Examples of org.apache.xerces.xni.parser.XMLParserConfiguration

org.apache.xerces.xni.parser.XMLParserConfiguration
Represents a parser configuration. The parser configuration maintains a table of recognized features and properties, assembles components for the parsing pipeline, and is responsible for initiating parsing of an XML document.
By separating the configuration of a parser from the specific parser instance, applications can create new configurations and re-use the existing parser components and external API generators (e.g. the DOMParser and SAXParser).
The internals of any specific parser configuration instance are hidden. Therefore, each configuration may implement the parsing mechanism any way necessary. However, the parser configuration should follow these guidelines:
- Call the reset method on each component before parsing. This is only required if the configuration is re-using existing components that conform to the XMLComponent interface. If the configuration uses all custom parts, then it is free to implement everything as it sees fit as long as it follows the other guidelines.
- Call the setFeature and setProperty method on each component during parsing to propagate features and properties that have changed. This is only required if the configuration is re-using existing components that conform to the XMLComponent interface. If the configuration uses all custom parts, then it is free to implement everything as it sees fit as long as it follows the other guidelines.
- Pass the same unique String references for all symbols that are propagated to the registered handlers. Symbols include, but may not be limited to, the names of elements and attributes (including their uri, prefix, and localpart). This is suggested but not an absolute must. However, the standard parser components may require access to the same symbol table for creation of unique symbol references to be propagated in the XNI pipeline.
@author Arnaud Le Hors, IBM @author Andy Clark, IBM @version $Id: XMLParserConfiguration.java,v 1.3 2001/10/01 10:13:21 andyc Exp $

    {
      StreamSource localStreamSource = (StreamSource)paramSource;
      XMLInputSource localXMLInputSource = new XMLInputSource(localStreamSource.getPublicId(), localStreamSource.getSystemId(), null);
      localXMLInputSource.setByteStream(localStreamSource.getInputStream());
      localXMLInputSource.setCharacterStream(localStreamSource.getReader());
      XMLParserConfiguration localXMLParserConfiguration = (XMLParserConfiguration)this.fConfiguration.get();
      if (localXMLParserConfiguration == null)
      {
        localXMLParserConfiguration = initialize();
      }
      else if (this.fComponentManager.getFeature("http://apache.org/xml/features/internal/parser-settings"))
      {
        localXMLParserConfiguration.setProperty("http://apache.org/xml/properties/internal/entity-resolver", this.fComponentManager.getProperty("http://apache.org/xml/properties/internal/entity-resolver"));
        localXMLParserConfiguration.setProperty("http://apache.org/xml/properties/internal/error-handler", this.fComponentManager.getProperty("http://apache.org/xml/properties/internal/error-handler"));
      }
      this.fComponentManager.reset();
      this.fSchemaValidator.setDocumentHandler(null);
      try
      {
        localXMLParserConfiguration.parse(localXMLInputSource);
      }
      catch (XMLParseException localXMLParseException)
      {
        throw Util.toSAXParseException(localXMLParseException);
      }

View Full Code Here

   public static void main(String[] argv) throws Exception {
      if (argv.length == 0) {
         printUsage();
         System.exit(1);
      }
      XMLParserConfiguration parser = new HTMLConfiguration();
      parser.setFeature(NOTIFY_CHAR_REFS, true);
      parser.setFeature(NOTIFY_HTML_BUILTIN_REFS, true);
      String iencoding = null;
      String oencoding = "Windows-1252";
      boolean identity = false;
      boolean purify = false;
      for (int i = 0; i < argv.length; i++) {
         String arg = argv[i];
         if (arg.equals("-ie")) {
            iencoding = argv[++i];
            continue;
         }
         if (arg.equals("-e") || arg.equals("-oe")) {
            oencoding = argv[++i];
            continue;
         }
         if (arg.equals("-i")) {
            identity = true;
            continue;
         }
         if (arg.equals("-p")) {
            purify = true;
            continue;
         }
         if (arg.equals("-h")) {
            printUsage();
            System.exit(1);
         }
         java.util.Vector filtersVector = new java.util.Vector(2);
         if (identity) {
            filtersVector.addElement(new Identity());
         } else if (purify) {
            filtersVector.addElement(new Purifier());
         }
         filtersVector.addElement(new Writer(System.out, oencoding));
         XMLDocumentFilter[] filters = new XMLDocumentFilter[filtersVector.size()];
         filtersVector.copyInto(filters);
         parser.setProperty(FILTERS, filters);
         XMLInputSource source = new XMLInputSource(null, arg, null);
         source.setEncoding(iencoding);
         parser.parse(source);
      }
   } // main(String[])

