Examples of BitField


Examples of net.yacy.kelondro.order.Bitfield

        // if addMedia == true, then all the media links are also parsed and added to the words
        // added media words are flagged with the appropriate media flag
        this.intStringFormatter.setMinimumIntegerDigits(numlength);
        this.intStringFormatter.setMaximumIntegerDigits(numlength);
        this.words = new HashMap<String, Word>();
        this.RESULT_FLAGS = new Bitfield(4);

        // construct flag set for document
        if (!document.getImages().isEmpty())     RESULT_FLAGS.set(flag_cat_hasimage, true);
        if (!document.getAudiolinks().isEmpty()) RESULT_FLAGS.set(flag_cat_hasaudio, true);
        if (!document.getVideolinks().isEmpty()) RESULT_FLAGS.set(flag_cat_hasvideo, true);
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

     * @return a ranking: the higher the number, the better is the ranking
     */
    public long cardinal(final WordReferenceVars t) {
        //return Long.MAX_VALUE - preRanking(ranking, iEntry, this.entryMin, this.entryMax, this.searchWords);
        // the normalizedEntry must be a normalized indexEntry
        final Bitfield flags = t.flags();
        assert this.min != null;
        assert this.max != null;
        assert t != null;
        assert this.ranking != null;
        final long tf = ((this.max.termFrequency() == this.min.termFrequency()) ? 0 : (((int)(((t.termFrequency()-this.min.termFrequency())*256.0)/(this.max.termFrequency() - this.min.termFrequency())))) << this.ranking.coeff_termfrequency);
        //System.out.println("tf(" + t.urlHash + ") = " + Math.floor(1000 * t.termFrequency()) + ", min = " + Math.floor(1000 * min.termFrequency()) + ", max = " + Math.floor(1000 * max.termFrequency()) + ", tf-normed = " + tf);
        final int maxmaxpos = this.max.maxposition();
        final int minminpos = this.min.minposition();
        final long r =
             ((256 - DigestURI.domLengthNormalized(t.urlhash())) << this.ranking.coeff_domlength)
           + ((this.ranking.coeff_ybr > 12) ? ((256 - (BlockRank.ranking(t.urlhash()) << 4)) << this.ranking.coeff_ybr) : 0)
           + ((this.max.urlcomps()      == this.min.urlcomps()   )   ? 0 : (256 - (((t.urlcomps()     - this.min.urlcomps()     ) << 8) / (this.max.urlcomps()     - this.min.urlcomps())     )) << this.ranking.coeff_urlcomps)
           + ((this.max.urllength()     == this.min.urllength()  )   ? 0 : (256 - (((t.urllength()    - this.min.urllength()    ) << 8) / (this.max.urllength()    - this.min.urllength())    )) << this.ranking.coeff_urllength)
                                        + ((maxmaxpos == minminpos) ? 0 : (256 - (((t.minposition() - minminpos) << 8) / (maxmaxpos - minminpos))) << this.ranking.coeff_posintext)
           + ((this.max.posofphrase()   == this.min.posofphrase())   ? 0 : (256 - (((t.posofphrase()  - this.min.posofphrase()  ) << 8) / (this.max.posofphrase()  - this.min.posofphrase())  )) << this.ranking.coeff_posofphrase)
           + ((this.max.posinphrase()   == this.min.posinphrase())   ? 0 : (256 - (((t.posinphrase()  - this.min.posinphrase()  ) << 8) / (this.max.posinphrase()  - this.min.posinphrase())  )) << this.ranking.coeff_posinphrase)
           + ((this.max.distance()      == this.min.distance()   )   ? 0 : (256 - (((t.distance()     - this.min.distance()     ) << 8) / (this.max.distance()     - this.min.distance())     )) << this.ranking.coeff_worddistance)
           + ((this.max.virtualAge()    == this.min.virtualAge())    ? 0 :        (((t.virtualAge()   - this.min.virtualAge()   ) << 8) / (this.max.virtualAge()   - this.min.virtualAge())    ) << this.ranking.coeff_date)
           + ((this.max.wordsintitle()  == this.min.wordsintitle())  ? 0 : (((t.wordsintitle() - this.min.wordsintitle()  ) << 8) / (this.max.wordsintitle() - this.min.wordsintitle())  ) << this.ranking.coeff_wordsintitle)
           + ((this.max.wordsintext()   == this.min.wordsintext())   ? 0 : (((t.wordsintext()  - this.min.wordsintext()   ) << 8) / (this.max.wordsintext()  - this.min.wordsintext())   ) << this.ranking.coeff_wordsintext)
           + ((this.max.phrasesintext() == this.min.phrasesintext()) ? 0 : (((t.phrasesintext()- this.min.phrasesintext() ) << 8) / (this.max.phrasesintext()- this.min.phrasesintext()) ) << this.ranking.coeff_phrasesintext)
           + ((this.max.llocal()        == this.min.llocal())        ? 0 : (((t.llocal()       - this.min.llocal()        ) << 8) / (this.max.llocal()       - this.min.llocal())        ) << this.ranking.coeff_llocal)
           + ((this.max.lother()        == this.min.lother())        ? 0 : (((t.lother()       - this.min.lother()        ) << 8) / (this.max.lother()       - this.min.lother())        ) << this.ranking.coeff_lother)
           + ((this.max.hitcount()      == this.min.hitcount())      ? 0 : (((t.hitcount()     - this.min.hitcount()      ) << 8) / (this.max.hitcount()     - this.min.hitcount())      ) << this.ranking.coeff_hitcount)
           + tf
           + ((this.ranking.coeff_authority > 12) ? (authority(t.hosthash()) << this.ranking.coeff_authority) : 0)
           + ((flags.get(WordReferenceRow.flag_app_dc_identifier))  ? 255 << this.ranking.coeff_appurl             : 0)
           + ((flags.get(WordReferenceRow.flag_app_dc_title))       ? 255 << this.ranking.coeff_app_dc_title       : 0)
           + ((flags.get(WordReferenceRow.flag_app_dc_creator))     ? 255 << this.ranking.coeff_app_dc_creator     : 0)
           + ((flags.get(WordReferenceRow.flag_app_dc_subject))     ? 255 << this.ranking.coeff_app_dc_subject     : 0)
           + ((flags.get(WordReferenceRow.flag_app_dc_description)) ? 255 << this.ranking.coeff_app_dc_description : 0)
           + ((flags.get(WordReferenceRow.flag_app_emphasized))     ? 255 << this.ranking.coeff_appemph            : 0)
           + ((flags.get(Condenser.flag_cat_indexof))      ? 255 << this.ranking.coeff_catindexof         : 0)
           + ((flags.get(Condenser.flag_cat_hasimage))     ? 255 << this.ranking.coeff_cathasimage        : 0)
           + ((flags.get(Condenser.flag_cat_hasaudio))     ? 255 << this.ranking.coeff_cathasaudio        : 0)
           + ((flags.get(Condenser.flag_cat_hasvideo))     ? 255 << this.ranking.coeff_cathasvideo        : 0)
           + ((flags.get(Condenser.flag_cat_hasapp))       ? 255 << this.ranking.coeff_cathasapp          : 0)
           + ((ByteBuffer.equals(t.language, this.language)) ? 255 << this.ranking.coeff_language           : 0)
           + ((DigestURI.probablyRootURL(t.urlhash())) 15 << this.ranking.coeff_urllength          : 0);

