src/main/java/net/pterodactylus/sone/web/SearchPage.java

   1 /*
   2  * Sone - SearchPage.java - Copyright © 2010–2012 David Roden
   3  *
   4  * This program is free software: you can redistribute it and/or modify
   5  * it under the terms of the GNU General Public License as published by
   6  * the Free Software Foundation, either version 3 of the License, or
   7  * (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16  */
  17
  18 package net.pterodactylus.sone.web;
  19
  20 import java.util.ArrayList;
  21 import java.util.Collection;
  22 import java.util.Collections;
  23 import java.util.Comparator;
  24 import java.util.HashSet;
  25 import java.util.List;
  26 import java.util.Set;
  27 import java.util.logging.Level;
  28 import java.util.logging.Logger;
  29
  30 import net.pterodactylus.sone.data.Post;
  31 import net.pterodactylus.sone.data.PostReply;
  32 import net.pterodactylus.sone.data.Profile;
  33 import net.pterodactylus.sone.data.Profile.Field;
  34 import net.pterodactylus.sone.data.Reply;
  35 import net.pterodactylus.sone.data.Sone;
  36 import net.pterodactylus.sone.web.page.FreenetRequest;
  37 import net.pterodactylus.util.cache.Cache;
  38 import net.pterodactylus.util.cache.CacheException;
  39 import net.pterodactylus.util.cache.CacheItem;
  40 import net.pterodactylus.util.cache.DefaultCacheItem;
  41 import net.pterodactylus.util.cache.MemoryCache;
  42 import net.pterodactylus.util.cache.ValueRetriever;
  43 import net.pterodactylus.util.collection.Mapper;
  44 import net.pterodactylus.util.collection.Mappers;
  45 import net.pterodactylus.util.collection.Pagination;
  46 import net.pterodactylus.util.collection.TimedMap;
  47 import net.pterodactylus.util.filter.Filter;
  48 import net.pterodactylus.util.filter.Filters;
  49 import net.pterodactylus.util.logging.Logging;
  50 import net.pterodactylus.util.number.Numbers;
  51 import net.pterodactylus.util.template.Template;
  52 import net.pterodactylus.util.template.TemplateContext;
  53 import net.pterodactylus.util.text.StringEscaper;
  54 import net.pterodactylus.util.text.TextException;
  55
  56 /**
  57  * This page lets the user search for posts and replies that contain certain
  58  * words.
  59  *
  60  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  61  */
  62 public class SearchPage extends SoneTemplatePage {
  63
  64         /** The logger. */
  65         private static final Logger logger = Logging.getLogger(SearchPage.class);
  66
  67         /** Short-term cache. */
  68         private final Cache<List<Phrase>, Set<Hit<Post>>> hitCache = new MemoryCache<List<Phrase>, Set<Hit<Post>>>(new ValueRetriever<List<Phrase>, Set<Hit<Post>>>() {
  69
  70                 @Override
  71                 @SuppressWarnings("synthetic-access")
  72                 public CacheItem<Set<Hit<Post>>> retrieve(List<Phrase> phrases) throws CacheException {
  73                         Set<Post> posts = new HashSet<Post>();
  74                         for (Sone sone : webInterface.getCore().getSones()) {
  75                                 posts.addAll(sone.getPosts());
  76                         }
  77                         return new DefaultCacheItem<Set<Hit<Post>>>(getHits(Filters.filteredSet(posts, Post.FUTURE_POSTS_FILTER), phrases, new PostStringGenerator()));
  78                 }
  79
  80         }, new TimedMap<List<Phrase>, CacheItem<Set<Hit<Post>>>>(300000));
  81
  82         /**
  83          * Creates a new search page.
