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