr);      // remove braces
        String query = String.format("boolean(%s)", expr);
        return SaxonUtil.evaluateQuery(query, new Document(data.clone()));
    }
    catch (SaxonApiException sae) {
        throw new RdrConditionException("Invalid XPath expression (" + expr + ").");
    }
}
 

@Test
public void testExecutionException() throws SaxonApiException {
    final Processor p = mock(Processor.class);
    final DocumentBuilder documentBuilder = mock(DocumentBuilder.class);
    this.action = new CreateReportAction(p, new ConversionService(), null);

    when(p.newDocumentBuilder()).thenReturn(documentBuilder);
    when(documentBuilder.build(any(Source.class))).thenThrow(new SaxonApiException("mocked"));
    final Bag bag = TestBagBuilder.createBag(InputFactory.read(Simple.SIMPLE_VALID), true);
    this.action.check(bag);
    assertThat(bag.isStopped()).isTrue();

}
 

private List evaluateListQuery(String query, Element element)
        throws YQueryException {

    try {
        logger.debug("Evaluating XQuery: " + query);
        return SaxonUtil.evaluateListQuery(query, element);
    } catch (SaxonApiException e) {
        YQueryException de = new YQueryException(e.getMessage());
        de.setStackTrace(e.getStackTrace());
        throw de;
    }
}
 

public boolean isHideIf(String data) {
    boolean hide = false;
    String query = getValue("hideIf");
    if (query != null) {
        try {
            Document dataDoc = JDOMUtil.stringToDocument(data);
            String queryResult = SaxonUtil.evaluateQuery(query, dataDoc);
            hide = queryResult.equalsIgnoreCase("true");
        }
        catch (SaxonApiException sae) {
            // nothing to do, will default to false
        }
    }
    return hide;
}
 

protected XdmNode parseXmlDocument(InputStream stream) throws XMLStreamException, SaxonApiException {
  XMLStreamReader reader = inputFactory.createXMLStreamReader(null, stream);
  BuildingStreamWriterImpl writer = documentBuilder.newBuildingStreamWriter();      
  new XMLStreamCopier(reader, writer).copy(false); // push XML into Saxon and build TinyTree
  reader.close();
  writer.close();
  XdmNode document = writer.getDocumentNode();
  return document;
}
 

@Override
protected boolean doProcess2(Record inputRecord, InputStream stream) throws SaxonApiException, XMLStreamException {
  incrementNumRecords();      
  for (Fragment fragment : fragments) {
    Record template = inputRecord.copy();
    removeAttachments(template);
    XdmNode document = parseXmlDocument(stream);
    LOG.trace("XQuery input document: {}", document);
    XQueryEvaluator evaluator = fragment.xQueryEvaluator;
    evaluator.setContextItem(document);
    
    int i = 0;
    for (XdmItem item : evaluator) {
      i++;
      if (LOG.isTraceEnabled()) {
        LOG.trace("XQuery result sequence item #{} is of class: {} with value: {}", new Object[] { i,
            item.getUnderlyingValue().getClass().getName(), item });
      }
      if (item.isAtomicValue()) {
        LOG.debug("Ignoring atomic value in result sequence: {}", item);
        continue;
      }
      XdmNode node = (XdmNode) item;
      Record outputRecord = template.copy();
      boolean isNonEmpty = addRecordValues(node, Axis.SELF, XdmNodeKind.ATTRIBUTE, outputRecord);
      isNonEmpty = addRecordValues(node, Axis.ATTRIBUTE, XdmNodeKind.ATTRIBUTE, outputRecord) || isNonEmpty;
      isNonEmpty = addRecordValues(node, Axis.CHILD, XdmNodeKind.ELEMENT, outputRecord) || isNonEmpty;
      if (isNonEmpty) { // pass record to next command in chain   
        if (!getChild().process(outputRecord)) { 
          return false;
        }
      }
    }
  }      
  return true;
}
 

