src/com/hughes/android/dictionary/parser/wiktionary/EnFunctionCallbacks.java

   1 // Copyright 2012 Google Inc. All Rights Reserved.
   2 //
   3 // Licensed under the Apache License, Version 2.0 (the "License");
   4 // you may not use this file except in compliance with the License.
   5 // You may obtain a copy of the License at
   6 //
   7 //     http://www.apache.org/licenses/LICENSE-2.0
   8 //
   9 // Unless required by applicable law or agreed to in writing, software
  10 // distributed under the License is distributed on an "AS IS" BASIS,
  11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  12 // See the License for the specific language governing permissions and
  13 // limitations under the License.
  14
  15 package com.hughes.android.dictionary.parser.wiktionary;
  16
  17 import java.util.ArrayList;
  18 import java.util.Arrays;
  19 import java.util.LinkedHashMap;
  20 import java.util.LinkedHashSet;
  21 import java.util.List;
  22 import java.util.Map;
  23 import java.util.Set;
  24 import java.util.concurrent.atomic.AtomicInteger;
  25
  26 import com.hughes.android.dictionary.engine.EntryTypeName;
  27 import com.hughes.android.dictionary.engine.IndexBuilder;
  28 import com.hughes.android.dictionary.parser.WikiTokenizer;
  29 import com.hughes.android.dictionary.parser.wiktionary.AbstractWiktionaryParser.AppendAndIndexWikiCallback;
  30 import com.hughes.android.dictionary.parser.wiktionary.AbstractWiktionaryParser.NameAndArgs;
  31 import com.hughes.util.ListUtil;
  32 import com.hughes.util.MapUtil;
  33 import com.hughes.util.StringUtil;
  34
  35 class EnFunctionCallbacks {
  36
  37     static final Map<String,FunctionCallback<EnParser>> DEFAULT = new LinkedHashMap<String, FunctionCallback<EnParser>>();
  38
  39     static <T extends AbstractWiktionaryParser> void addGenericCallbacks(Map<String, FunctionCallback<T>> callbacks) {
  40         FunctionCallback<T> callback = new Gender<T>();
  41         callbacks.put("m", callback);
  42         callbacks.put("f", callback);
  43         callbacks.put("n", callback);
  44         callbacks.put("p", callback);
  45         callbacks.put("g", callback);
  46
  47         callbacks.put("etyl", new etyl<T>());
  48         callbacks.put("term", new term<T>());
  49
  50         callback = new EncodingCallback<T>();
  51         Set<String> encodings = new LinkedHashSet<String>(Arrays.asList(
  52                     "IPA", "IPAchar",  // Not really encodings, but it works.
  53                     "zh-ts", "zh-tsp",
  54                     "sd-Arab", "ku-Arab", "Arab", "unicode", "Laoo", "ur-Arab", "Thai",
  55                     "fa-Arab", "Khmr", "Cyrl", "ug-Arab", "ko-inline",
  56                     "Jpan", "Kore", "Hebr", "rfscript", "Beng", "Mong", "Knda", "Cyrs",
  57                     "yue-tsj", "Mlym", "Tfng", "Grek", "yue-yue-j"));
  58         for (final String encoding : encodings) {
  59             callbacks.put(encoding, callback);
  60         }
  61
  62         callback = new Ignore<T>();
  63         callbacks.put("trreq", callback);
  64         callbacks.put("t-image", callback);
  65         callbacks.put("defn", callback);
  66         callbacks.put("rfdef", callback);
  67         callbacks.put("rfdate", callback);
  68         callbacks.put("rfex", callback);
  69         callbacks.put("rfquote", callback);
  70         callbacks.put("attention", callback);
  71         callbacks.put("zh-attention", callback);
  72         callbacks.put("top2", callback);
  73         callbacks.put("mid2", callback);
  74         callbacks.put("top3", callback);
  75         callbacks.put("mid3", callback);
  76         callbacks.put("bottom", callback);
  77         callbacks.put("rel-mid", callback);
  78         callbacks.put("rel-mid3", callback);
  79         callbacks.put("rel-mid4", callback);
  80         callbacks.put("rel-bottom", callback);
  81         callbacks.put("der-top", callback);
  82         callbacks.put("der-mid", callback);
  83         callbacks.put("der-mid3", callback);
  84         callbacks.put("der-bottom", callback);
  85
  86         callback = new AppendName<T>();
  87         callbacks.put("...", callback);
  88
  89         callbacks.put("qualifier", new QualifierCallback<T>());
  90         callbacks.put("italbrac", new italbrac<T>());
  91         callbacks.put("gloss", new gloss<T>());
  92         callbacks.put("not used", new not_used<T>());
  93         callbacks.put("wikipedia", new wikipedia<T>());
  94
  95         final it_conj<T> it_conj_cb = new it_conj<T>();
  96         callbacks.put("it-conj", it_conj_cb);
  97         callbacks.put("it-conj-are", new it_conj_are<T>(it_conj_cb));
  98         callbacks.put("it-conj-arsi", new it_conj_are<T>(it_conj_cb));
  99         callbacks.put("it-conj-care", new it_conj_are<T>(it_conj_cb));
 100         callbacks.put("it-conj-carsi", new it_conj_are<T>(it_conj_cb));
 101         callbacks.put("it-conj-ciare", new it_conj_are<T>(it_conj_cb));
 102         callbacks.put("it-conj-ciarsi", new it_conj_are<T>(it_conj_cb));
 103         callbacks.put("it-conj-iare", new it_conj_are<T>(it_conj_cb));
 104         callbacks.put("it-conj-iarsi", new it_conj_are<T>(it_conj_cb));
 105         callbacks.put("it-conj-iare-b", new it_conj_are<T>(it_conj_cb));
 106         callbacks.put("it-conj-iarsi-b", new it_conj_are<T>(it_conj_cb));
 107         callbacks.put("it-conj-ire", new it_conj_ire<T>(it_conj_cb));
 108         callbacks.put("it-conj-irsi", new it_conj_ire<T>(it_conj_cb));
 109         callbacks.put("it-conj-ire-b", new it_conj_ire<T>(it_conj_cb));
 110         callbacks.put("it-conj-irsi-b", new it_conj_ire<T>(it_conj_cb));
 111         callbacks.put("it-conj-cire", new it_conj_ire<T>(it_conj_cb));
 112         callbacks.put("it-conj-cirsi", new it_conj_ire<T>(it_conj_cb));
 113         callbacks.put("it-conj-ire", new it_conj_ire<T>(it_conj_cb));
 114         callbacks.put("it-conj-ere", new it_conj_ere<T>(it_conj_cb));
 115         callbacks.put("it-conj-ersi", new it_conj_ere<T>(it_conj_cb));
 116         callbacks.put("it-conj-urre", new it_conj_urre<T>(it_conj_cb));
 117         callbacks.put("it-conj-ursi", new it_conj_urre<T>(it_conj_cb));
 118         callbacks.put("it-conj-fare", new it_conj_fare<T>(it_conj_cb));
 119
 120
 121         //"{{it-conj-fare|putre|avere}}\n" +
 122
 123
 124     }
 125
 126     static {
 127         addGenericCallbacks(DEFAULT);
 128
 129         FunctionCallback<EnParser> callback = new TranslationCallback<EnParser>();
 130         DEFAULT.put("t", callback);
 131         DEFAULT.put("t+", callback);
 132         DEFAULT.put("t-", callback);
 133         DEFAULT.put("tø", callback);
 134         DEFAULT.put("apdx-t", callback);
 135
 136         callback = new l_term();
 137         DEFAULT.put("l", callback);
 138         DEFAULT.put("term", callback);
 139
 140         //callback = new AppendArg0();
 141
 142         callback = new FormOf();
 143         DEFAULT.put("form of", callback);
 144         DEFAULT.put("conjugation of", callback);
 145         DEFAULT.put("participle of", callback);
 146         DEFAULT.put("present participle of", callback);
 147         DEFAULT.put("past participle of", callback);
 148         DEFAULT.put("feminine past participle of", callback);
 149         DEFAULT.put("gerund of", callback);
 150         DEFAULT.put("feminine of", callback);
 151         DEFAULT.put("plural of", callback);
 152         DEFAULT.put("feminine plural of", callback);
 153         DEFAULT.put("inflected form of", callback);
 154         DEFAULT.put("alternative form of", callback);
 155         DEFAULT.put("dated form of", callback);
 156         DEFAULT.put("apocopic form of", callback);
 157
 158         callback = new InflOrHead();
 159         DEFAULT.put("infl", callback);
 160         DEFAULT.put("head", callback);
 161     }
 162
 163     static final NameAndArgs<EnParser> NAME_AND_ARGS = new NameAndArgs<EnParser>();
 164
 165     // ------------------------------------------------------------------
 166
 167     static final class TranslationCallback<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 168         @Override
 169         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 170                                       final Map<String, String> namedArgs, final T parser,
 171                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 172
 173             final String transliteration = namedArgs.remove("tr");
 174             final String alt = namedArgs.remove("alt");
 175             namedArgs.keySet().removeAll(EnParser.USELESS_WIKI_ARGS);
 176             if (args.size() < 2) {
 177                 if (!name.equals("ttbc")) {
 178                     EnParser.LOG.warning("{{t...}} with wrong args: title=" + parser.title + ", " + wikiTokenizer.token());
 179                 }
 180                 return false;
 181             }
 182             final String langCode = ListUtil.get(args, 0);
 183             if (!appendAndIndexWikiCallback.langCodeToTCount.containsKey(langCode)) {
 184                 appendAndIndexWikiCallback.langCodeToTCount.put(langCode, new AtomicInteger());
 185             }
 186             appendAndIndexWikiCallback.langCodeToTCount.get(langCode).incrementAndGet();
 187             final String word = ListUtil.get(args, 1);
 188             appendAndIndexWikiCallback.dispatch(alt != null ? alt : word, EntryTypeName.WIKTIONARY_TITLE_MULTI);
 189
 190             // Genders...
