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

   1 /*
   2  * Sone - SearchPage.java - Copyright © 2010–2013 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.concurrent.TimeUnit;
  28 import java.util.logging.Level;
  29 import java.util.logging.Logger;
  30
  31 import net.pterodactylus.sone.data.Post;
  32 import net.pterodactylus.sone.data.PostReply;
  33 import net.pterodactylus.sone.data.Profile;
  34 import net.pterodactylus.sone.data.Profile.Field;
  35 import net.pterodactylus.sone.data.Reply;
  36 import net.pterodactylus.sone.data.Sone;
  37 import net.pterodactylus.sone.web.page.FreenetRequest;
  38 import net.pterodactylus.util.collection.Pagination;
  39 import net.pterodactylus.util.logging.Logging;
  40 import net.pterodactylus.util.number.Numbers;
  41 import net.pterodactylus.util.template.Template;
  42 import net.pterodactylus.util.template.TemplateContext;
  43 import net.pterodactylus.util.text.StringEscaper;
  44 import net.pterodactylus.util.text.TextException;
  45
  46 import com.google.common.base.Function;
  47 import com.google.common.base.Predicate;
  48 import com.google.common.cache.CacheBuilder;
  49 import com.google.common.cache.CacheLoader;
  50 import com.google.common.cache.LoadingCache;
  51 import com.google.common.collect.Collections2;
  52 import com.google.common.collect.FluentIterable;
  53 import com.google.common.collect.Ordering;
  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 LoadingCache<List<Phrase>, Set<Hit<Post>>> hitCache = CacheBuilder.newBuilder().expireAfterWrite(5, TimeUnit.MINUTES).build(new CacheLoader<List<Phrase>, Set<Hit<Post>>>() {
  68
  69                 @Override
  70                 @SuppressWarnings("synthetic-access")
  71                 public Set<Hit<Post>> load(List<Phrase> phrases) {
  72                         Set<Post> posts = new HashSet<Post>();
  73                         for (Sone sone : webInterface.getCore().getSones()) {
  74                                 posts.addAll(sone.getPosts());
  75                         }
  76                         return getHits(Collections2.filter(posts, Post.FUTURE_POSTS_FILTER), phrases, new PostStringGenerator());
  77                 }
  78         });
  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         @SuppressWarnings("synthetic-access")
 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                 /* check for a couple of shortcuts. */
 114                 if (phrases.size() == 1) {
 115                         String phrase = phrases.get(0).getPhrase();
 116
 117                         /* is it a Sone ID? */
 118                         redirectIfNotNull(getSoneId(phrase), "viewSone.html?sone=");
 119
 120                         /* is it a post ID? */
 121                         redirectIfNotNull(getPostId(phrase), "viewPost.html?post=");
 122
 123                         /* is it a reply ID? show the post. */
 124                         redirectIfNotNull(getReplyPostId(phrase), "viewPost.html?post=");
 125
 126                         /* is it an album ID? */
 127                         redirectIfNotNull(getAlbumId(phrase), "imageBrowser.html?album=");
 128
 129                         /* is it an image ID? */
 130                         redirectIfNotNull(getImageId(phrase), "imageBrowser.html?image=");
 131                 }
 132
 133                 Set<Sone> sones = webInterface.getCore().getSones();
 134                 Collection<Hit<Sone>> soneHits = getHits(sones, phrases, SoneStringGenerator.COMPLETE_GENERATOR);
 135
 136                 Collection<Hit<Post>> postHits = hitCache.getUnchecked(phrases);
 137
 138                 /* now filter. */
 139                 soneHits = Collections2.filter(soneHits, Hit.POSITIVE_FILTER);
 140                 postHits = Collections2.filter(postHits, Hit.POSITIVE_FILTER);
 141
 142                 /* now sort. */
 143                 List<Hit<Sone>> sortedSoneHits = Ordering.from(Hit.DESCENDING_COMPARATOR).sortedCopy(soneHits);
 144                 List<Hit<Post>> sortedPostHits = Ordering.from(Hit.DESCENDING_COMPARATOR).sortedCopy(postHits);
 145
 146                 /* extract Sones and posts. */
 147                 List<Sone> resultSones = FluentIterable.from(sortedSoneHits).transform(new HitMapper<Sone>()).toList();
 148                 List<Post> resultPosts = FluentIterable.from(sortedPostHits).transform(new HitMapper<Post>()).toList();
 149
 150                 /* pagination. */
 151                 Pagination<Sone> sonePagination = new Pagination<Sone>(resultSones, webInterface.getCore().getPreferences().getPostsPerPage()).setPage(Numbers.safeParseInteger(request.getHttpRequest().getParam("sonePage"), 0));
 152                 Pagination<Post> postPagination = new Pagination<Post>(resultPosts, webInterface.getCore().getPreferences().getPostsPerPage()).setPage(Numbers.safeParseInteger(request.getHttpRequest().getParam("postPage"), 0));
 153
 154                 templateContext.set("sonePagination", sonePagination);
 155                 templateContext.set("soneHits", sonePagination.getItems());
 156                 templateContext.set("postPagination", postPagination);
 157                 templateContext.set("postHits", postPagination.getItems());
 158         }
 159
 160         //
 161         // PRIVATE METHODS
 162         //
 163
 164         /**
 165          * Collects hit information for the given objects. The objects are converted
 166          * to a {@link String} using the given {@link StringGenerator}, and the
 167          * {@link #calculateScore(List, String) calculated score} is stored together
 168          * with the object in a {@link Hit}, and all resulting {@link Hit}s are then
 169          * returned.
