From bc6dd8194bac0613d028be00fd002773c49d75f6 Mon Sep 17 00:00:00 2001 From: orbiter Date: Mon, 15 Jun 2009 11:49:00 +0000 Subject: [PATCH] refactoring: moved search query class to new search package git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6075 6c8d7289-2bf4-0310-a012-ef5d649a1542 --- htroot/AccessTracker_p.java | 8 ++-- htroot/ContentIntegrationPHPBB3_p.html | 7 +++ htroot/IndexControlRWIs_p.java | 4 +- htroot/Ranking_p.java | 6 +-- htroot/api/timeline.java | 6 +-- htroot/index.java | 24 +++++----- htroot/yacy/search.java | 38 +++++++-------- htroot/yacysearch.java | 32 ++++++------- htroot/yacysearchitem.java | 14 +++--- htroot/yacysearchtrailer.java | 16 +++---- .../metadataPrototype/URLMetadataRow.java | 4 +- source/de/anomic/plasma/plasmaGrafics.java | 3 +- .../de/anomic/plasma/plasmaSearchEvent.java | 47 ++++++++++--------- .../plasma/plasmaSearchRankingProcess.java | 23 ++++----- .../plasma/plasmaSearchRankingProfile.java | 14 +++--- .../de/anomic/plasma/plasmaSnippetCache.java | 19 ++++---- .../de/anomic/plasma/plasmaSwitchboard.java | 11 +++-- .../Query.java} | 11 +++-- source/de/anomic/yacy/yacySearch.java | 4 +- 19 files changed, 153 insertions(+), 138 deletions(-) rename source/de/anomic/{plasma/plasmaSearchQuery.java => search/Query.java} (95%) diff --git a/htroot/AccessTracker_p.java b/htroot/AccessTracker_p.java index 3ddab339f..36ed8a0e7 100644 --- a/htroot/AccessTracker_p.java +++ b/htroot/AccessTracker_p.java @@ -37,8 +37,8 @@ import java.text.SimpleDateFormat; import de.anomic.http.httpRequestHeader; import de.anomic.net.natLib; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSwitchboard; +import de.anomic.search.Query; import de.anomic.server.serverCore; import de.anomic.server.serverObjects; import de.anomic.server.serverSwitch; @@ -138,8 +138,8 @@ public class AccessTracker_p { prop.put("page_num", entCount); } if ((page == 2) || (page == 4)) { - final ArrayList array = (page == 2) ? sb.localSearches : sb.remoteSearches; - plasmaSearchQuery searchProfile; + final ArrayList array = (page == 2) ? sb.localSearches : sb.remoteSearches; + Query searchProfile; int m = Math.min(maxCount, array.size()); long qcountSum = 0; long rcountSum = 0; @@ -167,7 +167,7 @@ public class AccessTracker_p { } else { // remote search prop.putHTML("page_list_" + entCount + "_peername", (searchProfile.remotepeer == null) ? "" : searchProfile.remotepeer.getName()); - prop.put("page_list_" + entCount + "_queryhashes", plasmaSearchQuery.anonymizedQueryHashes(searchProfile.queryHashes)); + prop.put("page_list_" + entCount + "_queryhashes", Query.anonymizedQueryHashes(searchProfile.queryHashes)); } prop.putNum("page_list_" + entCount + "_querycount", searchProfile.linesPerPage); prop.putNum("page_list_" + entCount + "_resultcount", searchProfile.resultcount); diff --git a/htroot/ContentIntegrationPHPBB3_p.html b/htroot/ContentIntegrationPHPBB3_p.html index 6c165e26e..d4b137fe8 100644 --- a/htroot/ContentIntegrationPHPBB3_p.html +++ b/htroot/ContentIntegrationPHPBB3_p.html @@ -13,6 +13,13 @@ Each extraction is specific to the data that is hosted in the database. This interface gives you access to the phpBB3 forums software content.

+

+ If you read from a imported database, here are some hints to get around problems when importing dumps in phpMyAdmin: +

    +
  • before importing large database dumps, set 
    $cfg['UploadDir'] = '/tmp';
    in phpmyadmin/config.inc.php and place your dump file in /tmp. Otherwise it is not possible to upload files larger than 2MB.
    • +
  • deselect the partial import flag/li> +
+

When a export is started, surrogate files are generated into DATA/SURROGATE/in which are automatically fetched by an indexer thread. All indexed surrogate files are then moved to DATA/SURROGATE/out and can be re-cycled when a index is deleted. diff --git a/htroot/IndexControlRWIs_p.java b/htroot/IndexControlRWIs_p.java index e37f6eac4..90b5d9150 100644 --- a/htroot/IndexControlRWIs_p.java +++ b/htroot/IndexControlRWIs_p.java @@ -50,11 +50,11 @@ import de.anomic.kelondro.text.referencePrototype.WordReference; import de.anomic.kelondro.text.referencePrototype.WordReferenceRow; import de.anomic.kelondro.util.DateFormatter; import de.anomic.plasma.plasmaSearchEvent; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProcess; import de.anomic.plasma.plasmaSwitchboard; import de.anomic.plasma.parser.Condenser; import de.anomic.plasma.parser.Word; +import de.anomic.search.Query; import de.anomic.server.serverObjects; import de.anomic.server.serverSwitch; import de.anomic.yacy.yacyClient; @@ -478,7 +478,7 @@ public class IndexControlRWIs_p { } public static plasmaSearchRankingProcess genSearchresult(final serverObjects prop, final plasmaSwitchboard sb, final byte[] keyhash, final Bitfield filter) { - final plasmaSearchQuery query = new plasmaSearchQuery(new String(keyhash), -1, sb.getRanking(), filter); + final Query query = new Query(new String(keyhash), -1, sb.getRanking(), filter); final plasmaSearchRankingProcess ranked = new plasmaSearchRankingProcess(sb.indexSegment, query, Integer.MAX_VALUE, 1); ranked.execQuery(); diff --git a/htroot/Ranking_p.java b/htroot/Ranking_p.java index 3768cefcf..16f29cbb7 100644 --- a/htroot/Ranking_p.java +++ b/htroot/Ranking_p.java @@ -31,9 +31,9 @@ import java.util.Map.Entry; import de.anomic.http.httpRequestHeader; import de.anomic.plasma.plasmaSearchEvent; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProfile; import de.anomic.plasma.plasmaSwitchboard; +import de.anomic.search.Query; import de.anomic.server.serverObjects; import de.anomic.server.serverSwitch; import de.anomic.tools.crypt; @@ -139,7 +139,7 @@ public class Ranking_p { // we create empty entries for template strings final serverObjects prop = defaultValues(); final plasmaSearchRankingProfile ranking; - if(sb == null) ranking = new plasmaSearchRankingProfile(plasmaSearchQuery.CONTENTDOM_TEXT); + if(sb == null) ranking = new plasmaSearchRankingProfile(Query.CONTENTDOM_TEXT); else ranking = sb.getRanking(); putRanking(prop, ranking, "local"); return prop; @@ -156,7 +156,7 @@ public class Ranking_p { if (post.containsKey("ResetRanking")) { sb.setConfig("rankingProfile", ""); - final plasmaSearchRankingProfile ranking = new plasmaSearchRankingProfile(plasmaSearchQuery.CONTENTDOM_TEXT); + final