From c726154a59994137053d3c557a73e18315dcbbfc Mon Sep 17 00:00:00 2001
From: luccioman <luccioman@users.noreply.github.com>
Date: Thu, 5 Jul 2018 09:36:36 +0200
Subject: [PATCH] Fixed removal of URLs from the delegatedURL remote crawl
 stack

URLs were removed from the stack using their hash as a bytes array,
whereas the hash is stored in the stack as String instance.
---
 htroot/yacy/crawlReceipt.java                 | 4 ++--
 source/net/yacy/crawler/data/CrawlQueues.java | 8 ++++++--
 2 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/htroot/yacy/crawlReceipt.java b/htroot/yacy/crawlReceipt.java
index 553b6bf6a..9bb321f96 100644
--- a/htroot/yacy/crawlReceipt.java
+++ b/htroot/yacy/crawlReceipt.java
@@ -147,7 +147,7 @@ public final class crawlReceipt {
             // put new entry into database
             sb.index.fulltext().putMetadata(entry);
             ResultURLs.stack(ASCII.String(entry.url().hash()), entry.url().getHost(), youare.getBytes(), iam.getBytes(), EventOrigin.REMOTE_RECEIPTS);
-            sb.crawlQueues.delegatedURL.remove(entry.hash()); // the delegated work has been done
+            sb.crawlQueues.delegatedURL.remove(ASCII.String(entry.hash())); // the delegated work has been done
             if (log.isInfo()) log.info("crawlReceipt: RECEIVED RECEIPT from " + otherPeerName + " for URL " + ASCII.String(entry.hash()) + ":" + entry.url().toNormalform(false));
 
             // ready for more
@@ -160,7 +160,7 @@ public final class crawlReceipt {
         }
 
         if (sb.crawlQueues.delegatedURL != null) { // the delegated work is transformed into an error case
-            sb.crawlQueues.delegatedURL.remove(entry.hash());
+            sb.crawlQueues.delegatedURL.remove(ASCII.String(entry.hash()));
             sb.crawlQueues.errorURL.push(entry.url(), 997, null, FailCategory.FINAL_LOAD_CONTEXT, result + ":" + reason, -1);
         }
         //switchboard.noticeURL.remove(receivedUrlhash);
diff --git a/source/net/yacy/crawler/data/CrawlQueues.java b/source/net/yacy/crawler/data/CrawlQueues.java
index 0425ecc50..8c0fb2d20 100644
--- a/source/net/yacy/crawler/data/CrawlQueues.java
+++ b/source/net/yacy/crawler/data/CrawlQueues.java
@@ -82,6 +82,8 @@ public class CrawlQueues {
 
     public  NoticedURL noticeURL;
     public  ErrorCache errorURL;
+    
+    /** URLs pulled by remote peers in order to crawl them for us */
     public Map<String, DigestURL> delegatedURL;
 
     public CrawlQueues(final Switchboard sb, final File queuePath) {
@@ -107,7 +109,7 @@ public class CrawlQueues {
         if (this.remoteCrawlProviderHashes == null) this.remoteCrawlProviderHashes = new ArrayList<String>();
         if (this.delegatedURL == null) {
             this.delegatedURL = new ConcurrentHashMap<String, DigestURL>();
-            log.config("Finishted Startup of Crawling Management");
+            log.config("Finished Startup of Crawling Management");
         }
     }
     /**
@@ -205,7 +207,9 @@ public class CrawlQueues {
     public void removeURL(final byte[] hash) {
         assert hash != null && hash.length == 12;
         this.noticeURL.removeByURLHash(hash);
-        if (this.delegatedURL != null) this.delegatedURL.remove(hash);
+        if (this.delegatedURL != null) {
+        	this.delegatedURL.remove(ASCII.String(hash));
+        }
     }
     
     public int removeHosts(final Set<String> hosthashes) {