TextExtractor


1   package org.jahia.services.fileextraction;
2   
3   import java.io.IOException  ;
4   import java.io.InputStream  ;
5   import java.io.InputStreamReader  ;
6   
7   import org.jahia.utils.FileUtils;
8   
9   
10  /**
11   * <p>Title: HDF Text content extractor file</p>
12   * <p>Description: This classes uses Jakarta's POI library to extract content
13   * from HDF formatted files. Some of the code in this class is not used
14   * currently but might be needed to process other types of files based on
15   * the POI File system.</p>
16   * <p>Copyright: Copyright (c) 2003</p>
17   * <p>Company: Jahia Ltd</p>
18   * @author Serge Huber
19   * @version 1.0
20   */
21  
22  public class TextExtractor implements FileExtractor {
23  
24      public TextExtractor(){
25      }
26  
27      /**
28      *
29      * @param path String
30      * @param lastModified long
31      * @param fileStream InputStream
32      * @throws Exception
33      * @return String
34      */
35     public ExtractedDocument getExtractedDocument(
36                                               String   path,
37                                               long lastModified,
38                                               InputStream   fileStream)
39     throws Exception  {
40          return getExtractedDocument(path,lastModified,fileStream,null);
41     }
42  
43      /**
44      *
45      * @param path String
46      * @param lastModified long
47      * @param fileStream InputStream
48      * @param charSet String
49      * @throws Exception
50      * @return String
51      */
52     public ExtractedDocument getExtractedDocument(
53                                               String   path,
54                                               long lastModified,
55                                               InputStream   fileStream,
56                                               String   charSet)
57     throws Exception  {
58         ExtractedDocumentImpl extDoc = new ExtractedDocumentImpl();
59         extDoc.setContent(this.getContentAsString(path,lastModified,fileStream,charSet));
60         return extDoc;
61     }
62      
63      /**
64       * This method returns all the text of an HDF file.
65       *
66       * @param path String
67       * @param lastModified long
68       * @param fileStream InputStream
69       * @throws IOException thrown if there was an error while parsing the
70       * file format, notably if the file is an RTF file instead of a HDF file.
71       * @return String
72       */
73      public String   getContentAsString(String   path, long lastModified,
74                                       InputStream   fileStream)
75      throws IOException   {
76          return getContentAsString(path, lastModified, fileStream, null);
77      }
78  
79      /**
80       * This method returns all the text of an HDF file.
81       *
82       * @param path String
83       * @param lastModified long
84       * @param fileStream InputStream
85       * @param charSet String
86       * @throws IOException thrown if there was an error while parsing the
87       * file format, notably if the file is an RTF file instead of a HDF file.
88       * @return String
89       */
90      public String   getContentAsString(String   path, long lastModified,
91                                       InputStream   fileStream, String   charSet)
92      throws IOException   {
93          InputStreamReader   reader = null;
94          if ( charSet != null ){
95              reader = new InputStreamReader  (fileStream, charSet);
96          } else {
97              reader = new InputStreamReader  (fileStream);
98          }
99          return FileUtils.readerToString(reader);
100     }
101 }
102
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags