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 static com.google.common.collect.FluentIterable.from;
  21
  22 import java.util.ArrayList;
  23 import java.util.Collection;
  24 import java.util.Collections;
  25 import java.util.Comparator;
  26 import java.util.HashSet;
  27 import java.util.List;
  28 import java.util.Set;
  29 import java.util.concurrent.TimeUnit;
  30 import java.util.logging.Level;
  31 import java.util.logging.Logger;
  32
  33 import net.pterodactylus.sone.data.Post;
  34 import net.pterodactylus.sone.data.PostReply;
  35 import net.pterodactylus.sone.data.Profile;
  36 import net.pterodactylus.sone.data.Profile.Field;
  37 import net.pterodactylus.sone.data.Reply;
  38 import net.pterodactylus.sone.data.Sone;
  39 import net.pterodactylus.sone.web.page.FreenetRequest;
  40 import net.pterodactylus.util.collection.Pagination;
  41 import net.pterodactylus.util.logging.Logging;
  42 import net.pterodactylus.util.number.Numbers;
  43 import net.pterodactylus.util.template.Template;
  44 import net.pterodactylus.util.template.TemplateContext;
  45 import net.pterodactylus.util.text.StringEscaper;
  46 import net.pterodactylus.util.text.TextException;
  47
  48 import com.google.common.base.Function;
  49 import com.google.common.base.Optional;
  50 import com.google.common.base.Predicate;
  51 import com.google.common.cache.CacheBuilder;
  52 import com.google.common.cache.CacheLoader;
  53 import com.google.common.cache.LoadingCache;
  54 import com.google.common.collect.Collections2;
  55 import com.google.common.collect.Ordering;
  56
  57 /**
  58  * This page lets the user search for posts and replies that contain certain
  59  * words.
  60  *
  61  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  62  */
  63 public class SearchPage extends SoneTemplatePage {
  64
  65         /** The logger. */
  66         private static final Logger logger = Logging.getLogger(SearchPage.class);
  67
  68         /** Short-term cache. */
  69         private final LoadingCache<List<Phrase>, Set<Hit<Post>>> hitCache = CacheBuilder.newBuilder().expireAfterWrite(5, TimeUnit.MINUTES).build(new CacheLoader<List<Phrase>, Set<Hit<Post>>>() {
  70
  71                 @Override
  72                 @SuppressWarnings("synthetic-access")
  73                 public Set<Hit<Post>> load(List<Phrase> phrases) {
  74                         Set<Post> posts = new HashSet<Post>();
  75                         for (Sone sone : webInterface.getCore().getSones()) {
  76                                 posts.addAll(sone.getPosts());
  77                         }
  78                         return getHits(Collections2.filter(posts, Post.FUTURE_POSTS_FILTER), phrases, new PostStringGenerator());
  79                 }
  80         });
  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         @Override
  99         @SuppressWarnings("synthetic-access")
 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                 /* check for a couple of shortcuts. */
 113                 if (phrases.size() == 1) {
 114                         String phrase = phrases.get(0).getPhrase();
 115
 116                         /* is it a Sone ID? */
 117                         redirectIfNotNull(getSoneId(phrase), "viewSone.html?sone=");
 118
 119                         /* is it a post ID? */
 120                         redirectIfNotNull(getPostId(phrase), "viewPost.html?post=");
 121
 122                         /* is it a reply ID? show the post. */
 123                         redirectIfNotNull(getReplyPostId(phrase), "viewPost.html?post=");
 124
 125                         /* is it an album ID? */
 126                         redirectIfNotNull(getAlbumId(phrase), "imageBrowser.html?album=");
 127
 128                         /* is it an image ID? */
 129                         redirectIfNotNull(getImageId(phrase), "imageBrowser.html?image=");
 130                 }
 131
 132                 Collection<Sone> sones = webInterface.getCore().getSones();
 133                 Collection<Hit<Sone>> soneHits = getHits(sones, phrases, SoneStringGenerator.COMPLETE_GENERATOR);
 134
 135                 Collection<Hit<Post>> postHits = hitCache.getUnchecked(phrases);
 136
 137                 /* now filter. */
 138                 soneHits = Collections2.filter(soneHits, Hit.POSITIVE_FILTER);
 139                 postHits = Collections2.filter(postHits, Hit.POSITIVE_FILTER);
 140
 141                 /* now sort. */
 142                 List<Hit<Sone>> sortedSoneHits = Ordering.from(Hit.DESCENDING_COMPARATOR).sortedCopy(soneHits);
 143                 List<Hit<Post>> sortedPostHits = Ordering.from(Hit.DESCENDING_COMPARATOR).sortedCopy(postHits);
 144
 145                 /* extract Sones and posts. */