        //if (searchWords != null) r += (yacyURL.probablyWordURL(t.urlHash(), searchWords) != null) ? 256 << ranking.coeff_appurl : 0;

View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

                            } catch (final NumberFormatException e) {
                                Log.logSevere("kelondroRow", "NumberFormatException for celltype_binary; row = " + i + ", celltype = " + col.celltype + ", encoder = " + col.encoder + ", value = '" + elts[i].substring(p + 1).trim() + "'");
                                setCol(clstrt, col.cellwidth, new byte[]{0});
                            }
                        } else if ((decimalCardinal) && (col.celltype == Column.celltype_bitfield)) {
                            setCol(clstrt, col.cellwidth, (new Bitfield(col.cellwidth, elts[i].substring(p + 1).trim())).bytes());
                        } else {
                            setCol(clstrt, col.cellwidth, UTF8.getBytes(elts[i].substring(p + 1).trim()));
                        }
                    }
                }
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

                bb.append(propertySymbol);
                if (quotes) bb.append('"');
                if ((decimalCardinal) && (row[i].celltype == Column.celltype_cardinal)) {
                    bb.append(Long.toString(getColLong(i)));
                } else if ((decimalCardinal) && (row[i].celltype == Column.celltype_bitfield)) {
                    bb.append((new Bitfield(getColBytes(i, true))).exportB64());
                } else if ((decimalCardinal) && (row[i].celltype == Column.celltype_binary)) {
                    assert row[i].cellwidth == 1 : toString();
                    bb.append(Integer.toString((0xff & getColByte(i))));
                } else {
                    bb.append(rowinstance, offset + colstart[i], row[i].cellwidth);
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

    public int urlcomps() {
        return (int) this.entry.getColLong(col_urlComps);
    }
   
    public Bitfield flags() {
        return new Bitfield(this.entry.getColBytes(col_flags, false));
    }
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

     * @param args
     */
    public static void main(final String[] args) {
        try {
            final DigestURI url = new DigestURI("http", "www.yacy.net", 80, "/");
            final URIMetadataRow urlRef = new URIMetadataRow(url, "YaCy Homepage", "", "", "", 0.0f, 0.0f, new Date(), new Date(), new Date(), "", new byte[] {}, 123, 42, '?', new Bitfield(), UTF8.getBytes("de"), 0, 0, 0, 0, 0, 0);
            EventOrigin stackNo = EventOrigin.LOCAL_CRAWLING;
            System.out.println("valid test:\n=======");
            // add
            stack(urlRef, urlRef.hash(), url.hash(), stackNo);
            // size
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

        this.appdate       = (appdate == null) ? 0 : appdate.getTime();
        this.profileHandle = profileHandle; // must not be null
        this.depth         = depth;
        this.anchors       = anchors;
        this.forkfactor    = forkfactor;
        this.flags         = new Bitfield(rowdef.width(10));
        this.statusMessage = "loaded(args)";
        this.initialHash   = url.hashCode();
        this.status        = WorkflowJob.STATUS_INITIATED;
        this.size          = size;
    }
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

        this.appdate = entry.getColLong(5);
        this.profileHandle = (entry.empty(6)) ? null : entry.getColString(6).trim();
        this.depth = (int) entry.getColLong(7);
        this.anchors = (int) entry.getColLong(8);
        this.forkfactor = (int) entry.getColLong(9);
        this.flags = new Bitfield(entry.getColBytes(10, true));
        //this.loaddate = entry.getColLong(12);
        //this.lastmodified = entry.getColLong(13);
        this.size = entry.getColLong(14);
        this.statusMessage        = "loaded(kelondroRow.Entry)";
        this.initialHash   = url.hashCode();
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

