Complex classes like PhoneNumberMatcherTest often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes. You can also have a look at the cohesion graph to spot any un-connected, or weakly-connected components.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use PhoneNumberMatcherTest, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 14 | class PhoneNumberMatcherTest extends \PHPUnit_Framework_TestCase |
||
| 15 | { |
||
| 16 | /** |
||
| 17 | * @var PhoneNumberUtil |
||
| 18 | */ |
||
| 19 | protected $phoneUtil; |
||
| 20 | |||
| 21 | public function setUp() |
||
| 22 | { |
||
| 23 | PhoneNumberUtil::resetInstance(); |
||
| 24 | $this->phoneUtil = PhoneNumberUtil::getInstance( |
||
| 25 | PhoneNumberUtilTest::TEST_META_DATA_FILE_PREFIX, |
||
| 26 | CountryCodeToRegionCodeMapForTesting::$countryCodeToRegionCodeMapForTesting |
||
| 27 | ); |
||
| 28 | } |
||
| 29 | |||
| 30 | public function testContainsMoreThanOneSlashInNationalNumber() |
||
| 31 | { |
||
| 32 | // A date should return true. |
||
| 33 | $number = new PhoneNumber(); |
||
| 34 | $number->setCountryCode(1); |
||
| 35 | $number->setCountryCodeSource(CountryCodeSource::FROM_DEFAULT_COUNTRY); |
||
| 36 | $candidate = "1/05/2013"; |
||
| 37 | $this->assertTrue(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 38 | |||
| 39 | // Here, the country code source thinks it started with a country calling code, but this is not |
||
| 40 | // the same as the part before the slash, so it's still true. |
||
| 41 | $number = new PhoneNumber(); |
||
| 42 | $number->setCountryCodeSource(274); |
||
| 43 | $number->setCountryCodeSource(CountryCodeSource::FROM_NUMBER_WITHOUT_PLUS_SIGN); |
||
| 44 | $candidate = "27/4/2013"; |
||
| 45 | $this->assertTrue(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 46 | |||
| 47 | // Now it should be false, because the first slash is after the country calling code. |
||
| 48 | $number = new PhoneNumber(); |
||
| 49 | $number->setCountryCode(49); |
||
| 50 | $number->setCountryCodeSource(CountryCodeSource::FROM_NUMBER_WITH_PLUS_SIGN); |
||
| 51 | $candidate = "49/69/2013"; |
||
| 52 | $this->assertFalse(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 53 | |||
| 54 | $number = new PhoneNumber(); |
||
| 55 | $number->setCountryCode(49); |
||
| 56 | $number->setCountryCodeSource(CountryCodeSource::FROM_NUMBER_WITHOUT_PLUS_SIGN); |
||
| 57 | $candidate = "+49/69/2013"; |
||
| 58 | $this->assertFalse(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 59 | |||
| 60 | $candidate = "+ 49/69/2013"; |
||
| 61 | $this->assertFalse(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 62 | |||
| 63 | $candidate = "+ 49/69/20/13"; |
||
| 64 | $this->assertTrue(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 65 | |||
| 66 | // Here, the first group is not assumed to be the country calling code, even though it is the |
||
| 67 | // same as it, so this should return true. |
||
| 68 | $number = new PhoneNumber(); |
||
| 69 | $number->setCountryCode(49); |
||
| 70 | $number->setCountryCodeSource(CountryCodeSource::FROM_DEFAULT_COUNTRY); |
||
| 71 | $candidate = "49/69/2013"; |
||
| 72 | $this->assertTrue(PhoneNumberMatcher::containsMoreThanOneSlashInNationalNumber($number, $candidate)); |
||
| 73 | } |
||
| 74 | |||
| 75 | public function testFindNationalNumber() |
||
| 76 | { |
||
| 77 | // same cases as in testParseNationalNumber |
||
| 78 | $this->doTestFindInContext("033316005", RegionCode::NZ); |