 146                 List<Sone> resultSones = from(sortedSoneHits).transform(new HitMapper<Sone>()).toList();
 147                 List<Post> resultPosts = from(sortedPostHits).transform(new HitMapper<Post>()).toList();
 148
 149                 /* pagination. */
 150                 Pagination<Sone> sonePagination = new Pagination<Sone>(resultSones, webInterface.getCore().getPreferences().getPostsPerPage()).setPage(Numbers.safeParseInteger(request.getHttpRequest().getParam("sonePage"), 0));
 151                 Pagination<Post> postPagination = new Pagination<Post>(resultPosts, webInterface.getCore().getPreferences().getPostsPerPage()).setPage(Numbers.safeParseInteger(request.getHttpRequest().getParam("postPage"), 0));
 152
 153                 templateContext.set("sonePagination", sonePagination);
 154                 templateContext.set("soneHits", sonePagination.getItems());
 155                 templateContext.set("postPagination", postPagination);
 156                 templateContext.set("postHits", postPagination.getItems());
 157         }
 158
 159         //
 160         // PRIVATE METHODS
 161         //
 162
 163         /**
 164          * Collects hit information for the given objects. The objects are converted
 165          * to a {@link String} using the given {@link StringGenerator}, and the
 166          * {@link #calculateScore(List, String) calculated score} is stored together
 167          * with the object in a {@link Hit}, and all resulting {@link Hit}s are then
 168          * returned.
 169          *
 170          * @param <T>
 171          *            The type of the objects
 172          * @param objects
 173          *            The objects to search over
 174          * @param phrases
 175          *            The phrases to search for
 176          * @param stringGenerator
 177          *            The string generator for the objects
 178          * @return The hits for the given phrases
 179          */
 180         private static <T> Set<Hit<T>> getHits(Collection<T> objects, List<Phrase> phrases, StringGenerator<T> stringGenerator) {
 181                 Set<Hit<T>> hits = new HashSet<Hit<T>>();
 182                 for (T object : objects) {
 183                         String objectString = stringGenerator.generateString(object);
 184                         double score = calculateScore(phrases, objectString);
 185                         hits.add(new Hit<T>(object, score));
 186                 }
 187                 return hits;
 188         }
 189
 190         /**
 191          * Parses the given query into search phrases. The query is split on
 192          * whitespace while allowing to group words using single or double quotes.
 193          * Isolated phrases starting with a “+” are
 194          * {@link Phrase.Optionality#REQUIRED}, phrases with a “-” are
 195          * {@link Phrase.Optionality#FORBIDDEN}.
 196          *
 197          * @param query
 198          *            The query to parse
 199          * @return The parsed phrases
 200          */
 201         private static List<Phrase> parseSearchPhrases(String query) {
 202                 List<String> parsedPhrases = null;
 203                 try {
 204                         parsedPhrases = StringEscaper.parseLine(query);
 205                 } catch (TextException te1) {
 206                         /* invalid query. */
 207                         return Collections.emptyList();
 208                 }
 209
 210                 List<Phrase> phrases = new ArrayList<Phrase>();
 211                 for (String phrase : parsedPhrases) {
 212                         if (phrase.startsWith("+")) {
 213                                 if (phrase.length() > 1) {
 214                                         phrases.add(new Phrase(phrase.substring(1), Phrase.Optionality.REQUIRED));
 215                                 } else {
 216                                         phrases.add(new Phrase("+", Phrase.Optionality.OPTIONAL));
 217                                 }
 218                         } else if (phrase.startsWith("-")) {
 219                                 if (phrase.length() > 1) {
 220                                         phrases.add(new Phrase(phrase.substring(1), Phrase.Optionality.FORBIDDEN));
 221                                 } else {
 222                                         phrases.add(new Phrase("-", Phrase.Optionality.OPTIONAL));
 223                                 }
 224                         } else {
 225                                 phrases.add(new Phrase(phrase, Phrase.Optionality.OPTIONAL));
 226                         }
 227                 }
 228                 return phrases;
 229         }
 230
 231         /**
 232          * Calculates the score for the given expression when using the given
 233          * phrases.