        }
        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
View Full Code Here

Examples of net.yacy.kelondro.order.Bitfield

        String prefermask = (post == null) ? "" : post.get("prefermaskfilter", "");
        if (!prefermask.isEmpty() && prefermask.indexOf(".*",0) < 0) {
            prefermask = ".*" + prefermask + ".*";
        }

        Bitfield constraint = (post != null && post.containsKey("constraint") && !post.get("constraint", "").isEmpty()) ? new Bitfield(4, post.get("constraint", "______")) : null;
        if (indexof) {
            constraint = new Bitfield(4);
            constraint.set(Condenser.flag_cat_indexof, true);
        }

        // SEARCH
        final boolean clustersearch = sb.isRobinsonMode() && (sb.getConfig("cluster.mode", "").equals("privatecluster") || sb.getConfig("cluster.mode", "").equals("publiccluster"));
        final boolean indexReceiveGranted = sb.getConfigBool(SwitchboardConstants.INDEX_RECEIVE_ALLOW, true) || sb.getConfigBool(SwitchboardConstants.INDEX_RECEIVE_AUTODISABLED, true) || clustersearch;
        global = global && indexReceiveGranted; // if the user does not want indexes from remote peers, it cannot be a global searchnn

        // increase search statistic counter
        if (!global) {
            // we count only searches on the local peer here, because global searches
            // are counted on the target peer to preserve privacy of the searcher
            if (authenticated) {
                // local or authenticated search requests are counted separately
                // because they are not part of a public available peer statistic
                sb.searchQueriesRobinsonFromLocal++;
            } else {
                // robinson-searches from non-authenticated requests are public
                // and may be part of the public available statistic
                sb.searchQueriesRobinsonFromRemote++;
            }
        }

        // find search domain
        final ContentDomain contentdom = ContentDomain.contentdomParser(post == null ? "text" : post.get("contentdom", "text"));

        // patch until better search profiles are available
        if (contentdom == ContentDomain.TEXT) {
            if (maximumRecords > 50) maximumRecords = 10;
        } else {
            if (maximumRecords <= 32) maximumRecords = 64;
        }

        // check the search tracker
        TreeSet<Long> trackerHandles = sb.localSearchTracker.get(client);
        if (trackerHandles == null) {
            trackerHandles = new TreeSet<Long>();
        }
        boolean block = false;
        if (Domains.matchesList(client, sb.networkBlacklist)) {
            global = false;
            if (snippetFetchStrategy != null) {
                snippetFetchStrategy = null;
            }
            block = true;
            Log.logWarning("LOCAL_SEARCH", "ACCESS CONTROL: BLACKLISTED CLIENT FROM " + client + " gets no permission to search");
        } else if (Domains.matchesList(client, sb.networkWhitelist)) {
            Log.logInfo("LOCAL_SEARCH", "ACCESS CONTROL: WHITELISTED CLIENT FROM " + client + " gets no search restrictions");
        } else if (!authenticated && !localhostAccess) {
            // in case that we do a global search or we want to fetch snippets, we check for DoS cases
            synchronized (trackerHandles) {
                final int accInThreeSeconds = trackerHandles.tailSet(Long.valueOf(System.currentTimeMillis() - 3000)).size();
                final int accInOneMinute = trackerHandles.tailSet(Long.valueOf(System.currentTimeMillis() - 60000)).size();
                final int accInTenMinutes = trackerHandles.tailSet(Long.valueOf(System.currentTimeMillis() - 600000)).size();
                // protections against too strong YaCy network load, reduces remote search
                if (global) {
                    if (accInTenMinutes >= 60 || accInOneMinute >= 6 || accInThreeSeconds >= 1) {
                        global = false;
                        Log.logWarning("LOCAL_SEARCH", "ACCESS CONTROL: CLIENT FROM " + client + ": " + accInThreeSeconds + "/3s, " + accInOneMinute + "/60s, " + accInTenMinutes + "/600s, " + " requests, disallowed global search");
                    }
                }
                // protection against too many remote server snippet loads (protects traffic on server)
                if (snippetFetchStrategy != null && snippetFetchStrategy.isAllowedToFetchOnline()) {
                    if (accInTenMinutes >= 20 || accInOneMinute >= 4 || accInThreeSeconds >= 1) {
                        snippetFetchStrategy = CacheStrategy.CACHEONLY;
                        Log.logWarning("LOCAL_SEARCH", "ACCESS CONTROL: CLIENT FROM " + client + ": " + accInThreeSeconds + "/3s, " + accInOneMinute + "/60s, " + accInTenMinutes + "/600s, " + " requests, disallowed remote snippet loading");
                    }
                }
                // general load protection
                if (accInTenMinutes >= 3000 || accInOneMinute >= 600 || accInThreeSeconds >= 60) {
                    block = true;
                    Log.logWarning("LOCAL_SEARCH", "ACCESS CONTROL: CLIENT FROM " + client + ": " + accInThreeSeconds + "/3s, " + accInOneMinute + "/60s, " + accInTenMinutes + "/600s, " + " requests, disallowed search");
                }
            }
        }

        if (!block && (post == null || post.get("cat", "href").equals("href"))) {
            String urlmask = null;

