src/chrome/browser/spellchecker/spellcheck_platform_mac_unittest.cc

   1 // Copyright (c) 2011 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "base/bind.h"
   6 #include "base/message_loop/message_loop.h"
   7 #include "base/strings/string_util.h"
   8 #include "base/strings/utf_string_conversions.h"
   9 #include "chrome/browser/spellchecker/spellcheck_platform_mac.h"
  10 #include "chrome/common/spellcheck_result.h"
  11 #include "content/public/test/test_utils.h"
  12 #include "testing/gtest/include/gtest/gtest.h"
  13
  14 namespace {
  15
  16 class SpellcheckMacTest: public testing::Test {
  17  public:
  18   SpellcheckMacTest()
  19       : callback_(base::Bind(&SpellcheckMacTest::CompletionCallback,
  20                              base::Unretained(this))),
  21         callback_finished_(false) {}
  22
  23   void WaitForCallback() {
  24     content::RunMessageLoop();
  25   }
  26
  27   std::vector<SpellCheckResult> results_;
  28   spellcheck_mac::TextCheckCompleteCallback callback_;
  29   bool callback_finished_;
  30
  31  private:
  32   void QuitMessageLoop() {
  33     CHECK(base::MessageLoop::current() == &message_loop_);
  34     base::MessageLoop::current()->Quit();
  35   }
  36
  37   void CompletionCallback(const std::vector<SpellCheckResult>& results) {
  38     results_ = results;
  39     callback_finished_ = true;
  40     message_loop_.PostTask(FROM_HERE,
  41                            base::Bind(&SpellcheckMacTest::QuitMessageLoop,
  42                                       base::Unretained(this)));
  43   }
  44
  45   base::MessageLoopForUI message_loop_;
  46   spellcheck_mac::ScopedEnglishLanguageForTest scoped_language_;
  47 };
  48
  49 // Tests that words are properly ignored. Currently only enabled on OS X as it
  50 // is the only platform to support ignoring words. Note that in this test, we
  51 // supply a non-zero doc_tag, in order to test that ignored words are matched to
  52 // the correct document.
  53 TEST_F(SpellcheckMacTest, IgnoreWords_EN_US) {
  54   const char* kTestCases[] = {
  55     "teh",
  56     "morblier",
  57     "watre",
  58     "noooen",
  59   };
  60
  61   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(kTestCases); ++i) {
  62     const base::string16 word(base::ASCIIToUTF16(kTestCases[i]));
  63     const int doc_tag = spellcheck_mac::GetDocumentTag();
  64
  65     // The word should show up as misspelled.
  66     EXPECT_FALSE(spellcheck_mac::CheckSpelling(word, doc_tag)) << word;
  67
  68     // Ignore the word.
  69     spellcheck_mac::IgnoreWord(word);
  70
  71     // The word should now show up as correctly spelled.
  72     EXPECT_TRUE(spellcheck_mac::CheckSpelling(word, doc_tag)) << word;
  73
  74     // Close the docuemnt. Any words that we had previously ignored should no
  75     // longer be ignored and thus should show up as misspelled.
  76     spellcheck_mac::CloseDocumentWithTag(doc_tag);
  77
  78     // The word should now show be spelled wrong again
  79     EXPECT_FALSE(spellcheck_mac::CheckSpelling(word, doc_tag)) << word;
  80   }
  81 }  // Test IgnoreWords_EN_US
  82
  83 TEST_F(SpellcheckMacTest, SpellCheckSuggestions_EN_US) {
  84   static const struct {
  85     const char* input;           // A string to be tested.
  86     const char* suggested_word;  // A suggested word that should occur.
  87   } kTestCases[] = {
  88     // We need to have separate test cases here, since hunspell and the OS X
  89     // spellchecking service occasionally differ on what they consider a valid
  90     // suggestion for a given word, although these lists could likely be
  91     // integrated somewhat. The test cases for non-Mac are in
  92     // chrome/renderer/spellcheck_unittest.cc
  93     // These words come from the wikipedia page of the most commonly
  94     // misspelled words in english.
