Examples with DigestingParser org.apache.tika.parser.DigestingParser used on opensource projects

Example 1 with DigestingParser

use of org.apache.tika.parser.DigestingParser in project tika by apache.

the class TikaResource method createParser.

@SuppressWarnings("serial")
public static Parser createParser() {
    final Parser parser = new AutoDetectParser(tikaConfig);
    Map<MediaType, Parser> parsers = ((AutoDetectParser) parser).getParsers();
    parsers.put(MediaType.APPLICATION_XML, new HtmlParser());
    ((AutoDetectParser) parser).setParsers(parsers);
    ((AutoDetectParser) parser).setFallback(new Parser() {

        public Set<MediaType> getSupportedTypes(ParseContext parseContext) {
            return parser.getSupportedTypes(parseContext);
        }

        public void parse(InputStream inputStream, ContentHandler contentHandler, Metadata metadata, ParseContext parseContext) {
            throw new WebApplicationException(Response.Status.UNSUPPORTED_MEDIA_TYPE);
        }
    });
    if (digester != null) {
        return new DigestingParser(parser, digester);
    }
    return parser;
}

Example 2 with DigestingParser

use of org.apache.tika.parser.DigestingParser in project tika by apache.

the class UnpackerResource method process.

private Map<String, byte[]> process(InputStream is, @Context HttpHeaders httpHeaders, @Context UriInfo info, boolean saveAll) throws Exception {
    Metadata metadata = new Metadata();
    ParseContext pc = new ParseContext();
    Parser parser = TikaResource.createParser();
    if (parser instanceof DigestingParser) {
        //no need to digest for unwrapping
        parser = ((DigestingParser) parser).getWrappedParser();
    }
    TikaResource.fillMetadata(parser, metadata, pc, httpHeaders.getRequestHeaders());
    TikaResource.logRequest(LOG, info, metadata);
    ContentHandler ch;
    ByteArrayOutputStream text = new ByteArrayOutputStream();
    if (saveAll) {
        ch = new BodyContentHandler(new RichTextContentHandler(new OutputStreamWriter(text, UTF_8)));
    } else {
        ch = new DefaultHandler();
    }
    Map<String, byte[]> files = new HashMap<>();
    MutableInt count = new MutableInt();
    pc.set(EmbeddedDocumentExtractor.class, new MyEmbeddedDocumentExtractor(count, files));
    TikaResource.parse(parser, LOG, info.getPath(), is, ch, metadata, pc);
    if (count.intValue() == 0 && !saveAll) {
        throw new WebApplicationException(Response.Status.NO_CONTENT);
    }
    if (saveAll) {
        files.put(TEXT_FILENAME, text.toByteArray());
        ByteArrayOutputStream metaStream = new ByteArrayOutputStream();
        metadataToCsv(metadata, metaStream);
        files.put(META_FILENAME, metaStream.toByteArray());
    }
    return files;
}

Example 3 with DigestingParser

use of org.apache.tika.parser.DigestingParser in project tika by apache.

the class TikaCLI method configure.

private void configure(String configFilePath) throws Exception {
    this.configFilePath = configFilePath;
    config = new TikaConfig(new File(configFilePath));
    parser = new AutoDetectParser(config);
    if (digester != null) {
        parser = new DigestingParser(parser, digester);
    }
    detector = config.getDetector();
    context.set(Parser.class, parser);
}

Example 4 with DigestingParser

use of org.apache.tika.parser.DigestingParser in project tika by apache.

the class DigestingAutoDetectParserFactory method getParser.

@Override
public Parser getParser(TikaConfig config) {
    Parser p = new AutoDetectParser(config);
    if (digester == null) {
        return p;
    }
    DigestingParser d = new DigestingParser(p, digester);
    return d;
}

Example 5 with DigestingParser

use of org.apache.tika.parser.DigestingParser in project tika by apache.

the class TikaCLI method process.