            // check available memory and clean up if necessary
            if (!MemoryControl.request(8000000L, false)) {
                indexSegment.urlMetadata().clearCache();
                SearchEventCache.cleanupEvents(true);
            }

            final RankingProfile ranking = sb.getRanking();
            final StringBuilder modifier = new StringBuilder(20);

            if (querystring.indexOf("/near",0) >= 0) {
              querystring = querystring.replace("/near", "");
              ranking.coeff_worddistance = RankingProfile.COEFF_MAX;
              modifier.append("/near ");
            }
            if (querystring.indexOf("/date",0) >= 0) {
                querystring = querystring.replace("/date", "");
                ranking.coeff_date = RankingProfile.COEFF_MAX;
                modifier.append("/date ");
            }
            if (querystring.indexOf("/http",0) >= 0) {
                querystring = querystring.replace("/http", "");
                urlmask = "https?://.*";
                modifier.append("/http ");
            }
            if (querystring.indexOf("/https",0) >= 0) {
                querystring = querystring.replace("/https", "");
                urlmask = "https?://.*";
                modifier.append("/https ");
            }
            if (querystring.indexOf("/ftp",0) >= 0) {
                querystring = querystring.replace("/ftp", "");
                urlmask = "ftp://.*";
                modifier.append("/ftp ");
            }
            if (querystring.indexOf("/smb",0) >= 0) {
                querystring = querystring.replace("/smb", "");
                urlmask = "smb://.*";
                modifier.append("/smb ");
            }
            if (querystring.indexOf("/file",0) >= 0) {
                querystring = querystring.replace("/file", "");
                urlmask = "file://.*";
                modifier.append("/file ");
            }
            if (querystring.indexOf("/location",0) >= 0) {
                querystring = querystring.replace("/location", "");
                if (constraint == null) {
                    constraint = new Bitfield(4);
                }
                constraint.set(Condenser.flag_cat_haslocation, true);
                modifier.append("/location ");
            }
            final int lrp = querystring.indexOf("/language/",0);
            String language = "";
            if (lrp >= 0) {
                if (querystring.length() >= (lrp + 12)) {
                    language = querystring.substring(lrp + 10, lrp + 12);
                }
                querystring = querystring.replace("/language/" + language, "");
                language = language.toLowerCase();
                modifier.append("/language/").append(language).append(" ");
            }
            final int inurl = querystring.indexOf("inurl:",0);
            if (inurl >= 0) {
                int ftb = querystring.indexOf(' ', inurl);
                if (ftb == -1) {
                    ftb = querystring.length();
                }
                final String urlstr = querystring.substring(inurl + 6, ftb);
                querystring = querystring.replace("inurl:" + urlstr, "");
                if (!urlstr.isEmpty()) {
                    urlmask = urlmask == null ? ".*" + urlstr + ".*" : urlmask + urlstr + ".*";
                }
                modifier.append("inurl:").append(urlstr).append(" ");
            }
            final int filetype = querystring.indexOf("filetype:",0);
            if (filetype >= 0) {
                int ftb = querystring.indexOf(' ', filetype);
                if (ftb == -1) {
                    ftb = querystring.length();
                }
                String ft = querystring.substring(filetype + 9, ftb);
                querystring = querystring.replace("filetype:" + ft, "");
                while (!ft.isEmpty() && ft.charAt(0) == '.') ft = ft.substring(1);
                if (!ft.isEmpty()) {
                    if (urlmask == null) {
                        urlmask = ".*\\." + ft;
                    } else {
                        urlmask = urlmask + ".*\\." + ft;
                    }
                }
                modifier.append("filetype:").append(ft).append(" ");
            }
            String tenant = null;
            if (post.containsKey("tenant")) {
                tenant = post.get("tenant");
                if (tenant != null && tenant.isEmpty()) {
                    tenant = null;
                }
                if (tenant != null) {
                    if (urlmask == null) {
                        urlmask = ".*" + tenant + ".*";
                    } else urlmask = ".*" + tenant + urlmask;
                }
            }
            final int site = querystring.indexOf("site:",0);
            String sitehash = null;
            String sitehost = null;
            if (site >= 0) {
                int ftb = querystring.indexOf(' ', site);
                if (ftb == -1) {
                    ftb = querystring.length();
                }
                sitehost = querystring.substring(site + 5, ftb);
                querystring = querystring.replace("site:" + sitehost, "");
                while (sitehost.length() > 0 && sitehost.charAt(0) == '.') {
                    sitehost = sitehost.substring(1);
                }
                while (sitehost.endsWith(".")) {
                    sitehost = sitehost.substring(0, sitehost.length() - 1);
                }
                sitehash = DigestURI.hosthash(sitehost);
                modifier.append("site:").append(sitehost).append(" ");
            }

            final int heuristicScroogle = querystring.indexOf("/heuristic/scroogle",0);
            if (heuristicScroogle >= 0) {
                querystring = querystring.replace("/heuristic/scroogle", "");
                modifier.append("/heuristic/scroogle ");
            }

            final int heuristicBlekko = querystring.indexOf("/heuristic/blekko",0);
            if (heuristicBlekko >= 0) {
                querystring = querystring.replace("/heuristic/blekko", "");
                modifier.append("/heuristic/blekko ");
            }

