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

   1 /*
   2  * Sone - SearchPage.java - Copyright © 2010–2012 David Roden
   3  *
   4  * This program is free software: you can redistribute it and/or modify
   5  * it under the terms of the GNU General Public License as published by
   6  * the Free Software Foundation, either version 3 of the License, or
   7  * (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16  */
  17
  18 package net.pterodactylus.sone.web;
  19
  20 import java.util.ArrayList;
  21 import java.util.Collection;
  22 import java.util.Collections;
  23 import java.util.Comparator;
  24 import java.util.HashSet;
  25 import java.util.List;
  26 import java.util.Set;
  27 import java.util.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         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                 Set<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 = FluentIterable.from(sortedSoneHits).transform(new HitMapper<Sone>()).toList();
 147                 List<Post> resultPosts = FluentIterable.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, false) != null) ? 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().getPost(postId, false) != null) ? 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                 return (webInterface.getCore().getPostReply(replyId, false) != null) ? webInterface.getCore().getPostReply(replyId, false).getPost().getId() : null;
 339         }
 340
 341         /**
 342          * If the given phrase contains an album ID (optionally prefixed by
 343          * “album://”), returns said album ID, otherwise return {@code null}.
 344          *
 345          * @param phrase
 346          *            The phrase that maybe is an album ID
 347          * @return The album ID, or {@code null}
 348          */
 349         private String getAlbumId(String phrase) {
 350                 String albumId = phrase.startsWith("album://") ? phrase.substring(8) : phrase;
 351                 return (webInterface.getCore().getAlbum(albumId, false) != null) ? albumId : null;
 352         }
 353
 354         /**
 355          * If the given phrase contains an image ID (optionally prefixed by
 356          * “image://”), returns said image ID, otherwise return {@code null}.
 357          *
 358          * @param phrase
 359          *            The phrase that maybe is an image ID
 360          * @return The image ID, or {@code null}
 361          */
 362         private String getImageId(String phrase) {
 363                 String imageId = phrase.startsWith("image://") ? phrase.substring(8) : phrase;
 364                 return (webInterface.getCore().getImage(imageId, false) != null) ? imageId : null;
 365         }
 366
 367         /**
 368          * Converts a given object into a {@link String}.
 369          *
 370          * @param <T>
 371          *            The type of the objects
 372          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 373          */
 374         private static interface StringGenerator<T> {
 375
 376                 /**
 377                  * Generates a {@link String} for the given object.
 378                  *
 379                  * @param object
 380                  *            The object to generate the {@link String} for
 381                  * @return The generated {@link String}
 382                  */
 383                 public String generateString(T object);
 384
 385         }
 386
 387         /**
 388          * Generates a {@link String} from a {@link Sone}, concatenating the name of
 389          * the Sone and all {@link Profile} {@link Field} values.
 390          *
 391          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 392          */
 393         private static class SoneStringGenerator implements StringGenerator<Sone> {
 394
 395                 /** A static instance of a complete Sone string generator. */
 396                 public static final SoneStringGenerator COMPLETE_GENERATOR = new SoneStringGenerator(true);
 397
 398                 /**
 399                  * A static instance of a Sone string generator that will only use the
 400                  * name of the Sone.
 401                  */
 402                 public static final SoneStringGenerator NAME_GENERATOR = new SoneStringGenerator(false);
 403
 404                 /** Whether to generate a string from all data of a Sone. */
 405                 private final boolean complete;
 406
 407                 /**
 408                  * Creates a new Sone string generator.
 409                  *
 410                  * @param complete
 411                  *            {@code true} to use the profile’s fields, {@code false} to
 412                  *            not to use the profile‘s fields
 413                  */
 414                 private SoneStringGenerator(boolean complete) {
 415                         this.complete = complete;
 416                 }
 417
 418                 /**
 419                  * {@inheritDoc}
 420                  */
 421                 @Override
 422                 public String generateString(Sone sone) {
 423                         StringBuilder soneString = new StringBuilder();
 424                         soneString.append(sone.getName());
 425                         Profile soneProfile = sone.getProfile();
 426                         if (soneProfile.getFirstName() != null) {
 427                                 soneString.append(' ').append(soneProfile.getFirstName());
 428                         }
 429                         if (soneProfile.getMiddleName() != null) {
 430                                 soneString.append(' ').append(soneProfile.getMiddleName());
 431                         }
 432                         if (soneProfile.getLastName() != null) {
 433                                 soneString.append(' ').append(soneProfile.getLastName());
 434                         }
 435                         if (complete) {
 436                                 for (Field field : soneProfile.getFields()) {
 437                                         soneString.append(' ').append(field.getValue());
 438                                 }
 439                         }
 440                         return soneString.toString();
 441                 }
 442
 443         }
 444
 445         /**
 446          * Generates a {@link String} from a {@link Post}, concatenating the text of
 447          * the post, the text of all {@link Reply}s, and the name of all
 448          * {@link Sone}s that have replied.