@Override
public XdmValue call(XdmValue[] arguments) throws SaxonApiException {
  StringBuilder buf = new StringBuilder();
  for (XdmValue argument : arguments) {
    XdmSequenceIterator iter = argument.iterator();
    while (iter.hasNext()) {
      buf.append(iter.next().getStringValue());
    }
  }
  return new XdmAtomicValue(buf.toString());
}
 

private XdmNode createErrorInformation(final Collection<XMLSyntaxError> errors) throws SaxonApiException, SAXException {
    final BuildingContentHandler contentHandler = this.processor.newDocumentBuilder().newBuildingContentHandler();
    contentHandler.startDocument();
    contentHandler.startElement(EngineInformation.getFrameworkNamespace(), ERROR_MESSAGE_ELEMENT, ERROR_MESSAGE_ELEMENT,
            new AttributesImpl());
    final String message = errors.stream().map(XMLSyntaxError::getMessage).collect(Collectors.joining());
    contentHandler.characters(message.toCharArray(), 0, message.length());
    contentHandler.endElement(EngineInformation.getFrameworkNamespace(), ERROR_MESSAGE_ELEMENT, ERROR_MESSAGE_ELEMENT);
    return contentHandler.getDocumentNode();
}
 

@Override
public void serialize(final XdmNode node) throws SaxonApiException, IOException {
    try ( final ByteArrayOutputStream out = new ByteArrayOutputStream() ) {
        final Serializer serializer = this.processor.newSerializer();
        serializer.setOutputStream(out);
        serializer.serializeNode(node);
        serializer.close();
        this.bytes = out.toByteArray();
    }
}
 

@Override
public void serialize(final XdmNode node) throws SaxonApiException, IOException {
    try ( final OutputStream out = Files.newOutputStream(this.file) ) {
        final Serializer serializer = this.processor.newSerializer();
        serializer.setOutputStream(out);
        serializer.serializeNode(node);
        serializer.close();
    }
}
 

private SourceProvider resolveSource(final Bag results) throws IOException, SaxonApiException {
    final SourceProvider source;
    if (results.getInput() instanceof AbstractInput && (((AbstractInput) results.getInput()).supportsMultipleReads())) {
        source = () -> results.getInput().getSource();
    } else {
        source = serialize(results.getInput(), results.getParserResult().getObject());
    }
    return source;

}
 

private SerializedDocument serialize(final Input input, final XdmNode object) throws IOException, SaxonApiException {
    final SerializedDocument doc;
    if (input instanceof AbstractInput && ((AbstractInput) input).getLength() < getInMemoryLimit()) {
        doc = new ByteArraySerializedDocument(this.processor);
    } else {
        doc = new FileSerializedDocument(this.processor);
    }
    doc.serialize(object);
    return doc;
}
 

private static void evaluateAcceptanceMatch(final Bag results, final XPathSelector selector) {
    try {
        selector.setContextItem(results.getReport());
        results.setAcceptStatus(selector.effectiveBooleanValue() ? AcceptRecommendation.ACCEPTABLE : AcceptRecommendation.REJECT);
    } catch (final SaxonApiException e) {
        final String msg = String.format("Error evaluating accept recommendation: %s", selector.getUnderlyingXPathContext().toString());
        log.error(msg, e);
        results.stopProcessing(msg);
    }
}
 

private byte[] serialize(final Result result) {
    try ( final ByteArrayOutputStream out = new ByteArrayOutputStream() ) {
        final Serializer serializer = this.processor.newSerializer(out);
        serializer.serializeNode(result.getReport());
        return out.toByteArray();
    } catch (final SaxonApiException | IOException e) {
        log.error("Error serializing result", e);
        throw new IllegalStateException("Can not serialize result", e);
    }
}
 

private static boolean match(final XdmNode document, final Scenario scenario) {
    try {
        final XPathSelector selector = scenario.getMatchSelector();
        selector.setContextItem(document);
        return selector.effectiveBooleanValue();
    } catch (final SaxonApiException e) {
        log.error("Error evaluating xpath expression", e);
    }
    return false;
}