[Archive-crawler-cvs] ArchiveOpenCrawler/src/org/archive/crawler/datamodel CrawlURI.java,1.37,1.38

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

Update of /cvsroot/archive-crawler/ArchiveOpenCrawler/src/org/archive/crawler/datamodel
In directory sc8-pr-cvs1:/tmp/cvs-serv28250/src/org/archive/crawler/datamodel

Modified Files:
	CrawlURI.java 
Log Message:
improved link-hop, embed-hop handling: embeds get a separate leash distance from last link

Index: CrawlURI.java
===================================================================
RCS file: /cvsroot/archive-crawler/ArchiveOpenCrawler/src/org/archive/crawler/datamodel/CrawlURI.java,v
retrieving revision 1.37
retrieving revision 1.38
diff -C2 -d -r1.37 -r1.38
*** CrawlURI.java	6 Sep 2003 02:00:13 -0000	1.37
--- CrawlURI.java	12 Sep 2003 02:02:24 -0000	1.38
***************
*** 62,65 ****
--- 62,68 ----
  	private int chaffness = 0; // suspiciousness of being of chaff
  	
+ 	private int linkHopCount = -1; // from seeds
+ 	private int embedHopCount = -1; // from a sure link
+ 	
  	private int threadNumber;
  	
***************
*** 461,470 ****
  	 * @param sourceCuri
  	 */
! 	public void setVia(CrawlURI sourceCuri) {
  		via = sourceCuri;
  	}
  	
  /*	public boolean isFubared(){
  		return ( fetchStatus < 0 && numberOfFetchAttempts >= 3);
  	}*/
  }
--- 464,518 ----
  	 * @param sourceCuri
  	 */
! 	public void setViaLinkFrom(CrawlURI sourceCuri) {
  		via = sourceCuri;
+ 		int candidateLinkHopCount = sourceCuri.getLinkHopCount()+1;
+ 		embedHopCount = 0;
+ 		if (linkHopCount == -1) {
+ 			linkHopCount = candidateLinkHopCount;
+ 			return;
+ 		}
+ 		if (linkHopCount > candidateLinkHopCount) {
+ 			linkHopCount = candidateLinkHopCount; 
+ 		}
+ 	}
+ 	
+ 	/**
+ 	 * @param sourceCuri
+ 	 */
+ 	public void setViaEmbedFrom(CrawlURI sourceCuri) {
+ 		via = sourceCuri;
+ 		int candidateLinkHopCount = sourceCuri.getLinkHopCount();
+ 		if (linkHopCount == -1) {
+ 			linkHopCount = candidateLinkHopCount;
+ 		} else if (linkHopCount > candidateLinkHopCount) {
+ 			linkHopCount = candidateLinkHopCount; 
+ 		}
+ 		int candidateEmbedHopCount = sourceCuri.getEmbedHopCount()+1;
+ 		if (embedHopCount == -1) {
+ 			embedHopCount = candidateEmbedHopCount;
+ 		} else if (embedHopCount > candidateEmbedHopCount) {
+ 			embedHopCount = candidateEmbedHopCount; 
+ 		}
  	}
+ 
  	
  /*	public boolean isFubared(){
  		return ( fetchStatus < 0 && numberOfFetchAttempts >= 3);
  	}*/
+ 
+ 	
+ 	/**
+ 	 * @return
+ 	 */
+ 	public int getEmbedHopCount() {
+ 		return embedHopCount;
+ 	}
+ 
+ 	/**
+ 	 * @return
+ 	 */
+ 	public int getLinkHopCount() {
+ 		return linkHopCount;
+ 	}
+ 
  }