1 31 32 package org.opencms.search.extractors; 33 34 import org.opencms.main.OpenCms; 35 import org.opencms.util.CmsHtmlExtractor; 36 import org.opencms.util.CmsStringUtil; 37 38 import java.io.InputStream ; 39 40 49 public final class CmsExtractorHtml extends A_CmsTextExtractor { 50 51 52 private static final CmsExtractorHtml INSTANCE = new CmsExtractorHtml(); 53 54 57 private CmsExtractorHtml() { 58 59 } 61 62 67 public static I_CmsTextExtractor getExtractor() { 68 69 return INSTANCE; 70 } 71 72 75 public I_CmsExtractionResult extractText(InputStream in, String encoding) throws Exception { 76 77 if (CmsStringUtil.isEmpty(encoding)) { 78 encoding = OpenCms.getSystemInfo().getDefaultEncoding(); 79 } 80 81 String result = CmsHtmlExtractor.extractText(in, encoding); 82 result = removeControlChars(result); 83 return new CmsExtractionResult(result); 84 } 85 } | Popular Tags |