 191             if (args.size() > 2) {
 192                 appendAndIndexWikiCallback.builder.append(" {");
 193                 for (int i = 2; i < args.size(); ++i) {
 194                     if (i > 2) {
 195                         appendAndIndexWikiCallback.builder.append("|");
 196                     }
 197                     appendAndIndexWikiCallback.builder.append(args.get(i));
 198                 }
 199                 appendAndIndexWikiCallback.builder.append("}");
 200             }
 201
 202             if (transliteration != null) {
 203                 appendAndIndexWikiCallback.builder.append(" (");
 204                 appendAndIndexWikiCallback.dispatch(transliteration, EntryTypeName.WIKTIONARY_TRANSLITERATION);
 205                 appendAndIndexWikiCallback.builder.append(")");
 206             }
 207
 208             if (alt != null) {
 209                 // If alt wasn't null, we appended alt instead of the actual word
 210                 // we're filing under..
 211                 appendAndIndexWikiCallback.builder.append(" (");
 212                 appendAndIndexWikiCallback.dispatch(word, EntryTypeName.WIKTIONARY_TITLE_MULTI);
 213                 appendAndIndexWikiCallback.builder.append(")");
 214             }
 215
 216             // Catch-all for anything else...
 217             if (!namedArgs.isEmpty()) {
 218                 appendAndIndexWikiCallback.builder.append(" {");
 219                 EnParser.appendNamedArgs(namedArgs, appendAndIndexWikiCallback);
 220                 appendAndIndexWikiCallback.builder.append("}");
 221             }
 222
 223             return true;
 224         }
 225     }
 226
 227     // ------------------------------------------------------------------
 228
 229     static final class QualifierCallback<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 230         @Override
 231         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 232                                       final Map<String, String> namedArgs,
 233                                       final T parser,
 234                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 235             namedArgs.remove("lang");
 236             if (!namedArgs.isEmpty()) {
 237                 EnParser.LOG.warning("weird qualifier: " + wikiTokenizer.token());
 238                 return false;
 239             }
 240             appendAndIndexWikiCallback.builder.append("(");
 241             for (int i = 0; i < args.size(); ++i) {
 242                 appendAndIndexWikiCallback.dispatch(args.get(i), null);
 243                 if (i > 0) {
 244                     appendAndIndexWikiCallback.builder.append(", ");
 245                 }
 246             }
 247             appendAndIndexWikiCallback.builder.append(")");
 248             return true;
 249         }
 250     }
 251
 252     // ------------------------------------------------------------------
 253
 254     static final class EncodingCallback<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 255         @Override
 256         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 257                                       final Map<String, String> namedArgs,
 258                                       final T parser,
 259                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 260             namedArgs.remove("lang");
 261             if (!namedArgs.isEmpty()) {
 262                 EnParser.LOG.warning("weird encoding: " + wikiTokenizer.token());
 263                 return false;
 264             }
 265             if (args.size() == 0) {
 266                 // Things like "{{Jpan}}" exist.
 267                 return true;
 268             }
 269
 270             if (name.equals("IPA")) {
 271                 appendAndIndexWikiCallback.dispatch("IPA: ", null);
 272             }
 273
 274             for (int i = 0; i < args.size(); ++i) {
 275                 if (i > 0) {
 276                     appendAndIndexWikiCallback.builder.append(", ");
 277                 }
 278                 final String arg = args.get(i);
 279 //        if (arg.equals(parser.title)) {
 280 //          parser.titleAppended = true;
 281 //        }
 282                 appendAndIndexWikiCallback.dispatch(arg, appendAndIndexWikiCallback.entryTypeName);
 283             }
 284
 285             return true;
 286         }
 287     }
 288
 289     // ------------------------------------------------------------------
 290
 291     static final class Gender<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 292         @Override
 293         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 294                                       final Map<String, String> namedArgs,
 295                                       final T parser,
 296                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 297             if (!namedArgs.isEmpty()) {
 298                 return false;
 299             }
 300             appendAndIndexWikiCallback.builder.append("{");
 301             appendAndIndexWikiCallback.builder.append(name);
 302             for (int i = 0; i < args.size(); ++i) {
 303                 appendAndIndexWikiCallback.builder.append("|").append(args.get(i));
 304             }
 305             appendAndIndexWikiCallback.builder.append("}");
 306             return true;
 307         }
 308     }
 309
 310     // ------------------------------------------------------------------
 311
 312     static final class l_term implements FunctionCallback<EnParser> {
 313         @Override
 314         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 315                                       final Map<String, String> namedArgs,
 316                                       final EnParser parser,
 317                                       final AppendAndIndexWikiCallback<EnParser> appendAndIndexWikiCallback) {
 318
 319             // for {{l}}, lang is arg 0, but not for {{term}}
 320             if (name.equals("term")) {
 321                 args.add(0, "");
 322             }
 323
 324             final EntryTypeName entryTypeName;
 325             switch (parser.state) {
 326             case TRANSLATION_LINE:
 327                 entryTypeName = EntryTypeName.WIKTIONARY_TRANSLATION_OTHER_TEXT;
 328                 break;
 329             case ENGLISH_DEF_OF_FOREIGN:
 330                 entryTypeName = EntryTypeName.WIKTIONARY_ENGLISH_DEF_WIKI_LINK;
 331                 break;
 332             default:
 333                 throw new IllegalStateException("Invalid enum value: " + parser.state);
 334             }
 335
 336             final String langCode = args.get(0);
 337             final IndexBuilder indexBuilder;
 338             if ("".equals(langCode)) {
 339                 indexBuilder = parser.foreignIndexBuilder;
 340             } else if ("en".equals(langCode)) {
 341                 indexBuilder = parser.enIndexBuilder;
 342             } else {
 343                 indexBuilder = parser.foreignIndexBuilder;
 344             }
 345
 346             String displayText = ListUtil.get(args, 2, "");
 347             if (displayText.equals("")) {
 348                 displayText = ListUtil.get(args, 1, null);
 349             }
 350
 351             if (displayText != null) {
 352                 appendAndIndexWikiCallback.dispatch(displayText, indexBuilder, entryTypeName);
 353             } else {
 354                 EnParser.LOG.warning("no display text: " + wikiTokenizer.token());
 355             }
 356
 357             final String tr = namedArgs.remove("tr");
 358             if (tr != null) {
 359                 appendAndIndexWikiCallback.builder.append(" (");
 360                 appendAndIndexWikiCallback.dispatch(tr, indexBuilder, EntryTypeName.WIKTIONARY_TRANSLITERATION);
 361                 appendAndIndexWikiCallback.builder.append(")");
 362             }
 363
 364             final String gloss = ListUtil.get(args, 3, "");
 365             if (!gloss.equals("")) {
 366                 appendAndIndexWikiCallback.builder.append(" (");
 367                 appendAndIndexWikiCallback.dispatch(gloss, parser.enIndexBuilder, EntryTypeName.WIKTIONARY_ENGLISH_DEF);
 368                 appendAndIndexWikiCallback.builder.append(")");
 369             }
