[Archive-access-cvs] SF.net SVN: archive-access: [2129] trunk/archive-access/projects/wayback/ wayback-core/src/main/java/org/archive/wayback/resourcestore/ WARCRecordToSearchResultAdapter.java

[<bra...@us...>]

Revision: 2129
          http://archive-access.svn.sourceforge.net/archive-access/?rev=2129&view=rev
Author:   bradtofel
Date:     2008-01-14 17:46:54 -0800 (Mon, 14 Jan 2008)

Log Message:
-----------
BUGFIX: was not correctly resolving URLs in Location: HTTP headers.
BUGFIX: was not using WARC payload digest... which should be trusted.

Modified Paths:
--------------
    trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WARCRecordToSearchResultAdapter.java

Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WARCRecordToSearchResultAdapter.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WARCRecordToSearchResultAdapter.java	2008-01-15 01:43:29 UTC (rev 2128)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/resourcestore/WARCRecordToSearchResultAdapter.java	2008-01-15 01:46:54 UTC (rev 2129)
@@ -121,7 +121,7 @@
 		return result;
 	}
 	
-	private void addUrlDataToSearchResult(SearchResult result, String urlStr)
+	private UURI addUrlDataToSearchResult(SearchResult result, String urlStr)
 	throws IOException {
 
 		result.put(WaybackConstants.RESULT_URL, urlStr);
@@ -141,6 +141,8 @@
 
 		String urlKey = canonicalizer.urlStringToKey(urlStr);
 		result.put(WaybackConstants.RESULT_URL_KEY, urlKey);
+
+		return uri;
 	}
 
 	private SearchResult adaptDNS(ArchiveRecordHeader header, WARCRecord rec) 
@@ -218,7 +220,7 @@
 				String.valueOf(header.getOffset()));
 		
 		String origUrl = header.getUrl();
-		addUrlDataToSearchResult(result,origUrl);
+		UURI uri = addUrlDataToSearchResult(result,origUrl);
 
 		// need to parse the documents HTTP message and headers here: WARCReader
 		// does not implement this... yet..
@@ -243,7 +245,9 @@
                 ARCConstants.DEFAULT_ENCODING);
 
 		rec.close();
-		result.put(WaybackConstants.RESULT_MD5_DIGEST, rec.getDigestStr());
+		result.put(WaybackConstants.RESULT_MD5_DIGEST, 
+				transformDigest(header.getHeaderValue(
+						WARCRecord.HEADER_KEY_PAYLOAD_DIGEST)));
 
 		if (headers != null) {
 	
@@ -263,7 +267,7 @@
 					// should we prefer one over the other?
 					// right now, we're ignoring "Content-Location"
 					try {
-						UURI uriRedirect = UURIFactory.getInstance(origUrl,
+						UURI uriRedirect = UURIFactory.getInstance(uri,
 								locationStr);
 						result.put(WaybackConstants.RESULT_REDIRECT_URL,
 								uriRedirect.getEscapedURI());


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.