plasmaSearchRankingProfile ranking = new plasmaSearchRankingProfile(Query.CONTENTDOM_TEXT); final serverObjects prop = defaultValues(); //prop.putAll(ranking.toExternalMap("local")); putRanking(prop, ranking, "local"); diff --git a/htroot/api/timeline.java b/htroot/api/timeline.java index 239f40f27..553f5ff84 100644 --- a/htroot/api/timeline.java +++ b/htroot/api/timeline.java @@ -34,9 +34,9 @@ import de.anomic.kelondro.text.Segment; import de.anomic.kelondro.text.TermSearch; import de.anomic.kelondro.text.referencePrototype.WordReference; import de.anomic.kelondro.util.DateFormatter; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSwitchboard; import de.anomic.plasma.parser.Word; +import de.anomic.search.Query; import de.anomic.server.serverObjects; import de.anomic.server.serverSwitch; import de.anomic.tools.iso639; @@ -63,7 +63,7 @@ public final class timeline { language = (agent == null) ? "en" : iso639.userAgentLanguageDetection(agent); if (language == null) language = "en"; } - final TreeSet[] query = plasmaSearchQuery.cleanQuery(querystring); // converts also umlaute + final TreeSet[] query = Query.cleanQuery(querystring); // converts also umlaute TreeSet q = Word.words2hashes(query[0]); // tell all threads to do nothing for a specific time @@ -108,7 +108,7 @@ public final class timeline { // log yacyCore.log.logInfo("EXIT TIMELINE SEARCH: " + - plasmaSearchQuery.anonymizedQueryHashes(q) + " - " + joincount + " links found, " + + Query.anonymizedQueryHashes(q) + " - " + joincount + " links found, " + prop.get("linkcount", "?") + " links selected, " + indexabstractContainercount + " index abstracts, " + (System.currentTimeMillis() - timestamp) + " milliseconds"); diff --git a/htroot/index.java b/htroot/index.java index ae68473fc..b6e906df2 100644 --- a/htroot/index.java +++ b/htroot/index.java @@ -30,9 +30,9 @@ import de.anomic.http.httpRequestHeader; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSwitchboard; import de.anomic.plasma.plasmaSwitchboardConstants; +import de.anomic.search.Query; import de.anomic.server.serverObjects; import de.anomic.server.serverSwitch; @@ -79,13 +79,13 @@ public class index { //global = global && indexDistributeGranted && indexReceiveGranted; // search domain - int contentdom = plasmaSearchQuery.CONTENTDOM_TEXT; + int contentdom = Query.CONTENTDOM_TEXT; final String cds = (post == null) ? "text" : post.get("contentdom", "text"); - if (cds.equals("text")) contentdom = plasmaSearchQuery.CONTENTDOM_TEXT; - if (cds.equals("audio")) contentdom = plasmaSearchQuery.CONTENTDOM_AUDIO; - if (cds.equals("video")) contentdom = plasmaSearchQuery.CONTENTDOM_VIDEO; - if (cds.equals("image")) contentdom = plasmaSearchQuery.CONTENTDOM_IMAGE; - if (cds.equals("app")) contentdom = plasmaSearchQuery.CONTENTDOM_APP; + if (cds.equals("text")) contentdom = Query.CONTENTDOM_TEXT; + if (cds.equals("audio")) contentdom = Query.CONTENTDOM_AUDIO; + if (cds.equals("video")) contentdom = Query.CONTENTDOM_VIDEO; + if (cds.equals("image")) contentdom = Query.CONTENTDOM_IMAGE; + if (cds.equals("app")) contentdom = Query.CONTENTDOM_APP; // we create empty entries for template strings String promoteSearchPageGreeting = env.getConfig(plasmaSwitchboardConstants.GREETING, ""); @@ -120,11 +120,11 @@ public class index { prop.put("display", display); prop.putHTML("constraint", constraint); prop.put("searchoptions_display", display); - prop.put("contentdomCheckText", (contentdom == plasmaSearchQuery.CONTENTDOM_TEXT) ? "1" : "0"); - prop.put("contentdomCheckAudio", (contentdom == plasmaSearchQuery.CONTENTDOM_AUDIO) ? "1" : "0"); - prop.put("contentdomCheckVideo", (contentdom == plasmaSearchQuery.CONTENTDOM_VIDEO) ? "1" : "0"); - prop.put("contentdomCheckImage", (contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) ? "1" : "0"); - prop.put("contentdomCheckApp", (contentdom == plasmaSearchQuery.CONTENTDOM_APP) ? "1" : "0"); + prop.put("contentdomCheckText", (contentdom == Query.CONTENTDOM_TEXT) ? "1" : "0"); + prop.put("contentdomCheckAudio", (contentdom == Query.CONTENTDOM_AUDIO) ? "1" : "0"); + prop.put("contentdomCheckVideo", (contentdom == Query.CONTENTDOM_VIDEO) ? "1" : "0"); + prop.put("contentdomCheckImage", (contentdom == Query.CONTENTDOM_IMAGE) ? "1" : "0"); + prop.put("contentdomCheckApp", (contentdom == Query.CONTENTDOM_APP) ? "1" : "0"); // online caution timing sb.localSearchLastAccess = System.currentTimeMillis(); diff --git a/htroot/yacy/search.java b/htroot/yacy/search.java index 4d36a065c..143fa3480 100644 --- a/htroot/yacy/search.java +++ b/htroot/yacy/search.java @@ -44,11 +44,11 @@ import de.anomic.kelondro.util.SortStack; import de.anomic.net.natLib; import de.anomic.plasma.plasmaProfiling; import de.anomic.plasma.plasmaSearchEvent; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProfile; import de.anomic.plasma.plasmaSwitchboard; import de.anomic.plasma.plasmaSearchEvent.ResultEntry; import de.anomic.plasma.plasmaSearchRankingProcess.NavigatorEntry; +import de.anomic.search.Query; import de.anomic.server.serverCore; import de.anomic.server.serverObjects; import de.anomic.server.serverProfiling; @@ -157,7 +157,7 @@ public final class search { // tell all threads to do nothing for a specific time sb.intermissionAllThreads(3000); - final TreeSet abstractSet = ((abstracts.length() == 0) || (abstracts.equals("auto"))) ? null : plasmaSearchQuery.hashes2Set(abstracts); + final TreeSet abstractSet = ((abstracts.length() == 0) || (abstracts.equals("auto"))) ? null : Query.hashes2Set(abstracts); // store accessing peer final yacySeed remoteSeed = yacySeed.genRemoteSeed(oseed, key, false); @@ -168,23 +168,23 @@ public final class search { } // prepare search - final TreeSet queryhashes = plasmaSearchQuery.hashes2Set(query); - final TreeSet excludehashes = (exclude.length() == 0) ? new TreeSet(Base64Order.enhancedCoder) : plasmaSearchQuery.hashes2Set(exclude); + final TreeSet queryhashes = Query.hashes2Set(query); + final TreeSet excludehashes = (exclude.length() == 0) ? new TreeSet(Base64Order.enhancedCoder) : Query.hashes2Set(exclude); final long timestamp = System.currentTimeMillis(); // prepare a search profile - final plasmaSearchRankingProfile rankingProfile = (profile.length() == 0) ? new plasmaSearchRankingProfile(plasmaSearchQuery.contentdomParser(contentdom)) : new