 234          *
 235          * @param phrases
 236          *            The phrases to search for
 237          * @param expression
 238          *            The expression to search
 239          * @return The score of the expression
 240          */
 241         private static double calculateScore(List<Phrase> phrases, String expression) {
 242                 logger.log(Level.FINEST, String.format("Calculating Score for “%s”…", expression));
 243                 double optionalHits = 0;
 244                 double requiredHits = 0;
 245                 int forbiddenHits = 0;
 246                 int requiredPhrases = 0;
 247                 for (Phrase phrase : phrases) {
 248                         String phraseString = phrase.getPhrase().toLowerCase();
 249                         if (phrase.getOptionality() == Phrase.Optionality.REQUIRED) {
 250                                 ++requiredPhrases;
 251                         }
 252                         int matches = 0;
 253                         int index = 0;
 254                         double score = 0;
 255                         while (index < expression.length()) {
 256                                 int position = expression.toLowerCase().indexOf(phraseString, index);
 257                                 if (position == -1) {
 258                                         break;
 259                                 }
 260                                 score += Math.pow(1 - position / (double) expression.length(), 2);
 261                                 index = position + phraseString.length();
 262                                 logger.log(Level.FINEST, String.format("Got hit at position %d.", position));
 263                                 ++matches;
 264                         }
 265                         logger.log(Level.FINEST, String.format("Score: %f", score));
 266                         if (matches == 0) {
 267                                 continue;
 268                         }
 269                         if (phrase.getOptionality() == Phrase.Optionality.REQUIRED) {
 270                                 requiredHits += score;
 271                         }
 272                         if (phrase.getOptionality() == Phrase.Optionality.OPTIONAL) {
 273                                 optionalHits += score;
 274                         }
 275                         if (phrase.getOptionality() == Phrase.Optionality.FORBIDDEN) {
 276                                 forbiddenHits += matches;
 277                         }
 278                 }
 279                 return requiredHits * 3 + optionalHits + (requiredHits - requiredPhrases) * 5 - (forbiddenHits * 2);
 280         }
 281
 282         /**
 283          * Throws a
 284          * {@link net.pterodactylus.sone.web.page.FreenetTemplatePage.RedirectException}
 285          * if the given object is not {@code null}, appending the object to the
 286          * given target URL.
 287          *
 288          * @param object
 289          *            The object on which to redirect
 290          * @param target
 291          *            The target of the redirect
 292          * @throws RedirectException
 293          *             if {@code object} is not {@code null}
 294          */
 295         private static void redirectIfNotNull(String object, String target) throws RedirectException {
 296                 if (object != null) {
 297                         throw new RedirectException(target + object);
 298                 }
 299         }
 300
 301         /**
 302          * If the given phrase contains a Sone ID (optionally prefixed by
 303          * “sone://”), returns said Sone ID, otherwise return {@code null}.
 304          *
 305          * @param phrase
 306          *            The phrase that maybe is a Sone ID
 307          * @return The Sone ID, or {@code null}
 308          */
 309         private String getSoneId(String phrase) {
 310                 String soneId = phrase.startsWith("sone://") ? phrase.substring(7) : phrase;
 311                 return (webInterface.getCore().getSone(soneId).isPresent()) ? soneId : null;
 312         }
 313
 314         /**
 315          * If the given phrase contains a post ID (optionally prefixed by
 316          * “post://”), returns said post ID, otherwise return {@code null}.
 317          *
 318          * @param phrase
 319          *            The phrase that maybe is a post ID
 320          * @return The post ID, or {@code null}
 321          */
 322         private String getPostId(String phrase) {
 323                 String postId = phrase.startsWith("post://") ? phrase.substring(7) : phrase;
 324                 return (webInterface.getCore().getDatabase().getPost(postId).isPresent()) ? postId : null;
 325         }
 326
 327         /**
 328          * If the given phrase contains a reply ID (optionally prefixed by
 329          * “reply://”), returns the ID of the post the reply belongs to, otherwise
 330          * return {@code null}.