View Full Code Here

        if (fFileSets.size() == 0) {
            throw new BuildException("must specify at least one fileset");
        }


        // create parser
        XMLParserConfiguration parser = new HTMLConfiguration();


        // parse input files and produce output files
        log("Parsing test files and generating output...");
        File outdir = new File(outputdir);
        int size = fFileSets.size();
        for (int i = 0; i < size; i++) {
            FileSet fileset = (FileSet)fFileSets.elementAt(i);
            DirectoryScanner dirscanner = fileset.getDirectoryScanner(project);
            File indir = dirscanner.getBasedir();
            String[] files = dirscanner.getIncludedFiles();
            for (int j = 0; j < files.length; j++) {
                File infile = new File(indir, files[j]);
                File outfile = new File(outdir, files[j]);
                log("  "+outfile, Project.MSG_VERBOSE);
                OutputStream out = null;
                try {
                    out = new FileOutputStream(outfile);
                    XMLDocumentFilter[] filters = { new Writer(out) };
                    parser.setProperty("http://cyberneko.org/html/properties/filters", filters);
                    parser.parse(new XMLInputSource(null, infile.toString(), null));
                }
                catch (Exception e) {
                    log("  error parsing input file, "+infile);
                    throw new BuildException(e);
                }

View Full Code Here

    public static void main(String[] argv) throws Exception {
        if (argv.length == 0) {
            printUsage();
            System.exit(1);
        }
        XMLParserConfiguration parser = new HTMLConfiguration();
        parser.setFeature(NOTIFY_CHAR_REFS, true);
        parser.setFeature(NOTIFY_HTML_BUILTIN_REFS, true);
        String encoding = "Windows-1252";
        boolean identity = false;
        boolean purify = false;
        for (int i = 0; i < argv.length; i++) {
            String arg = argv[i];
            if (arg.equals("-e")) {
                encoding = argv[++i];
                continue;
            }
            if (arg.equals("-i")) {
                identity = true;
                continue;
            }
            if (arg.equals("-p")) {
                purify = true;
                continue;
            }
            if (arg.equals("-h")) {
                printUsage();
                System.exit(1);
            }
            java.util.Vector filtersVector = new java.util.Vector(2);
            if (identity) {
                filtersVector.addElement(new Identity());
            }
            else if (purify) {
                filtersVector.addElement(new Purifier());
            }
            filtersVector.addElement(new Writer(System.out, encoding));
            XMLDocumentFilter[] filters = 
                new XMLDocumentFilter[filtersVector.size()];
            filtersVector.copyInto(filters);
            parser.setProperty(FILTERS, filters);
            parser.parse(new XMLInputSource(null, arg, null));
        }
    } // main(String[])

View Full Code Here

            remover,
            writer,
        };


        // create HTML parser
        XMLParserConfiguration parser = new HTMLConfiguration();
        parser.setProperty("http://cyberneko.org/html/properties/filters", filters);


        // parse documents
        for (int i = 0; i < argv.length; i++) {
            String systemId = argv[i];
            XMLInputSource source = new XMLInputSource(null, systemId, null);
            parser.parse(source);
        }


    } // main(String[])

View Full Code Here

        }


        // variables
        PrintWriter out = new PrintWriter(System.out);
        XMLDocumentParser parser = null;
        XMLParserConfiguration parserConfig = null;
        int repetition = DEFAULT_REPETITION;
        boolean namespaces = DEFAULT_NAMESPACES;
        boolean namespacePrefixes = DEFAULT_NAMESPACE_PREFIXES;
        boolean validation = DEFAULT_VALIDATION;
        boolean schemaValidation = DEFAULT_SCHEMA_VALIDATION;
        boolean memoryUsage = DEFAULT_MEMORY_USAGE;
        boolean tagginess = DEFAULT_TAGGINESS;
        
