Examples with FSDocumentSelector org.apache.tika.batch.fs.FSDocumentSelector used on opensource projects

Example 1 with FSDocumentSelector

use of org.apache.tika.batch.fs.FSDocumentSelector in project tika by apache.

the class FSCrawlerBuilder method buildSelector.

private DocumentSelector buildSelector(Map<String, String> attributes) {
    String includeString = attributes.get(INCLUDE_FILE_PAT_ATTR);
    String excludeString = attributes.get(EXCLUDE_FILE_PAT_ATTR);
    long maxFileSize = PropsUtil.getLong(attributes.get(MAX_FILE_SIZE_BYTES_ATTR), -1L);
    long minFileSize = PropsUtil.getLong(attributes.get(MIN_FILE_SIZE_BYTES_ATTR), -1L);
    Pattern includePat = (includeString != null && includeString.length() > 0) ? Pattern.compile(includeString) : null;
    Pattern excludePat = (excludeString != null && excludeString.length() > 0) ? Pattern.compile(excludeString) : null;
    return new FSDocumentSelector(includePat, excludePat, minFileSize, maxFileSize);
}