  95     // (http://en.wikipedia.org/wiki/Commonly_misspelled_words).
  96     // However, 10.6 loads multiple dictionaries and enables many non-English
  97     // dictionaries by default. As a result, we have removed from the list any
  98     // word that is marked as correct because it is correct in another
  99     // language.
 100     {"absense", "absence"},
 101     {"acceptible", "acceptable"},
 102     {"accidentaly", "accidentally"},
 103     {"acheive", "achieve"},
 104     {"acknowlege", "acknowledge"},
 105     {"acquaintence", "acquaintance"},
 106     {"aquire", "acquire"},
 107     {"aquit", "acquit"},
 108     {"acrage", "acreage"},
 109     {"adultary", "adultery"},
 110     {"advertize", "advertise"},
 111     {"adviseable", "advisable"},
 112     {"alchohol", "alcohol"},
 113     {"alege", "allege"},
 114     {"allegaince", "allegiance"},
 115     {"allmost", "almost"},
 116     // Ideally, this test should pass. It works in firefox, but not in hunspell
 117     // or OS X.
 118     // {"alot", "a lot"},
 119     {"amatuer", "amateur"},
 120     {"ammend", "amend"},
 121     {"amung", "among"},
 122     {"anually", "annually"},
 123     {"apparant", "apparent"},
 124     {"artic", "arctic"},
 125     {"arguement", "argument"},
 126     {"athiest", "atheist"},
 127     {"athelete", "athlete"},
 128     {"avrage", "average"},
 129     {"awfull", "awful"},
 130     {"ballance", "balance"},
 131     {"basicly", "basically"},
 132     {"becuase", "because"},
 133     {"becomeing", "becoming"},
 134     {"befor", "before"},
 135     {"begining", "beginning"},
 136     {"beleive", "believe"},
 137     {"bellweather", "bellwether"},
 138     {"benifit", "benefit"},
 139     {"bouy", "buoy"},
 140     {"briliant", "brilliant"},
 141     {"burgler", "burglar"},
 142     {"camoflage", "camouflage"},
 143     {"carefull", "careful"},
 144     {"Carribean", "Caribbean"},
 145     {"catagory", "category"},
 146     {"cauhgt", "caught"},
 147     {"cieling", "ceiling"},
 148     {"cemetary", "cemetery"},
 149     {"certin", "certain"},
 150     {"changable", "changeable"},
 151     {"cheif", "chief"},
 152     {"citezen", "citizen"},
 153     {"collaegue", "colleague"},
 154     {"colum", "column"},
 155     {"comming", "coming"},
 156     {"commited", "committed"},
 157     {"compitition", "competition"},
 158     {"conceed", "concede"},
 159     {"congradulate", "congratulate"},
 160     {"consciencious", "conscientious"},
 161     {"concious", "conscious"},
 162     {"concensus", "consensus"},
 163     {"contraversy", "controversy"},
 164     {"conveniance", "convenience"},
 165     {"critecize", "criticize"},
 166     {"dacquiri", "daiquiri"},
 167     {"decieve", "deceive"},
 168     {"dicide", "decide"},
 169     {"definate", "definite"},
 170     {"definitly", "definitely"},
 171     {"desparate", "desperate"},
 172     {"develope", "develop"},
 173     {"diffrence", "difference"},
 174     {"disapear", "disappear"},
 175     {"disapoint", "disappoint"},
 176     {"disasterous", "disastrous"},
 177     {"disipline", "discipline"},
 178     {"drunkeness", "drunkenness"},
 179     {"dumbell", "dumbbell"},
 180     {"easely", "easily"},
 181     {"eigth", "eight"},
 182     {"embarass", "embarrass"},
 183     {"enviroment", "environment"},
 184     {"equiped", "equipped"},
 185     {"equiptment", "equipment"},
 186     {"exagerate", "exaggerate"},
 187     {"exellent", "excellent"},
 188     {"exsept", "except"},
 189     {"exercize", "exercise"},
 190     {"exilerate", "exhilarate"},
 191     {"existance", "existence"},
 192     {"experiance", "experience"},