 331          *
 332          * @param phrase
 333          *            The phrase that maybe is a reply ID
 334          * @return The reply’s post ID, or {@code null}
 335          */
 336         private String getReplyPostId(String phrase) {
 337                 String replyId = phrase.startsWith("reply://") ? phrase.substring(8) : phrase;
 338                 Optional<PostReply> postReply = webInterface.getCore().getDatabase().getPostReply(replyId);
 339                 if (!postReply.isPresent()) {
 340                         return null;
 341                 }
 342                 return postReply.get().getPostId();
 343         }
 344
 345         /**
 346          * If the given phrase contains an album ID (optionally prefixed by
 347          * “album://”), returns said album ID, otherwise return {@code null}.
 348          *
 349          * @param phrase
 350          *            The phrase that maybe is an album ID
 351          * @return The album ID, or {@code null}
 352          */
 353         private String getAlbumId(String phrase) {
 354                 String albumId = phrase.startsWith("album://") ? phrase.substring(8) : phrase;
 355                 return webInterface.getCore().getAlbum(albumId).isPresent() ? albumId : null;
 356         }
 357
 358         /**
 359          * If the given phrase contains an image ID (optionally prefixed by
 360          * “image://”), returns said image ID, otherwise return {@code null}.
 361          *
 362          * @param phrase
 363          *            The phrase that maybe is an image ID
 364          * @return The image ID, or {@code null}
 365          */
 366         private String getImageId(String phrase) {
 367                 String imageId = phrase.startsWith("image://") ? phrase.substring(8) : phrase;
 368                 return webInterface.getCore().getImage(imageId).isPresent() ? imageId : null;
 369         }
 370
 371         /**
 372          * Converts a given object into a {@link String}.
 373          *
 374          * @param <T>
 375          *            The type of the objects
 376          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 377          */
 378         private static interface StringGenerator<T> {
 379
 380                 /**
 381                  * Generates a {@link String} for the given object.
 382                  *
 383                  * @param object
 384                  *            The object to generate the {@link String} for
 385                  * @return The generated {@link String}
 386                  */
 387                 public String generateString(T object);
 388
 389         }
 390
 391         /**
 392          * Generates a {@link String} from a {@link Sone}, concatenating the name of
 393          * the Sone and all {@link Profile} {@link Field} values.
 394          *
 395          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 396          */
 397         private static class SoneStringGenerator implements StringGenerator<Sone> {
 398
 399                 /** A static instance of a complete Sone string generator. */
 400                 public static final SoneStringGenerator COMPLETE_GENERATOR = new SoneStringGenerator(true);
 401
 402                 /**
 403                  * A static instance of a Sone string generator that will only use the
 404                  * name of the Sone.
 405                  */
 406                 public static final SoneStringGenerator NAME_GENERATOR = new SoneStringGenerator(false);
 407
 408                 /** Whether to generate a string from all data of a Sone. */
 409                 private final boolean complete;
 410
 411                 /**
 412                  * Creates a new Sone string generator.
 413                  *
 414                  * @param complete
 415                  *            {@code true} to use the profile’s fields, {@code false} to
 416                  *            not to use the profile‘s fields
 417                  */
 418                 private SoneStringGenerator(boolean complete) {
 419                         this.complete = complete;
 420                 }
 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                 @Override
 456                 public String generateString(Post post) {
 457                         StringBuilder postString = new StringBuilder();
 458                         postString.append(post.getText());
 459                         if (post.getRecipient().isPresent()) {
 460                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(post.getRecipient().get()));
 461                         }
 462                         for (PostReply reply : from(post.getReplies()).filter(Reply.FUTURE_REPLY_FILTER)) {
 463                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(reply.getSone()));
 464                                 postString.append(' ').append(reply.getText());
 465                         }
 466                         return postString.toString();
 467                 }
 468
 469         }
 470
 471         /**
 472          * A search phrase.
 473          *
 474          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 475          */
 476         private static class Phrase {
 477
 478                 /**
 479                  * The optionality of a search phrase.