 370
 371             namedArgs.keySet().removeAll(EnParser.USELESS_WIKI_ARGS);
 372             if (!namedArgs.isEmpty()) {
 373                 appendAndIndexWikiCallback.builder.append(" {").append(name);
 374                 EnParser.appendNamedArgs(namedArgs, appendAndIndexWikiCallback);
 375                 appendAndIndexWikiCallback.builder.append("}");
 376             }
 377
 378             return true;
 379         }
 380     }
 381
 382     // ------------------------------------------------------------------
 383
 384     static final class AppendArg0<T extends AbstractWiktionaryParser> implements FunctionCallback<EnParser> {
 385         @Override
 386         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 387                                       final Map<String, String> namedArgs,
 388                                       final EnParser parser,
 389                                       final AppendAndIndexWikiCallback<EnParser> appendAndIndexWikiCallback) {
 390             if (args.size() != 1 || !namedArgs.isEmpty()) {
 391                 return false;
 392             }
 393             appendAndIndexWikiCallback.dispatch(args.get(0), EntryTypeName.WIKTIONARY_TRANSLATION_OTHER_TEXT);
 394
 395             final String tr = namedArgs.remove("tr");
 396             if (tr != null) {
 397                 appendAndIndexWikiCallback.builder.append(" (");
 398                 appendAndIndexWikiCallback.dispatch(tr, EntryTypeName.WIKTIONARY_TRANSLATION_OTHER_TEXT);
 399                 appendAndIndexWikiCallback.builder.append(")");
 400                 parser.wordForms.add(tr);
 401             }
 402
 403             return true;
 404         }
 405     }
 406
 407     // ------------------------------------------------------------------
 408
 409     static final class italbrac<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 410         @Override
 411         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 412                                       final Map<String, String> namedArgs,
 413                                       final T parser,
 414                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 415             if (args.size() != 1 || !namedArgs.isEmpty()) {
 416                 return false;
 417             }
 418             appendAndIndexWikiCallback.builder.append("(");
 419             appendAndIndexWikiCallback.dispatch(args.get(0), EntryTypeName.WIKTIONARY_TRANSLATION_OTHER_TEXT);
 420             appendAndIndexWikiCallback.builder.append(")");
 421             return true;
 422         }
 423     }
 424
 425     // ------------------------------------------------------------------
 426
 427     static final class gloss<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 428         @Override
 429         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 430                                       final Map<String, String> namedArgs,
 431                                       final T parser,
 432                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 433             if (args.size() != 1 || !namedArgs.isEmpty()) {
 434                 return false;
 435             }
 436             appendAndIndexWikiCallback.builder.append("(");
 437             appendAndIndexWikiCallback.dispatch(args.get(0), EntryTypeName.WIKTIONARY_TRANSLATION_OTHER_TEXT);
 438             appendAndIndexWikiCallback.builder.append(")");
 439             return true;
 440         }
 441     }
 442
 443     // ------------------------------------------------------------------
 444
 445     static final class Ignore<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 446         @Override
 447         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 448                                       final Map<String, String> namedArgs,
 449                                       final T parser,
 450                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 451             return true;
 452         }
 453     }
 454
 455     // ------------------------------------------------------------------
 456
 457     static final class not_used<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 458         @Override
 459         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 460                                       final Map<String, String> namedArgs,
 461                                       final T parser,
 462                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 463             appendAndIndexWikiCallback.builder.append("(not used)");
 464             return true;
 465         }
 466     }
 467
 468
 469     // ------------------------------------------------------------------
 470
 471     static final class AppendName<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 472         @Override
 473         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 474                                       final Map<String, String> namedArgs,
 475                                       final T parser,
 476                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 477             if (!args.isEmpty() || !namedArgs.isEmpty()) {
 478                 return false;
 479             }
 480             appendAndIndexWikiCallback.builder.append(name);
 481             return true;
 482         }
 483     }
 484
 485     // --------------------------------------------------------------------
 486     // --------------------------------------------------------------------
 487
 488
 489     static final class FormOf implements FunctionCallback<EnParser> {
 490         @Override
 491         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 492                                       final Map<String, String> namedArgs,
 493                                       final EnParser parser,
 494                                       final AppendAndIndexWikiCallback<EnParser> appendAndIndexWikiCallback) {
 495             parser.entryIsFormOfSomething = true;
 496             String formName = name;
 497             if (name.equals("form of")) {
 498                 formName = ListUtil.remove(args, 0, null);
 499             }
 500             if (formName == null) {
 501                 EnParser.LOG.warning("Missing form name: " + parser.title);
 502                 formName = "form of";
 503             }
 504             String baseForm = ListUtil.get(args, 1, "");
 505             if ("".equals(baseForm)) {
 506                 baseForm = ListUtil.get(args, 0, null);
 507                 ListUtil.remove(args, 1, "");
 508             } else {
 509                 ListUtil.remove(args, 0, null);
 510             }
 511             namedArgs.keySet().removeAll(EnParser.USELESS_WIKI_ARGS);
 512
 513             appendAndIndexWikiCallback.builder.append("{");
 514             NAME_AND_ARGS.onWikiFunction(wikiTokenizer, formName, args, namedArgs, parser, appendAndIndexWikiCallback);
 515             appendAndIndexWikiCallback.builder.append("}");
 516             if (baseForm != null && appendAndIndexWikiCallback.indexedEntry != null) {
 517                 parser.foreignIndexBuilder.addEntryWithString(appendAndIndexWikiCallback.indexedEntry, baseForm, EntryTypeName.WIKTIONARY_BASE_FORM_MULTI);
 518             } else {
 519                 // null baseForm happens in Danish.
 520                 EnParser.LOG.warning("Null baseform: " + parser.title);
 521             }
 522             return true;
 523         }
 524     }
 525
 526     static final EnFunctionCallbacks.FormOf FORM_OF = new FormOf();
 527
 528
 529     // --------------------------------------------------------------------
 530     // --------------------------------------------------------------------
 531
 532     static final class wikipedia<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 533         @Override
 534         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 535                                       final Map<String, String> namedArgs,
 536                                       final T parser,
 537                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 538             namedArgs.remove("lang");
 539             if (args.size() > 1 || !namedArgs.isEmpty()) {
 540                 // Unindexed!