 170          *
 171          * @param <T>
 172          *            The type of the objects
 173          * @param objects
 174          *            The objects to search over
 175          * @param phrases
 176          *            The phrases to search for
 177          * @param stringGenerator
 178          *            The string generator for the objects
 179          * @return The hits for the given phrases
 180          */
 181         private static <T> Set<Hit<T>> getHits(Collection<T> objects, List<Phrase> phrases, StringGenerator<T> stringGenerator) {
 182                 Set<Hit<T>> hits = new HashSet<Hit<T>>();
 183                 for (T object : objects) {
 184                         String objectString = stringGenerator.generateString(object);
 185                         double score = calculateScore(phrases, objectString);
 186                         hits.add(new Hit<T>(object, score));
 187                 }
 188                 return hits;
 189         }
 190
 191         /**
 192          * Parses the given query into search phrases. The query is split on
 193          * whitespace while allowing to group words using single or double quotes.
 194          * Isolated phrases starting with a “+” are
 195          * {@link Phrase.Optionality#REQUIRED}, phrases with a “-” are
 196          * {@link Phrase.Optionality#FORBIDDEN}.
 197          *
 198          * @param query
 199          *            The query to parse
 200          * @return The parsed phrases
 201          */
 202         private static List<Phrase> parseSearchPhrases(String query) {
 203                 List<String> parsedPhrases = null;
 204                 try {
 205                         parsedPhrases = StringEscaper.parseLine(query);
 206                 } catch (TextException te1) {
 207                         /* invalid query. */
 208                         return Collections.emptyList();
 209                 }
 210
 211                 List<Phrase> phrases = new ArrayList<Phrase>();
 212                 for (String phrase : parsedPhrases) {
 213                         if (phrase.startsWith("+")) {
 214                                 if (phrase.length() > 1) {
 215                                         phrases.add(new Phrase(phrase.substring(1), Phrase.Optionality.REQUIRED));
 216                                 } else {
 217                                         phrases.add(new Phrase("+", Phrase.Optionality.OPTIONAL));
 218                                 }
 219                         } else if (phrase.startsWith("-")) {
 220                                 if (phrase.length() > 1) {
 221                                         phrases.add(new Phrase(phrase.substring(1), Phrase.Optionality.FORBIDDEN));
 222                                 } else {
 223                                         phrases.add(new Phrase("-", Phrase.Optionality.OPTIONAL));
 224                                 }
 225                         } else {
 226                                 phrases.add(new Phrase(phrase, Phrase.Optionality.OPTIONAL));
 227                         }
 228                 }
 229                 return phrases;
 230         }
 231
 232         /**
 233          * Calculates the score for the given expression when using the given
 234          * phrases.
