[Archive-crawler-cvs] ArchiveOpenCrawler/src/org/archive/crawler/datamodel UURI.java,1.20,1.21

[<go...@us...>]

Update of /cvsroot/archive-crawler/ArchiveOpenCrawler/src/org/archive/crawler/datamodel
In directory sc8-pr-cvs1:/tmp/cvs-serv28332/src/org/archive/crawler/datamodel

Modified Files:
	UURI.java 
Log Message:
strip leading (/..)+

Index: UURI.java
===================================================================
RCS file: /cvsroot/archive-crawler/ArchiveOpenCrawler/src/org/archive/crawler/datamodel/UURI.java,v
retrieving revision 1.20
retrieving revision 1.21
diff -C2 -d -r1.20 -r1.21
*** UURI.java	6 Sep 2003 02:00:12 -0000	1.20
--- UURI.java	12 Sep 2003 02:03:01 -0000	1.21
***************
*** 89,96 ****
  		if (u.getSchemeSpecificPart().startsWith("/")) {
  			// hierarchical URI
! 			if ("".equals(u.getPath())) {
! 				u = u.resolve("/"); // ensure root URLs end with '/'
  			}
- 			u = u.normalize(); // factor out path cruft
  			String canonizedAuthority = u.getAuthority();
  			if(canonizedAuthority==null) {
--- 89,99 ----
  		if (u.getSchemeSpecificPart().startsWith("/")) {
  			// hierarchical URI
! 			u = u.normalize(); // factor out path cruft, according to official spec
! 			// now, go further and eliminate extra '..' segments
! 			String fixedPath = u.getPath().replaceFirst("^(/\\.\\.)+","");
! 			if ("".equals(fixedPath)) {
! //				ensure root URLs end with '/'
! 				fixedPath = "/"; 
  			}
  			String canonizedAuthority = u.getAuthority();
  			if(canonizedAuthority==null) {
***************
*** 137,141 ****
  			u = new URI(u.getScheme().toLowerCase(), // case-flatten scheme
  			            canonizedAuthority, // case and port flatten
! 			            u.getPath(), // leave alone
  			            u.getQuery(), // leave alone
  		                null); // drop fragment
--- 140,144 ----
  			u = new URI(u.getScheme().toLowerCase(), // case-flatten scheme
  			            canonizedAuthority, // case and port flatten
! 			            fixedPath, // leave alone
  			            u.getQuery(), // leave alone
  		                null); // drop fragment