 541                 return false;
 542             } else if (args.size() == 1) {
 543                 return false;
 544             } else {
 545                 return true;
 546             }
 547         }
 548     }
 549
 550     static final class InflOrHead implements FunctionCallback<EnParser> {
 551         @Override
 552         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 553                                       final Map<String, String> namedArgs,
 554                                       final EnParser parser,
 555                                       final AppendAndIndexWikiCallback<EnParser> appendAndIndexWikiCallback) {
 556             // See: http://en.wiktionary.org/wiki/Template:infl
 557             // TODO: Actually these functions should start a new WordPOS:
 558             // See: http://en.wiktionary.org/wiki/quattro
 559             final String langCode = ListUtil.get(args, 0);
 560             String head = namedArgs.remove("head");
 561             if (head == null) {
 562                 head = namedArgs.remove("title"); // Bug
 563             }
 564             if (head == null) {
 565                 head = parser.title;
 566             }
 567
 568             namedArgs.keySet().removeAll(EnParser.USELESS_WIKI_ARGS);
 569
 570             final String tr = namedArgs.remove("tr");
 571             String g = namedArgs.remove("g");
 572             if (g == null) {
 573                 g = namedArgs.remove("gender");
 574             }
 575             final String g2 = namedArgs.remove("g2");
 576             final String g3 = namedArgs.remove("g3");
 577
 578             // We might have already taken care of this in a generic way...
 579             if (!parser.titleAppended) {
 580                 appendAndIndexWikiCallback.dispatch(head, EntryTypeName.WIKTIONARY_TITLE_MULTI);
 581                 parser.titleAppended = true;
 582             }
 583
 584             if (g != null) {
 585                 appendAndIndexWikiCallback.builder.append(" {").append(g);
 586                 if (g2 != null) {
 587                     appendAndIndexWikiCallback.builder.append("|").append(g2);
 588                 }
 589                 if (g3 != null) {
 590                     appendAndIndexWikiCallback.builder.append("|").append(g3);
 591                 }
 592                 appendAndIndexWikiCallback.builder.append("}");
 593             }
 594
 595             if (tr != null) {
 596                 appendAndIndexWikiCallback.builder.append(" (");
 597                 appendAndIndexWikiCallback.dispatch(tr, EntryTypeName.WIKTIONARY_TITLE_MULTI);
 598                 appendAndIndexWikiCallback.builder.append(")");
 599                 parser.wordForms.add(tr);
 600             }
 601
 602             final String pos = ListUtil.get(args, 1);
 603             if (pos != null) {
 604                 appendAndIndexWikiCallback.builder.append(" (").append(pos).append(")");
 605             }
 606             for (int i = 2; i < args.size(); i += 2) {
 607                 final String inflName = ListUtil.get(args, i);
 608                 final String inflValue = ListUtil.get(args, i + 1);
 609                 appendAndIndexWikiCallback.builder.append(", ");
 610                 appendAndIndexWikiCallback.dispatch(inflName, null, null);
 611                 if (inflValue != null && inflValue.length() > 0) {
 612                     appendAndIndexWikiCallback.builder.append(": ");
 613                     appendAndIndexWikiCallback.dispatch(inflValue, null, null);
 614                     parser.wordForms.add(inflValue);
 615                 }
 616             }
 617             for (final String key : namedArgs.keySet()) {
 618                 final String value = WikiTokenizer.toPlainText(namedArgs.get(key));
 619                 appendAndIndexWikiCallback.builder.append(" ");
 620                 appendAndIndexWikiCallback.dispatch(key, null, null);
 621                 appendAndIndexWikiCallback.builder.append("=");
 622                 appendAndIndexWikiCallback.dispatch(value, null, null);
 623                 parser.wordForms.add(value);
 624             }
 625             return true;
 626         }
 627     }
 628
 629     static final class etyl<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 630         @Override
 631         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 632                                       final Map<String, String> namedArgs,
 633                                       final T parser,
 634                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 635             final String langCode = ListUtil.get(args, 0);
 636             if (langCode == null) {
 637                 return false;
 638             }
 639             String langName = WiktionaryLangs.getEnglishName(langCode);
 640             if (langName != null) {
 641                 appendAndIndexWikiCallback.dispatch(langName, null);
 642             } else {
 643                 appendAndIndexWikiCallback.dispatch("lang:" + langCode, null);
 644             }
 645             return true;
 646         }
 647     }
 648
 649     static final class term<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 650         @Override
 651         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 652                                       final Map<String, String> namedArgs,
 653                                       final T parser,
 654                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 655             namedArgs.remove("sc");
 656
 657             // Main text.
 658             final String lang = namedArgs.remove("lang");
 659             String head = ListUtil.get(args, 0);
 660             String display = ListUtil.get(args, 1);
 661             if (StringUtil.isNullOrEmpty(head) && StringUtil.isNullOrEmpty(display)) {
 662                 head = display = parser.title;
 663             }
 664             if (StringUtil.isNullOrEmpty(head)) {
 665                 // Dispatches formatted wiki text.
 666                 appendAndIndexWikiCallback.dispatch(display, null);
 667             } else {
 668                 if (StringUtil.isNullOrEmpty(display)) {
 669                     display = head;
 670                 }
 671                 appendAndIndexWikiCallback.dispatch(String.format("[[%s|%s]]", display, head), null);
 672             }
 673
 674             // Stuff in ()s.
 675             final String tr = namedArgs.remove("tr");
 676             final String pos = namedArgs.remove("pos");
 677             String gloss = ListUtil.get(args, 2);
 678             String literally = namedArgs.remove("lit");
 679             if (!StringUtil.isNullOrEmpty(gloss)) {
 680                 gloss = String.format("\"%s\"", gloss);
 681             }
 682             if (!StringUtil.isNullOrEmpty(literally)) {
 683                 literally = String.format("literally %s", literally);
 684             }
 685             final List<String> inParens = new ArrayList<String>(Arrays.asList(tr, pos, gloss, literally));
 686             cleanList(inParens);
 687             appendCommaSeparatedList(appendAndIndexWikiCallback, inParens);
 688
 689             if (tr != null) {
 690                 parser.addLinkToCurrentEntry(tr, lang, EntryTypeName.WIKTIONARY_MENTIONED);
 691             }
 692             return namedArgs.isEmpty();
 693         }
 694
 695         private void appendCommaSeparatedList(
 696             final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback,
 697             final List<String> inParens) {
 698             if (!inParens.isEmpty()) {
 699                 appendAndIndexWikiCallback.dispatch(" (", null);
 700                 for (int i = 0; i < inParens.size(); ++i) {
 701                     if (i > 0) {
 702                         appendAndIndexWikiCallback.dispatch(", ", null);
 703                     }
