From 5746aae3db80fd4b4629e5e4a71cd2a1989bbb9c Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Fri, 18 Apr 2014 06:51:46 +0200
Subject: [PATCH] add canonical links to the same crawldepth, not the next
 crawldepth

---
 source/net/yacy/document/Document.java  | 6 +++++-
 source/net/yacy/search/Switchboard.java | 3 ++-
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/source/net/yacy/document/Document.java b/source/net/yacy/document/Document.java
index 71b8b79e2..1d901f99c 100644
--- a/source/net/yacy/document/Document.java
+++ b/source/net/yacy/document/Document.java
@@ -907,6 +907,8 @@ dc_rights
         return newDoc;
     }
 
+    public final static String CANONICAL_MARKER = "canonical";
+    
     public static Map<DigestURL, String> getHyperlinks(final Document[] documents) {
         final Map<DigestURL, String> result = new HashMap<DigestURL, String>();
         for (final Document d: documents) {
@@ -917,7 +919,9 @@ dc_rights
                 String refresh = html.getRefreshPath();
                 if (refresh != null && refresh.length() > 0) try {result.put(new DigestURL(refresh), "refresh");} catch (final MalformedURLException e) {}
                 DigestURL canonical = html.getCanonical();
-                if (canonical != null) result.put(canonical, "canonical");
+                if (canonical != null) {
+                    result.put(canonical, CANONICAL_MARKER);
+                }
             }
         }
         return result;
diff --git a/source/net/yacy/search/Switchboard.java b/source/net/yacy/search/Switchboard.java
index a7705abdd..111c716d0 100644
--- a/source/net/yacy/search/Switchboard.java
+++ b/source/net/yacy/search/Switchboard.java
@@ -2634,6 +2634,7 @@ public final class Switchboard extends serverSwitch {
                 //if (m.matches()) u = m.replaceAll("");
                 
                 // enqueue the hyperlink into the pre-notice-url db
+                int nextdepth = nextEntry.getValue() != null && nextEntry.getValue().equals(Document.CANONICAL_MARKER) ? response.depth() : response.depth() + 1; // canonical documents are on the same depth
                 try {
                     this.crawlStacker.enqueueEntry(new Request(
                         response.initiator(),
@@ -2642,7 +2643,7 @@ public final class Switchboard extends serverSwitch {
                         nextEntry.getValue(),
                         new Date(),
                         response.profile().handle(),
-                        response.depth() + 1,
+                        nextdepth,
                         0,
                         0));
                 } catch (final MalformedURLException e ) {