            final int authori = querystring.indexOf("author:",0);
          String authorhash = null;
            if (authori >= 0) {
              // check if the author was given with single quotes or without
              final boolean quotes = (querystring.charAt(authori + 7) == (char) 39);
              String author;
              if (quotes) {
                    int ftb = querystring.indexOf((char) 39, authori + 8);
                    if (ftb == -1) {
                        ftb = querystring.length() + 1;
                    }
                    author = querystring.substring(authori + 8, ftb);
                    querystring = querystring.replace("author:'" + author + "'", "");
                    modifier.append("author:'").append(author).append("' ");
              } else {
                    int ftb = querystring.indexOf(' ', authori);
                    if (ftb == -1) {
                        ftb = querystring.length();
                    }
                    author = querystring.substring(authori + 7, ftb);
                    querystring = querystring.replace("author:" + author, "");
                    modifier.append("author:").append(author).append(" ");
              }
              authorhash = ASCII.String(Word.word2hash(author));
            }
            final int tld = querystring.indexOf("tld:",0);
            if (tld >= 0) {
                int ftb = querystring.indexOf(' ', tld);
                if (ftb == -1) {
                    ftb = querystring.length();
                }
                String domain = querystring.substring(tld + 4, ftb);
                querystring = querystring.replace("tld:" + domain, "");
                modifier.append("tld:").append(domain).append(" ");
                while (domain.length() > 0 && domain.charAt(0) == '.') {
                    domain = domain.substring(1);
                }
                if (domain.indexOf('.',0) < 0) {
                    domain = "\\." + domain;
                } // is tld
                if (domain.length() > 0) {
                    urlmask = "[a-zA-Z]*://[^/]*" + domain + "/.*" + ((urlmask != null) ? urlmask : "");
                }
            }
            if (urlmask == null || urlmask.isEmpty()) {
                urlmask = originalUrlMask;
            } //if no urlmask was given

            // read the language from the language-restrict option 'lr'
            // if no one is given, use the user agent or the system language as default
            language = (post == null) ? language : post.get("lr", language);
            if (language.startsWith("lang_")) {
                language = language.substring(5);
            }
            if (!ISO639.exists(language)) {
                // find out language of the user by reading of the user-agent string
                String agent = header.get(HeaderFramework.ACCEPT_LANGUAGE);
                if (agent == null) {
                    agent = System.getProperty("user.language");
                }
                language = (agent == null) ? "en" : ISO639.userAgentLanguageDetection(agent);
                if (language == null) {
                    language = "en";
                }
            }

            // navigation
            final String navigation = (post == null) ? sb.getConfig("search.navigation", "all") : post.get("nav", "");

            // the query
            final TreeSet<String>[] query = QueryParams.cleanQuery(querystring.trim()); // converts also umlaute

            final int maxDistance = (querystring.indexOf('"',0) >= 0) ? query.length - 1 : Integer.MAX_VALUE;

            // filter out stopwords
            final SortedSet<String> filtered = SetTools.joinConstructive(query[0], Switchboard.stopwords);
            if (!filtered.isEmpty()) {
                SetTools.excludeDestructive(query[0], Switchboard.stopwords);
            }

            // if a minus-button was hit, remove a special reference first
            if (post != null && post.containsKey("deleteref")) {
                try {
                    if (!sb.verifyAuthentication(header)) {
                        prop.put("AUTHENTICATE", "admin log-in"); // force log-in
                        return prop;
                    }

                    // delete the index entry locally
                    final String delHash = post.get("deleteref", ""); // urlhash
                    indexSegment.termIndex().remove(Word.words2hashesHandles(query[0]), delHash.getBytes());

                    // make new news message with negative voting
                    if (!sb.isRobinsonMode()) {
                        final Map<String, String> map = new HashMap<String, String>();
                        map.put("urlhash", delHash);
                        map.put("vote", "negative");
                        map.put("refid", "");
                        sb.peers.newsPool.publishMyNews(sb.peers.mySeed(), NewsPool.CATEGORY_SURFTIPP_VOTE_ADD, map);
                    }

                    // delete the search history since this still shows the entry
                    SearchEventCache.delete(delHash);
                } catch (final IOException e) {
                    Log.logException(e);
                }
            }

            // if a plus-button was hit, create new voting message
            if (post != null && post.containsKey("recommendref")) {
                if (!sb.verifyAuthentication(header)) {
                    prop.put("AUTHENTICATE", "admin log-in"); // force log-in
                    return prop;
                }
                final String recommendHash = post.get("recommendref", ""); // urlhash
                final URIMetadataRow urlentry = indexSegment.urlMetadata().load(UTF8.getBytes(recommendHash));
                if (urlentry != null) {
                    final URIMetadataRow.Components metadata = urlentry.metadata();
                    Document[] documents = null;
                    try {
                        documents = sb.loader.loadDocuments(sb.loader.request(metadata.url(), true, false), CacheStrategy.IFEXIST, 5000, Integer.MAX_VALUE);
                    } catch (final IOException e) {
                    } catch (final Parser.Failure e) {
                    }
                    if (documents != null) {
                        // create a news message
                        final Map<String, String> map = new HashMap<String, String>();
                        map.put("url", metadata.url().toNormalform(false, true).replace(',', '|'));
                        map.put("title", metadata.dc_title().replace(',', ' '));
                        map.put("description", documents[0].dc_title().replace(',', ' '));
                        map.put("author", documents[0].dc_creator());
                        map.put("tags", documents[0].dc_subject(' '));
                        sb.peers.newsPool.publishMyNews(sb.peers.mySeed(), NewsPool.CATEGORY_SURFTIPP_ADD, map);
                        documents[0].close();
                    }
                }
            }