 704                     appendAndIndexWikiCallback.dispatch(inParens.get(i), null);
 705                 }
 706                 appendAndIndexWikiCallback.dispatch(")", null);
 707             }
 708         }
 709
 710     }
 711
 712     private static void cleanList(List<String> asList) {
 713         int pos;
 714         while ((pos = asList.indexOf("")) != -1) {
 715             asList.remove(pos);
 716         }
 717         while ((pos = asList.indexOf(null)) != -1) {
 718             asList.remove(pos);
 719         }
 720     }
 721
 722
 723     static {
 724         DEFAULT.put("it-noun", new it_noun());
 725     }
 726     static final class it_noun implements FunctionCallback<EnParser> {
 727         @Override
 728         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 729                                       final Map<String, String> namedArgs,
 730                                       final EnParser parser,
 731                                       final AppendAndIndexWikiCallback<EnParser> appendAndIndexWikiCallback) {
 732             parser.titleAppended = true;
 733             final String gender = ListUtil.get(args, 0);
 734             final String singular = parser.title;
 735             final String plural = ListUtil.get(args, 1, null);
 736             appendAndIndexWikiCallback.builder.append(" ");
 737             appendAndIndexWikiCallback.dispatch(singular, null, null);
 738             appendAndIndexWikiCallback.builder.append(" {").append(gender).append("}, ");
 739             if (plural != null) {
 740                 appendAndIndexWikiCallback.dispatch(plural, null, null);
 741                 appendAndIndexWikiCallback.builder.append(" {pl}");
 742                 parser.wordForms.add(plural);
 743             }
 744             final String f = namedArgs.remove("f");
 745             if (f != null) {
 746                 appendAndIndexWikiCallback.builder.append(", ");
 747                 appendAndIndexWikiCallback.dispatch(f, null, null);
 748                 appendAndIndexWikiCallback.builder.append(" {f}");
 749             }
 750             final String m = namedArgs.remove("m");
 751             if (m != null) {
 752                 appendAndIndexWikiCallback.builder.append(", ");
 753                 appendAndIndexWikiCallback.dispatch(m, null, null);
 754                 appendAndIndexWikiCallback.builder.append(" {m}");
 755             }
 756             parser.wordForms.add(singular);
 757             if (!namedArgs.isEmpty() || args.size() > 4) {
 758                 EnParser.LOG.warning("Invalid it-noun: " + wikiTokenizer.token());
 759             }
 760             return true;
 761         }
 762     }
 763
 764     static {
 765         DEFAULT.put("it-proper noun", new it_proper_noun<EnParser>());
 766     }
 767     static final class it_proper_noun<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 768         @Override
 769         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 770                                       final Map<String, String> namedArgs,
 771                                       final T parser,
 772                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 773             return false;
 774         }
 775     }
 776
 777     // -----------------------------------------------------------------------
 778     // Italian stuff
 779     // -----------------------------------------------------------------------
 780
 781     static final class it_conj_are<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 782         final it_conj<T> dest;
 783         it_conj_are(it_conj<T> dest) {
 784             this.dest = dest;
 785         }
 786         @Override
 787         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 788                                       final Map<String, String> namedArgs,
 789                                       final T parser,
 790                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 791             final String h = name.equals("it-conj-care") || name.equals("it-conj-carsi") ? "h" : "";
 792             final String i = name.equals("it-conj-ciare") || name.equals("it-conj-ciarsi") ? "i" : "";
 793             final String i2 = name.equals("it-conj-iare") || name.equals("it-conj-iarsi") ? "" : "i";
 794             final boolean si = name.equals("it-conj-arsi") || name.equals("it-conj-iarsi") || name.equals("it-conj-iarsi-b") || name.equals("it-conj-carsi") || name.equals("it-conj-ciarsi");
 795             final String root = args.get(0);
 796             passThroughOrFillIn(namedArgs, "inf", root + i + (si ? "arsi" : "are"), false);
 797             namedArgs.put("aux", ListUtil.get(args, 1, ""));
 798             passThroughOrFillIn(namedArgs, "ger", root + i + "ando" + (si ? "si" : ""), true);
 799             passThroughOrFillIn(namedArgs, "presp", root + i + "ante"+ (si ? "si" : ""), true);
 800             passThroughOrFillIn(namedArgs, "pastp", root + i + "ato", true);
 801             if (si) {
 802                 passThroughOrFillIn(namedArgs, "pastp2", root + i + "atosi", true);
 803             }
 804             final String i2b = (name.equals("it-conj-iare-b") || name.equals("it-conj-iarsi-b")) ? "" : i2;
 805
 806             it_conj_passMood(namedArgs, "pres", false, root, Arrays.asList(i + "o", h + i2, i + "a", h + i2 + "amo", i + "ate", i + "ano"));
 807             it_conj_passMood(namedArgs, "imperf", false, root, Arrays.asList(i + "avo", i + "avi", i + "ava", i + "avamo", i + "avate", i + "avano"));
 808             it_conj_passMood(namedArgs, "prem", false, root, Arrays.asList(i + "ai", i + "asti", i + "ò", i + "ammo", i + "aste", i + "arono"));
 809             it_conj_passMood(namedArgs, "fut", true, root, Arrays.asList(h + "erò", h + "erai", h + "erà", h + "eremo", h + "erete", h + "eranno"));
 810             it_conj_passMood(namedArgs, "cond", true, root, Arrays.asList(h + "erei", h + "eresti", h + "erebbe", h + "eremmo", h + "ereste", h + "erebbero"));
 811
 812             passThroughOrFillIn(namedArgs, "sub123s", root + h + i2, false);
 813             passThroughOrFillIn(namedArgs, "sub1p", root + h + i2b + "amo", false);
 814             passThroughOrFillIn(namedArgs, "sub2p", root + h + i2b + "ate", false);
 815             passThroughOrFillIn(namedArgs, "sub3p", root + h + i2 + "no", false);
 816
 817             passThroughOrFillIn(namedArgs, "impsub12s", root + i + "assi", false);
 818             passThroughOrFillIn(namedArgs, "impsub3s", root + i + "asse", false);
 819             passThroughOrFillIn(namedArgs, "impsub1p", root + i + "assimo", false);
 820             passThroughOrFillIn(namedArgs, "impsub2p", root + i + "aste", false);
 821             passThroughOrFillIn(namedArgs, "impsub3p", root + i + "assero", false);
 822
 823             passThroughOrFillIn(namedArgs, "imp2s", root + i + "a" + (si ? "ti" : ""), true);
 824             passThroughOrFillIn(namedArgs, "imp3s", (si ? "si " : "") + root + h + i2, true);
 825             passThroughOrFillIn(namedArgs, "imp1p", root + h + i2b + "amo" + (si ? "ci" : ""), true);
 826             passThroughOrFillIn(namedArgs, "imp2p", root + i + "ate" + (si ? "vi" : ""), true);
 827             passThroughOrFillIn(namedArgs, "imp3p", (si ? "si " : "") + root + h + i2 + "no", true);
 828
 829             return dest.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, appendAndIndexWikiCallback);
 830         }
 831     }
 832
 833     static final class it_conj_ire<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 834         final it_conj<T> dest;
 835         it_conj_ire(it_conj<T> dest) {
 836             this.dest = dest;
 837         }
 838         @Override
 839         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 840                                       final Map<String, String> namedArgs,
 841                                       final T parser,
 842                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 843             final String root = args.get(0);
