- 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/HTMLRecogniser.java
r5800 r5944 9 9 public class HTMLRecogniser implements RecogniserInterface 10 10 { 11 11 DocumentList listRepository; 12 12 13 public HTMLRecogniser(DocumentList listRepository) 14 { this.listRepository = listRepository; 15 } 13 public HTMLRecogniser(DocumentList listRepository) 14 { this.listRepository = listRepository; 15 } 16 17 public boolean parseDocument(METSFile file) 18 { 19 String MIMEType = file.getMIMEType(); 20 if (MIMEType == null || 21 MIMEType.equals("text/html")) { 22 URL location = file.getLocation(); 23 return this.parseDocument(location); 24 } 25 return false; 26 } 16 27 17 public boolean parseDocument(METSFile file) 18 { 19 String MIMEType = file.getMIMEType(); 20 if (MIMEType == null || 21 MIMEType.equals("text/html")) { 22 URL location = file.getLocation(); 23 return this.parseDocument(location); 24 } 25 return false; 26 } 28 public boolean parseDocument(URL url) 29 { String fileName = null; 27 30 28 public boolean parseDocument(URL url) 29 { if (url.toString().startsWith("file://")) { 30 String fileName = url.toString().substring(7); 31 if (fileName.endsWith(".htm") || 32 fileName.endsWith(".html")) 33 { System.out.println("Posting HTML Document " + fileName); 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 } 34 37 35 HTMLDocument doc = new HTMLDocument(url); 36 this.listRepository.addDocument(doc); 37 return true; 38 } 39 } 40 else { 41 // Get Mime type remotely, and then proceed if required 42 String mimeType = HTTPTools.getMIMEType(url); 38 if (fileName != null) { 39 if (fileName.endsWith(".htm") || 40 fileName.endsWith(".html")) 41 { System.out.println("Posting HTML Document " + fileName); 43 42 44 if (mimeType == "text/html") 45 { System.out.println("Posting HTML Document " + url.toString()); 43 HTMLDocument doc = new HTMLDocument(url); 44 this.listRepository.addDocument(doc); 45 return true; 46 } 47 } 48 else { 49 // Get Mime type remotely, and then proceed if required 50 String mimeType = HTTPTools.getMIMEType(url); 51 52 if (mimeType == "text/html") 53 { System.out.println("Posting HTML Document " + url.toString()); 46 54 47 48 49 50 51 52 53 55 HTMLDocument doc = new HTMLDocument(url); 56 this.listRepository.addDocument(doc); 57 return true; 58 } 59 } 60 return false; 61 } 54 62 }
Note:
See TracChangeset
for help on using the changeset viewer.