 449          *
 450          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 451          */
 452         private class PostStringGenerator implements StringGenerator<Post> {
 453
 454                 /**
 455                  * {@inheritDoc}
 456                  */
 457                 @Override
 458                 public String generateString(Post post) {
 459                         StringBuilder postString = new StringBuilder();
 460                         postString.append(post.getText());
 461                         if (post.getRecipient() != null) {
 462                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(post.getRecipient()));
 463                         }
 464                         for (PostReply reply : Collections2.filter(webInterface.getCore().getReplies(post), Reply.FUTURE_REPLY_FILTER)) {
 465                                 postString.append(' ').append(SoneStringGenerator.NAME_GENERATOR.generateString(reply.getSone()));
 466                                 postString.append(' ').append(reply.getText());
 467                         }
 468                         return postString.toString();
 469                 }
 470
 471         }
 472
 473         /**
 474          * A search phrase.
 475          *
 476          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 477          */
 478         private static class Phrase {
 479
 480                 /**
 481                  * The optionality of a search phrase.
 482                  *
 483                  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’
 484                  *         Roden</a>
 485                  */
 486                 public enum Optionality {
 487
 488                         /** The phrase is optional. */
 489                         OPTIONAL,
 490
 491                         /** The phrase is required. */
 492                         REQUIRED,
 493
 494                         /** The phrase is forbidden. */
 495                         FORBIDDEN
 496
 497                 }
 498
 499                 /** The phrase to search for. */
 500                 private final String phrase;
 501
 502                 /** The optionality of the phrase. */
 503                 private final Optionality optionality;
 504
 505                 /**
 506                  * Creates a new phrase.
 507                  *
 508                  * @param phrase
 509                  *            The phrase to search for
 510                  * @param optionality
 511                  *            The optionality of the phrase
 512                  */
 513                 public Phrase(String phrase, Optionality optionality) {
 514                         this.optionality = optionality;
 515                         this.phrase = phrase;
 516                 }
 517
 518                 /**
 519                  * Returns the phrase to search for.
 520                  *
 521                  * @return The phrase to search for
 522                  */
 523                 public String getPhrase() {
 524                         return phrase;
 525                 }
 526
 527                 /**
 528                  * Returns the optionality of the phrase.
 529                  *
 530                  * @return The optionality of the phrase
 531                  */
 532                 public Optionality getOptionality() {
 533                         return optionality;
 534                 }
 535
 536                 //
 537                 // OBJECT METHODS
 538                 //
 539
 540                 /**
 541                  * {@inheritDoc}
 542                  */
 543                 @Override
 544                 public int hashCode() {
 545                         return phrase.hashCode() ^ ((optionality == Optionality.FORBIDDEN) ? (0xaaaaaaaa) : ((optionality == Optionality.REQUIRED) ? 0x55555555 : 0));
 546                 }
 547
 548                 /**
 549                  * {@inheritDoc}
 550                  */
 551                 @Override
 552                 public boolean equals(Object object) {
 553                         if (!(object instanceof Phrase)) {
 554                                 return false;
 555                         }
 556                         Phrase phrase = (Phrase) object;
 557                         return (this.optionality == phrase.optionality) && this.phrase.equals(phrase.phrase);
 558                 }
 559
 560         }
 561
 562         /**
 563          * A hit consists of a searched object and the score it got for the phrases
 564          * of the search.
 565          *
 566          * @see SearchPage#calculateScore(List, String)
 567          * @param <T>
 568          *            The type of the searched object
 569          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 570          */
 571         private static class Hit<T> {
 572
 573                 /** Filter for {@link Hit}s with a score of more than 0. */
 574                 public static final Predicate<Hit<?>> POSITIVE_FILTER = new Predicate<Hit<?>>() {
 575
 576                         @Override
 577                         public boolean apply(Hit<?> hit) {
 578                                 return hit.getScore() > 0;
 579                         }
 580
 581                 };
 582
 583                 /** Comparator that sorts {@link Hit}s descending by score. */
 584                 public static final Comparator<Hit<?>> DESCENDING_COMPARATOR = new Comparator<Hit<?>>() {
 585
 586                         @Override
 587                         public int compare(Hit<?> leftHit, Hit<?> rightHit) {
 588                                 return (rightHit.getScore() < leftHit.getScore()) ? -1 : ((rightHit.getScore() > leftHit.getScore()) ? 1 : 0);
 589                         }
 590
 591                 };
 592
 593                 /** The object that was searched. */
 594                 private final T object;
 595
 596                 /** The score of the object. */
 597                 private final double score;
 598
 599                 /**
 600                  * Creates a new hit.
 601                  *
 602                  * @param object
 603                  *            The object that was searched
 604                  * @param score
 605                  *            The score of the object
 606                  */
 607                 public Hit(T object, double score) {
 608                         this.object = object;
 609                         this.score = score;
 610                 }
 611
 612                 /**
 613                  * Returns the object that was searched.
 614                  *
 615                  * @return The object that was searched
 616                  */
 617                 public T getObject() {
 618                         return object;
 619                 }
 620
 621                 /**
 622                  * Returns the score of the object.
 623                  *
 624                  * @return The score of the object
 625                  */
 626                 public double getScore() {
 627                         return score;
 628                 }
 629
 630         }
 631
 632         /**
 633          * Extracts the object from a {@link Hit}.
 634          *
 635          * @param <T>
 636          *            The type of the object to extract
 637          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 638          */
 639         private static class HitMapper<T> implements Function<Hit<T>, T> {
 640
 641                 /**
 642                  * {@inheritDoc}
 643                  */
 644                 @Override
 645                 public T apply(Hit<T> input) {
 646                         return input.getObject();
 647                 }
 648
 649         }
 650
 651 }