 844             final String i = name.equals("it-conj-cire") || name.equals("it-conj-cirsi") ? "i" : "";
 845             final boolean si = name.equals("it-conj-irsi") || name.equals("it-conj-irsi-b") || name.equals("it-conj-cirsi");
 846
 847             passThroughOrFillIn(namedArgs, "inf", root + (si ? "irsi" : "ire"), false);
 848             namedArgs.put("aux", ListUtil.get(args, 1, ""));
 849             passThroughOrFillIn(namedArgs, "ger", root + "endo" + (si ? "si" : ""), true);
 850             passThroughOrFillIn(namedArgs, "presp", root + "ente" + (si ? "si" : ""), true);
 851             passThroughOrFillIn(namedArgs, "pastp", root + "ito", true);
 852             if (si) {
 853                 passThroughOrFillIn(namedArgs, "pastp2", root + "itosi", true);
 854             }
 855             if (!name.endsWith("-b")) {
 856                 it_conj_passMood(namedArgs, "pres", false, root, Arrays.asList(i + "o", "i", "e", "iamo", "ite", i + "ono"));
 857             } else {
 858                 it_conj_passMood(namedArgs, "pres", false, root, Arrays.asList("isco", "isci", "isce", "iamo", "ite", "iscono"));
 859             }
 860             it_conj_passMood(namedArgs, "imperf", false, root, Arrays.asList("ivo", "ivi", "iva", "ivamo", "ivate", "ivano"));
 861             it_conj_passMood(namedArgs, "prem", false, root, Arrays.asList("ii", "isti", "ì", "immo", "iste", "irono"));
 862             // Regular past historic synonyms:
 863             passThroughOrFillIn(namedArgs, "prem3s2", root + "é", true);
 864             passThroughOrFillIn(namedArgs, "prem3p2", root + "erono", true);
 865             it_conj_passMood(namedArgs, "fut", true, root, Arrays.asList("irò", "irai", "irà", "iremo", "irete", "iranno"));
 866             it_conj_passMood(namedArgs, "cond", true, root, Arrays.asList("irei", "iresti", "irebbe", "iremmo", "ireste", "irebbero"));
 867
 868             if (!name.endsWith("-b")) {
 869                 passThroughOrFillIn(namedArgs, "sub123s", root + i + "a", false);
 870                 passThroughOrFillIn(namedArgs, "sub3p", root + i + "ano", false);
 871             } else {
 872                 passThroughOrFillIn(namedArgs, "sub123s", root + "isca", false);
 873                 passThroughOrFillIn(namedArgs, "sub3p", root + "iscano", false);
 874             }
 875             passThroughOrFillIn(namedArgs, "sub1p", root + "iamo", false);
 876             passThroughOrFillIn(namedArgs, "sub2p", root + "iate", false);
 877
 878             passThroughOrFillIn(namedArgs, "impsub12s", root + "issi", false);
 879             passThroughOrFillIn(namedArgs, "impsub3s", root + "isse", false);
 880             passThroughOrFillIn(namedArgs, "impsub1p", root + "issimo", false);
 881             passThroughOrFillIn(namedArgs, "impsub2p", root + "iste", false);
 882             passThroughOrFillIn(namedArgs, "impsub3p", root + "issero", false);
 883
 884             if (!name.endsWith("-b")) {
 885                 passThroughOrFillIn(namedArgs, "imp2s", root + "i" + (si ? "ti" : ""), true);
 886                 passThroughOrFillIn(namedArgs, "imp3s", (si ? "si " : "") + root + i + "a", true);
 887                 passThroughOrFillIn(namedArgs, "imp3p", (si ? "si " : "") + root + i + "ano", true);
 888             } else {
 889                 passThroughOrFillIn(namedArgs, "imp2s", root + "isci" + (si ? "ti" : ""), true);
 890                 passThroughOrFillIn(namedArgs, "imp3s", (si ? "si " : "") + root + "isca", true);
 891                 passThroughOrFillIn(namedArgs, "imp3p", (si ? "si " : "") + root + "iscano", true);
 892             }
 893             passThroughOrFillIn(namedArgs, "imp1p", root + "iamo" + (si ? "ci" : ""), true);
 894             passThroughOrFillIn(namedArgs, "imp2p", root + "ite" + (si ? "vi" : ""), true);
 895
 896             return dest.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, appendAndIndexWikiCallback);
 897         }
 898     }
 899
 900
 901     static final class it_conj_ere<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 902         final it_conj<T> dest;
 903         it_conj_ere(it_conj<T> dest) {
 904             this.dest = dest;
 905         }
 906         @Override
 907         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 908                                       final Map<String, String> namedArgs,
 909                                       final T parser,
 910                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 911             final String root = args.get(0);
 912             final boolean si = name.equals("it-conj-ersi");
 913
 914             passThroughOrFillIn(namedArgs, "inf", root + (si ? "ersi" : "ere"), false);
 915             namedArgs.put("aux", ListUtil.get(args, 1, ""));
 916             passThroughOrFillIn(namedArgs, "ger", root + "endo" + (si ? "si" : ""), true);
 917             passThroughOrFillIn(namedArgs, "presp", root + "ente" + (si ? "si" : ""), true);
 918             passThroughOrFillIn(namedArgs, "pastp", root + "uto", true);
 919             if (si) {
 920                 passThroughOrFillIn(namedArgs, "pastp2", root + "utosi", true);
 921             }
 922             it_conj_passMood(namedArgs, "pres", false, root, Arrays.asList("o", "i", "e", "iamo", "ete", "ono"));
 923             it_conj_passMood(namedArgs, "imperf", false, root, Arrays.asList("evo", "evi", "eva", "evamo", "evate", "evano"));
 924             it_conj_passMood(namedArgs, "prem", false, root, Arrays.asList("ei", "esti", "ette", "emmo", "este", "ettero"));
 925             // Regular past historic synonyms:
 926             passThroughOrFillIn(namedArgs, "prem3s2", root + "é", true);
 927             passThroughOrFillIn(namedArgs, "prem3p2", root + "erono", true);
 928             it_conj_passMood(namedArgs, "fut", true, root, Arrays.asList("erò", "erai", "erà", "eremo", "erete", "eranno"));
 929             it_conj_passMood(namedArgs, "cond", true, root, Arrays.asList("erei", "eresti", "erebbe", "eremmo", "ereste", "erebbero"));
 930
 931             passThroughOrFillIn(namedArgs, "sub123s", root + "a", false);
 932             passThroughOrFillIn(namedArgs, "sub1p", root + "iamo", false);
 933             passThroughOrFillIn(namedArgs, "sub2p", root + "iate", false);
 934             passThroughOrFillIn(namedArgs, "sub3p", root + "ano", false);
 935
 936             passThroughOrFillIn(namedArgs, "impsub12s", root + "essi", false);
 937             passThroughOrFillIn(namedArgs, "impsub3s", root + "esse", false);
 938             passThroughOrFillIn(namedArgs, "impsub1p", root + "essimo", false);
 939             passThroughOrFillIn(namedArgs, "impsub2p", root + "este", false);
 940             passThroughOrFillIn(namedArgs, "impsub3p", root + "essero", false);
 941
 942             passThroughOrFillIn(namedArgs, "imp2s", root + "i" + (si ? "ti" : ""), true);
 943             passThroughOrFillIn(namedArgs, "imp3s", (si ? "si " : "") + root + "a", true);
 944             passThroughOrFillIn(namedArgs, "imp1p", root + "iamo" + (si ? "ci" : ""), true);
 945             passThroughOrFillIn(namedArgs, "imp2p", root + "ete" + (si ? "vi" : ""), true);
 946             passThroughOrFillIn(namedArgs, "imp3p", (si ? "si " : "") + root + "ano", true);
 947
 948             return dest.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, appendAndIndexWikiCallback);
 949         }
 950     }
 951
 952     static final class it_conj_urre<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
 953         final it_conj<T> dest;
 954         it_conj_urre(it_conj<T> dest) {
 955             this.dest = dest;
 956         }
 957         @Override
 958         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
 959                                       final Map<String, String> namedArgs,
 960                                       final T parser,