  84          *
  85          * @param template
  86          *            The template to render
  87          * @param webInterface
  88          *            The Sone web interface
  89          */
  90         public SearchPage(Template template, WebInterface webInterface) {
  91                 super("search.html", template, "Page.Search.Title", webInterface);
  92         }
  93
  94         //
  95         // SONETEMPLATEPAGE METHODS
  96         //
  97
  98         /**
  99          * {@inheritDoc}
 100          */
 101         @Override
 102         protected void processTemplate(FreenetRequest request, TemplateContext templateContext) throws RedirectException {
 103                 super.processTemplate(request, templateContext);
 104                 String query = request.getHttpRequest().getParam("query").trim();
 105                 if (query.length() == 0) {
 106                         throw new RedirectException("index.html");
 107                 }
 108
 109                 List<Phrase> phrases = parseSearchPhrases(query);
 110                 if (phrases.isEmpty()) {
 111                         throw new RedirectException("index.html");
 112                 }
 113
 114                 Set<Sone> sones = webInterface.getCore().getSones();
 115                 Set<Hit<Sone>> soneHits = getHits(sones, phrases, SoneStringGenerator.COMPLETE_GENERATOR);
 116
 117                 Set<Hit<Post>> postHits;
 118                 try {
 119                         postHits = hitCache.get(phrases);
 120                 } catch (CacheException ce1) {
 121                         /* should never happen. */
 122                         logger.log(Level.SEVERE, "Could not get search results from cache!", ce1);
 123                         postHits = Collections.emptySet();
 124                 }
 125
 126                 /* now filter. */
 127                 soneHits = Filters.filteredSet(soneHits, Hit.POSITIVE_FILTER);
 128                 postHits = Filters.filteredSet(postHits, Hit.POSITIVE_FILTER);
 129
 130                 /* now sort. */
 131                 List<Hit<Sone>> sortedSoneHits = new ArrayList<Hit<Sone>>(soneHits);
 132                 Collections.sort(sortedSoneHits, Hit.DESCENDING_COMPARATOR);
 133                 List<Hit<Post>> sortedPostHits = new ArrayList<Hit<Post>>(postHits);
 134                 Collections.sort(sortedPostHits, Hit.DESCENDING_COMPARATOR);
 135
 136                 /* extract Sones and posts. */
 137                 List<Sone> resultSones = Mappers.mappedList(sortedSoneHits, new HitMapper<Sone>());
 138                 List<Post> resultPosts = Mappers.mappedList(sortedPostHits, new HitMapper<Post>());
 139
 140                 /* pagination. */
 141                 Pagination<Sone> sonePagination = new Pagination<Sone>(resultSones, webInterface.getCore().getPreferences().getPostsPerPage()).setPage(Numbers.safeParseInteger(request.getHttpRequest().getParam("sonePage"), 0));
 142                 Pagination<Post> postPagination = new Pagination<Post>(resultPosts, webInterface.getCore().getPreferences().getPostsPerPage()).setPage(Numbers.safeParseInteger(request.getHttpRequest().getParam("postPage"), 0));
 143
 144                 templateContext.set("sonePagination", sonePagination);
 145                 templateContext.set("soneHits", sonePagination.getItems());
 146                 templateContext.set("postPagination", postPagination);
 147                 templateContext.set("postHits", postPagination.getItems());
 148         }
 149
 150         //
 151         // PRIVATE METHODS
 152         //
 153
 154         /**
 155          * Collects hit information for the given objects. The objects are converted
 156          * to a {@link String} using the given {@link StringGenerator}, and the
 157          * {@link #calculateScore(List, String) calculated score} is stored together
 158          * with the object in a {@link Hit}, and all resulting {@link Hit}s are then
 159          * returned.
 160          *
 161          * @param <T>
 162          *            The type of the objects
 163          * @param objects
 164          *            The objects to search over
 165          * @param phrases
 166          *            The phrases to search for
 167          * @param stringGenerator
 168          *            The string generator for the objects
 169          * @return The hits for the given phrases
 170          */
 171         private <T> Set<Hit<T>> getHits(Collection<T> objects, List<Phrase> phrases, StringGenerator<T> stringGenerator) {
 172                 Set<Hit<T>> hits = new HashSet<Hit<T>>();
 173                 for (T object : objects) {
 174                         String objectString = stringGenerator.generateString(object);
 175                         double score = calculateScore(phrases, objectString);
 176                         hits.add(new Hit<T>(object, score));
 177                 }
 178                 return hits;
 179         }
 180
 181         /**
 182          * Parses the given query into search phrases. The query is split on
 183          * whitespace while allowing to group words using single or double quotes.