 480                  *
 481                  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’
 482                  *         Roden</a>
 483                  */
 484                 public enum Optionality {
 485
 486                         /** The phrase is optional. */
 487                         OPTIONAL,
 488
 489                         /** The phrase is required. */
 490                         REQUIRED,
 491
 492                         /** The phrase is forbidden. */
 493                         FORBIDDEN
 494
 495                 }
 496
 497                 /** The phrase to search for. */
 498                 private final String phrase;
 499
 500                 /** The optionality of the phrase. */
 501                 private final Optionality optionality;
 502
 503                 /**
 504                  * Creates a new phrase.
 505                  *
 506                  * @param phrase
 507                  *            The phrase to search for
 508                  * @param optionality
 509                  *            The optionality of the phrase
 510                  */
 511                 public Phrase(String phrase, Optionality optionality) {
 512                         this.optionality = optionality;
 513                         this.phrase = phrase;
 514                 }
 515
 516                 /**
 517                  * Returns the phrase to search for.
 518                  *
 519                  * @return The phrase to search for
 520                  */
 521                 public String getPhrase() {
 522                         return phrase;
 523                 }
 524
 525                 /**
 526                  * Returns the optionality of the phrase.
 527                  *
 528                  * @return The optionality of the phrase
 529                  */
 530                 public Optionality getOptionality() {
 531                         return optionality;
 532                 }
 533
 534                 //
 535                 // OBJECT METHODS
 536                 //
 537
 538                 @Override
 539                 public int hashCode() {
 540                         return phrase.hashCode() ^ ((optionality == Optionality.FORBIDDEN) ? (0xaaaaaaaa) : ((optionality == Optionality.REQUIRED) ? 0x55555555 : 0));
 541                 }
 542
 543                 @Override
 544                 public boolean equals(Object object) {
 545                         if (!(object instanceof Phrase)) {
 546                                 return false;
 547                         }
 548                         Phrase phrase = (Phrase) object;
 549                         return (this.optionality == phrase.optionality) && this.phrase.equals(phrase.phrase);
 550                 }
 551
 552         }
 553
 554         /**
 555          * A hit consists of a searched object and the score it got for the phrases
 556          * of the search.
 557          *
 558          * @see SearchPage#calculateScore(List, String)
 559          * @param <T>
 560          *            The type of the searched object
 561          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 562          */
 563         private static class Hit<T> {
 564
 565                 /** Filter for {@link Hit}s with a score of more than 0. */
 566                 public static final Predicate<Hit<?>> POSITIVE_FILTER = new Predicate<Hit<?>>() {
 567
 568                         @Override
 569                         public boolean apply(Hit<?> hit) {
 570                                 return (hit == null) ? false : hit.getScore() > 0;
 571                         }
 572
 573                 };
 574
 575                 /** Comparator that sorts {@link Hit}s descending by score. */
 576                 public static final Comparator<Hit<?>> DESCENDING_COMPARATOR = new Comparator<Hit<?>>() {
 577
 578                         @Override
 579                         public int compare(Hit<?> leftHit, Hit<?> rightHit) {
 580                                 return (rightHit.getScore() < leftHit.getScore()) ? -1 : ((rightHit.getScore() > leftHit.getScore()) ? 1 : 0);
 581                         }
 582
 583                 };
 584
 585                 /** The object that was searched. */
 586                 private final T object;
 587
 588                 /** The score of the object. */
 589                 private final double score;
 590
 591                 /**
 592                  * Creates a new hit.
 593                  *
 594                  * @param object
 595                  *            The object that was searched
 596                  * @param score
 597                  *            The score of the object
 598                  */
 599                 public Hit(T object, double score) {
 600                         this.object = object;
 601                         this.score = score;
 602                 }
 603
 604                 /**
 605                  * Returns the object that was searched.
 606                  *
 607                  * @return The object that was searched
 608                  */
 609                 public T getObject() {
 610                         return object;
 611                 }
 612
 613                 /**
 614                  * Returns the score of the object.
 615                  *
 616                  * @return The score of the object
 617                  */
 618                 public double getScore() {
 619                         return score;
 620                 }
 621
 622         }
 623
 624         /**
 625          * Extracts the object from a {@link Hit}.
 626          *
 627          * @param <T>
 628          *            The type of the object to extract
 629          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 630          */
 631         private static class HitMapper<T> implements Function<Hit<T>, T> {
 632
 633                 @Override
 634                 public T apply(Hit<T> input) {
 635                         return input.getObject();
 636                 }
 637
 638         }
 639
 640 }