plasmaSearchRankingProfile("", profile); + final plasmaSearchRankingProfile rankingProfile = (profile.length() == 0) ? new plasmaSearchRankingProfile(Query.contentdomParser(contentdom)) : new plasmaSearchRankingProfile("", profile); // prepare an abstract result final StringBuilder indexabstract = new StringBuilder(); int indexabstractContainercount = 0; int joincount = 0; - plasmaSearchQuery theQuery = null; + Query theQuery = null; ArrayList.stackElement> accu = null; plasmaSearchEvent theSearch = null; if ((query.length() == 0) && (abstractSet != null)) { // this is _not_ a normal search, only a request for index abstracts - theQuery = new plasmaSearchQuery( + theQuery = new Query( null, abstractSet, new TreeSet(Base64Order.enhancedCoder), @@ -192,14 +192,14 @@ public final class search { rankingProfile, maxdist, prefer, - plasmaSearchQuery.contentdomParser(contentdom), + Query.contentdomParser(contentdom), language, "", // no navigation false, count, 0, filter, - plasmaSearchQuery.SEARCHDOM_LOCAL, + Query.SEARCHDOM_LOCAL, null, -1, null, @@ -209,12 +209,12 @@ public final class search { yacyURL.TLD_any_zone_filter, client, false); - theQuery.domType = plasmaSearchQuery.SEARCHDOM_LOCAL; - yacyCore.log.logInfo("INIT HASH SEARCH (abstracts only): " + plasmaSearchQuery.anonymizedQueryHashes(theQuery.queryHashes) + " - " + theQuery.displayResults() + " links"); + theQuery.domType = Query.SEARCHDOM_LOCAL; + yacyCore.log.logInfo("INIT HASH SEARCH (abstracts only): " + Query.anonymizedQueryHashes(theQuery.queryHashes) + " - " + theQuery.displayResults() + " links"); final long timer = System.currentTimeMillis(); //final Map>[] containers = sb.indexSegment.index().searchTerm(theQuery.queryHashes, theQuery.excludeHashes, plasmaSearchQuery.hashes2StringSet(urls)); - final HashMap> incc = sb.indexSegment.termIndex().searchConjunction(theQuery.queryHashes, plasmaSearchQuery.hashes2StringSet(urls)); + final HashMap> incc = sb.indexSegment.termIndex().searchConjunction(theQuery.queryHashes, Query.hashes2StringSet(urls)); serverProfiling.update("SEARCH", new plasmaProfiling.searchEvent(theQuery.id(true), plasmaSearchEvent.COLLECTION, incc.size(), System.currentTimeMillis() - timer), false); if (incc != null) { @@ -236,7 +236,7 @@ public final class search { } else { // retrieve index containers from search request - theQuery = new plasmaSearchQuery( + theQuery = new Query( null, queryhashes, excludehashes, @@ -244,7 +244,7 @@ public final class search { rankingProfile, maxdist, prefer, - plasmaSearchQuery. + Query. contentdomParser(contentdom), language, "", // no navigation @@ -252,7 +252,7 @@ public final class search { count, 0, filter, - plasmaSearchQuery.SEARCHDOM_LOCAL, + Query.SEARCHDOM_LOCAL, null, -1, constraint, @@ -262,9 +262,9 @@ public final class search { yacyURL.TLD_any_zone_filter, client, false); - theQuery.domType = plasmaSearchQuery.SEARCHDOM_LOCAL; - yacyCore.log.logInfo("INIT HASH SEARCH (query-" + abstracts + "): " + plasmaSearchQuery.anonymizedQueryHashes(theQuery.queryHashes) + " - " + theQuery.displayResults() + " links"); - RSSFeed.channels(RSSFeed.REMOTESEARCH).addMessage(new RSSMessage("Remote Search Request from " + ((remoteSeed == null) ? "unknown" : remoteSeed.getName()), plasmaSearchQuery.anonymizedQueryHashes(theQuery.queryHashes), "")); + theQuery.domType = Query.SEARCHDOM_LOCAL; + yacyCore.log.logInfo("INIT HASH SEARCH (query-" + abstracts + "): " + Query.anonymizedQueryHashes(theQuery.queryHashes) + " - " + theQuery.displayResults() + " links"); + RSSFeed.channels(RSSFeed.REMOTESEARCH).addMessage(new RSSMessage("Remote Search Request from " + ((remoteSeed == null) ? "unknown" : remoteSeed.getName()), Query.anonymizedQueryHashes(theQuery.queryHashes), "")); // make event theSearch = plasmaSearchEvent.getEvent(theQuery, rankingProfile, sb.indexSegment, sb.peers, sb.crawlResults, null, true); @@ -382,7 +382,7 @@ public final class search { // log yacyCore.log.logInfo("EXIT HASH SEARCH: " + - plasmaSearchQuery.anonymizedQueryHashes(theQuery.queryHashes) + " - " + joincount + " links found, " + + Query.anonymizedQueryHashes(theQuery.queryHashes) + " - " + joincount + " links found, " + prop.get("linkcount", "?") + " links selected, " + indexabstractContainercount + " index abstracts, " + (System.currentTimeMillis() - timestamp) + " milliseconds"); diff --git a/htroot/yacysearch.java b/htroot/yacysearch.java index d1b019560..0fbf119aa 100644 --- a/htroot/yacysearch.java +++ b/htroot/yacysearch.java @@ -42,13 +42,13 @@ import de.anomic.kelondro.util.Log; import de.anomic.plasma.plasmaParserDocument; import de.anomic.plasma.plasmaProfiling; import de.anomic.plasma.plasmaSearchEvent; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProfile; import de.anomic.plasma.plasmaSnippetCache; import de.anomic.plasma.plasmaSwitchboard; import de.anomic.plasma.plasmaSwitchboardConstants; import de.anomic.plasma.parser.Word; import de.anomic.plasma.parser.Condenser; +import de.anomic.search.Query; import de.anomic.server.serverCore; import de.anomic.server.serverDomains; import de.anomic.server.serverObjects; @@ -177,10 +177,10 @@ public class yacysearch { if (clustersearch) global = true; // switches search on, but search target is limited to cluster nodes // find search domain - final int contentdomCode = plasmaSearchQuery.contentdomParser((post == null ? "text" : post.get("contentdom", "text"))); + final int contentdomCode = Query.contentdomParser((post == null ? "text" : post.get("contentdom", "text"))); // patch until better search profiles are available - if ((contentdomCode != plasmaSearchQuery.CONTENTDOM_TEXT) && (itemsPerPage <= 32)) itemsPerPage = 32; + if ((contentdomCode != Query.CONTENTDOM_TEXT) && (itemsPerPage <= 32)) itemsPerPage = 32; // check the search tracker TreeSet trackerHandles = sb.localSearchTracker.get(client); @@ -340,7 +340,7 @@ public class yacysearch { String navigation = (post == null) ? "" : post.get("nav", ""); // the query - final TreeSet[] query = plasmaSearchQuery.cleanQuery(querystring.trim()); // converts also umlaute + final TreeSet[] query = Query.cleanQuery(querystring.trim()); // converts also umlaute int maxDistance = (querystring.indexOf('"') >= 0) ? maxDistance = query.length - 1 : Integer.MAX_VALUE; @@ -403,7 +403,7 @@ public class yacysearch { // do the