            // if a bookmarks-button was hit, create new bookmark entry
            if (post != null && post.containsKey("bookmarkref")) {
                if (!sb.verifyAuthentication(header)) {
                    prop.put("AUTHENTICATE", "admin log-in"); // force log-in
                    return prop;
                }
                final String bookmarkHash = post.get("bookmarkref", ""); // urlhash
                final URIMetadataRow urlentry = indexSegment.urlMetadata().load(UTF8.getBytes(bookmarkHash));
                if (urlentry != null) {
                    final URIMetadataRow.Components metadata = urlentry.metadata();
                    try {
                        sb.tables.bookmarks.createBookmark(sb.loader, metadata.url(), YMarkTables.USER_ADMIN, true, "searchresult", "/search");
                    } catch (final Throwable e) {
                    }
                }
            }

            // do the search
            final HandleSet queryHashes = Word.words2hashesHandles(query[0]);
            final Pattern snippetPattern = QueryParams.stringSearchPattern(originalquerystring);

            // check filters
            try {
                Pattern.compile(urlmask);
            } catch (final PatternSyntaxException ex) {
                Log.logWarning("SEARCH", "Illegal URL mask, not a valid regex: " + urlmask);
                prop.put("urlmaskerror", 1);
                prop.putHTML("urlmaskerror_urlmask", urlmask);
                urlmask = ".*";
            }

            try {
                Pattern.compile(prefermask);
            } catch (final PatternSyntaxException ex) {
                Log.logWarning("SEARCH", "Illegal prefer mask, not a valid regex: " + prefermask);
                prop.put("prefermaskerror", 1);
                prop.putHTML("prefermaskerror_prefermask", prefermask);
                prefermask = "";
            }

            final QueryParams theQuery = new QueryParams(
                    originalquerystring,
                    queryHashes,
                    Word.words2hashesHandles(query[1]),
                    Word.words2hashesHandles(query[2]),
                    snippetPattern,
                    tenant,
                    modifier.toString().trim(),
                    maxDistance,
                    prefermask,
                    contentdom,
                    language,
                    navigation,
                    snippetFetchStrategy,
                    maximumRecords,
                    startRecord,
                    urlmask,
                    clustersearch && global ? QueryParams.Searchdom.CLUSTER :
                    (global && indexReceiveGranted ? QueryParams.Searchdom.GLOBAL : QueryParams.Searchdom.LOCAL),
                    20,
                    constraint,
                    true,
                    sitehash,
                    authorhash,
                    DigestURI.TLD_any_zone_filter,
                    client,
                    authenticated,
                    indexSegment,
                    ranking,
                    header.get(RequestHeader.USER_AGENT, ""),
                    sb.getConfigBool(SwitchboardConstants.SEARCH_VERIFY_DELETE, false) && sb.getConfigBool(SwitchboardConstants.NETWORK_SEARCHVERIFY, false) && sb.peers.mySeed().getFlagAcceptRemoteIndex());
            EventTracker.delete(EventTracker.EClass.SEARCH);
            EventTracker.update(EventTracker.EClass.SEARCH, new ProfilingGraph.EventSearch(theQuery.id(true), SearchEvent.Type.INITIALIZATION, "", 0, 0), false);

            // tell all threads to do nothing for a specific time
            sb.intermissionAllThreads(3000);

            // filter out words that appear in bluelist
            theQuery.filterOut(Switchboard.blueList);

            // log
            Log.logInfo("LOCAL_SEARCH", "INIT WORD SEARCH: " + theQuery.queryString + ":" + QueryParams.hashSet2hashString(theQuery.queryHashes) + " - " + theQuery.neededResults() + " links to be computed, " + theQuery.displayResults() + " lines to be displayed");
            EventChannel.channels(EventChannel.LOCALSEARCH).addMessage(new RSSMessage("Local Search Request", theQuery.queryString, ""));
            final long timestamp = System.currentTimeMillis();

            // create a new search event
            if (SearchEventCache.getEvent(theQuery.id(false)) == null) {
                theQuery.setOffset(0); // in case that this is a new search, always start without a offset
                startRecord = 0;
            }
            final SearchEvent theSearch = SearchEventCache.getEvent(
                theQuery, sb.peers, sb.tables, (sb.isRobinsonMode()) ? sb.clusterhashes : null, false, sb.loader,
                (int) sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXCOUNT_USER, sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXCOUNT_DEFAULT, 10)),
                      sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXTIME_USER, sb.getConfigLong(SwitchboardConstants.REMOTESEARCH_MAXTIME_DEFAULT, 3000)),
                (int) sb.getConfigLong(SwitchboardConstants.DHT_BURST_ROBINSON, 0),
                (int) sb.getConfigLong(SwitchboardConstants.DHT_BURST_MULTIWORD, 0));

            if (startRecord == 0) {
                if (sitehost != null && sb.getConfigBool("heuristic.site", false) && authenticated) {
                    sb.heuristicSite(theSearch, sitehost);
                }
                if ((heuristicScroogle >= || sb.getConfigBool("heuristic.scroogle", false)) && authenticated) {
                    sb.heuristicScroogle(theSearch);
                }
                if ((heuristicBlekko >= || sb.getConfigBool("heuristic.blekko", false)) && authenticated) {
                    sb.heuristicRSS("http://blekko.com/ws/$+/rss", theSearch, "blekko");
                }
            }

