From c6c61be3f0814e3e777ef485575e39b29ce5e081 Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Fri, 24 Feb 2012 00:38:57 +0100
Subject: [PATCH] fix for http://bugs.yacy.net/view.php?id=148

---
 htroot/CrawlResults.java                            | 2 +-
 source/de/anomic/crawler/CrawlStacker.java          | 6 ++++--
 source/net/yacy/cora/document/MultiProtocolURI.java | 5 +++--
 3 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/htroot/CrawlResults.java b/htroot/CrawlResults.java
index 519f8cd0a..e2edb10c0 100644
--- a/htroot/CrawlResults.java
+++ b/htroot/CrawlResults.java
@@ -253,7 +253,7 @@ public class CrawlResults {
 
                     if (showIP && urle != null) {
                         prop.put("table_indexed_" + cnt + "_showIP", "1");
-                        prop.put("table_indexed_" + cnt + "_showIP_ip", urle.url().getInetAddress().getHostAddress());
+                        prop.put("table_indexed_" + cnt + "_showIP_ip", urle.url().getHost() == null ? "" : urle.url().getInetAddress().getHostAddress());
                     } else
                         prop.put("table_indexed_" + cnt + "_showIP", "0");
 
diff --git a/source/de/anomic/crawler/CrawlStacker.java b/source/de/anomic/crawler/CrawlStacker.java
index 9a0c6c557..57aa22143 100644
--- a/source/de/anomic/crawler/CrawlStacker.java
+++ b/source/de/anomic/crawler/CrawlStacker.java
@@ -186,6 +186,7 @@ public final class CrawlStacker {
     }
     public void enqueueEntriesAsynchronous(final byte[] initiator, final String profileHandle, final Map<MultiProtocolURI, Properties> hyperlinks, final boolean replace) {
         new Thread() {
+            @Override
             public void run() {
                 enqueueEntries(initiator, profileHandle, hyperlinks, true);
             }
@@ -240,6 +241,7 @@ public final class CrawlStacker {
     public void enqueueEntriesFTP(final byte[] initiator, final String profileHandle, final String host, final int port, final boolean replace) {
         final CrawlQueues cq = this.nextQueue;
         new Thread() {
+            @Override
             public void run() {
                 BlockingQueue<FTPClient.entryInfo> queue;
                 try {
@@ -487,13 +489,13 @@ public final class CrawlStacker {
         // this is expensive and those filters are check at the end of all other tests
 
         // filter with must-match for IPs
-        if ((depth > 0) && profile.ipMustMatchPattern() != CrawlProfile.MATCH_ALL_PATTERN && !profile.ipMustMatchPattern().matcher(url.getInetAddress().getHostAddress()).matches()) {
+        if ((depth > 0) && profile.ipMustMatchPattern() != CrawlProfile.MATCH_ALL_PATTERN && url.getHost() != null && !profile.ipMustMatchPattern().matcher(url.getInetAddress().getHostAddress()).matches()) {
             if (this.log.isFine()) this.log.logFine("IP " + url.getInetAddress().getHostAddress() + " of URL '" + urlstring + "' does not match must-match crawling filter '" + profile.ipMustMatchPattern().toString() + "'.");
             return "ip " + url.getInetAddress().getHostAddress() + " of url does not match must-match filter";
         }
 
         // filter with must-not-match for IPs
-        if ((depth > 0) && profile.ipMustMatchPattern() != CrawlProfile.MATCH_NEVER_PATTERN && profile.ipMustNotMatchPattern().matcher(url.getInetAddress().getHostAddress()).matches()) {
+        if ((depth > 0) && profile.ipMustMatchPattern() != CrawlProfile.MATCH_NEVER_PATTERN && url.getHost() != null && profile.ipMustNotMatchPattern().matcher(url.getInetAddress().getHostAddress()).matches()) {
             if (this.log.isFine()) this.log.logFine("IP " + url.getInetAddress().getHostAddress() + " of URL '" + urlstring + "' matches must-not-match crawling filter '" + profile.ipMustMatchPattern().toString() + "'.");
             return "ip " + url.getInetAddress().getHostAddress() + " of url matches must-not-match filter";
         }
diff --git a/source/net/yacy/cora/document/MultiProtocolURI.java b/source/net/yacy/cora/document/MultiProtocolURI.java
index f73a1d254..d8ce9b393 100644
--- a/source/net/yacy/cora/document/MultiProtocolURI.java
+++ b/source/net/yacy/cora/document/MultiProtocolURI.java
@@ -702,6 +702,7 @@ public class MultiProtocolURI implements Serializable, Comparable<MultiProtocolU
 
     public InetAddress getInetAddress() {
         if (this.hostAddress != null) return this.hostAddress;
+        if (this.host == null) return null; // this may happen for file:// urls
         this.hostAddress = Domains.dnsResolve(this.host.toLowerCase());
         return this.hostAddress;
     }
@@ -1050,7 +1051,7 @@ public class MultiProtocolURI implements Serializable, Comparable<MultiProtocolU
 			        } else if (host_tld.equals("cr")) {//Costa Rica /2,060,000
 			        	language = "es";//spanish; spa
 			        } else if (host_tld.equals("cy")) {//Cyprus /2,500,000
-			        	language = "el";//greek; gre (ell); ell	
+			        	language = "el";//greek; gre (ell); ell
 			        } else if (host_tld.equals("cu")) {//Cuba /2,040,000
 			        	language = "es";//spanish; spa
 			        } else if (host_tld.equals("cx")) {//Christmas Island /1,830,000
@@ -1323,7 +1324,7 @@ public class MultiProtocolURI implements Serializable, Comparable<MultiProtocolU
 			        } else if (host_tld.equals("mg")) {//Madagascar /255,000
 			        	language = "mg";//malagasy; mlg (mlg); mlg (macrolanguage): plt
 			        	//language = "fr";//french; fre (fra); fra
-			        	//malagasy is native language, but elite want to french 
+			        	//malagasy is native language, but elite want to french
 			        } else if (host_tld.equals("mr")) {//Mauritania /210,000
 			        	language = "ar";//arabic; ara; mey
 			        	//language = "fr";//french; fre (fra); fra