||
| 79 | // ("33316005", RegionCode.NZ) is omitted since the national prefix is obligatory for these |
||
| 80 | // types of numbers in New Zealand. |
||
| 81 | // National prefix attached and some formatting present. |
||
| 82 | $this->doTestFindInContext("03-331 6005", RegionCode::NZ); |
||
| 83 | $this->doTestFindInContext("03 331 6005", RegionCode::NZ); |
||
| 84 | // Testing international prefixes. |
||
| 85 | // Should strip country code. |
||
| 86 | $this->doTestFindInContext("0064 3 331 6005", RegionCode::NZ); |
||
| 87 | // Try again, but this time we have an international number with Region Code US. It should |
||
| 88 | // recognize the country code and parse accordingly. |
||
| 89 | $this->doTestFindInContext("01164 3 331 6005", RegionCode::US); |
||
| 90 | $this->doTestFindInContext("+64 3 331 6005", RegionCode::US); |
||
| 91 | |||
| 92 | $this->doTestFindInContext("64(0)64123456", RegionCode::NZ); |
||
| 93 | // Check that using a "/" is fine in a phone number. |
||
| 94 | // Note that real Polish numbers do *not* start with a 0. |
||
| 95 | $this->doTestFindInContext("0123/456789", RegionCode::PL); |
||
| 96 | $this->doTestFindInContext("123-456-7890", RegionCode::US); |
||
| 97 | } |
||
| 98 | |||
| 99 | public function testFindWithInternationalPrefixes() |
||
| 100 | { |
||
| 101 | $this->doTestFindInContext("+1 (650) 333-6000", RegionCode::NZ); |
||
| 102 | $this->doTestFindInContext("1-650-333-6000", RegionCode::US); |
||
| 103 | // Calling the US number from Singapore by using different service providers |
||
| 104 | // 1st test: calling using SingTel IDD service (IDD is 001) |
||
| 105 | $this->doTestFindInContext("0011-650-333-6000", RegionCode::SG); |
||
| 106 | // 2nd test: calling using StarHub IDD service (IDD is 008) |
||
| 107 | $this->doTestFindInContext("0081-650-333-6000", RegionCode::SG); |
||
| 108 | // 3rd test: calling using SingTel V019 service (IDD is 019) |
||
| 109 | $this->doTestFindInContext("0191-650-333-6000", RegionCode::SG); |
||
| 110 | // Calling the US number from Poland |
||
| 111 | $this->doTestFindInContext("0~01-650-333-6000", RegionCode::PL); |
||
| 112 | // Using "++" at the start. |
||
| 113 | $this->doTestFindInContext("++1 (650) 333-6000", RegionCode::PL); |
||
| 114 | // Using a full-width plus sign. |
||
| 115 | $this->doTestFindInContext("\xEF\xBC\x8B1 (650) 333-6000", RegionCode::SG); |
||
| 116 | // The whole number, including punctuation, is here represented in full-width form. |
||
| 117 | $this->doTestFindInContext("+1 (650) 333-6000", RegionCode::SG); |
||
| 118 | } |
||
| 119 | |||
| 120 | public function testFindWithLeadingZero() |
||
| 121 | { |
||
| 122 | $this->doTestFindInContext("+39 02-36618 300", RegionCode::NZ); |
||
| 123 | $this->doTestFindInContext("02-36618 300", RegionCode::IT); |
||
| 124 | $this->doTestFindInContext("312 345 678", RegionCode::IT); |
||
| 125 | } |
||
| 126 | |||
| 127 | public function testFindNationalNumberArgentina() |
||
| 128 | { |
||
| 129 | // Test parsing mobile numbers of Argentina. |
||
| 130 | $this->doTestFindInContext("+54 9 343 555 1212", RegionCode::AR); |
||
| 131 | $this->doTestFindInContext("0343 15 555 1212", RegionCode::AR); |
||
| 132 | |||
| 133 | $this->doTestFindInContext("+54 9 3715 65 4320", RegionCode::AR); |
||
| 134 | $this->doTestFindInContext("03715 15 65 4320", RegionCode::AR); |
||
| 135 | |||
| 136 | // Test parsing fixed-line numbers of Argentina. |