search final TreeSet queryHashes = Word.words2hashes(query[0]); - final plasmaSearchQuery theQuery = new plasmaSearchQuery( + final Query theQuery = new Query( originalquerystring, queryHashes, Word.words2hashes(query[1]), @@ -418,8 +418,8 @@ public class yacysearch { itemsPerPage, offset, urlmask, - (clustersearch && globalsearch) ? plasmaSearchQuery.SEARCHDOM_CLUSTERALL : - ((globalsearch) ? plasmaSearchQuery.SEARCHDOM_GLOBALDHT : plasmaSearchQuery.SEARCHDOM_LOCAL), + (clustersearch && globalsearch) ? Query.SEARCHDOM_CLUSTERALL : + ((globalsearch) ? Query.SEARCHDOM_GLOBALDHT : Query.SEARCHDOM_LOCAL), "", 20, constraint, @@ -438,7 +438,7 @@ public class yacysearch { theQuery.filterOut(plasmaSwitchboard.blueList); // log - Log.logInfo("LOCAL_SEARCH", "INIT WORD SEARCH: " + theQuery.queryString + ":" + plasmaSearchQuery.hashSet2hashString(theQuery.queryHashes) + " - " + theQuery.neededResults() + " links to be computed, " + theQuery.displayResults() + " lines to be displayed"); + Log.logInfo("LOCAL_SEARCH", "INIT WORD SEARCH: " + theQuery.queryString + ":" + Query.hashSet2hashString(theQuery.queryHashes) + " - " + theQuery.neededResults() + " links to be computed, " + theQuery.displayResults() + " lines to be displayed"); RSSFeed.channels(RSSFeed.LOCALSEARCH).addMessage(new RSSMessage("Local Search Request", theQuery.queryString, "")); final long timestamp = System.currentTimeMillis(); @@ -535,7 +535,7 @@ public class yacysearch { resnav.append(" "); } else { resnav.append(" "); } final int numberofpages = Math.min(10, Math.min(thispage + 2, totalcount / theQuery.displayResults())); @@ -546,7 +546,7 @@ public class yacysearch { resnav.append(".gif\" width=\"16\" height=\"16\"> "); } else { resnav.append(" "); @@ -556,7 +556,7 @@ public class yacysearch { resnav.append(""); } else { resnav.append(""); } prop.put("num-results_resnav", resnav.toString()); @@ -612,11 +612,11 @@ public class yacysearch { prop.put("constraint", (constraint == null) ? "" : constraint.exportB64()); prop.put("verify", (fetchSnippets) ? "true" : "false"); prop.put("contentdom", (post == null ? "text" : post.get("contentdom", "text"))); - prop.put("contentdomCheckText", (contentdomCode == plasmaSearchQuery.CONTENTDOM_TEXT) ? "1" : "0"); - prop.put("contentdomCheckAudio", (contentdomCode == plasmaSearchQuery.CONTENTDOM_AUDIO) ? "1" : "0"); - prop.put("contentdomCheckVideo", (contentdomCode == plasmaSearchQuery.CONTENTDOM_VIDEO) ? "1" : "0"); - prop.put("contentdomCheckImage", (contentdomCode == plasmaSearchQuery.CONTENTDOM_IMAGE) ? "1" : "0"); - prop.put("contentdomCheckApp", (contentdomCode == plasmaSearchQuery.CONTENTDOM_APP) ? "1" : "0"); + prop.put("contentdomCheckText", (contentdomCode == Query.CONTENTDOM_TEXT) ? "1" : "0"); + prop.put("contentdomCheckAudio", (contentdomCode == Query.CONTENTDOM_AUDIO) ? "1" : "0"); + prop.put("contentdomCheckVideo", (contentdomCode == Query.CONTENTDOM_VIDEO) ? "1" : "0"); + prop.put("contentdomCheckImage", (contentdomCode == Query.CONTENTDOM_IMAGE) ? "1" : "0"); + prop.put("contentdomCheckApp", (contentdomCode == Query.CONTENTDOM_APP) ? "1" : "0"); // for RSS: don't HTML encode some elements prop.putXML("rss_query", originalquerystring); diff --git a/htroot/yacysearchitem.java b/htroot/yacysearchitem.java index be555599e..f3158a58e 100644 --- a/htroot/yacysearchitem.java +++ b/htroot/yacysearchitem.java @@ -33,10 +33,10 @@ import java.util.TreeSet; import de.anomic.http.httpRequestHeader; import de.anomic.plasma.plasmaProfiling; import de.anomic.plasma.plasmaSearchEvent; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProcess; import de.anomic.plasma.plasmaSnippetCache; import de.anomic.plasma.plasmaSwitchboard; +import de.anomic.search.Query; import de.anomic.server.serverObjects; import de.anomic.server.serverProfiling; import de.anomic.server.serverSwitch; @@ -77,7 +77,7 @@ public class yacysearchitem { // the event does not exist, show empty page return prop; } - final plasmaSearchQuery theQuery = theSearch.getQuery(); + final Query theQuery = theSearch.getQuery(); // dynamically update count values final int offset = theQuery.neededResults() - theQuery.displayResults() + 1; @@ -89,7 +89,7 @@ public class yacysearchitem { prop.put("remoteIndexCount", Formatter.number(theSearch.getRankingResult().getRemoteIndexCount(), true)); prop.put("remotePeerCount", Formatter.number(theSearch.getRankingResult().getRemotePeerCount(), true)); - if (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_TEXT) { + if (theQuery.contentdom == Query.CONTENTDOM_TEXT) { // text search // generate result object @@ -151,7 +151,7 @@ public class yacysearchitem { return prop; } - if (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) { + if (theQuery.contentdom == Query.CONTENTDOM_IMAGE) { // image search; shows thumbnails prop.put("content", theQuery.contentdom + 1); // switch on specific content @@ -171,9 +171,9 @@ public class yacysearchitem { return prop; } - if ((theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_AUDIO) || - (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_VIDEO) || - (theQuery.contentdom == plasmaSearchQuery.CONTENTDOM_APP)) { + if ((theQuery.contentdom == Query.CONTENTDOM_AUDIO) || + (theQuery.contentdom == Query.CONTENTDOM_VIDEO) || + (theQuery.contentdom == Query.CONTENTDOM_APP)) { // any other media content // generate result object diff --git a/htroot/yacysearchtrailer.java b/htroot/yacysearchtrailer.java index 87e0fd903..b0a7ecba4 100644 --- a/htroot/yacysearchtrailer.java +++ b/htroot/yacysearchtrailer.java @@ -30,8 +30,8 @@ import java.util.Iterator; import de.anomic.http.httpRequestHeader; import de.anomic.plasma.plasmaProfiling; import de.anomic.plasma.plasmaSearchEvent; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProcess.NavigatorEntry; +import de.anomic.search.Query; import de.anomic.server.serverObjects; import de.anomic.server.serverProfiling; import de.anomic.server.serverSwitch; @@ -53,7 +53,7 @@ public class yacysearchtrailer { // the event does not exist, show empty page return prop; } - final plasmaSearchQuery theQuery = theSearch.getQuery(); + final Query theQuery = theSearch.getQuery(); // compose search navigation @@ -68,8 +68,8 @@ public class yacysearchtrailer { for (i = 0; i < hostNavigator.size(); i++) { entry = hostNavigator.get(i); prop.put("nav-domains_element_" + i + "_name", entry.name); - prop.put("nav-domains_element_" + i + "_url", "" + entry.name + " (" + entry.count + ")"); - prop.putJSON("nav-domains_element_" + i + "_url-json", plasmaSearchQuery.navurl("json", 0, display, theQuery, theQuery.urlMask, "site:" + entry.name, theQuery.navigators)); + prop.put("nav-domains_element_" + i + "_url", "" + entry.name + " (" + entry.count + ")"); + prop.putJSON("nav-domains_element_" + i + "_url-json", Query.navurl("json", 0, display, theQuery, theQuery.urlMask, "site:" + entry.name, theQuery.navigators)); prop.put("nav-domains_element_" + i + "_count", entry.count); prop.put("nav-domains_element_" + i + "_modifier", "site:" + entry.name); prop.put("nav-domains_element_" + i + "_nl", 1); @@ -92,8 +92,8 @@ public class yacysearchtrailer { entry = authorNavigator.get(i); anav = (entry.name.indexOf(' ') < 0) ? "author:" + entry.name : "author:'" + entry.name + "'"; prop.put("nav-authors_element_" + i + "_name", entry.name); - prop.put("nav-authors_element_" + i + "_url", "" + entry.name + " (" + entry.count + ")"); - prop.putJSON("nav-authors_element_" + i + "_url-json", plasmaSearchQuery.navurl("json", 0, display, theQuery, theQuery.urlMask, anav, theQuery.navigators)); + prop.put("nav-authors_element_" + i + "_url", "" + entry.name + " (" + entry.count + ")"); + prop.putJSON("nav-authors_element_" + i + "_url-json", Query.navurl("json", 0, display, theQuery, theQuery.urlMask, anav, theQuery.navigators)); prop.put("nav-authors_element_" + i + "_count", entry.count); prop.put("nav-authors_element_" + i + "_modifier", "author:'" + entry.name + "'"); prop.put("nav-authors_element_" + i + "_nl", 1); @@ -114,8 +114,8 @@ public class yacysearchtrailer { if (/*(theQuery == null) ||*/ (theQuery.queryString == null)) break; if (e != null && e.name != null) { prop.putHTML("nav-topics_element_" + i + "_name", e.name); - prop.put("nav-topics_element_" + i + "_url", "" + e.name + " (" + e.count + ")"); - prop.putJSON("nav-topics_element_" + i + "_url-json", plasmaSearchQuery.navurl("json", 0, display, theQuery, theQuery.urlMask, e.name, theQuery.navigators)); + prop.put("nav-topics_element_" + i + "_url", "" + e.name + " (" + e.count + ")"); + prop.putJSON("nav-topics_element_" + i + "_url-json", Query.navurl("json", 0, display, theQuery, theQuery.urlMask, e.name, theQuery.navigators)); prop.put("nav-topics_element_" + i + "_count", e.count); prop.put("nav-topics_element_" + i + "_modifier", e.name); prop.put("nav-topics_element_" + i + "_nl", (iter.hasNext() && i < MAX_TOPWORDS) ? 1 : 0); diff --git a/source/de/anomic/kelondro/text/metadataPrototype/URLMetadataRow.java b/source/de/anomic/kelondro/text/metadataPrototype/URLMetadataRow.java index a5c96cf13..6ed485c74 100644 --- a/source/de/anomic/kelondro/text/metadataPrototype/URLMetadataRow.java +++ b/source/de/anomic/kelondro/text/metadataPrototype/URLMetadataRow.java @@ -45,7 +45,7 @@ import de.anomic.kelondro.text.referencePrototype.WordReferenceRow; import de.anomic.kelondro.util.DateFormatter; import de.anomic.kelondro.util.FileUtils; import de.anomic.kelondro.util.kelondroException; -import de.anomic.plasma.plasmaSearchQuery; +import de.anomic.search.Query; import de.anomic.server.serverCharBuffer; import de.anomic.server.serverCodings; import de.anomic.tools.crypt; @@ -254,7 +254,7 @@ public class URLMetadataRow implements Metadata { this.entry.setCol(col_wc, Integer.parseInt(prop.getProperty("wc", "0"))); this.entry.setCol(col_dt, new byte[]{(byte) prop.getProperty("dt", "t").charAt(0)}); final String flags = prop.getProperty("flags", "AAAAAA"); - this.entry.setCol(col_flags, (flags.length() > 6) ? plasmaSearchQuery.empty_constraint.bytes() : (new Bitfield(4, flags)).bytes()); + this.entry.setCol(col_flags, (flags.length() > 6) ? Query.empty_constraint.bytes() : (new Bitfield(4, flags)).bytes()); try { this.entry.setCol(col_lang, prop.getProperty("lang", "uk").getBytes("UTF-8")); } catch (UnsupportedEncodingException e) { diff --git a/source/de/anomic/plasma/plasmaGrafics.java b/source/de/anomic/plasma/plasmaGrafics.java index 7e8a0e383..f6556a878 100644 --- a/source/de/anomic/plasma/plasmaGrafics.java +++ b/source/de/anomic/plasma/plasmaGrafics.java @@ -35,6 +35,7 @@ import java.awt.image.BufferedImage; import java.util.Date; import java.util.Iterator; +import de.anomic.search.Query; import de.anomic.yacy.yacySearch; import de.anomic.yacy.yacySeed; import de.anomic.yacy.yacySeedDB; @@ -144,7 +145,7 @@ public class plasmaGrafics { } // draw in the search target - final plasmaSearchQuery query = event.getQuery(); + final Query query = event.getQuery(); final Iterator i = query.queryHashes.iterator(); eventPicture.setColor(ymageMatrix.GREY); while (i.hasNext()) { diff --git a/source/de/anomic/plasma/plasmaSearchEvent.java b/source/de/anomic/plasma/plasmaSearchEvent.java index f6a53c989..399f853b2 100644 --- a/source/de/anomic/plasma/plasmaSearchEvent.java +++ b/source/de/anomic/plasma/plasmaSearchEvent.java @@ -54,6 +54,7 @@ import de.anomic.plasma.parser.Word; import de.anomic.plasma.parser.Condenser; import de.anomic.plasma.plasmaSearchRankingProcess.NavigatorEntry; import de.anomic.plasma.plasmaSnippetCache.MediaSnippet; +import de.anomic.search.Query; import de.anomic.server.serverProfiling; import de.anomic.yacy.yacySearch; import de.anomic.yacy.yacySeed; @@ -75,10 +76,10 @@ public final class plasmaSearchEvent { public static String lastEventID = ""; private static ConcurrentHashMap lastEvents = new ConcurrentHashMap(); // a cache for objects from this class: re-use old search requests public static final long eventLifetime = 60000; // the time an event will stay in the cache, 1 Minute - private static final int max_results_preparation = 300; + private static final int max_results_preparation = 1000; private long eventTime; - plasmaSearchQuery query; + Query query; private final Segment indexSegment; private final yacySeedDB peers; plasmaSearchRankingProcess rankedCache; // ordered search results, grows dynamically as all the query threads enrich this container @@ -100,7 +101,7 @@ public final class plasmaSearchEvent { public ResultURLs crawlResults; @SuppressWarnings("unchecked") - private plasmaSearchEvent(final plasmaSearchQuery query, + private plasmaSearchEvent(final Query query, final Segment