        // process arguments
        for (int i = 0; i < argv.length; i++) {
            String arg = argv[i];
            if (arg.startsWith("-")) {
                String option = arg.substring(1);
                if (option.equals("p")) {
                    // get parser name
                    if (++i == argv.length) {
                        System.err.println("error: Missing argument to -p option.");
                        continue;
                    }
                    String parserName = argv[i];


                    // create parser
                    try {
                        parserConfig = (XMLParserConfiguration)Class.forName(parserName).newInstance();
                        parserConfig.addRecognizedFeatures(new String[] {
                            NAMESPACE_PREFIXES_FEATURE_ID,
                        });
                        parser = null;
                    }
                    catch (Exception e) {
                        parserConfig = null;
                        System.err.println("error: Unable to instantiate parser configuration ("+parserName+")");
                    }
                    continue;
                }
                if (option.equals("x")) {
                    if (++i == argv.length) {
                        System.err.println("error: Missing argument to -x option.");
                        continue;
                    }
                    String number = argv[i];
                    try {
                        int value = Integer.parseInt(number);
                        if (value < 1) {
                            System.err.println("error: Repetition must be at least 1.");
                            continue;
                        }
                        repetition = value;
                    }
                    catch (NumberFormatException e) {
                        System.err.println("error: invalid number ("+number+").");
                    }
                    continue;
                }
                if (option.equalsIgnoreCase("n")) {
                    namespaces = option.equals("n");
                    continue;
                }
                if (option.equalsIgnoreCase("np")) {
                    namespacePrefixes = option.equals("np");
                    continue;
                }
                if (option.equalsIgnoreCase("v")) {
                    validation = option.equals("v");
                    continue;
                }
                if (option.equalsIgnoreCase("s")) {
                    schemaValidation = option.equals("s");
                    continue;
                }
                if (option.equalsIgnoreCase("m")) {
                    memoryUsage = option.equals("m");
                    continue;
                }
                if (option.equalsIgnoreCase("t")) {
                    tagginess = option.equals("t");
                    continue;
                }
                if (option.equals("-rem")) {
                    if (++i == argv.length) {
                        System.err.println("error: Missing argument to -# option.");
                        continue;
                    }
                    System.out.print("# ");
                    System.out.println(argv[i]);
                    continue;
                }
                if (option.equals("h")) {
                    printUsage();
                    continue;
                }
                System.err.println("error: unknown option ("+option+").");
                continue;
            }


            // use default parser?
            if (parserConfig == null) {


                // create parser
                try {
                    parserConfig = (XMLParserConfiguration)Class.forName(DEFAULT_PARSER_CONFIG).newInstance();
                    parserConfig.addRecognizedFeatures(new String[] {
                        NAMESPACE_PREFIXES_FEATURE_ID,
                    });
                }
                catch (Exception e) {
                    System.err.println("error: Unable to instantiate parser configuration ("+DEFAULT_PARSER_CONFIG+")");
                    continue;
                }
            }
        
            // set parser features
            if (parser == null) {
                parser = new Counter(parserConfig);
            }
            try {
                parserConfig.setFeature(NAMESPACES_FEATURE_ID, namespaces);
            }
            catch (XMLConfigurationException e) {
                System.err.println("warning: Parser does not support feature ("+NAMESPACES_FEATURE_ID+")");
            }
            try {
                parserConfig.setFeature(VALIDATION_FEATURE_ID, validation);
            }
            catch (XMLConfigurationException e) {
                System.err.println("warning: Parser does not support feature ("+VALIDATION_FEATURE_ID+")");
            }
            try {
                parserConfig.setFeature(SCHEMA_VALIDATION_FEATURE_ID, schemaValidation);
            }
            catch (XMLConfigurationException e) {
                if (e.getType() == XMLConfigurationException.NOT_SUPPORTED) {
                    System.err.println("warning: Parser does not support feature ("+SCHEMA_VALIDATION_FEATURE_ID+")");
                }

View Full Code Here

            System.exit(1);
        }


        // variables
        Writer writer = null;
        XMLParserConfiguration parserConfig = null;
        boolean namespaces = DEFAULT_NAMESPACES;
        boolean validation = DEFAULT_VALIDATION;
        boolean schemaValidation = DEFAULT_SCHEMA_VALIDATION;
        boolean canonical = DEFAULT_CANONICAL;
        
