public final class search {
public static serverObjects respond(final RequestHeader header, final serverObjects post, final serverSwitch env) {
// return variable that accumulates replacements
final Switchboard sb = (Switchboard) env;
sb.remoteSearchLastAccess = System.currentTimeMillis();
final serverObjects prop = new serverObjects();
// set nice default values for error cases
prop.put("searchtime", "0");
prop.put("references", "");
prop.put("joincount", "0");
prop.put("linkcount", "0");
prop.put("links", "");
prop.put("indexcount", "");
prop.put("indexabstract", "");
prop.put("fwhop", ""); // hops (depth) of forwards that had been performed to construct this result
prop.put("fwsrc", ""); // peers that helped to construct this result
prop.put("fwrec", ""); // peers that would have helped to construct this result (recommendations)
if (post == null || env == null) return prop;
if (!Protocol.authentifyRequest(post, env)) return prop;
final String client = header.get(HeaderFramework.CONNECTION_PROP_CLIENTIP);
//System.out.println("yacy: search received request = " + post.toString());
final String oseed = post.get("myseed", ""); // complete seed of the requesting peer
// final String youare = post.get("youare", ""); // seed hash of the target peer, used for testing network stability
final String key = post.get("key", ""); // transmission key for response
final String query = post.get("query", ""); // a string of word hashes that shall be searched and combined
final String exclude= post.get("exclude", "");// a string of word hashes that shall not be within the search result
final String urls = post.get("urls", ""); // a string of url hashes that are preselected for the search: no other may be returned
final String abstracts = post.get("abstracts", ""); // a string of word hashes for abstracts that shall be generated, or 'auto' (for maxcount-word), or '' (for none)
// final String fwdep = post.get("fwdep", ""); // forward depth. if "0" then peer may NOT ask another peer for more results
// final String fwden = post.get("fwden", ""); // forward deny, a list of seed hashes. They may NOT be target of forward hopping
final int count = Math.min((int) sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXCOUNT_DEFAULT, 100), post.getInt("count", 10)); // maximum number of wanted results
final long maxtime = Math.min((int) sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXTIME_DEFAULT, 3000), post.getLong("time", 3000)); // maximum waiting time
final int maxdist= post.getInt("maxdist", Integer.MAX_VALUE);
final String prefer = post.get("prefer", "");
final String modifier = post.get("modifier", "").trim();
final String contentdom = post.get("contentdom", "text");
final String filter = post.get("filter", ".*"); // a filter on the url
final Pattern snippetPattern = Pattern.compile(post.get("snippet", ".*")); // a filter on the snippet
String sitehash = post.get("sitehash", ""); if (sitehash.length() == 0) sitehash = null;
String authorhash = post.get("authorhash", ""); if (authorhash.length() == 0) authorhash = null;
String language = post.get("language", "");
if (language == null || language.length() == 0 || !ISO639.exists(language)) {
// take language from the user agent
String agent = header.get("User-Agent");
if (agent == null) agent = System.getProperty("user.language");
language = (agent == null) ? "en" : ISO639.userAgentLanguageDetection(agent);
if (language == null) language = "en";
}
final int partitions = post.getInt("partitions", 30);
String profile = post.get("profile", ""); // remote profile hand-over
if (profile.length() > 0) profile = crypt.simpleDecode(profile, null);
//final boolean includesnippet = post.get("includesnippet", "false").equals("true");