            // log
            Log.logInfo("LOCAL_SEARCH", "EXIT WORD SEARCH: " + theQuery.queryString + " - " +
                    "local-unfiltered(" + theSearch.getRankingResult().getLocalIndexCount() + "), " +
                    "local_miss(" + theSearch.getRankingResult().getMissCount() + "), " +
                    "local_sortout(" + theSearch.getRankingResult().getSortOutCount() + "), " +
                    "remote(" + theSearch.getRankingResult().getRemoteResourceSize() + ") links found, " +
                    (System.currentTimeMillis() - timestamp) + " ms");

            // prepare search statistics
            theQuery.resultcount = theSearch.getRankingResult().getLocalIndexCount() - theSearch.getRankingResult().getMissCount() - theSearch.getRankingResult().getSortOutCount() + theSearch.getRankingResult().getRemoteIndexCount();
            theQuery.searchtime = System.currentTimeMillis() - timestamp;
            theQuery.urlretrievaltime = theSearch.result().getURLRetrievalTime();
            theQuery.snippetcomputationtime = theSearch.result().getSnippetComputationTime();
            AccessTracker.add(AccessTracker.Location.local, theQuery);

            // check suggestions
            final int meanMax = (post != null) ? post.getInt("meanCount", 0) : 0;

            prop.put("meanCount", meanMax);
            if (meanMax > 0 && !json && !rss) {
                final DidYouMean didYouMean = new DidYouMean(indexSegment.termIndex(), new StringBuilder(querystring));
              final Iterator<StringBuilder> meanIt = didYouMean.getSuggestions(100, 5).iterator();
                int meanCount = 0;
                String suggestion;
                while( meanCount<meanMax && meanIt.hasNext()) {
                    suggestion = meanIt.next().toString();
                    prop.put("didYouMean_suggestions_"+meanCount+"_word", suggestion);
                    prop.put("didYouMean_suggestions_"+meanCount+"_url",
                            QueryParams.navurl("html", 0, theQuery, suggestion, originalUrlMask.toString(), theQuery.navigators).toString()
                   );
                    prop.put("didYouMean_suggestions_"+meanCount+"_sep","|");
                    meanCount++;
                }
                prop.put("didYouMean_suggestions_"+(meanCount-1)+"_sep","");
                prop.put("didYouMean", meanCount>0 ? 1:0);
                prop.put("didYouMean_suggestions", meanCount);
            } else {
                prop.put("didYouMean", 0);
            }

            // find geographic info
            final SortedSet<Location> coordinates = LibraryProvider.geoLoc.find(originalquerystring, false);
            if (coordinates == null || coordinates.isEmpty() || startRecord > 0) {
                prop.put("geoinfo", "0");
            } else {
                int i = 0;
                for (final Location c: coordinates) {
                    prop.put("geoinfo_loc_" + i + "_lon", Math.round(c.lon() * 10000.0f) / 10000.0f);
                    prop.put("geoinfo_loc_" + i + "_lat", Math.round(c.lat() * 10000.0f) / 10000.0f);
                    prop.put("geoinfo_loc_" + i + "_name", c.getName());
                    i++;
                    if (i >= 10) break;
                }
                prop.put("geoinfo_loc", i);
                prop.put("geoinfo", "1");
            }

            // update the search tracker
            try {
                synchronized (trackerHandles) {
                    trackerHandles.add(theQuery.time);
                    while (trackerHandles.size() > 600) {
                        if (!trackerHandles.remove(trackerHandles.first())) break;
                    }
                }
                sb.localSearchTracker.put(client, trackerHandles);
              if (sb.localSearchTracker.size() > 100) {
                    sb.localSearchTracker.remove(sb.localSearchTracker.keys().nextElement());
                }
              if (MemoryControl.shortStatus()) sb.localSearchTracker.clear();
            } catch (final Exception e) {
                Log.logException(e);
            }

            final int indexcount = theSearch.getRankingResult().getLocalIndexCount() - theSearch.getRankingResult().getMissCount() - theSearch.getRankingResult().getSortOutCount() + theSearch.getRankingResult().getRemoteIndexCount();
            prop.put("num-results_offset", startRecord == 0 ? 0 : startRecord + 1);
            prop.put("num-results_itemscount", Formatter.number(startRecord + theSearch.getQuery().itemsPerPage > indexcount ? startRecord + indexcount % theSearch.getQuery().itemsPerPage : startRecord + theSearch.getQuery().itemsPerPage, true));
            prop.put("num-results_itemsPerPage", maximumRecords);
            prop.put("num-results_totalcount", Formatter.number(indexcount, true));
            prop.put("num-results_globalresults", global && (indexReceiveGranted || clustersearch) ? "1" : "0");
            prop.put("num-results_globalresults_localResourceSize", Formatter.number(theSearch.getRankingResult().getLocalIndexCount(), true));
            prop.put("num-results_globalresults_localMissCount", Formatter.number(theSearch.getRankingResult().getMissCount(), true));
            prop.put("num-results_globalresults_remoteResourceSize", Formatter.number(theSearch.getRankingResult().getRemoteResourceSize(), true));
            prop.put("num-results_globalresults_remoteIndexCount", Formatter.number(theSearch.getRankingResult().getRemoteIndexCount(), true));
            prop.put("num-results_globalresults_remotePeerCount", Formatter.number(theSearch.getRankingResult().getRemotePeerCount(), true));

