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

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