 961                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
 962             final String root = args.get(0);
 963             final boolean si = name.equals("it-conj-ursi");
 964
 965             passThroughOrFillIn(namedArgs, "inf", root + (si ? "ursi" : "urre"), false);
 966             namedArgs.put("aux", ListUtil.get(args, 1, ""));
 967             passThroughOrFillIn(namedArgs, "ger", root + "ucendo" + (si ? "si" : ""), true);
 968             passThroughOrFillIn(namedArgs, "presp", root + "ucente" + (si ? "si" : ""), true);
 969             passThroughOrFillIn(namedArgs, "pastp", root + "otto", true);
 970             if (si) {
 971                 passThroughOrFillIn(namedArgs, "pastp2", root + "ottosi", true);
 972             }
 973             it_conj_passMood(namedArgs, "pres", false, root, Arrays.asList("uco", "uci", "uce", "uciamo", "ucete", "ucono"));
 974             it_conj_passMood(namedArgs, "imperf", false, root, Arrays.asList("ucevo", "ucevi", "uceva", "ucevamo", "ucevate", "ucevano"));
 975             it_conj_passMood(namedArgs, "prem", false, root, Arrays.asList("ussi", "ucesti", "usse", "ucemmo", "uceste", "ussero"));
 976             it_conj_passMood(namedArgs, "fut", true, root, Arrays.asList("urrò", "urrai", "urrà", "urremo", "urrete", "urranno"));
 977             it_conj_passMood(namedArgs, "cond", true, root, Arrays.asList("urrei", "urresti", "urrebbe", "urremmo", "urreste", "urrebbero"));
 978
 979             passThroughOrFillIn(namedArgs, "sub123s", root + "uca", false);
 980             passThroughOrFillIn(namedArgs, "sub1p", root + "uciamo", false);
 981             passThroughOrFillIn(namedArgs, "sub2p", root + "uciate", false);
 982             passThroughOrFillIn(namedArgs, "sub3p", root + "ucano", false);
 983
 984             passThroughOrFillIn(namedArgs, "impsub12s", root + "ucessi", false);
 985             passThroughOrFillIn(namedArgs, "impsub3s", root + "ucesse", false);
 986             passThroughOrFillIn(namedArgs, "impsub1p", root + "ucessimo", false);
 987             passThroughOrFillIn(namedArgs, "impsub2p", root + "uceste", false);
 988             passThroughOrFillIn(namedArgs, "impsub3p", root + "ucessero", false);
 989
 990             passThroughOrFillIn(namedArgs, "imp2s", root + "uci" + (si ? "ti" : ""), true);
 991             passThroughOrFillIn(namedArgs, "imp3s", (si ? "si" : "") + root + "uca", true);
 992             passThroughOrFillIn(namedArgs, "imp1p", root + "uciamo" + (si ? "ci" : ""), true);
 993             passThroughOrFillIn(namedArgs, "imp2p", root + "ucete" + (si ? "vi" : ""), true);
 994             passThroughOrFillIn(namedArgs, "imp3p", (si ? "si" : "") + root + "ucano", true);
 995
 996             return dest.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, appendAndIndexWikiCallback);
 997         }
 998     }
 999
1000     static final class it_conj_fare<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
1001         final it_conj<T> dest;
1002         it_conj_fare(it_conj<T> dest) {
1003             this.dest = dest;
1004         }
1005         @Override
1006         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
1007                                       final Map<String, String> namedArgs,
1008                                       final T parser,
1009                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
1010             final String root = args.get(0);
1011             passThroughOrFillIn(namedArgs, "inf", root + "fare", false);
1012             namedArgs.put("aux", ListUtil.get(args, 1, ""));
1013             passThroughOrFillIn(namedArgs, "ger", root + "facendo", true);
1014             passThroughOrFillIn(namedArgs, "presp", root + "facente", true);
1015             passThroughOrFillIn(namedArgs, "pastp", root + "fatto", true);
1016             it_conj_passMood(namedArgs, "pres", false, root, Arrays.asList("faccio", "fai", "fà", "facciamo", "fate", "fanno"));
1017             passThroughOrFillIn(namedArgs, "pres1s2", root + "fò", true);
1018             it_conj_passMood(namedArgs, "imperf", false, root, Arrays.asList("facevo", "facevi", "faceva", "facevamo", "facevate", "facevano"));
1019             it_conj_passMood(namedArgs, "prem", false, root, Arrays.asList("feci", "facesti", "fece", "facemmo", "faceste", "fecero"));
1020             it_conj_passMood(namedArgs, "fut", true, root, Arrays.asList("farò", "farai", "farà", "faremo", "farete", "faranno"));
1021             it_conj_passMood(namedArgs, "cond", true, root, Arrays.asList("farei", "faresti", "farebbe", "faremmo", "fareste", "farebbero"));
1022
1023             passThroughOrFillIn(namedArgs, "sub123s", root + "faccia", false);
1024             passThroughOrFillIn(namedArgs, "sub1p", root + "facciamo", false);
1025             passThroughOrFillIn(namedArgs, "sub2p", root + "facciate", false);
1026             passThroughOrFillIn(namedArgs, "sub3p", root + "facciano", false);
1027
1028             passThroughOrFillIn(namedArgs, "impsub12s", root + "facessi", false);
1029             passThroughOrFillIn(namedArgs, "impsub3s", root + "facesse", false);
1030             passThroughOrFillIn(namedArgs, "impsub1p", root + "facessimo", false);
1031             passThroughOrFillIn(namedArgs, "impsub2p", root + "faceste", false);
1032             passThroughOrFillIn(namedArgs, "impsub3p", root + "facessero", false);
1033
1034             passThroughOrFillIn(namedArgs, "imp2s", root + "fa", true);
1035             passThroughOrFillIn(namedArgs, "imp3s", root + "faccia", true);
1036             passThroughOrFillIn(namedArgs, "imp1p", root + "facciamo", true);
1037             passThroughOrFillIn(namedArgs, "imp2p", root + "fate", true);
1038             passThroughOrFillIn(namedArgs, "imp3p", root + "facciano", true);
1039
1040             return dest.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, appendAndIndexWikiCallback);
1041         }
1042     }
1043
1044     static final Map<String,String> it_indicativePronouns = new LinkedHashMap<String, String>();
1045     static {
1046         it_indicativePronouns.put("1s", "io");
1047         it_indicativePronouns.put("2s", "tu");
1048         it_indicativePronouns.put("3s", "lui/lei");
1049         it_indicativePronouns.put("1p", "noi");
1050         it_indicativePronouns.put("2p", "voi");
1051         it_indicativePronouns.put("3p", "essi/esse");
1052     }
1053
1054     static final Map<String,String> it_subjunctivePronouns = new LinkedHashMap<String, String>();
1055     static {
1056         it_subjunctivePronouns.put("1s", "che io");
1057         it_subjunctivePronouns.put("2s", "che tu");
1058         it_subjunctivePronouns.put("3s", "che lui/lei");
1059         it_subjunctivePronouns.put("1p", "che noi");
1060         it_subjunctivePronouns.put("2p", "che voi");
1061         it_subjunctivePronouns.put("3p", "che essi/esse");
1062     }
1063
1064     static final Map<String,String> it_imperativePronouns = new LinkedHashMap<String, String>();
1065     static {
1066         it_imperativePronouns.put("1s", "-");
1067         it_imperativePronouns.put("2s", "tu");
1068         it_imperativePronouns.put("3s", "lui/lei");
1069         it_imperativePronouns.put("1p", "noi");
1070         it_imperativePronouns.put("2p", "voi");
1071         it_imperativePronouns.put("3p", "essi/esse");
1072     }
1073
1074
1075     static final class it_conj<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
1076         @Override
1077         public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
1078                                       final Map<String, String> namedArgs,
1079                                       final T parser,
1080                                       final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
1081
1082             final StringBuilder builder = appendAndIndexWikiCallback.builder;
1083
1084             final String inf = namedArgs.get("inf");
1085
1086             // TODO: center everything horizontally.