 235          *
 236          * @param phrases
 237          *            The phrases to search for
 238          * @param expression
 239          *            The expression to search
 240          * @return The score of the expression
 241          */
 242         private static double calculateScore(List<Phrase> phrases, String expression) {
 243                 logger.log(Level.FINEST, String.format("Calculating Score for “%s”…", expression));
 244                 double optionalHits = 0;
 245                 double requiredHits = 0;
 246                 int forbiddenHits = 0;
 247                 int requiredPhrases = 0;
 248                 for (Phrase phrase : phrases) {
 249                         String phraseString = phrase.getPhrase().toLowerCase();
 250                         if (phrase.getOptionality() == Phrase.Optionality.REQUIRED) {
 251                                 ++requiredPhrases;
 252                         }
 253                         int matches = 0;
 254                         int index = 0;
 255                         double score = 0;
 256                         while (index < expression.length()) {
 257                                 int position = expression.toLowerCase().indexOf(phraseString, index);
 258                                 if (position == -1) {
 259                                         break;
 260                                 }
 261                                 score += Math.pow(1 - position / (double) expression.length(), 2);
 262                                 index = position + phraseString.length();
 263                                 logger.log(Level.FINEST, String.format("Got hit at position %d.", position));
 264                                 ++matches;
 265                         }
 266                         logger.log(Level.FINEST, String.format("Score: %f", score));
 267                         if (matches == 0) {
 268                                 continue;
 269                         }
 270                         if (phrase.getOptionality() == Phrase.Optionality.REQUIRED) {
 271                                 requiredHits += score;
 272                         }
 273                         if (phrase.getOptionality() == Phrase.Optionality.OPTIONAL) {
 274                                 optionalHits += score;
 275                         }
 276                         if (phrase.getOptionality() == Phrase.Optionality.FORBIDDEN) {
 277                                 forbiddenHits += matches;
 278                         }
 279                 }
 280                 return requiredHits * 3 + optionalHits + (requiredHits - requiredPhrases) * 5 - (forbiddenHits * 2);
 281         }
 282
 283         /**
 284          * Throws a
 285          * {@link net.pterodactylus.sone.web.page.FreenetTemplatePage.RedirectException}
 286          * if the given object is not {@code null}, appending the object to the
 287          * given target URL.
 288          *
 289          * @param object
 290          *            The object on which to redirect
 291          * @param target
 292          *            The target of the redirect
 293          * @throws RedirectException
 294          *             if {@code object} is not {@code null}
 295          */
 296         private static void redirectIfNotNull(String object, String target) throws RedirectException {
 297                 if (object != null) {
 298                         throw new RedirectException(target + object);
 299                 }
 300         }
 301
 302         /**
 303          * If the given phrase contains a Sone ID (optionally prefixed by
 304          * “sone://”), returns said Sone ID, otherwise return {@code null}.
 305          *
 306          * @param phrase
 307          *            The phrase that maybe is a Sone ID
 308          * @return The Sone ID, or {@code null}
 309          */
 310         private String getSoneId(String phrase) {
 311                 String soneId = phrase.startsWith("sone://") ? phrase.substring(7) : phrase;
 312                 return (webInterface.getCore().getSone(soneId, false) != null) ? soneId : null;
 313         }
 314
 315         /**
 316          * If the given phrase contains a post ID (optionally prefixed by
 317          * “post://”), returns said post ID, otherwise return {@code null}.
 318          *
 319          * @param phrase
 320          *            The phrase that maybe is a post ID
 321          * @return The post ID, or {@code null}
 322          */
 323         private String getPostId(String phrase) {
 324                 String postId = phrase.startsWith("post://") ? phrase.substring(7) : phrase;
 325                 return (webInterface.getCore().getPost(postId) != null) ? postId : null;
 326         }
 327
 328         /**
 329          * If the given phrase contains a reply ID (optionally prefixed by
 330          * “reply://”), returns the ID of the post the reply belongs to, otherwise
 331          * return {@code null}.
 332          *
 333          * @param phrase
 334          *            The phrase that maybe is a reply ID
 335          * @return The reply’s post ID, or {@code null}
 336          */
 337         private String getReplyPostId(String phrase) {
 338                 String replyId = phrase.startsWith("reply://") ? phrase.substring(8) : phrase;
 339                 return (webInterface.getCore().getPostReply(replyId) != null) ? webInterface.getCore().getPostReply(replyId).getPost().getId() : null;
 340         }
 341
 342         /**
 343          * If the given phrase contains an album ID (optionally prefixed by
 344          * “album://”), returns said album ID, otherwise return {@code null}.
 345          *
 346          * @param phrase
 347          *            The phrase that maybe is an album ID
 348          * @return The album ID, or {@code null}
 349          */
 350         private String getAlbumId(String phrase) {
 351                 String albumId = phrase.startsWith("album://") ? phrase.substring(8) : phrase;
 352                 return (webInterface.getCore().getAlbum(albumId, false) != null) ? albumId : null;
 353         }
 354
 355         /**
 356          * If the given phrase contains an image ID (optionally prefixed by
 357          * “image://”), returns said image ID, otherwise return {@code null}.