Bitfield constraint = ((post.containsKey("constraint")) && (post.get("constraint", "").length() > 0)) ? new Bitfield(4, post.get("constraint", "______")) : null;
if (constraint != null) {
// check bad handover parameter from older versions
boolean allon = true;
for (int i = 0; i < 32; i++) {
if (!constraint.get(i)) {allon = false; break;}
}
if (allon) constraint = null;
}
// final boolean global = ((String) post.get("resource", "global")).equals("global"); // if true, then result may consist of answers from other peers
// Date remoteTime = yacyCore.parseUniversalDate((String) post.get(yacySeed.MYTIME)); // read remote time
// test:
// http://localhost:8090/yacy/search.html?query=4galTpdpDM5Q (search for linux)
// http://localhost:8090/yacy/search.html?query=gh8DKIhGKXws (search for book)
// http://localhost:8090/yacy/search.html?query=UEhMGfGv2vOE (search for kernel)
// http://localhost:8090/yacy/search.html?query=ZX-LjaYo74PP (search for help)
// http://localhost:8090/yacy/search.html?query=uDqIalxDfM2a (search for mail)
// http://localhost:8090/yacy/search.html?query=4galTpdpDM5Qgh8DKIhGKXws&abstracts=auto (search for linux and book, generate abstract automatically)
// http://localhost:8090/yacy/search.html?query=&abstracts=4galTpdpDM5Q (only abstracts for linux)
if (sb.isRobinsonMode() && !sb.isPublicRobinson()) {
// if we are a robinson cluster, answer only if this client is known by our network definition
return prop;
}
// check the search tracker
TreeSet<Long> trackerHandles = sb.remoteSearchTracker.get(client);
if (trackerHandles == null) trackerHandles = new TreeSet<Long>();
boolean block = false;
synchronized (trackerHandles) {
if (trackerHandles.tailSet(Long.valueOf(System.currentTimeMillis() - 3000)).size() > 1) {
block = true;
}
}
if (!block) synchronized (trackerHandles) {
if (trackerHandles.tailSet(Long.valueOf(System.currentTimeMillis() - 60000)).size() > 12) {
block = true;
}
}
if (!block) synchronized (trackerHandles) {
if (trackerHandles.tailSet(Long.valueOf(System.currentTimeMillis() - 600000)).size() > 36) {
block = true;
}
}
if (block && Domains.isLocal(client, null)) block = false; // check isLocal here to prevent dns lookup for client
if (block) {
return prop;
}
// tell all threads to do nothing for a specific time
sb.intermissionAllThreads(100);
EventTracker.delete(EventTracker.EClass.SEARCH);
final HandleSet abstractSet = (abstracts.length() == 0 || abstracts.equals("auto")) ? null : QueryParams.hashes2Set(abstracts);
// store accessing peer
Seed remoteSeed;
try {
remoteSeed = Seed.genRemoteSeed(oseed, key, false, client);
} catch (final IOException e) {
Network.log.logInfo("yacy.search: access with bad seed: " + e.getMessage());
remoteSeed = null;
}
if (sb.peers == null) {
Network.log.logSevere("yacy.search: seed cache not initialized");
} else {
sb.peers.peerActions.peerArrival(remoteSeed, true);
}
// prepare search
final HandleSet queryhashes = QueryParams.hashes2Set(query);
final HandleSet excludehashes = (exclude.length() == 0) ? new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0) : QueryParams.hashes2Set(exclude);
final long timestamp = System.currentTimeMillis();
// prepare a search profile
final RankingProfile rankingProfile = (profile.length() == 0) ? new RankingProfile(ContentDomain.contentdomParser(contentdom)) : new RankingProfile("", profile);
// prepare an abstract result
final StringBuilder indexabstract = new StringBuilder(6000);
int indexabstractContainercount = 0;
int joincount = 0;
QueryParams theQuery = null;
SearchEvent theSearch = null;
ArrayList<WeakPriorityBlockingQueue.Element<ResultEntry>> accu = null;
if (query.length() == 0 && abstractSet != null) {