 184          * Isolated phrases starting with a “+” are
 185          * {@link Phrase.Optionality#REQUIRED}, phrases with a “-” are
 186          * {@link Phrase.Optionality#FORBIDDEN}.
 187          *
 188          * @param query
 189          *            The query to parse
 190          * @return The parsed phrases
 191          */
 192         private List<Phrase> parseSearchPhrases(String query) {
 193                 List<String> parsedPhrases = null;
 194                 try {
 195                         parsedPhrases = StringEscaper.parseLine(query);
 196                 } catch (TextException te1) {
 197                         /* invalid query. */
 198                         return Collections.emptyList();
 199                 }
 200
 201                 List<Phrase> phrases = new ArrayList<Phrase>();
 202                 for (String phrase : parsedPhrases) {
 203                         if (phrase.startsWith("+")) {
 204                                 if (phrase.length() > 1) {
 205                                         phrases.add(new Phrase(phrase.substring(1), Phrase.Optionality.REQUIRED));
 206                                 } else {
 207                                         phrases.add(new Phrase("+", Phrase.Optionality.OPTIONAL));
 208                                 }
 209                         } else if (phrase.startsWith("-")) {
 210                                 if (phrase.length() > 1) {
 211                                         phrases.add(new Phrase(phrase.substring(1), Phrase.Optionality.FORBIDDEN));
 212                                 } else {
 213                                         phrases.add(new Phrase("-", Phrase.Optionality.OPTIONAL));
 214                                 }
 215                         } else {
 216                                 phrases.add(new Phrase(phrase, Phrase.Optionality.OPTIONAL));
 217                         }
 218                 }
 219                 return phrases;
 220         }
 221
 222         /**
 223          * Calculates the score for the given expression when using the given
 224          * phrases.
 225          *
 226          * @param phrases
 227          *            The phrases to search for
 228          * @param expression
 229          *            The expression to search
 230          * @return The score of the expression
 231          */
 232         private double calculateScore(List<Phrase> phrases, String expression) {
 233                 logger.log(Level.FINEST, "Calculating Score for “%s”…", expression);
 234                 double optionalHits = 0;
 235                 double requiredHits = 0;
 236                 int forbiddenHits = 0;
 237                 int requiredPhrases = 0;
 238                 for (Phrase phrase : phrases) {
 239                         String phraseString = phrase.getPhrase().toLowerCase();
 240                         if (phrase.getOptionality() == Phrase.Optionality.REQUIRED) {
 241                                 ++requiredPhrases;
 242                         }
 243                         int matches = 0;
 244                         int index = 0;
 245                         double score = 0;
 246                         while (index < expression.length()) {
 247                                 int position = expression.toLowerCase().indexOf(phraseString, index);
 248                                 if (position == -1) {
 249                                         break;
 250                                 }
 251                                 score += Math.pow(1 - position / (double) expression.length(), 2);
 252                                 index = position + phraseString.length();
 253                                 logger.log(Level.FINEST, "Got hit at position %d.", position);
 254                                 ++matches;
 255                         }
 256                         logger.log(Level.FINEST, "Score: %f", score);
 257                         if (matches == 0) {
 258                                 continue;
 259                         }
 260                         if (phrase.getOptionality() == Phrase.Optionality.REQUIRED) {
 261                                 requiredHits += score;
 262                         }
 263                         if (phrase.getOptionality() == Phrase.Optionality.OPTIONAL) {
 264                                 optionalHits += score;
 265                         }
 266                         if (phrase.getOptionality() == Phrase.Optionality.FORBIDDEN) {
 267                                 forbiddenHits += matches;
 268                         }
 269                 }
 270                 return requiredHits * 3 + optionalHits + (requiredHits - requiredPhrases) * 5 - (forbiddenHits * 2);
 271         }
 272
 273         /**
 274          * Converts a given object into a {@link String}.
 275          *
 276          * @param <T>
 277          *            The type of the objects
 278          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 279          */
 280         private static interface StringGenerator<T> {
 281
 282                 /**
 283                  * Generates a {@link String} for the given object.
