src/applications/search/compiler/__tests__/PhutilSearchQueryCompilerTestCase.php

   1 <?php
   2
   3 final class PhutilSearchQueryCompilerTestCase
   4   extends PhutilTestCase {
   5
   6   public function testCompileQueries() {
   7     $tests = array(
   8       '' => null,
   9       'cat dog' => '+"cat" +"dog"',
  10       'cat -dog' => '+"cat" -"dog"',
  11       'cat-dog' => '+"cat-dog"',
  12
  13       // Double quotes serve as delimiters even if there is no whitespace
  14       // between terms.
  15       '"cat"dog' => '+"cat" +"dog"',
  16
  17       // This query is too long.
  18       str_repeat('x', 2048) => false,
  19
  20       // Multiple operators are not permitted.
  21       '++cat' => false,
  22       '+-cat' => false,
  23       '--cat' => false,
  24
  25       // Stray operators are not permitted.
  26       '+' => false,
  27       'cat +' => false,
  28
  29       // Double quotes must be paired.
  30       '"' => false,
  31       'cat "' => false,
  32       '"cat' => false,
  33       'A"' => false,
  34       'A"B"' => '+"A" +"B"',
  35
  36       // Trailing whitespace should be discarded.
  37       'a b ' => '+"a" +"b"',
  38
  39       // Tokens must have search text.
  40       '""' => false,
  41       '-' => false,
  42
  43       // Previously, we permitted spaces to appear inside or after operators.
  44
  45       // Now that "title:-" is now a valid construction meaning "title is
  46       // absent", this had to be tightened. We want "title:- duck" to mean
  47       // "title is absent, and any other field matches 'duck'".
  48       'cat - dog' => false,
  49     );
  50
  51     $this->assertCompileQueries($tests);
  52
  53     // Test that we compile queries correctly if the operators have been
  54     // swapped to use "AND" by default.
  55     $operator_tests = array(
  56       'cat dog' => '"cat" "dog"',
  57       'cat -dog' => '"cat" -"dog"',
  58     );
  59     $this->assertCompileQueries($operator_tests, ' |-><()~*:""&\'');
  60
  61
  62     // Test that we compile queries correctly if the quote operators have
  63     // been swapped to differ.
  64     $quote_tests = array(
  65       'cat dog' => '+[cat] +[dog]',
  66       'cat -dog' => '+[cat] -[dog]',
  67     );
  68     $this->assertCompileQueries($quote_tests, '+ -><()~*:[]&|');
  69   }
  70
  71   public function testCompileQueriesWithStemming() {
  72     $stemming_tests = array(
  73       'cat dog' => array(
  74         null,
  75         '+"cat" +"dog"',
  76       ),
  77       'cats dogs' => array(
  78         null,
  79         '+"cat" +"dog"',
  80       ),
  81       'cats "dogs"' => array(
  82         '+"dogs"',
  83         '+"cat"',
  84       ),
  85       '"blessed blade" of the windseeker' => array(
  86         '+"blessed blade"',
  87         '+"of" +"the" +"windseek"',
  88       ),
  89       'mailing users for mentions on tasks' => array(
  90         null,
  91         '+"mail" +"user" +"for" +"mention" +"on" +"task"',
  92       ),
  93     );
  94
  95     $stemmer = new PhutilSearchStemmer();
  96     $this->assertCompileQueries($stemming_tests, null, $stemmer);
  97   }
  98
  99   public function testCompileQueriesWithFunctions() {
 100     $op_and = PhutilSearchQueryCompiler::OPERATOR_AND;
 101     $op_sub = PhutilSearchQueryCompiler::OPERATOR_SUBSTRING;
 102     $op_exact = PhutilSearchQueryCompiler::OPERATOR_EXACT;
 103     $op_present = PhutilSearchQueryCompiler::OPERATOR_PRESENT;
 104     $op_absent = PhutilSearchQueryCompiler::OPERATOR_ABSENT;
 105
 106     $mao = "\xE7\x8C\xAB";
 107
 108     $function_tests = array(
 109       'cat' => array(
 110         array(null, $op_and, 'cat'),
 111       ),
 112       ':cat' => array(
 113         array(null, $op_and, 'cat'),
 114       ),
 115       'title:cat' => array(
 116         array('title', $op_and, 'cat'),
 117       ),
 118       'title:cat:dog' => array(
 119         array('title', $op_and, 'cat:dog'),
 120       ),
 121       'title:~cat' => array(
 122         array('title', $op_sub, 'cat'),
 123       ),
 124       'cat title:="Meow Meow"' => array(
 125         array(null, $op_and, 'cat'),
 126         array('title', $op_exact, 'Meow Meow'),
 127       ),
 128       'title:cat title:dog' => array(
 129         array('title', $op_and, 'cat'),
 130         array('title', $op_and, 'dog'),
 131       ),
 132       '~"core and seven years ag"' => array(
 133         array(null, $op_sub, 'core and seven years ag'),
 134       ),
 135       $mao => array(
 136         array(null, $op_sub, $mao),
 137       ),
 138       '+'.$mao => array(
 139         array(null, $op_and, $mao),
 140       ),
 141       '~'.$mao => array(
 142         array(null, $op_sub, $mao),
 143       ),
 144       '"'.$mao.'"' => array(
 145         array(null, $op_and, $mao),
 146       ),
 147       'title:' => false,
 148       'title:+' => false,
 149       'title:+""' => false,
 150       'title:""' => false,
 151
 152       'title:~' => array(
 153         array('title', $op_present, null),
 154       ),
 155
 156       'title:-' => array(
 157         array('title', $op_absent, null),
 158       ),
 159
 160       '~' => false,
 161       '-' => false,
 162
 163       // Functions like "title:" apply to following terms if their term is
 164       // not specified with double quotes.