// this is _not_ a normal search, only a request for index abstracts
final Segment indexSegment = sb.indexSegments.segment(Segments.Process.PUBLIC);
theQuery = new QueryParams(
null,
abstractSet,
new HandleSet(WordReferenceRow.urlEntryRow.primaryKeyLength, WordReferenceRow.urlEntryRow.objectOrder, 0),
null,
snippetPattern,
null,
modifier,
maxdist,
prefer,
ContentDomain.contentdomParser(contentdom),
language,
"", // no navigation
CacheStrategy.CACHEONLY,
count,
0,
filter,
QueryParams.Searchdom.LOCAL,
-1,
null,
false,
sitehash,
authorhash,
DigestURI.TLD_any_zone_filter,
client,
false,
indexSegment,
rankingProfile,
header.get(RequestHeader.USER_AGENT, ""),
false
);
Network.log.logInfo("INIT HASH SEARCH (abstracts only): " + QueryParams.anonymizedQueryHashes(theQuery.queryHashes) + " - " + theQuery.displayResults() + " links");
final long timer = System.currentTimeMillis();
//final Map<byte[], ReferenceContainer<WordReference>>[] containers = sb.indexSegment.index().searchTerm(theQuery.queryHashes, theQuery.excludeHashes, plasmaSearchQuery.hashes2StringSet(urls));
final TreeMap<byte[], ReferenceContainer<WordReference>> incc = indexSegment.termIndex().searchConjunction(theQuery.queryHashes, QueryParams.hashes2Handles(urls));
EventTracker.update(EventTracker.EClass.SEARCH, new ProfilingGraph.EventSearch(theQuery.id(true), SearchEvent.Type.COLLECTION, "", incc.size(), System.currentTimeMillis() - timer), false);
if (incc != null) {
final Iterator<Map.Entry<byte[], ReferenceContainer<WordReference>>> ci = incc.entrySet().iterator();
Map.Entry<byte[], ReferenceContainer<WordReference>> entry;
byte[] wordhash;
while (ci.hasNext()) {
entry = ci.next();
wordhash = entry.getKey();
final ReferenceContainer<WordReference> container = entry.getValue();
indexabstractContainercount += container.size();
indexabstract.append("indexabstract.");
indexabstract.append(ASCII.String(wordhash));
indexabstract.append("=");
indexabstract.append(WordReferenceFactory.compressIndex(container, null, 1000).toString());
indexabstract.append(serverCore.CRLF_STRING);
}
}
prop.put("indexcount", "");
prop.put("joincount", "0");
prop.put("references", "");
} else {
// retrieve index containers from search request
theQuery = new QueryParams(
null,
queryhashes,
excludehashes,
null,
snippetPattern,
null,
modifier,
maxdist,
prefer,
ContentDomain.contentdomParser(contentdom),
language,
"", // no navigation
CacheStrategy.CACHEONLY,
count,
0,
filter,
QueryParams.Searchdom.LOCAL,
-1,
constraint,
false,
sitehash,
authorhash,
DigestURI.TLD_any_zone_filter,
client,
false,
sb.indexSegments.segment(Segments.Process.PUBLIC),
rankingProfile,
header.get(RequestHeader.USER_AGENT, ""),
false
);
Network.log.logInfo("INIT HASH SEARCH (query-" + abstracts + "): " + QueryParams.anonymizedQueryHashes(theQuery.queryHashes) + " - " + theQuery.displayResults() + " links");
EventChannel.channels(EventChannel.REMOTESEARCH).addMessage(new RSSMessage("Remote Search Request from " + ((remoteSeed == null) ? "unknown" : remoteSeed.getName()), QueryParams.anonymizedQueryHashes(theQuery.queryHashes), ""));
// make event
theSearch = SearchEventCache.getEvent(theQuery, sb.peers, sb.tables, null, abstracts.length() > 0, sb.loader, count, maxtime, (int) sb.getConfigLong(SwitchboardConstants.DHT_BURST_ROBINSON, 0), (int) sb.getConfigLong(SwitchboardConstants.DHT_BURST_MULTIWORD, 0));
// set statistic details of search result and find best result index set
joincount = theSearch.getRankingResult().getLocalIndexCount() - theSearch.getRankingResult().getMissCount() - theSearch.getRankingResult().getSortOutCount();
prop.put("joincount", Integer.toString(joincount));
if (joincount != 0) {