Examples with MimeTypeException org.apache.tika.mime.MimeTypeException used on opensource projects

Example 1 with MimeTypeException

use of org.apache.tika.mime.MimeTypeException in project lucene-solr by apache.

the class ExtractingRequestHandler method inform.

@Override
public void inform(SolrCore core) {
    if (initArgs != null) {
        //if relative,then relative to config dir, otherwise, absolute path
        String tikaConfigLoc = (String) initArgs.get(CONFIG_LOCATION);
        if (tikaConfigLoc != null) {
            File configFile = new File(tikaConfigLoc);
            if (configFile.isAbsolute() == false) {
                configFile = new File(core.getResourceLoader().getConfigDir(), configFile.getPath());
            }
            try {
                config = new TikaConfig(configFile);
            } catch (Exception e) {
                throw new SolrException(ErrorCode.SERVER_ERROR, e);
            }
        }
        String parseContextConfigLoc = (String) initArgs.get(PARSE_CONTEXT_CONFIG);
        if (parseContextConfigLoc != null) {
            try {
                parseContextConfig = new ParseContextConfig(core.getResourceLoader(), parseContextConfigLoc);
            } catch (Exception e) {
                throw new SolrException(ErrorCode.SERVER_ERROR, e);
            }
        }
        NamedList configDateFormats = (NamedList) initArgs.get(DATE_FORMATS);
        if (configDateFormats != null && configDateFormats.size() > 0) {
            dateFormats = new HashSet<>();
            Iterator<Map.Entry> it = configDateFormats.iterator();
            while (it.hasNext()) {
                String format = (String) it.next().getValue();
                log.info("Adding Date Format: " + format);
                dateFormats.add(format);
            }
        }
    }
    if (config == null) {
        try {
            config = getDefaultConfig(core.getResourceLoader().getClassLoader());
        } catch (MimeTypeException | IOException e) {
            throw new SolrException(ErrorCode.SERVER_ERROR, e);
        }
    }
    if (parseContextConfig == null) {
        parseContextConfig = new ParseContextConfig();
    }
    factory = createFactory();
}

Example 2 with MimeTypeException

use of org.apache.tika.mime.MimeTypeException in project sling by apache.

the class TikaMimeTypeProvider method getExtension.

public String getExtension(String mimeType) {
    try {
        MimeType type = types.forName(mimeType);
        String extension = type.getExtension();
        if (extension != null && extension.length() > 1) {
            // skip leading "."
            return extension.substring(1);
        }
    } catch (MimeTypeException e) {
    // ignore
    }
    // fall back
    return null;
}

Example 3 with MimeTypeException

use of org.apache.tika.mime.MimeTypeException in project nifi by apache.

the class IdentifyMimeType method onTrigger.

@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) {
    FlowFile flowFile = session.get();
    if (flowFile == null) {
        return;
    }
    final ComponentLog logger = getLogger();
    final AtomicReference<String> mimeTypeRef = new AtomicReference<>(null);
    final String filename = flowFile.getAttribute(CoreAttributes.FILENAME.key());
    session.read(flowFile, new InputStreamCallback() {

        @Override
        public void process(final InputStream stream) throws IOException {
            try (final InputStream in = new BufferedInputStream(stream)) {
                TikaInputStream tikaStream = TikaInputStream.get(in);
                Metadata metadata = new Metadata();
                if (filename != null && context.getProperty(USE_FILENAME_IN_DETECTION).asBoolean()) {
                    metadata.add(TikaMetadataKeys.RESOURCE_NAME_KEY, filename);
                }
                // Get mime type
                MediaType mediatype = detector.detect(tikaStream, metadata);
                mimeTypeRef.set(mediatype.toString());
            }
        }
    });
    String mimeType = mimeTypeRef.get();
    String extension = "";
    try {
        MimeType mimetype;
        mimetype = config.getMimeRepository().forName(mimeType);
        extension = mimetype.getExtension();
    } catch (MimeTypeException ex) {
        logger.warn("MIME type extension lookup failed: {}", new Object[] { ex });
    }
    // Workaround for bug in Tika - https://issues.apache.org/jira/browse/TIKA-1563
    if (mimeType != null && mimeType.equals("application/gzip") && extension.equals(".tgz")) {
        extension = ".gz";
    }
    if (mimeType == null) {
        flowFile = session.putAttribute(flowFile, CoreAttributes.MIME_TYPE.key(), "application/octet-stream");
        flowFile = session.putAttribute(flowFile, "mime.extension", "");
        logger.info("Unable to identify MIME Type for {}; setting to application/octet-stream", new Object[] { flowFile });
    } else {
        flowFile = session.putAttribute(flowFile, CoreAttributes.MIME_TYPE.key(), mimeType);
        flowFile = session.putAttribute(flowFile, "mime.extension", extension);
        logger.info("Identified {} as having MIME Type {}", new Object[] { flowFile, mimeType });
    }
    session.getProvenanceReporter().modifyAttributes(flowFile);
    session.transfer(flowFile, REL_SUCCESS);
}

Example 4 with MimeTypeException

use of org.apache.tika.mime.MimeTypeException in project ddf by codice.

the class DownloadManager method run.

