- Timestamp:
- 2003-11-24T14:26:35+13:00 (20 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/TextRecogniser.java
r5800 r5944 5 5 6 6 import org.greenstone.gsdl3.gs3build.metadata.*; 7 import org.greenstone.gsdl3.gs3build.util.HTTPTools; 7 8 8 9 public class TextRecogniser implements RecogniserInterface 9 10 { 10 DocumentList listRepository; 11 DocumentList listRepository; 12 13 public TextRecogniser(DocumentList listRepository) 14 { this.listRepository = listRepository; 15 } 11 16 12 public TextRecogniser(DocumentList listRepository) 13 { this.listRepository = listRepository; 14 } 17 public boolean parseDocument(METSFile file) 18 { 19 String MIMEType = file.getMIMEType(); 20 if (MIMEType == null || 21 MIMEType.equals("text/plain")) { 22 URL location = file.getLocation(); 23 return this.parseDocument(location); 24 } 25 return false; 26 } 15 27 16 public boolean parseDocument(METSFile file) 17 { 18 String MIMEType = file.getMIMEType(); 19 if (MIMEType == null || 20 MIMEType.equals("text/plain")) { 21 URL location = file.getLocation(); 22 return this.parseDocument(location); 23 } 24 return false; 25 } 28 public boolean parseDocument(URL url) 29 { String fileName = null; 26 30 27 public boolean parseDocument(URL url) 28 { if (url.toString().startsWith("file://")) { 29 String fileName = url.toString().substring(7); 31 if (url.toString().startsWith("file://")) { 32 fileName = url.toString().substring(7); 33 } 34 else if (url.toString().startsWith("file:/")) { 35 fileName = url.toString().substring(5); 36 } 30 37 31 if (fileName.endsWith(".txt") || 32 fileName.endsWith(".text")) 33 { this.listRepository.addDocument(new TextDocument(url)); 34 // TODO: spawn knowledge of children too... 35 System.out.println(">>> Posting text document " + fileName); 36 return true; 37 } 38 } 39 else 40 { // Check MIME type 41 } 38 if (fileName != null) { 39 if (fileName.endsWith(".txt") || 40 fileName.endsWith(".text")) 41 { this.listRepository.addDocument(new TextDocument(url)); 42 // TODO: spawn knowledge of children too... 43 System.out.println(">>> Posting text document " + fileName); 44 return true; 45 } 46 } 47 else 48 { // Check MIME type 49 String mimeType = HTTPTools.getMIMEType(url); 50 51 if (mimeType == "text/plain") 52 { System.out.println("Posting Text document " + url.toString()); 53 54 TextDocument doc = new TextDocument(url); 55 this.listRepository.addDocument(doc); 56 return true; 57 } 58 } 42 59 43 44 60 return false; 61 } 45 62 }
Note:
See TracChangeset
for help on using the changeset viewer.