From c3bf17a3a19f55a958b5fa17f9a8378093be7c92 Mon Sep 17 00:00:00 2001
From: orbiter <orbiter@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Tue, 5 Oct 2010 00:05:08 +0000
Subject: [PATCH] fixed must-match filter for smb crawling

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7222 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 htroot/Crawler_p.java | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/htroot/Crawler_p.java b/htroot/Crawler_p.java
index 1f5f48877..03b205727 100644
--- a/htroot/Crawler_p.java
+++ b/htroot/Crawler_p.java
@@ -144,7 +144,13 @@ public class Crawler_p {
                 if (newcrawlingMustMatch.length() < 2) newcrawlingMustMatch = CrawlProfile.MATCH_ALL; // avoid that all urls are filtered out if bad value was submitted
                 // special cases:
                 if (crawlingStartURL!= null && fullDomain) {
-                    newcrawlingMustMatch = crawlingStartURL.isFile() ? "file://" + crawlingStartURL.getPath() + ".*" : crawlingStartURL.isSMB() ? "smb://" + crawlingStartURL.getPath() + ".*" : ".*" + crawlingStartURL.getHost() + ".*";
+                    if (crawlingStartURL.isFile()) {
+                        newcrawlingMustMatch = "file://" + crawlingStartURL.getPath() + ".*";
+                    } else if (crawlingStartURL.isSMB()) {
+                        newcrawlingMustMatch = "smb://.*" + crawlingStartURL.getHost() + ".*" + crawlingStartURL.getPath() + ".*";
+                    } else {
+                        newcrawlingMustMatch = ".*" + crawlingStartURL.getHost() + ".*";
+                    }
                 }
                 if (crawlingStart!= null && subPath && (pos = crawlingStart.lastIndexOf('/')) > 0) {
                     newcrawlingMustMatch = crawlingStart.substring(0, pos + 1) + ".*";