indexSegment, final yacySeedDB peers, final ResultURLs crawlResults, @@ -136,8 +137,8 @@ public final class plasmaSearchEvent { } final long start = System.currentTimeMillis(); - if ((query.domType == plasmaSearchQuery.SEARCHDOM_GLOBALDHT) || - (query.domType == plasmaSearchQuery.SEARCHDOM_CLUSTERALL)) { + if ((query.domType == Query.SEARCHDOM_GLOBALDHT) || + (query.domType == Query.SEARCHDOM_CLUSTERALL)) { // initialize a ranking process that is the target for data // that is generated concurrently from local and global search threads @@ -152,8 +153,8 @@ public final class plasmaSearchEvent { final int fetchpeers = 12; Log.logFine("SEARCH_EVENT", "STARTING " + fetchpeers + " THREADS TO CATCH EACH " + query.displayResults() + " URLs"); this.primarySearchThreads = yacySearch.primaryRemoteSearches( - plasmaSearchQuery.hashSet2hashString(query.queryHashes), - plasmaSearchQuery.hashSet2hashString(query.excludeHashes), + Query.hashSet2hashString(query.queryHashes), + Query.hashSet2hashString(query.excludeHashes), "", query.prefer, query.urlMask, @@ -171,7 +172,7 @@ public final class plasmaSearchEvent { plasmaSwitchboard.urlBlacklist, query.ranking, query.constraint, - (query.domType == plasmaSearchQuery.SEARCHDOM_GLOBALDHT) ? null : preselectedPeerHashes); + (query.domType == Query.SEARCHDOM_GLOBALDHT) ? null : preselectedPeerHashes); serverProfiling.update("SEARCH", new plasmaProfiling.searchEvent(query.id(true), "remote search thread start", this.primarySearchThreads.length, System.currentTimeMillis() - timer), false); // finished searching @@ -303,9 +304,9 @@ public final class plasmaSearchEvent { final long dbRetrievalTime = System.currentTimeMillis() - startTime; // check exclusion - if ((plasmaSearchQuery.matches(pagetitle, query.excludeHashes)) || - (plasmaSearchQuery.matches(pageurl, query.excludeHashes)) || - (plasmaSearchQuery.matches(pageauthor, query.excludeHashes))) { + if ((Query.matches(pagetitle, query.excludeHashes)) || + (Query.matches(pageurl, query.excludeHashes)) || + (Query.matches(pageauthor, query.excludeHashes))) { return null; } @@ -324,19 +325,19 @@ public final class plasmaSearchEvent { return null; } - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_AUDIO) && (page.laudio() == 0)) { + if ((query.contentdom == Query.CONTENTDOM_AUDIO) && (page.laudio() == 0)) { registerFailure(page.hash(), "contentdom-audio constraint not fullfilled"); return null; } - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_VIDEO) && (page.lvideo() == 0)) { + if ((query.contentdom == Query.CONTENTDOM_VIDEO) && (page.lvideo() == 0)) { registerFailure(page.hash(), "contentdom-video constraint not fullfilled"); return null; } - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) && (page.limage() == 0)) { + if ((query.contentdom == Query.CONTENTDOM_IMAGE) && (page.limage() == 0)) { registerFailure(page.hash(), "contentdom-image constraint not fullfilled"); return null; } - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_APP) && (page.lapp() == 0)) { + if ((query.contentdom == Query.CONTENTDOM_APP) && (page.lapp() == 0)) { registerFailure(page.hash(), "contentdom-app constraint not fullfilled"); return null; } @@ -346,7 +347,7 @@ public final class plasmaSearchEvent { } // load snippet - if (query.contentdom == plasmaSearchQuery.CONTENTDOM_TEXT) { + if (query.contentdom == Query.CONTENTDOM_TEXT) { // attach text snippet startTime = System.currentTimeMillis(); final plasmaSnippetCache.TextSnippet snippet = plasmaSnippetCache.retrieveTextSnippet(metadata, snippetFetchWordHashes, (snippetFetchMode == 2), ((query.constraint != null) && (query.constraint.get(Condenser.flag_cat_indexof))), 180, 3000, (snippetFetchMode == 2) ? Integer.MAX_VALUE : 30000, query.isGlobal()); @@ -429,7 +430,7 @@ public final class plasmaSearchEvent { return count; } - public plasmaSearchQuery getQuery() { + public Query getQuery() { return query; } @@ -458,7 +459,7 @@ public final class plasmaSearchEvent { } public static plasmaSearchEvent getEvent( - final plasmaSearchQuery query, + final Query query, final plasmaSearchRankingProfile ranking, final Segment indexSegment, final yacySeedDB peers, @@ -488,7 +489,7 @@ public final class plasmaSearchEvent { } else { // if worker threads had been alive, but did not succeed, start them again to fetch missing links if ((!event.anyWorkerAlive()) && - (((query.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) && (event.images.size() + 30 < query.neededResults())) || + (((query.contentdom == Query.CONTENTDOM_IMAGE) && (event.images.size() + 30 < query.neededResults())) || (event.result.size() < query.neededResults() + 10)) && //(event.query.onlineSnippetFetch) && (event.getRankingResult().getLocalResourceSize() + event.getRankingResult().getRemoteResourceSize() > event.result.size())) { @@ -533,8 +534,8 @@ public final class plasmaSearchEvent { this.lastLifeSign = System.currentTimeMillis(); // check if we have enough - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) && (images.size() >= query.neededResults() + fetchAhead)) break; - if ((query.contentdom != plasmaSearchQuery.CONTENTDOM_IMAGE) && (result.size() >= query.neededResults() + fetchAhead)) break; + if ((query.contentdom == Query.CONTENTDOM_IMAGE) && (images.size() >= query.neededResults() + fetchAhead)) break; + if ((query.contentdom != Query.CONTENTDOM_IMAGE) && (result.size() >= query.neededResults() + fetchAhead)) break; // get next entry page = rankedCache.bestURL(true); @@ -601,8 +602,8 @@ public final class plasmaSearchEvent { // we have the wanted result already in the result array .. return that return this.result.element(item).element; } - if ((query.domType == plasmaSearchQuery.SEARCHDOM_GLOBALDHT) || - (query.domType == plasmaSearchQuery.SEARCHDOM_CLUSTERALL)) { + if ((query.domType == Query.SEARCHDOM_GLOBALDHT) || + (query.domType == Query.SEARCHDOM_CLUSTERALL)) { // this is a search using remote search threads. Also the local // search thread is started as background process if ((localSearchThread != null) && (localSearchThread.isAlive())) { diff --git a/source/de/anomic/plasma/plasmaSearchRankingProcess.java b/source/de/anomic/plasma/plasmaSearchRankingProcess.java index d0022938b..5b10757e3 100644 --- a/source/de/anomic/plasma/plasmaSearchRankingProcess.java +++ b/source/de/anomic/plasma/plasmaSearchRankingProcess.java @@ -54,6 +54,7 @@ import de.anomic.kelondro.util.SortStack; import de.anomic.kelondro.util.FileUtils; import de.anomic.plasma.parser.Word; import de.anomic.plasma.parser.Condenser; +import de.anomic.search.Query; import de.anomic.server.serverProfiling; import de.anomic.yacy.yacyURL; @@ -67,7 +68,7 @@ public final class plasmaSearchRankingProcess { private final SortStack stack; private final HashMap> doubleDomCache; // key = domhash (6 bytes); value = like stack private final HashSet handover; // key = urlhash; used for double-check of urls that had been handed over to search process - private final plasmaSearchQuery query; + private final Query query; private final int maxentries; private int remote_peerCount, remote_indexCount, remote_resourceSize, local_resourceSize; private final ReferenceOrder order; @@ -83,7 +84,7 @@ public final class plasmaSearchRankingProcess { public plasmaSearchRankingProcess( final Segment indexSegment, - final plasmaSearchQuery query, + final Query query, final int maxentries, final int concurrency) { // we collect the urlhashes and construct a list with urlEntry objects @@ -184,11 +185,11 @@ public final class plasmaSearchRankingProcess { if (!testFlags(iEntry)) continue; // check document domain - if (query.contentdom != plasmaSearchQuery.CONTENTDOM_TEXT) { - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_AUDIO) && (!(iEntry.flags().get(Condenser.flag_cat_hasaudio)))) continue; - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_VIDEO) && (!(iEntry.flags().get(Condenser.flag_cat_hasvideo)))) continue; - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) && (!(iEntry.flags().get(Condenser.flag_cat_hasimage)))) continue; - if ((query.contentdom == plasmaSearchQuery.CONTENTDOM_APP ) && (!(iEntry.flags().get(Condenser.flag_cat_hasapp )))) continue; + if (query.contentdom != Query.CONTENTDOM_TEXT) { + if ((query.contentdom == Query.CONTENTDOM_AUDIO) && (!(iEntry.flags().get(Condenser.flag_cat_hasaudio)))) continue; + if ((query.contentdom == Query.CONTENTDOM_VIDEO) && (!(iEntry.flags().get(Condenser.flag_cat_hasvideo)))) continue; + if ((query.contentdom == Query.CONTENTDOM_IMAGE) && (!(iEntry.flags().get(Condenser.flag_cat_hasimage)))) continue; + if ((query.contentdom == Query.CONTENTDOM_APP ) && (!(iEntry.flags().get(Condenser.flag_cat_hasapp )))) continue; } // check tld domain @@ -620,10 +621,10 @@ public final class plasmaSearchRankingProcess { long r = (255 - position) << 8; // for media search: prefer pages with many links - if (query.contentdom == plasmaSearchQuery.CONTENTDOM_IMAGE) r += rentry.limage() << query.ranking.coeff_cathasimage; - if (query.contentdom == plasmaSearchQuery.CONTENTDOM_AUDIO) r += rentry.laudio() << query.ranking.coeff_cathasaudio; - if (query.contentdom == plasmaSearchQuery.CONTENTDOM_VIDEO) r += rentry.lvideo() << query.ranking.coeff_cathasvideo; - if (query.contentdom == plasmaSearchQuery.CONTENTDOM_APP ) r += rentry.lapp() << query.ranking.coeff_cathasapp; + if (query.contentdom == Query.CONTENTDOM_IMAGE) r += rentry.limage() << query.ranking.coeff_cathasimage; + if (query.contentdom == Query.CONTENTDOM_AUDIO) r += rentry.laudio() << query.ranking.coeff_cathasaudio; + if (query.contentdom == Query.CONTENTDOM_VIDEO) r += rentry.lvideo() << query.ranking.coeff_cathasvideo; + if (query.contentdom == Query.CONTENTDOM_APP ) r += rentry.lapp() << query.ranking.coeff_cathasapp; // prefer hit with 'prefer' pattern if (rentry.url().toNormalform(true, true).matches(query.prefer)) r += 256 << query.ranking.coeff_prefer; diff --git a/source/de/anomic/plasma/plasmaSearchRankingProfile.java b/source/de/anomic/plasma/plasmaSearchRankingProfile.java index bb3ab9907..6f5f8f3d4 100644 --- a/source/de/anomic/plasma/plasmaSearchRankingProfile.java +++ b/source/de/anomic/plasma/plasmaSearchRankingProfile.java @@ -26,6 +26,8 @@ import java.util.HashMap; import java.util.Iterator; import java.util.Map; +import de.anomic.search.Query; + public class plasmaSearchRankingProfile { // pre-sort attributes @@ -101,11 +103,11 @@ public class plasmaSearchRankingProfile { coeff_app_dc_subject = 13; coeff_app_dc_description = 13; coeff_appemph = 10; - coeff_catindexof = (mediatype == plasmaSearchQuery.CONTENTDOM_TEXT) ? 0 : 15; - coeff_cathasimage = (mediatype == plasmaSearchQuery.CONTENTDOM_IMAGE) ? 15 : 0; - coeff_cathasaudio = (mediatype == plasmaSearchQuery.CONTENTDOM_AUDIO) ? 15 : 0; - coeff_cathasvideo = (mediatype == plasmaSearchQuery.CONTENTDOM_VIDEO) ? 15 : 0; - coeff_cathasapp = (mediatype == plasmaSearchQuery.CONTENTDOM_APP) ? 15 : 0; + coeff_catindexof = (mediatype == Query.CONTENTDOM_TEXT) ? 0 : 15; + coeff_cathasimage = (mediatype == Query.CONTENTDOM_IMAGE) ? 15 : 0; + coeff_cathasaudio = (mediatype == Query.CONTENTDOM_AUDIO) ? 15 : 0; + coeff_cathasvideo = (mediatype == Query.CONTENTDOM_VIDEO) ? 15 : 0; + coeff_cathasapp = (mediatype == Query.CONTENTDOM_APP) ? 15 : 0; coeff_termfrequency = 14; coeff_urlcompintoplist = 3; coeff_descrcompintoplist = 2; @@ -114,7 +116,7 @@ public class plasmaSearchRankingProfile { } public plasmaSearchRankingProfile(final String prefix, final String profile) { - this(plasmaSearchQuery.CONTENTDOM_TEXT); // set defaults + this(Query.CONTENTDOM_TEXT); // set defaults if ((profile != null) && (profile.length() > 0)) { //parse external form final HashMap coeff = new HashMap(); diff --git a/source/de/anomic/plasma/plasmaSnippetCache.java b/source/de/anomic/plasma/plasmaSnippetCache.java index 280b2f7bf..73bb1ffd8 100644 --- a/source/de/anomic/plasma/plasmaSnippetCache.java +++ b/source/de/anomic/plasma/plasmaSnippetCache.java @@ -51,6 +51,7 @@ import de.anomic.plasma.parser.Document; import de.anomic.plasma.parser.ParserException; import de.anomic.plasma.parser.Word; import de.anomic.plasma.parser.Condenser; +import de.anomic.search.Query; import de.anomic.yacy.yacySearch; import de.anomic.yacy.yacyURL; @@ -707,10 +708,10 @@ public class plasmaSnippetCache { final plasmaParserDocument document = retrieveDocument(url, fetchOnline, timeout, false, reindexing); final ArrayList a = new ArrayList(); if (document != null) { - if ((mediatype == plasmaSearchQuery.CONTENTDOM_ALL) || (mediatype == plasmaSearchQuery.CONTENTDOM_AUDIO)) a.addAll(computeMediaSnippets(document, queryhashes, plasmaSearchQuery.CONTENTDOM_AUDIO)); - if ((mediatype == plasmaSearchQuery.CONTENTDOM_ALL) || (mediatype == plasmaSearchQuery.CONTENTDOM_VIDEO)) a.addAll(computeMediaSnippets(document, queryhashes, plasmaSearchQuery.CONTENTDOM_VIDEO)); - if ((mediatype == plasmaSearchQuery.CONTENTDOM_ALL) || (mediatype == plasmaSearchQuery.CONTENTDOM_APP)) a.addAll(computeMediaSnippets(document, queryhashes, plasmaSearchQuery.CONTENTDOM_APP)); - if ((mediatype == plasmaSearchQuery.CONTENTDOM_ALL) || (mediatype == plasmaSearchQuery.CONTENTDOM_IMAGE)) a.addAll(computeImageSnippets(document, queryhashes)); + if ((mediatype == Query.CONTENTDOM_ALL) || (mediatype == Query.CONTENTDOM_AUDIO)) a.addAll(computeMediaSnippets(document, queryhashes, Query.CONTENTDOM_AUDIO)); + if ((mediatype == Query.CONTENTDOM_ALL) || (mediatype == Query.CONTENTDOM_VIDEO)) a.addAll(computeMediaSnippets(document, queryhashes, Query.CONTENTDOM_VIDEO)); + if ((mediatype == Query.CONTENTDOM_ALL) || (mediatype == Query.CONTENTDOM_APP)) a.addAll(computeMediaSnippets(document, queryhashes, Query.CONTENTDOM_APP)); + if ((mediatype == Query.CONTENTDOM_ALL) || (mediatype == Query.CONTENTDOM_IMAGE)) a.addAll(computeImageSnippets(document, queryhashes)); } return a; } @@ -719,9 +720,9 @@ public class plasmaSnippetCache { if (document == null) return new ArrayList(); Map media = null; - if (mediatype == plasmaSearchQuery.CONTENTDOM_AUDIO) media = document.getAudiolinks(); - else if (mediatype == plasmaSearchQuery.CONTENTDOM_VIDEO) media = document.getVideolinks(); - else if (mediatype == plasmaSearchQuery.CONTENTDOM_APP) media = document.getApplinks(); + if (mediatype == Query.CONTENTDOM_AUDIO) media = document.getAudiolinks(); + else if (mediatype == Query.CONTENTDOM_VIDEO) media = document.getVideolinks(); + else if (mediatype == Query.CONTENTDOM_APP) media = document.getApplinks(); if (media == null) return null; final Iterator> i = media.entrySet().iterator(); @@ -767,13 +768,13 @@ public class plasmaSnippetCache { s = removeAppearanceHashes(url.toNormalform(false, false), queryhashes); if (s.size() == 0) { final int ranking = ientry.hashCode(); - result.add(new MediaSnippet(plasmaSearchQuery.CONTENTDOM_IMAGE, url, desc, ientry.width() + " x " + ientry.height(), ranking, document.dc_source())); + result.add(new MediaSnippet(Query.CONTENTDOM_IMAGE, url, desc, ientry.width() + " x " + ientry.height(), ranking, document.dc_source())); continue; } s = removeAppearanceHashes(desc, s); if (s.size() == 0) { final int ranking = ientry.hashCode(); - result.add(new MediaSnippet(plasmaSearchQuery.CONTENTDOM_IMAGE, url, desc, ientry.width() + " x " + ientry.height(), ranking, document.dc_source())); + result.add(new MediaSnippet(Query.CONTENTDOM_IMAGE, url, desc, ientry.width() + " x " + ientry.height(), ranking, document.dc_source())); continue; } } diff --git a/source/de/anomic/plasma/plasmaSwitchboard.java b/source/de/anomic/plasma/plasmaSwitchboard.java index 86c03b1a1..6a777739b 100644 --- a/source/de/anomic/plasma/plasmaSwitchboard.java +++ b/source/de/anomic/plasma/plasmaSwitchboard.java @@ -164,6 +164,7 @@ import de.anomic.plasma.parser.Document; import de.anomic.plasma.parser.ParserException; import de.anomic.plasma.parser.Word; import de.anomic.plasma.parser.Condenser; +import de.anomic.search.Query; import de.anomic.server.serverAbstractSwitch; import de.anomic.server.serverBusyThread; import de.anomic.server.serverCore; @@ -246,8 +247,8 @@ public final class plasmaSwitchboard extends serverAbstractSwitch localSearches; // array of search result properties as HashMaps - public ArrayList remoteSearches; // array of search result properties as HashMaps + public ArrayList localSearches; // array of search result properties as HashMaps + public ArrayList remoteSearches; // array of search result properties as HashMaps public ConcurrentHashMap> localSearchTracker, remoteSearchTracker; // mappings from requesting host to a TreeSet of Long(access time) public long indexedPages = 0; public double requestedQueries = 0d; @@ -554,8 +555,8 @@ public final class plasmaSwitchboard extends serverAbstractSwitch>(); // String:TreeSet - IP:set of Long(accessTime) this.remoteSearchTracker = new ConcurrentHashMap>(); - this.localSearches = new ArrayList(); // contains search result properties as HashMaps - this.remoteSearches = new ArrayList(); + this.localSearches = new ArrayList(); // contains search result properties as HashMaps + this.remoteSearches = new ArrayList(); // init messages: clean up message symbol final File notifierSource = new File(getRootPath(), getConfig(plasmaSwitchboardConstants.HTROOT_PATH, plasmaSwitchboardConstants.HTROOT_PATH_DEFAULT) + "/env/grafics/empty.gif"); @@ -1019,7 +1020,7 @@ public final class plasmaSwitchboard extends serverAbstractSwitch queryHashes, final TreeSet excludeHashes, final TreeSet fullqueryHashes, @@ -348,7 +349,7 @@ public final class plasmaSearchQuery { * @param addToQuery * @return */ - public static String navurl(String ext, final int page, final int display, final plasmaSearchQuery theQuery, final String originalUrlMask, String addToQuery, String nav) { + public static String navurl(String ext, final int page, final int display, final Query theQuery, final String originalUrlMask, String addToQuery, String nav) { return "/yacysearch." + ext + "?display=" + display + "&search=" + theQuery.queryString(true) + ((addToQuery == null) ? "" : "+" + addToQuery) + diff --git a/source/de/anomic/yacy/yacySearch.java b/source/de/anomic/yacy/yacySearch.java index 91c01a3c4..235f5f449 100644 --- a/source/de/anomic/yacy/yacySearch.java +++ b/source/de/anomic/yacy/yacySearch.java @@ -37,9 +37,9 @@ import de.anomic.kelondro.order.Bitfield; import de.anomic.kelondro.text.Segment; import de.anomic.kelondro.util.ScoreCluster; import de.anomic.kelondro.util.Log; -import de.anomic.plasma.plasmaSearchQuery; import de.anomic.plasma.plasmaSearchRankingProcess; import de.anomic.plasma.plasmaSearchRankingProfile; +import de.anomic.search.Query; import de.anomic.yacy.dht.PeerSelection; public class yacySearch extends Thread { @@ -270,7 +270,7 @@ public class yacySearch extends Thread { (clusterselection == null) ? selectSearchTargets( peers, - plasmaSearchQuery.hashes2Set(wordhashes), + Query.hashes2Set(wordhashes), targets, peers.redundancy()) : selectClusterPeers(peers, clusterselection);