 284                  *
 285                  * @param object
 286                  *            The object to generate the {@link String} for
 287                  * @return The generated {@link String}
 288                  */
 289                 public String generateString(T object);
 290
 291         }
 292
 293         /**
 294          * Generates a {@link String} from a {@link Sone}, concatenating the name of
 295          * the Sone and all {@link Profile} {@link Field} values.
 296          *
 297          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 298          */
 299         private static class SoneStringGenerator implements StringGenerator<Sone> {
 300
 301                 /** A static instance of a complete Sone string generator. */
 302                 public static final SoneStringGenerator COMPLETE_GENERATOR = new SoneStringGenerator(true);
 303
 304                 /**
 305                  * A static instance of a Sone string generator that will only use the
 306                  * name of the Sone.
 307                  */
 308                 public static final SoneStringGenerator NAME_GENERATOR = new SoneStringGenerator(false);
 309
 310                 /** Whether to generate a string from all data of a Sone. */
 311                 private final boolean complete;
 312
 313                 /**
 314                  * Creates a new Sone string generator.
 315                  *
 316                  * @param complete
 317                  *            {@code true} to use the profile’s fields, {@code false} to
 318                  *            not to use the profile‘s fields
 319                  */
 320                 private SoneStringGenerator(boolean complete) {
 321                         this.complete = complete;
 322                 }
 323
 324                 /**
 325                  * {@inheritDoc}
 326                  */
 327                 @Override
 328                 public String generateString(Sone sone) {
 329                         StringBuilder soneString = new StringBuilder();
 330                         soneString.append(sone.getName());
 331                         Profile soneProfile = sone.getProfile();
 332                         if (soneProfile.getFirstName() != null) {
 333                                 soneString.append(' ').append(soneProfile.getFirstName());
 334                         }
 335                         if (soneProfile.getMiddleName() != null) {
 336                                 soneString.append(' ').append(soneProfile.getMiddleName());
 337                         }
 338                         if (soneProfile.getLastName() != null) {
 339                                 soneString.append(' ').append(soneProfile.getLastName());
 340                         }
 341                         if (complete) {
 342                                 for (Field field : soneProfile.getFields()) {
 343                                         soneString.append(' ').append(field.getValue());
 344                                 }
 345                         }
 346                         return soneString.toString();
 347                 }
 348
 349         }
 350
 351         /**
 352          * Generates a {@link String} from a {@link Post}, concatenating the text of
 353          * the post, the text of all {@link Reply}s, and the name of all
 354          * {@link Sone}s that have replied.
 355          *
 356          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 357          */
 358         private class PostStringGenerator implements StringGenerator<Post> {
 359
 360                 /**
 361                  * {@inheritDoc}
 362                  */
 363                 @Override
 364                 public String generateString(Post post) {
 365                         StringBuilder postString = new StringBuilder();
 366                         postString.append(post.getText());
 367                         if (post.getRecipient() != null) {
 368                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(post.getRecipient()));
 369                         }
 370                         for (PostReply reply : Filters.filteredList(webInterface.getCore().getReplies(post), Reply.FUTURE_REPLY_FILTER)) {
 371                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(reply.getSone()));
 372                                 postString.append(' ').append(reply.getText());
 373                         }
 374                         return postString.toString();
 375                 }
 376
 377         }
 378
 379         /**
 380          * A search phrase.
 381          *
 382          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 383          */
 384         private static class Phrase {
 385
 386                 /**
 387                  * The optionality of a search phrase.
 388                  *
 389                  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’
 390                  *         Roden</a>
 391                  */
 392                 public enum Optionality {
 393
 394                         /** The phrase is optional. */
 395                         OPTIONAL,
 396
 397                         /** The phrase is required. */
 398                         REQUIRED,
 399
 400                         /** The phrase is forbidden. */
 401                         FORBIDDEN
 402
 403                 }
 404
 405                 /** The phrase to search for. */
 406                 private final String phrase;
 407
 408                 /** The optionality of the phrase. */
 409                 private final Optionality optionality;
 410
 411                 /**
 412                  * Creates a new phrase.