 193     {"experament", "experiment"},
 194     {"explaination", "explanation"},
 195     {"facinating", "fascinating"},
 196     {"firey", "fiery"},
 197     {"finaly", "finally"},
 198     {"flourescent", "fluorescent"},
 199     {"foriegn", "foreign"},
 200     {"fourty", "forty"},
 201     {"foreward", "forward"},
 202     {"freind", "friend"},
 203     {"fundemental", "fundamental"},
 204     {"guage", "gauge"},
 205     {"generaly", "generally"},
 206     {"goverment", "government"},
 207     {"gratefull", "grateful"},
 208     {"garantee", "guarantee"},
 209     {"guidence", "guidance"},
 210     {"happyness", "happiness"},
 211     {"harrass", "harass"},
 212     {"heighth", "height"},
 213     {"heirarchy", "hierarchy"},
 214     {"humerous", "humorous"},
 215     {"hygene", "hygiene"},
 216     {"hipocrit", "hypocrite"},
 217     {"idenity", "identity"},
 218     {"ignorence", "ignorance"},
 219     {"imaginery", "imaginary"},
 220     {"immitate", "imitate"},
 221     {"immitation", "imitation"},
 222     {"imediately", "immediately"},
 223     {"incidently", "incidentally"},
 224     {"independant", "independent"},
 225     {"indispensible", "indispensable"},
 226     {"innoculate", "inoculate"},
 227     {"inteligence", "intelligence"},
 228     {"intresting", "interesting"},
 229     {"interuption", "interruption"},
 230     {"irrelevent", "irrelevant"},
 231     {"irritible", "irritable"},
 232     {"jellous", "jealous"},
 233     {"knowlege", "knowledge"},
 234     {"labratory", "laboratory"},
 235     {"lenght", "length"},
 236     {"liason", "liaison"},
 237     {"libary", "library"},
 238     {"lisence", "license"},
 239     {"lonelyness", "loneliness"},
 240     {"lieing", "lying"},
 241     {"maintenence", "maintenance"},
 242     {"manuever", "maneuver"},
 243     {"marrige", "marriage"},
 244     {"mathmatics", "mathematics"},
 245     {"medcine", "medicine"},
 246     {"miniture", "miniature"},
 247     {"minite", "minute"},
 248     {"mischevous", "mischievous"},
 249     {"mispell", "misspell"},
 250     // Maybe this one should pass, as it works in hunspell, but not in firefox.
 251     // {"misterius", "mysterious"},
 252     {"naturaly", "naturally"},
 253     {"neccessary", "necessary"},
 254     {"neice", "niece"},
 255     {"nieghbor", "neighbor"},
 256     {"nieghbour", "neighbor"},
 257     {"niether", "neither"},
 258     {"noticable", "noticeable"},
 259     {"occassion", "occasion"},
 260     {"occasionaly", "occasionally"},
 261     {"occurrance", "occurrence"},
 262     {"occured", "occurred"},
 263     {"ommision", "omission"},
 264     {"oppurtunity", "opportunity"},
 265     {"outragous", "outrageous"},
 266     {"parrallel", "parallel"},
 267     {"parliment", "parliament"},
 268     {"particurly", "particularly"},
 269     {"passtime", "pastime"},
 270     {"peculier", "peculiar"},
 271     {"percieve", "perceive"},
 272     {"pernament", "permanent"},
 273     {"perseverence", "perseverance"},
 274     {"personaly", "personally"},
 275     {"persaude", "persuade"},
 276     {"pichure", "picture"},
 277     {"peice", "piece"},
 278     {"plagerize", "plagiarize"},
 279     {"playright", "playwright"},
 280     {"plesant", "pleasant"},
 281     {"pollitical", "political"},
 282     {"posession", "possession"},
 283     {"potatos", "potatoes"},
 284     {"practicle", "practical"},
 285     {"preceed", "precede"},
 286     {"predjudice", "prejudice"},
 287     {"presance", "presence"},
 288     {"privelege", "privilege"},
 289     // This one should probably work. It does in FF and Hunspell.
