Class MagicDetector

tika-core/src/main/java/org/apache/tika/detect/MagicDetector.java:45–438

Content type detection based on magic bytes, i.e. type-specific patterns near the beginning of the document input stream. Because this works on bytes, not characters, by default any string matching is done as ISO_8859_1. To use an explicit different encoding, supply a type other than "string" /

 * @since Apache Tika 0.3
 */
public class MagicDetector implements Detector {

    /**
     * The matching media type. Returned by the
     * {@link #detect(InputStream, Metadata)} method if a match is found.
     */
    private final MediaType type;
    /**
     * Length of the comparison window.
     */
    private final int length;
    /**
     * The magic match pattern. If this byte pattern is equal to the
     * possibly bit-masked bytes from the input stream, then the type
     * detection succeeds and the configured {@link #type} is returned.
     */
    private final byte[] pattern;
    /**
     * Length of the pattern, which in the case of regular expressions will
     * not be the same as the comparison window length.
     */
    private final int patternLength;
    /**
     * True if pattern is a regular expression, false otherwise.
     */
    private final boolean isRegex;
    /**
     * True if we're doing a case-insensitive string match, false otherwise.
     */
    private final boolean isStringIgnoreCase;
    /**
     * Bit mask that is applied to the source bytes before pattern matching.
     */
    private final byte[] mask;
    /**
     * First offset (inclusive) of the comparison window within the
     * document input stream. Greater than or equal to zero.
     */
    private final int offsetRangeBegin;
    /**
     * Last offset (inclusive) of the comparison window within the document
     * input stream. Greater than or equal to the
     * {@link #offsetRangeBegin first offset}.
     * <p>
     * Note that this is <em>not</em> the offset of the last byte read from
     * the document stream. Instead, the last window of bytes to be compared
     * starts at this offset.
     */
    private final int offsetRangeEnd;

    /**
     * Creates a detector for input documents that have the exact given byte
     * pattern at the beginning of the document stream.
     *
     * @param type    matching media type
     * @param pattern magic match pattern
     */
    public MagicDetector(MediaType type, byte[] pattern) {