 358          *
 359          * @param phrase
 360          *            The phrase that maybe is an image ID
 361          * @return The image ID, or {@code null}
 362          */
 363         private String getImageId(String phrase) {
 364                 String imageId = phrase.startsWith("image://") ? phrase.substring(8) : phrase;
 365                 return (webInterface.getCore().getImage(imageId, false) != null) ? imageId : null;
 366         }
 367
 368         /**
 369          * Converts a given object into a {@link String}.
 370          *
 371          * @param <T>
 372          *            The type of the objects
 373          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 374          */
 375         private static interface StringGenerator<T> {
 376
 377                 /**
 378                  * Generates a {@link String} for the given object.
 379                  *
 380                  * @param object
 381                  *            The object to generate the {@link String} for
 382                  * @return The generated {@link String}
 383                  */
 384                 public String generateString(T object);
 385
 386         }
 387
 388         /**
 389          * Generates a {@link String} from a {@link Sone}, concatenating the name of
 390          * the Sone and all {@link Profile} {@link Field} values.
 391          *
 392          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 393          */
 394         private static class SoneStringGenerator implements StringGenerator<Sone> {
 395
 396                 /** A static instance of a complete Sone string generator. */
 397                 public static final SoneStringGenerator COMPLETE_GENERATOR = new SoneStringGenerator(true);
 398
 399                 /**
 400                  * A static instance of a Sone string generator that will only use the
 401                  * name of the Sone.
 402                  */
 403                 public static final SoneStringGenerator NAME_GENERATOR = new SoneStringGenerator(false);
 404
 405                 /** Whether to generate a string from all data of a Sone. */
 406                 private final boolean complete;
 407
 408                 /**
 409                  * Creates a new Sone string generator.
 410                  *
 411                  * @param complete
 412                  *            {@code true} to use the profile’s fields, {@code false} to
 413                  *            not to use the profile‘s fields
 414                  */
 415                 private SoneStringGenerator(boolean complete) {
 416                         this.complete = complete;
 417                 }
 418
 419                 /**
 420                  * {@inheritDoc}
 421                  */
 422                 @Override
 423                 public String generateString(Sone sone) {
 424                         StringBuilder soneString = new StringBuilder();
 425                         soneString.append(sone.getName());
 426                         Profile soneProfile = sone.getProfile();
 427                         if (soneProfile.getFirstName() != null) {
 428                                 soneString.append(' ').append(soneProfile.getFirstName());
 429                         }
 430                         if (soneProfile.getMiddleName() != null) {
 431                                 soneString.append(' ').append(soneProfile.getMiddleName());
 432                         }
 433                         if (soneProfile.getLastName() != null) {
 434                                 soneString.append(' ').append(soneProfile.getLastName());
 435                         }
 436                         if (complete) {
 437                                 for (Field field : soneProfile.getFields()) {
 438                                         soneString.append(' ').append(field.getValue());
 439                                 }
 440                         }
 441                         return soneString.toString();
 442                 }
 443
 444         }
 445
 446         /**
 447          * Generates a {@link String} from a {@link Post}, concatenating the text of
 448          * the post, the text of all {@link Reply}s, and the name of all
 449          * {@link Sone}s that have replied.
 450          *
 451          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 452          */
 453         private class PostStringGenerator implements StringGenerator<Post> {
 454
 455                 /**
 456                  * {@inheritDoc}
 457                  */
 458                 @Override
 459                 public String generateString(Post post) {
 460                         StringBuilder postString = new StringBuilder();
 461                         postString.append(post.getText());
 462                         if (post.getRecipient() != null) {
 463                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(post.getRecipient()));
 464                         }
 465                         for (PostReply reply : Collections2.filter(webInterface.getCore().getReplies(post), Reply.FUTURE_REPLY_FILTER)) {
 466                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(reply.getSone()));
 467                                 postString.append(' ').append(reply.getText());
 468                         }
 469                         return postString.toString();
 470                 }
 471
 472         }
 473
 474         /**
 475          * A search phrase.
 476          *
 477          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 478          */
 479         private static class Phrase {
 480
 481                 /**
 482                  * The optionality of a search phrase.