            // compose page navigation
            final StringBuilder resnav = new StringBuilder(200);
            final int thispage = startRecord / theQuery.displayResults();
            if (thispage == 0) {
              resnav.append("<img src=\"env/grafics/navdl.gif\" alt=\"arrowleft\" width=\"16\" height=\"16\" />&nbsp;");
            } else {
              resnav.append("<a id=\"prevpage\" href=\"");
                resnav.append(QueryParams.navurl("html", thispage - 1, theQuery, null, originalUrlMask, navigation).toString());
              resnav.append("\"><img src=\"env/grafics/navdl.gif\" alt=\"arrowleft\" width=\"16\" height=\"16\" /></a>&nbsp;");
            }
            final int numberofpages = Math.min(10, 1 + ((indexcount - 1) / theQuery.displayResults()));

            for (int i = 0; i < numberofpages; i++) {
                if (i == thispage) {
                    resnav.append("<img src=\"env/grafics/navs");
                    resnav.append(i + 1);
                    resnav.append(".gif\" alt=\"page");
                    resnav.append(i + 1);
                    resnav.append("\" width=\"16\" height=\"16\" />&nbsp;");
                } else {
                    resnav.append("<a href=\"");
                    resnav.append(QueryParams.navurl("html", i, theQuery, null, originalUrlMask, navigation).toString());
                    resnav.append("\"><img src=\"env/grafics/navd");
                    resnav.append(i + 1);
                    resnav.append(".gif\" alt=\"page");
                    resnav.append(i + 1);
                    resnav.append("\" width=\"16\" height=\"16\" /></a>&nbsp;");
                }
            }
            if (thispage >= numberofpages) {
              resnav.append("<img src=\"env/grafics/navdr.gif\" alt=\"arrowright\" width=\"16\" height=\"16\" />");
            } else {
                resnav.append("<a id=\"nextpage\" href=\"");
                resnav.append(QueryParams.navurl("html", thispage + 1, theQuery, null, originalUrlMask, navigation).toString());
                resnav.append("\"><img src=\"env/grafics/navdr.gif\" alt=\"arrowright\" width=\"16\" height=\"16\" /></a>");
            }
            final String resnavs = resnav.toString();
            prop.put("num-results_resnav", resnavs);
            prop.put("pageNavBottom", (indexcount - startRecord > 6) ? 1 : 0); // if there are more results than may fit on the page we add a navigation at the bottom
            prop.put("pageNavBottom_resnav", resnavs);

            // generate the search result lines; the content will be produced by another servlet
            for (int i = 0; i < theQuery.displayResults(); i++) {
                prop.put("results_" + i + "_item", startRecord + i);
                prop.put("results_" + i + "_eventID", theQuery.id(false));
            }
            prop.put("results", theQuery.displayResults());
            prop.put("resultTable", (contentdom == ContentDomain.APP || contentdom == ContentDomain.AUDIO || contentdom == ContentDomain.VIDEO) ? 1 : 0);
            prop.put("eventID", theQuery.id(false)); // for bottomline

            // process result of search
            if (!filtered.isEmpty()) {
                prop.put("excluded", "1");
                prop.putHTML("excluded_stopwords", filtered.toString());
            } else {
                prop.put("excluded", "0");
            }

            if (prop == null || prop.isEmpty()) {
                if (post.get("query", post.get("search", "")).length() < 2) {
                    prop.put("num-results", "2"); // no results - at least 2 chars
                } else {
                    prop.put("num-results", "1"); // no results
                }
            } else {
                prop.put("num-results", "3");
            }

            prop.put("cat", "href");
            prop.put("depth", "0");

            // adding some additional properties needed for the rss feed
            String hostName = header.get("Host", "localhost");
            if (hostName.indexOf(':',0) == -1) {
                hostName += ":" + serverCore.getPortNr(env.getConfig("port", "8090"));
            }
            prop.put("searchBaseURL", "http://" + hostName + "/yacysearch.html");
            prop.put("rssYacyImageURL", "http://" + hostName + "/env/grafics/yacy.gif");
        }

        prop.put("searchagain", global ? "1" : "0");
        prop.putHTML("former", originalquerystring);
        prop.put("count", maximumRecords);
        prop.put("offset", startRecord);
        prop.put("resource", global ? "global" : "local");
        prop.putHTML("urlmaskfilter", originalUrlMask);
        prop.putHTML("prefermaskfilter", prefermask);
        prop.put("indexof", (indexof) ? "on" : "off");
        prop.put("constraint", (constraint == null) ? "" : constraint.exportB64());
        prop.put("search.verify", snippetFetchStrategy == null ? sb.getConfig("search.verify", "iffresh") : snippetFetchStrategy.toName());
        prop.put("search.navigation", (post == null) ? sb.getConfig("search.navigation", "all") : post.get("nav", "all"));
        prop.put("contentdom", (post == null ? "text" : post.get("contentdom", "text")));
        prop.put("searchdomswitches", sb.getConfigBool("search.text", true) || sb.getConfigBool("search.audio", true) || sb.getConfigBool("search.video", true) || sb.getConfigBool("search.image", true) || sb.getConfigBool("search.app", true) ? 1 : 0);
        prop.put("searchdomswitches_searchtext", sb.getConfigBool("search.text", true) ? 1 : 0);
View Full Code Here
TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.