1087             builder.append("<table style=\"background:#F0F0F0\">");
1088
1089             builder.append("<tr>");
1090             builder.append("<th colspan=\"1\" style=\"background:#e2e4c0\">infinito</th>");
1091             builder.append("<td colspan=\"1\">");
1092             appendAndIndexWikiCallback.dispatch(MapUtil.safeRemove(namedArgs, "inf", "-"), null);
1093             builder.append("</td>");
1094             builder.append("</tr>\n");
1095
1096             builder.append("<tr>");
1097             builder.append("<th colspan=\"1\" style=\"background:#e2e4c0\">verbo ausiliare</th>");
1098             builder.append("<td colspan=\"1\">");
1099             appendAndIndexWikiCallback.dispatch(MapUtil.safeRemove(namedArgs, "aux", "-"), null);
1100             builder.append("</td>");
1101             builder.append("<th colspan=\"1\" style=\"background:#e2e4c0\">gerundio</th>");
1102             builder.append("<td colspan=\"1\">");
1103             outputKeyVariations(appendAndIndexWikiCallback, builder, "ger", namedArgs, true);
1104             builder.append("</td>");
1105             builder.append("</tr>\n");
1106
1107             builder.append("<tr>");
1108             builder.append("<th colspan=\"1\" style=\"background:#e2e4c0\">participio presente</th>");
1109             builder.append("<td colspan=\"1\">");
1110             outputKeyVariations(appendAndIndexWikiCallback, builder, "presp", namedArgs, true);
1111             builder.append("</td>");
1112             builder.append("<th colspan=\"1\" style=\"background:#e2e4c0\">participio passato</th>");
1113             builder.append("<td colspan=\"1\">");
1114             outputKeyVariations(appendAndIndexWikiCallback, builder, "pastp", namedArgs, true);
1115             builder.append("</td>");
1116             builder.append("</tr>\n");
1117
1118             final List<String> prefixes = (inf != null && inf.endsWith("si")) ? it_reflexive_pronouns : it_empty;
1119
1120             String style = " style=\"background:#c0cfe4\"";
1121             outputDataRow(appendAndIndexWikiCallback, style, "indicativo", style, "th", "", new LinkedHashMap<String, String>(it_indicativePronouns), it_empty, false);
1122             outputDataRow(appendAndIndexWikiCallback, style, "presente", "", "td", "pres", namedArgs, prefixes, true);
1123             outputDataRow(appendAndIndexWikiCallback, style, "imperfetto", "", "td", "imperf", namedArgs, prefixes, true);
1124             outputDataRow(appendAndIndexWikiCallback, style, "passato remoto", "", "td", "prem", namedArgs, prefixes, true);
1125             outputDataRow(appendAndIndexWikiCallback, style, "futuro", "", "td", "fut", namedArgs, prefixes, true);
1126
1127             style = " style=\"background:#c0d8e4\"";
1128             outputDataRow(appendAndIndexWikiCallback, style, "condizionale", style, "th", "", new LinkedHashMap<String, String>(it_indicativePronouns), it_empty, false);
1129             outputDataRow(appendAndIndexWikiCallback, style, "presente", "", "td", "cond", namedArgs, prefixes, true);
1130
1131             style = " style=\"background:#c0e4c0\"";
1132             outputDataRow(appendAndIndexWikiCallback, style, "congiuntivo", style, "th", "", new LinkedHashMap<String, String>(it_subjunctivePronouns), it_empty, false);
1133             namedArgs.put("sub3s2", namedArgs.remove("sub3s"));
1134             namedArgs.put("sub1s", namedArgs.get("sub123s"));
1135             namedArgs.put("sub2s", namedArgs.get("sub123s"));
1136             namedArgs.put("sub3s", namedArgs.remove("sub123s"));
1137             namedArgs.put("sub1s2", namedArgs.get("sub123s2"));
1138             namedArgs.put("sub2s2", namedArgs.get("sub123s2"));
1139             namedArgs.put("sub3s2", namedArgs.remove("sub123s2"));
1140             outputDataRow(appendAndIndexWikiCallback, style, "presente", "", "td", "sub", namedArgs, prefixes, true);
1141             namedArgs.put("impsub1s", namedArgs.get("impsub12s"));
1142             namedArgs.put("impsub2s", namedArgs.remove("impsub12s"));
1143             namedArgs.put("impsub1s2", namedArgs.get("impsub12s2"));
1144             namedArgs.put("impsub2s2", namedArgs.remove("impsub12s2"));
1145             outputDataRow(appendAndIndexWikiCallback, style, "imperfetto", "", "td", "impsub", namedArgs, prefixes, true);
1146
1147             style = " style=\"background:#e4d4c0\"";
1148             outputDataRow(appendAndIndexWikiCallback, style, "imperativo", style, "th", "", new LinkedHashMap<String, String>(it_imperativePronouns), it_empty, false);
1149             outputDataRow(appendAndIndexWikiCallback, style, "", "", "td", "imp", namedArgs, it_empty, false);  // these are attached to the stem.
1150
1151             builder.append("</table>\n");
1152
1153             if (!namedArgs.isEmpty()) {
1154                 System.err.println("NON-EMPTY namedArgs: " + namedArgs);
1155                 if ("muovesse".equals(namedArgs.get("impsib3s2"))) {
1156                     return false;
1157                 }
1158                 if ("percuotesse".equals(namedArgs.get("impsib3s2"))) {
1159                     return false;
1160                 }
1161                 // Too many to deal with:
1162                 //assert false;
1163                 return false;
1164             }
1165
1166             return true;
1167         }
1168
1169         private void outputDataRow(AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback,
1170                                    String col1Style, String headerName,
1171                                    String col2Style, final String type2,
1172                                    String moodName, Map<String, String> namedArgs, final List<String> prefixes, final boolean isForm) {
1173             final StringBuilder builder = appendAndIndexWikiCallback.builder;
1174             builder.append("<tr>");
1175             builder.append("<th colspan=\"1\"").append(col1Style).append(">").append(headerName).append("</th>");
1176             int i = 0;
1177             for (final String number : it_number_s_p) {
1178                 for (final String person : it_person_1_2_3) {
1179                     // Output <td> or <th>
1180                     builder.append("<").append(type2).append("").append(col2Style).append(">");
1181                     final String keyBase = String.format("%s%s%s", moodName, person, number);
1182                     appendAndIndexWikiCallback.dispatch(prefixes.get(i++), null);
1183                     outputKeyVariations(appendAndIndexWikiCallback, builder, keyBase, namedArgs, isForm);
1184                     // Output <td> or <th>
1185                     builder.append("</").append(type2).append(">");
1186                 }
1187             }
1188             builder.append("</tr>\n");
1189         }
1190     }
1191
1192     static void passThroughOrFillIn(final Map<String,String> namedArgs, final String key, final String fillIn, final boolean quoteToEmpty) {
1193         final String value = namedArgs.get(key);
1194         if (quoteToEmpty && "''".equals(value)) {
1195             namedArgs.put(key, "");
1196             return;
1197         }
1198         if (value == null || value.equals("")) {
1199             namedArgs.put(key, fillIn);
1200         }
1201     }
1202
1203     static final List<String> it_number_s_p = Arrays.asList("s", "p");
1204     static final List<String> it_person_1_2_3 = Arrays.asList("1", "2", "3");
1205     static final List<String> it_reflexive_pronouns = Arrays.asList("mi ", "ti ", "si ", "ci ", "vi ", "si ");
1206     static final List<String> it_empty = Arrays.asList("", "", "", "", "", "");
1207     static void it_conj_passMood(final Map<String,String> namedArgs, final String moodName, final boolean quoteToEmpty, final String root, final List<String> suffixes) {
1208         assert suffixes.size() == 6;
1209         int i = 0;
1210         for (final String number : it_number_s_p) {
1211             for (final String person : it_person_1_2_3) {
1212                 passThroughOrFillIn(namedArgs, String.format("%s%s%s", moodName, person, number), root + suffixes.get(i), quoteToEmpty);
1213                 ++i;
1214             }
1215         }
1216     }
1217
1218     private static <T extends AbstractWiktionaryParser> void outputKeyVariations(AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback,
1219             final StringBuilder builder, final String keyBase, Map<String, String> namedArgs, boolean isForm) {
1220         for (int suffix = 0; suffix <= 4; ++suffix) {
1221             final String key = suffix == 0 ? keyBase : keyBase + suffix;
1222             final String val = namedArgs.remove(key);
1223             if (val != null && !val.trim().equals("")) {
1224                 if (suffix > 0) {
1225                     builder.append(", ");
1226                 }
1227                 appendAndIndexWikiCallback.dispatch(val, null);
1228                 if (isForm) {
1229                     appendAndIndexWikiCallback.parser.addLinkToCurrentEntry(val, null, EntryTypeName.WIKTIONARY_INFLECTED_FORM_MULTI);
1230                 }
1231             }
1232         }
1233     }
1234
1235
1236 }