From f0041d504d314330cfe0a0f14c2d6e30e7682838 Mon Sep 17 00:00:00 2001 From: orbiter Date: Sat, 4 Mar 2006 00:10:49 +0000 Subject: [PATCH] remove of several results from a single domain is stopped if the result set is smaller than the wanted number of results git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1811 6c8d7289-2bf4-0310-a012-ef5d649a1542 --- source/de/anomic/plasma/plasmaSearchEvent.java | 2 +- source/de/anomic/plasma/plasmaSearchResult.java | 1 + yacy.init | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-) diff --git a/source/de/anomic/plasma/plasmaSearchEvent.java b/source/de/anomic/plasma/plasmaSearchEvent.java index 2d4fdb223..adbd10c5a 100644 --- a/source/de/anomic/plasma/plasmaSearchEvent.java +++ b/source/de/anomic/plasma/plasmaSearchEvent.java @@ -254,8 +254,8 @@ public final class plasmaSearchEvent extends Thread implements Runnable { // apply filter profileLocal.startTimer(); - acc.removeDoubleDom(); //acc.removeRedundant(); + acc.removeDoubleDom(); profileLocal.setYieldTime(plasmaSearchTimingProfile.PROCESS_FILTER); profileLocal.setYieldCount(plasmaSearchTimingProfile.PROCESS_FILTER, acc.sizeOrdered()); diff --git a/source/de/anomic/plasma/plasmaSearchResult.java b/source/de/anomic/plasma/plasmaSearchResult.java index 4f0c6acbd..2734e866f 100644 --- a/source/de/anomic/plasma/plasmaSearchResult.java +++ b/source/de/anomic/plasma/plasmaSearchResult.java @@ -172,6 +172,7 @@ public final class plasmaSearchResult { String dom; while (i.hasNext()) { + if (pageAcc.size() <= query.wantedResults) return; entry = (Map.Entry) i.next(); dom = ((plasmaCrawlLURL.Entry) entry.getValue()).url().getHost(); if (doms.contains(dom)) i.remove(); else doms.add(dom); diff --git a/yacy.init b/yacy.init index a1cb9e41b..4406794c0 100644 --- a/yacy.init +++ b/yacy.init @@ -417,7 +417,7 @@ xpstopw=true # the prereq-value is a memory pre-requisite: that much bytes must # be available/free in the heap; othervise the loop is not executed # and another idlesleep is performed -20_dhtdistribution_idlesleep=50000 +20_dhtdistribution_idlesleep=20000 20_dhtdistribution_busysleep=2000 20_dhtdistribution_memprereq=8388608 20_dhtdistribution_threads=1