From b1e08d354c59a54959f2ea1c5cfe88fd219444f6 Mon Sep 17 00:00:00 2001 From: orbiter Date: Wed, 12 Dec 2007 00:33:26 +0000 Subject: [PATCH] repaired indexing after search snippet loading git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4268 6c8d7289-2bf4-0310-a012-ef5d649a1542 --- htroot/WatchCrawler_p.java | 5 ++++- source/de/anomic/plasma/plasmaSnippetCache.java | 3 +++ 2 files changed, 7 insertions(+), 1 deletion(-) diff --git a/htroot/WatchCrawler_p.java b/htroot/WatchCrawler_p.java index b3ec7d1bf..6de31fac1 100644 --- a/htroot/WatchCrawler_p.java +++ b/htroot/WatchCrawler_p.java @@ -71,7 +71,10 @@ public class WatchCrawler_p { } else { prop.put("info", "0"); - if ((post.containsKey("autoforward")) && (switchboard.crawlQueues.coreCrawlJobSize() == 0) && (switchboard.crawlQueues.remoteTriggeredCrawlJobSize() == 0)) { + if ((post.containsKey("autoforward")) && + (switchboard.crawlQueues.coreCrawlJobSize() == 0) && + (switchboard.crawlQueues.remoteTriggeredCrawlJobSize() == 0) && + (switchboard.queueSize() < 30)) { prop.put("forwardToCrawlStart", "1"); } diff --git a/source/de/anomic/plasma/plasmaSnippetCache.java b/source/de/anomic/plasma/plasmaSnippetCache.java index 7b8628304..6c27e5773 100644 --- a/source/de/anomic/plasma/plasmaSnippetCache.java +++ b/source/de/anomic/plasma/plasmaSnippetCache.java @@ -284,6 +284,9 @@ public class plasmaSnippetCache { // download resource using the crawler and keep resource in memory if possible plasmaHTCache.Entry entry = plasmaSwitchboard.getSwitchboard().crawlQueues.loadResourceFromWeb(url, timeout, true, true); + // place entry on crawl queue + plasmaHTCache.push(entry); + // getting resource metadata (e.g. the http headers for http resources) if (entry != null) { resInfo = entry.getDocumentInfo();