 413                  *
 414                  * @param phrase
 415                  *            The phrase to search for
 416                  * @param optionality
 417                  *            The optionality of the phrase
 418                  */
 419                 public Phrase(String phrase, Optionality optionality) {
 420                         this.optionality = optionality;
 421                         this.phrase = phrase;
 422                 }
 423
 424                 /**
 425                  * Returns the phrase to search for.
 426                  *
 427                  * @return The phrase to search for
 428                  */
 429                 public String getPhrase() {
 430                         return phrase;
 431                 }
 432
 433                 /**
 434                  * Returns the optionality of the phrase.
 435                  *
 436                  * @return The optionality of the phrase
 437                  */
 438                 public Optionality getOptionality() {
 439                         return optionality;
 440                 }
 441
 442                 //
 443                 // OBJECT METHODS
 444                 //
 445
 446                 /**
 447                  * {@inheritDoc}
 448                  */
 449                 @Override
 450                 public int hashCode() {
 451                         return phrase.hashCode() ^ ((optionality == Optionality.FORBIDDEN) ? (0xaaaaaaaa) : ((optionality == Optionality.REQUIRED) ? 0x55555555 : 0));
 452                 }
 453
 454                 /**
 455                  * {@inheritDoc}
 456                  */
 457                 @Override
 458                 public boolean equals(Object object) {
 459                         if (!(object instanceof Phrase)) {
 460                                 return false;
 461                         }
 462                         Phrase phrase = (Phrase) object;
 463                         return (this.optionality == phrase.optionality) && this.phrase.equals(phrase.phrase);
 464                 }
 465
 466         }
 467
 468         /**
 469          * A hit consists of a searched object and the score it got for the phrases
 470          * of the search.
 471          *
 472          * @see SearchPage#calculateScore(List, String)
 473          * @param <T>
 474          *            The type of the searched object
 475          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 476          */
 477         private static class Hit<T> {
 478
 479                 /** Filter for {@link Hit}s with a score of more than 0. */
 480                 public static final Filter<Hit<?>> POSITIVE_FILTER = new Filter<Hit<?>>() {
 481
 482                         @Override
 483                         public boolean filterObject(Hit<?> hit) {
 484                                 return hit.getScore() > 0;
 485                         }
 486
 487                 };
 488
 489                 /** Comparator that sorts {@link Hit}s descending by score. */
 490                 public static final Comparator<Hit<?>> DESCENDING_COMPARATOR = new Comparator<Hit<?>>() {
 491
 492                         @Override
 493                         public int compare(Hit<?> leftHit, Hit<?> rightHit) {
 494                                 return (rightHit.getScore() < leftHit.getScore()) ? -1 : ((rightHit.getScore() > leftHit.getScore()) ? 1 : 0);
 495                         }
 496
 497                 };
 498
 499                 /** The object that was searched. */
 500                 private final T object;
 501
 502                 /** The score of the object. */
 503                 private final double score;
 504
 505                 /**
 506                  * Creates a new hit.
 507                  *
 508                  * @param object
 509                  *            The object that was searched
 510                  * @param score
 511                  *            The score of the object
 512                  */
 513                 public Hit(T object, double score) {
 514                         this.object = object;
 515                         this.score = score;
 516                 }
 517
 518                 /**
 519                  * Returns the object that was searched.
 520                  *
 521                  * @return The object that was searched
 522                  */
 523                 public T getObject() {
 524                         return object;
 525                 }
 526
 527                 /**
 528                  * Returns the score of the object.
 529                  *
 530                  * @return The score of the object
 531                  */
 532                 public double getScore() {
 533                         return score;
 534                 }
 535
 536         }
 537
 538         /**
 539          * Extracts the object from a {@link Hit}.
 540          *
 541          * @param <T>
 542          *            The type of the object to extract
 543          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 544          */
 545         public static class HitMapper<T> implements Mapper<Hit<T>, T> {
 546
 547                 /**
 548                  * {@inheritDoc}
 549                  */
 550                 @Override
 551                 public T map(Hit<T> input) {
 552                         return input.getObject();
 553                 }
 554
 555         }
 556
 557 }