        // process arguments
        for (int i = 0; i < argv.length; i++) {
            String arg = argv[i];
            if (arg.startsWith("-")) {
                String option = arg.substring(1);
                if (option.equals("p")) {
                    // get parser name
                    if (++i == argv.length) {
                        System.err.println("error: Missing argument to -p option.");
                    }
                    String parserName = argv[i];


                    // create parser
                    try {
                        parserConfig = (XMLParserConfiguration)Class.forName(parserName).newInstance();
                        /***
                        parserConfig.addRecognizedFeatures(new String[] {
                            NAMESPACE_PREFIXES_FEATURE_ID,
                        });
                        /***/
                        writer = null;
                    }
                    catch (Exception e) {
                        parserConfig = null;
                        System.err.println("error: Unable to instantiate parser configuration ("+parserName+")");
                    }
                    continue;
                }
                if (option.equalsIgnoreCase("n")) {
                    namespaces = option.equals("n");
                    continue;
                }
                if (option.equalsIgnoreCase("v")) {
                    validation = option.equals("v");
                    continue;
                }
                if (option.equalsIgnoreCase("s")) {
                    schemaValidation = option.equals("s");
                    continue;
                }
                if (option.equalsIgnoreCase("c")) {
                    canonical = option.equals("c");
                    continue;
                }
                if (option.equals("h")) {
                    printUsage();
                    continue;
                }
            }


            // use default parser?
            if (parserConfig == null) {


                // create parser
                try {
                    parserConfig = (XMLParserConfiguration)Class.forName(DEFAULT_PARSER_CONFIG).newInstance();
                    /***
                    parserConfig.addRecognizedFeatures(new String[] {
                        NAMESPACE_PREFIXES_FEATURE_ID,
                    });
                    /***/
                }
                catch (Exception e) {
                    System.err.println("error: Unable to instantiate parser configuration ("+DEFAULT_PARSER_CONFIG+")");
                    continue;
                }
            }
        
            // set parser features
            if (writer == null) {
                writer = new Writer(parserConfig);
            }
            try {
                parserConfig.setFeature(NAMESPACES_FEATURE_ID, namespaces);
            }
            catch (XMLConfigurationException e) {
                System.err.println("warning: Parser does not support feature ("+NAMESPACES_FEATURE_ID+")");
            }
            try {
                parserConfig.setFeature(VALIDATION_FEATURE_ID, validation);
            }
            catch (XMLConfigurationException e) {
                System.err.println("warning: Parser does not support feature ("+VALIDATION_FEATURE_ID+")");
            }
            try {
                parserConfig.setFeature(SCHEMA_VALIDATION_FEATURE_ID, schemaValidation);
            }
            catch (XMLConfigurationException e) {
                if (e.getType() == XMLConfigurationException.NOT_SUPPORTED) {
                    System.err.println("warning: Parser does not support feature ("+SCHEMA_VALIDATION_FEATURE_ID+")");
                }

View Full Code Here

            System.exit(1);
        }


        // variables
        XMLDocumentParser parser = null;
        XMLParserConfiguration parserConfig = null;
        boolean namespaces = DEFAULT_NAMESPACES;
        boolean validation = DEFAULT_VALIDATION;
        boolean schemaValidation = DEFAULT_SCHEMA_VALIDATION;
        boolean notifyCharRefs = DEFAULT_NOTIFY_CHAR_REFS;


        // process arguments
        for (int i = 0; i < argv.length; i++) {
            String arg = argv[i];
            if (arg.startsWith("-")) {
                String option = arg.substring(1);
                if (option.equals("p")) {
                    // get parser name
                    if (++i == argv.length) {
                        System.err.println("error: Missing argument to -p option.");
                        continue;
                    }
                    String parserName = argv[i];