 290     // {"probly", "probably"},
 291     {"proffesional", "professional"},
 292     {"promiss", "promise"},
 293     {"pronounciation", "pronunciation"},
 294     {"prufe", "proof"},
 295     {"psycology", "psychology"},
 296     {"publically", "publicly"},
 297     {"quanity", "quantity"},
 298     {"quarentine", "quarantine"},
 299     {"questionaire", "questionnaire"},
 300     {"readible", "readable"},
 301     {"realy", "really"},
 302     {"recieve", "receive"},
 303     {"reciept", "receipt"},
 304     {"reconize", "recognize"},
 305     {"recomend", "recommend"},
 306     {"refered", "referred"},
 307     {"referance", "reference"},
 308     {"relevent", "relevant"},
 309     {"religous", "religious"},
 310     {"repitition", "repetition"},
 311     {"restarant", "restaurant"},
 312     {"rythm", "rhythm"},
 313     {"rediculous", "ridiculous"},
 314     {"sacrefice", "sacrifice"},
 315     {"saftey", "safety"},
 316     {"sissors", "scissors"},
 317     {"secratary", "secretary"},
 318     {"seperate", "separate"},
 319     {"sargent", "sergeant"},
 320     {"shineing", "shining"},
 321     {"similer", "similar"},
 322     {"sinceerly", "sincerely"},
 323     {"speach", "speech"},
 324     {"strenght", "strength"},
 325     {"succesful", "successful"},
 326     {"supercede", "supersede"},
 327     {"surelly", "surely"},
 328     {"suprise", "surprise"},
 329     {"temperture", "temperature"},
 330     {"temprary", "temporary"},
 331     {"tommorrow", "tomorrow"},
 332     {"tounge", "tongue"},
 333     {"truely", "truly"},
 334     {"twelth", "twelfth"},
 335     {"tyrany", "tyranny"},
 336     {"underate", "underrate"},
 337     {"untill", "until"},
 338     {"unuseual", "unusual"},
 339     {"upholstry", "upholstery"},
 340     {"usible", "usable"},
 341     {"useing", "using"},
 342     {"usualy", "usually"},
 343     {"vaccuum", "vacuum"},
 344     {"vegatarian", "vegetarian"},
 345     {"vehical", "vehicle"},
 346     {"visious", "vicious"},
 347     {"villege", "village"},
 348     {"wierd", "weird"},
 349     {"wellcome", "welcome"},
 350     {"wellfare", "welfare"},
 351     {"wilfull", "willful"},
 352     {"withold", "withhold"},
 353     {"writting", "writing"},
 354   };
 355
 356   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(kTestCases); ++i) {
 357     const base::string16 word(base::ASCIIToUTF16(kTestCases[i].input));
 358     EXPECT_FALSE(spellcheck_mac::CheckSpelling(word, 0)) << word;
 359
 360     // Check if the suggested words occur.
 361     std::vector<base::string16> suggestions;
 362     spellcheck_mac::FillSuggestionList(word, &suggestions);
 363     bool suggested_word_is_present = false;
 364     const base::string16 suggested_word(
 365         base::ASCIIToUTF16(kTestCases[i].suggested_word));
 366     for (size_t j = 0; j < suggestions.size(); j++) {
 367       if (suggestions[j].compare(suggested_word) == 0) {
 368         suggested_word_is_present = true;
 369         break;
 370       }
 371     }
 372     EXPECT_TRUE(suggested_word_is_present) << suggested_word;
 373   }
 374 }
 375
 376 // The OSX spellchecker returns non-spellcheck results when invoked on a
 377 // sentence, specifically an NSTextCheckingTypeOrthography result indicating
 378 // the language used in that sentence. Test that it is filtered out from
 379 // RequestTextCheck results.
 380 TEST_F(SpellcheckMacTest, SpellCheckIgnoresOrthography)  {
 381   base::string16 test_string(base::ASCIIToUTF16("Icland is awesome."));
 382   spellcheck_mac::RequestTextCheck(0, test_string, callback_);
 383   WaitForCallback();
 384   EXPECT_TRUE(callback_finished_);
 385   EXPECT_EQ(1U, results_.size());
 386 }
 387
 388 }  // namespace