 165       'title:x y' => array(
 166         array('title', $op_and, 'x'),
 167         array('title', $op_and, 'y'),
 168       ),
 169       'title: x y' => array(
 170         array('title', $op_and, 'x'),
 171         array('title', $op_and, 'y'),
 172       ),
 173       'title:"x" y' => array(
 174         array('title', $op_and, 'x'),
 175         array(null, $op_and, 'y'),
 176       ),
 177
 178       // The "present" and "absent" functions are not sticky.
 179       'title:~ x' => array(
 180         array('title', $op_present, null),
 181         array(null, $op_and, 'x'),
 182       ),
 183       'title:- x' => array(
 184         array('title', $op_absent, null),
 185         array(null, $op_and, 'x'),
 186       ),
 187
 188       // Functions like "title:" continue to stick across quotes if the
 189       // quotes aren't the initial argument.
 190       'title:a "b c" d' => array(
 191         array('title', $op_and, 'a'),
 192         array('title', $op_and, 'b c'),
 193         array('title', $op_and, 'd'),
 194       ),
 195
 196       // These queries require a field be both present and absent, which is
 197       // impossible.
 198       'title:- title:x' => false,
 199       'title:- title:~' => false,
 200
 201       'abcdefghijklmnopqrstuvwxyz-ABCDEFGHIJKLMNOPQRSTUVWXYZ:xyz' => array(
 202         array(
 203           'abcdefghijklmnopqrstuvwxyz-ABCDEFGHIJKLMNOPQRSTUVWXYZ',
 204           $op_and,
 205           'xyz',
 206         ),
 207       ),
 208
 209       // See T12995. Interpret CJK tokens as substring queries since these
 210       // languages do not use spaces as word separators.
 211       "\xE7\x8C\xAB" => array(
 212         array(null, $op_sub, "\xE7\x8C\xAB"),
 213       ),
 214
 215       // See T13632. Interpret tokens that begin with "_" as substring tokens
 216       // if no function is specified.
 217       '_x _y_ "_z_"' => array(
 218         array(null, $op_sub, '_x'),
 219         array(null, $op_sub, '_y_'),
 220         array(null, $op_and, '_z_'),
 221       ),
 222     );
 223
 224     $this->assertCompileFunctionQueries($function_tests);
 225   }
 226
 227   private function assertCompileQueries(
 228     array $tests,
 229     $operators = null,
 230     PhutilSearchStemmer $stemmer = null) {
 231     foreach ($tests as $input => $expect) {
 232       $caught = null;
 233
 234       $query = null;
 235       $literal_query = null;
 236       $stemmed_query = null;
 237
 238       try {
 239         $compiler = new PhutilSearchQueryCompiler();
 240
 241         if ($operators !== null) {
 242           $compiler->setOperators($operators);
 243         }
 244
 245         if ($stemmer !== null) {
 246           $compiler->setStemmer($stemmer);
 247         }
 248
 249         $tokens = $compiler->newTokens($input);
 250
 251         if ($stemmer) {
 252           $literal_query = $compiler->compileLiteralQuery($tokens);
 253           $stemmed_query = $compiler->compileStemmedQuery($tokens);
 254         } else {
 255           $query = $compiler->compileQuery($tokens);
 256         }
 257       } catch (PhutilSearchQueryCompilerSyntaxException $ex) {
 258         $caught = $ex;
 259       }
 260
 261       if ($caught !== null) {
 262         $query = false;
 263         $literal_query = false;
 264         $stemmed_query = false;
 265       }
 266
 267       if (!$stemmer) {
 268         $this->assertEqual(
 269           $expect,
 270           $query,
 271           pht('Compilation of query: %s', $input));
 272       } else {
 273         $this->assertEqual(
 274           $expect,
 275           ($literal_query === false)
 276             ? false
 277             : array($literal_query, $stemmed_query),
 278           pht('Stemmed compilation of query: %s', $input));
 279       }
 280     }
 281   }
 282
 283   private function assertCompileFunctionQueries(array $tests) {
 284     foreach ($tests as $input => $expect) {
 285       $compiler = id(new PhutilSearchQueryCompiler())
 286         ->setEnableFunctions(true);
 287
 288       try {
 289         $tokens = $compiler->newTokens($input);
 290
 291         $result = array();
 292         foreach ($tokens as $token) {
 293           $result[] = array(
 294             $token->getFunction(),
 295             $token->getOperator(),
 296             $token->getValue(),
 297           );
 298         }
 299       } catch (PhutilSearchQueryCompilerSyntaxException $ex) {
 300         $result = false;
 301       }
 302
 303       $this->assertEqual(
 304         $expect,
 305         $result,
 306         pht('Function compilation of query: %s', $input));
 307     }
 308   }
 309
 310 }