                    // create parser
                    try {
                        parserConfig = (XMLParserConfiguration)Class.forName(parserName).newInstance();
                        parser = null;
                    }
                    catch (Exception e) {
                        parserConfig = null;
                        System.err.println("error: Unable to instantiate parser configuration ("+parserName+")");
                    }
                    continue;
                }
                if (option.equalsIgnoreCase("n")) {
                    namespaces = option.equals("n");
                    continue;
                }
                if (option.equalsIgnoreCase("v")) {
                    validation = option.equals("v");
                    continue;
                }
                if (option.equalsIgnoreCase("s")) {
                    schemaValidation = option.equals("s");
                    continue;
                }
                if (option.equalsIgnoreCase("c")) {
                    notifyCharRefs = option.equals("c");
                    continue;
                }
                if (option.equals("h")) {
                    printUsage();
                    continue;
                }
            }


            // use default parser?
            if (parserConfig == null) {


                // create parser
                try {
                    parserConfig = (XMLParserConfiguration)Class.forName(DEFAULT_PARSER_CONFIG).newInstance();
                }
                catch (Exception e) {
                    System.err.println("error: Unable to instantiate parser configuration ("+DEFAULT_PARSER_CONFIG+")");
                    continue;
                }
            }
        
            // set parser features
            if (parser == null) {
                parser = new DocumentTracer(parserConfig);
            }
            try {
                parserConfig.setFeature(NAMESPACES_FEATURE_ID, namespaces);
            }
            catch (XMLConfigurationException e) {
                System.err.println("warning: Parser does not support feature ("+NAMESPACES_FEATURE_ID+")");
            }
            try {
                parserConfig.setFeature(VALIDATION_FEATURE_ID, validation);
            }
            catch (XMLConfigurationException e) {
                System.err.println("warning: Parser does not support feature ("+VALIDATION_FEATURE_ID+")");
            }
            try {
                parserConfig.setFeature(SCHEMA_VALIDATION_FEATURE_ID, schemaValidation);
            }
            catch (XMLConfigurationException e) {
                if (e.getType() == XMLConfigurationException.NOT_SUPPORTED) {
                    System.err.println("warning: Parser does not support feature ("+SCHEMA_VALIDATION_FEATURE_ID+")");
                }
            }
            try {
                parserConfig.setFeature(NOTIFY_CHAR_REFS_FEATURE_ID, notifyCharRefs);
            }
            catch (XMLConfigurationException e) {
                if (e.getType() == XMLConfigurationException.NOT_RECOGNIZED) {
                    e.printStackTrace();
                }

View Full Code Here

        StringWriter sw = new StringWriter();
        Writer writer = new Writer(sw, null);
        XMLDocumentFilter[] filters = {
                remover, writer,
        };
        XMLParserConfiguration parser = new HTMLConfiguration();
        parser.setProperty("http://cyberneko.org/html/properties/filters", filters);


        XMLInputSource source = new XMLInputSource(null, null, null, new StringReader(sourceText), null);
        try {
            parser.parse(source);
            return sw.getBuffer()
                    .toString()
                    .replaceAll("&apos;", "'")
                    .replaceAll("&gt;", ">")
                    .replaceAll("&lt;", "<")

View Full Code Here

        }
    }


    @Override
    protected void initParser(Ruby runtime) {
        XMLParserConfiguration config = new HTMLConfiguration();
        XMLDocumentFilter removeNSAttrsFilter = new RemoveNSAttrsFilter();
        XMLDocumentFilter elementValidityCheckFilter = new ElementValidityCheckFilter(errorHandler);
        //XMLDocumentFilter[] filters = { removeNSAttrsFilter,  elementValidityCheckFilter};
        XMLDocumentFilter[] filters = { elementValidityCheckFilter};


        config.setErrorHandler(this.errorHandler);


        parser = new NokogiriDomParser(config);


        // see http://nekohtml.sourceforge.net/settings.html for details
        setProperty("http://cyberneko.org/html/properties/default-encoding", java_encoding);

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of org.apache.xerces.xni.parser.XMLParserConfiguration

com.enigmastation.extractors.impl.NekoHTMLStripper

com.googlecode.html.filters.Writer

nexj.core.util.XMLUtil

nokogiri.internals.HtmlDomParserContext

org.apache.lucene.gdata.search.analysis.HTMLStrategy

org.apache.maven.wagon.shared.http.HtmlFileListParser

org.apache.xerces.jaxp.DocumentBuilderImpl

org.apache.xerces.jaxp.SAXParserImpl

org.apache.xerces.jaxp.SAXParserMMImpl

org.apache.xerces.jaxp.validation.StreamValidatorHelper

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.