@Override
public void run() {
    String mimeType = null;
    try (ReadableByteChannel byteChannel = Channels.newChannel(url.openStream())) {
        mimeType = url.openConnection().getContentType();
        String fileExtension = allTypes.forName(mimeType).getExtension();
        LOGGER.debug("downloading product from: {}", url.toString());
        LOGGER.debug("mimetype is: {}", mimeType);
        LOGGER.debug("File Extension is: {}", fileExtension);
        try (FileOutputStream fileOutputStream = new FileOutputStream(outputFileName + fileExtension)) {
            fileOutputStream.getChannel().transferFrom(byteChannel, 0, Long.MAX_VALUE);
        } catch (IOException e) {
            LOGGER.info("Error opening stream for {}", outputFileName, e);
        }
    } catch (IOException e) {
        LOGGER.info("Error downloading file from url: {}", url, e);
    } catch (MimeTypeException e) {
        LOGGER.info("Error determining file extension from mimetype: {}", mimeType, e);
    }
}

Example 5 with MimeTypeException

use of org.apache.tika.mime.MimeTypeException in project tika by apache.

the class TikaCLI method compareFileMagic.

/**
     * Compares our mime types registry with the File(1) tool's 
     *  directory of (uncompiled) Magic entries. 
     * (Well, those with mimetypes anyway)
     * @param magicDir Path to the magic directory
     */
private void compareFileMagic(String magicDir) throws Exception {
    Set<String> tikaLacking = new TreeSet<String>();
    Set<String> tikaNoMagic = new TreeSet<String>();
    // Sanity check
    File dir = new File(magicDir);
    if ((new File(dir, "elf")).exists() && (new File(dir, "mime")).exists() && (new File(dir, "vorbis")).exists()) {
    // Looks plausible
    } else {
        throw new IllegalArgumentException(magicDir + " doesn't seem to hold uncompressed file magic entries");
    }
    // Find all the mimetypes in the directory
    Set<String> fileMimes = new HashSet<String>();
    for (File mf : dir.listFiles()) {
        if (mf.isFile()) {
            BufferedReader r = new BufferedReader(new InputStreamReader(new FileInputStream(mf), UTF_8));
            String line;
            while ((line = r.readLine()) != null) {
                if (line.startsWith("!:mime") || line.startsWith("#!:mime")) {
                    String mime = line.substring(7).trim();
                    fileMimes.add(mime);
                }
            }
            r.close();
        }
    }
    // See how those compare to the Tika ones
    TikaConfig config = TikaConfig.getDefaultConfig();
    MimeTypes mimeTypes = config.getMimeRepository();
    MediaTypeRegistry registry = config.getMediaTypeRegistry();
    for (String mime : fileMimes) {
        try {
            final MimeType type = mimeTypes.getRegisteredMimeType(mime);
            if (type == null) {
                // Tika doesn't know about this one
                tikaLacking.add(mime);
            } else {
                // Tika knows about this one!
                // Does Tika have magic for it?
                boolean hasMagic = type.hasMagic();
                // How about the children?
                if (!hasMagic) {
                    for (MediaType child : registry.getChildTypes(type.getType())) {
                        MimeType childType = mimeTypes.getRegisteredMimeType(child.toString());
                        if (childType != null && childType.hasMagic()) {
                            hasMagic = true;
                        }
                    }
                }
                // How about the parents?
                MimeType parentType = type;
                while (parentType != null && !hasMagic) {
                    if (parentType.hasMagic()) {
                        // Has magic, fine
                        hasMagic = true;
                    } else {
                        // Check the parent next
                        MediaType parent = registry.getSupertype(type.getType());
                        if (parent == MediaType.APPLICATION_XML || parent == MediaType.TEXT_PLAIN || parent == MediaType.OCTET_STREAM) {
                            // Stop checking parents if we hit a top level type
                            parent = null;
                        }
                        if (parent != null) {
                            parentType = mimeTypes.getRegisteredMimeType(parent.toString());
                        } else {
                            parentType = null;
                        }
                    }
                }
                if (!hasMagic) {
                    tikaNoMagic.add(mime);
                }
            }
        } catch (MimeTypeException e) {
        // Broken entry in the file magic directory
        // Silently skip
        }
    }
    // Check how many tika knows about
    int tikaTypes = 0;
    int tikaAliases = 0;
    for (MediaType type : registry.getTypes()) {
        tikaTypes++;
        tikaAliases += registry.getAliases(type).size();
    }
    // Report
    System.out.println("Tika knows about " + tikaTypes + " unique mime types");
    System.out.println("Tika knows about " + (tikaTypes + tikaAliases) + " mime types including aliases");
    System.out.println("The File Magic directory knows about " + fileMimes.size() + " unique mime types");
    System.out.println();
    System.out.println("The following mime types are known to File but not Tika:");
    for (String mime : tikaLacking) {
        System.out.println("  " + mime);
    }
    System.out.println();
    System.out.println("The following mime types from File have no Tika magic (but their children might):");
    for (String mime : tikaNoMagic) {
        System.out.println("  " + mime);
    }
}