 483                  *
 484                  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’
 485                  *         Roden</a>
 486                  */
 487                 public enum Optionality {
 488
 489                         /** The phrase is optional. */
 490                         OPTIONAL,
 491
 492                         /** The phrase is required. */
 493                         REQUIRED,
 494
 495                         /** The phrase is forbidden. */
 496                         FORBIDDEN
 497
 498                 }
 499
 500                 /** The phrase to search for. */
 501                 private final String phrase;
 502
 503                 /** The optionality of the phrase. */
 504                 private final Optionality optionality;
 505
 506                 /**
 507                  * Creates a new phrase.
 508                  *
 509                  * @param phrase
 510                  *            The phrase to search for
 511                  * @param optionality
 512                  *            The optionality of the phrase
 513                  */
 514                 public Phrase(String phrase, Optionality optionality) {
 515                         this.optionality = optionality;
 516                         this.phrase = phrase;
 517                 }
 518
 519                 /**
 520                  * Returns the phrase to search for.
 521                  *
 522                  * @return The phrase to search for
 523                  */
 524                 public String getPhrase() {
 525                         return phrase;
 526                 }
 527
 528                 /**
 529                  * Returns the optionality of the phrase.
 530                  *
 531                  * @return The optionality of the phrase
 532                  */
 533                 public Optionality getOptionality() {
 534                         return optionality;
 535                 }
 536
 537                 //
 538                 // OBJECT METHODS
 539                 //
 540
 541                 /**
 542                  * {@inheritDoc}
 543                  */
 544                 @Override
 545                 public int hashCode() {
 546                         return phrase.hashCode() ^ ((optionality == Optionality.FORBIDDEN) ? (0xaaaaaaaa) : ((optionality == Optionality.REQUIRED) ? 0x55555555 : 0));
 547                 }
 548
 549                 /**
 550                  * {@inheritDoc}
 551                  */
 552                 @Override
 553                 public boolean equals(Object object) {
 554                         if (!(object instanceof Phrase)) {
 555                                 return false;
 556                         }
 557                         Phrase phrase = (Phrase) object;
 558                         return (this.optionality == phrase.optionality) && this.phrase.equals(phrase.phrase);
 559                 }
 560
 561         }
 562
 563         /**
 564          * A hit consists of a searched object and the score it got for the phrases
 565          * of the search.
 566          *
 567          * @see SearchPage#calculateScore(List, String)
 568          * @param <T>
 569          *            The type of the searched object
 570          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 571          */
 572         private static class Hit<T> {
 573
 574                 /** Filter for {@link Hit}s with a score of more than 0. */
 575                 public static final Predicate<Hit<?>> POSITIVE_FILTER = new Predicate<Hit<?>>() {
 576
 577                         @Override
 578                         public boolean apply(Hit<?> hit) {
 579                                 return hit.getScore() > 0;
 580                         }
 581
 582                 };
 583
 584                 /** Comparator that sorts {@link Hit}s descending by score. */
 585                 public static final Comparator<Hit<?>> DESCENDING_COMPARATOR = new Comparator<Hit<?>>() {
 586
 587                         @Override
 588                         public int compare(Hit<?> leftHit, Hit<?> rightHit) {
 589                                 return (rightHit.getScore() < leftHit.getScore()) ? -1 : ((rightHit.getScore() > leftHit.getScore()) ? 1 : 0);
 590                         }
 591
 592                 };
 593
 594                 /** The object that was searched. */
 595                 private final T object;
 596
 597                 /** The score of the object. */
 598                 private final double score;
 599
 600                 /**
 601                  * Creates a new hit.
 602                  *
 603                  * @param object
 604                  *            The object that was searched
 605                  * @param score
 606                  *            The score of the object
 607                  */
 608                 public Hit(T object, double score) {
 609                         this.object = object;
 610                         this.score = score;
 611                 }
 612
 613                 /**
 614                  * Returns the object that was searched.
 615                  *
 616                  * @return The object that was searched
 617                  */
 618                 public T getObject() {
 619                         return object;
 620                 }
 621
 622                 /**
 623                  * Returns the score of the object.
 624                  *
 625                  * @return The score of the object
 626                  */
 627                 public double getScore() {
 628                         return score;
 629                 }
 630
 631         }
 632
 633         /**
 634          * Extracts the object from a {@link Hit}.
 635          *
 636          * @param <T>
 637          *            The type of the object to extract
 638          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 639          */
 640         private static class HitMapper<T> implements Function<Hit<T>, T> {
 641
 642                 /**
 643                  * {@inheritDoc}
 644                  */
 645                 @Override
 646                 public T apply(Hit<T> input) {
 647                         return input.getObject();
 648                 }
 649
 650         }
 651
 652 }