public void process(String arg) throws Exception {
    if (arg.equals("-?") || arg.equals("--help")) {
        pipeMode = false;
        usage();
    } else if (arg.equals("-V") || arg.equals("--version")) {
        pipeMode = false;
        version();
    } else if (arg.equals("-v") || arg.equals("--verbose")) {
        org.apache.log4j.Logger.getRootLogger().setLevel(Level.DEBUG);
    } else if (arg.equals("-g") || arg.equals("--gui")) {
        pipeMode = false;
        if (configFilePath != null) {
            TikaGUI.main(new String[] { configFilePath });
        } else {
            TikaGUI.main(new String[0]);
        }
    } else if (arg.equals("--list-parser") || arg.equals("--list-parsers")) {
        pipeMode = false;
        displayParsers(false, false);
    } else if (arg.equals("--list-detector") || arg.equals("--list-detectors")) {
        pipeMode = false;
        displayDetectors();
    } else if (arg.equals("--list-parser-detail") || arg.equals("--list-parser-details")) {
        pipeMode = false;
        displayParsers(true, false);
    } else if (arg.equals("--list-parser-detail-apt") || arg.equals("--list-parser-details-apt")) {
        pipeMode = false;
        displayParsers(true, true);
    } else if (arg.equals("--list-met-models")) {
        pipeMode = false;
        displayMetModels();
    } else if (arg.equals("--list-supported-types")) {
        pipeMode = false;
        displaySupportedTypes();
    } else if (arg.startsWith("--compare-file-magic=")) {
        pipeMode = false;
        compareFileMagic(arg.substring(arg.indexOf('=') + 1));
    } else if (arg.equals("--dump-minimal-config")) {
        pipeMode = false;
        dumpConfig(TikaConfigSerializer.Mode.MINIMAL);
    } else if (arg.equals("--dump-current-config")) {
        pipeMode = false;
        dumpConfig(TikaConfigSerializer.Mode.CURRENT);
    } else if (arg.equals("--dump-static-config")) {
        pipeMode = false;
        dumpConfig(TikaConfigSerializer.Mode.STATIC);
    } else if (arg.equals("--dump-static-full-config")) {
        pipeMode = false;
        dumpConfig(TikaConfigSerializer.Mode.STATIC_FULL);
    } else if (arg.equals("--container-aware") || arg.equals("--container-aware-detector")) {
    // ignore, as container-aware detectors are now always used
    } else if (arg.equals("-f") || arg.equals("--fork")) {
        fork = true;
    } else if (arg.startsWith("--config=")) {
        configure(arg.substring("--config=".length()));
    } else if (arg.startsWith("--digest=")) {
        CommonsDigester.DigestAlgorithm[] algos = CommonsDigester.parse(arg.substring("--digest=".length()));
        digester = new CommonsDigester(MAX_MARK, algos);
        parser = new DigestingParser(parser, digester);
    } else if (arg.startsWith("-e")) {
        encoding = arg.substring("-e".length());
    } else if (arg.startsWith("--encoding=")) {
        encoding = arg.substring("--encoding=".length());
    } else if (arg.startsWith("-p") && !arg.equals("-p")) {
        password = arg.substring("-p".length());
    } else if (arg.startsWith("--password=")) {
        password = arg.substring("--password=".length());
    } else if (arg.equals("-j") || arg.equals("--json")) {
        type = JSON;
    } else if (arg.equals("-J") || arg.equals("--jsonRecursive")) {
        recursiveJSON = true;
    } else if (arg.equals("-y") || arg.equals("--xmp")) {
        type = XMP;
    } else if (arg.equals("-x") || arg.equals("--xml")) {
        type = XML;
    } else if (arg.equals("-h") || arg.equals("--html")) {
        type = HTML;
    } else if (arg.equals("-t") || arg.equals("--text")) {
        type = TEXT;
    } else if (arg.equals("-T") || arg.equals("--text-main")) {
        type = TEXT_MAIN;
    } else if (arg.equals("-m") || arg.equals("--metadata")) {
        type = METADATA;
    } else if (arg.equals("-l") || arg.equals("--language")) {
        type = LANGUAGE;
    } else if (arg.equals("-d") || arg.equals("--detect")) {
        type = DETECT;
    } else if (arg.startsWith("--extract-dir=")) {
        extractDir = new File(arg.substring("--extract-dir=".length()));
    } else if (arg.equals("-z") || arg.equals("--extract")) {
        type = NO_OUTPUT;
        context.set(EmbeddedDocumentExtractor.class, new FileEmbeddedDocumentExtractor());
    } else if (arg.equals("-r") || arg.equals("--pretty-print")) {
        prettyPrint = true;
    } else if (arg.equals("-p") || arg.equals("--port") || arg.equals("-s") || arg.equals("--server")) {
        serverMode = true;
        pipeMode = false;
    } else if (arg.startsWith("-c")) {
        URI uri = new URI(arg.substring("-c".length()));
        parser = new NetworkParser(uri);
    } else if (arg.startsWith("--client=")) {
        URI uri = new URI(arg.substring("--client=".length()));
        parser = new NetworkParser(uri);
    } else {
        pipeMode = false;
        if (serverMode) {
            new TikaServer(Integer.parseInt(arg)).start();
        } else if (arg.equals("-")) {
            try (InputStream stream = TikaInputStream.get(new CloseShieldInputStream(System.in))) {
                type.process(stream, System.out, new Metadata());
            }
        } else {
            URL url;
            File file = new File(arg);
            if (file.isFile()) {
                url = file.toURI().toURL();
            } else {
                url = new URL(arg);
            }
            if (recursiveJSON) {
                handleRecursiveJson(url, System.out);
            } else {
                Metadata metadata = new Metadata();
                try (InputStream input = TikaInputStream.get(url, metadata)) {
                    type.process(input, System.out, metadata);
                } finally {
                    System.out.flush();
                }
            }
        }
    }
}