||
| 137 | $this->doTestFindInContext("+54 11 3797 0000", RegionCode::AR); |
||
| 138 | $this->doTestFindInContext("011 3797 0000", RegionCode::AR); |
||
| 139 | |||
| 140 | $this->doTestFindInContext("+54 3715 65 4321", RegionCode::AR); |
||
| 141 | $this->doTestFindInContext("03715 65 4321", RegionCode::AR); |
||
| 142 | |||
| 143 | $this->doTestFindInContext("+54 23 1234 0000", RegionCode::AR); |
||
| 144 | $this->doTestFindInContext("023 1234 0000", RegionCode::AR); |
||
| 145 | } |
||
| 146 | |||
| 147 | public function testFindWithXInNumber() |
||
| 148 | { |
||
| 149 | $this->doTestFindInContext("(0xx) 123456789", RegionCode::AR); |
||
| 150 | // A case where x denotes both carrier codes and extension symbol. |
||
| 151 | $this->doTestFindInContext("(0xx) 123456789 x 1234", RegionCode::AR); |
||
| 152 | |||
| 153 | // This test is intentionally constructed such that the number of digit after xx is larger than |
||
| 154 | // 7, so that the number won't be mistakenly treated as an extension, as we allow extensions up |
||
| 155 | // to 7 digits. This assumption is okay for now as all the countries where a carrier selection |
||
| 156 | // code is written in the form of xx have a national significant number of length larger than 7. |
||
| 157 | $this->doTestFindInContext("011xx5481429712", RegionCode::US); |
||
| 158 | } |
||
| 159 | |||
| 160 | public function testFindNumbersMexico() |
||
| 161 | { |
||
| 162 | // Test parsing fixed-line numbers of Mexico. |
||
| 163 | $this->doTestFindInContext("+52 (449)978-0001", RegionCode::MX); |
||
| 164 | $this->doTestFindInContext("01 (449)978-0001", RegionCode::MX); |
||
| 165 | $this->doTestFindInContext("(449)978-0001", RegionCode::MX); |
||
| 166 | |||
| 167 | // Test parsing mobile numbers of Mexico. |
||
| 168 | $this->doTestFindInContext("+52 1 33 1234-5678", RegionCode::MX); |
||
| 169 | $this->doTestFindInContext("044 (33) 1234-5678", RegionCode::MX); |
||
| 170 | $this->doTestFindInContext("045 33 1234-5678", RegionCode::MX); |
||
| 171 | } |
||
| 172 | |||
| 173 | public function testFindNumbersWithPlusWithNoRegion() |
||
| 174 | { |
||
| 175 | // RegionCode.ZZ is allowed only if the number starts with a '+' - then the country code can be |
||
| 176 | // calculated. |
||
| 177 | $this->doTestFindInContext("+64 3 331 6005", RegionCode::ZZ); |
||
| 178 | // Null is also allowed for the region code in these cases. |
||
| 179 | $this->doTestFindInContext("+64 3 331 6005", null); |
||
| 180 | } |
||
| 181 | |||
| 182 | public function testFindExtensions() |
||
| 183 | { |
||
| 184 | $this->doTestFindInContext("03 331 6005 ext 3456", RegionCode::NZ); |
||
| 185 | $this->doTestFindInContext("03-3316005x3456", RegionCode::NZ); |
||
| 186 | $this->doTestFindInContext("03-3316005 int.3456", RegionCode::NZ); |
||
| 187 | $this->doTestFindInContext("03 3316005 #3456", RegionCode::NZ); |
||
| 188 | $this->doTestFindInContext("0~0 1800 7493 524", RegionCode::PL); |
||
| 189 | $this->doTestFindInContext("(1800) 7493.524", RegionCode::US); |
||
| 190 | // Check that the last instance of an extension token is matched. |
||
| 191 | $this->doTestFindInContext("0~0 1800 7493 524 ~1234", RegionCode::PL); |
||
| 192 | // Verifying bug-fix where the last digit of a number was previously omitted if it was a 0 when |
||
| 193 | // extracting the extension. Also verifying a few different cases of extensions. |
||
| 194 | $this->doTestFindInContext("+44 2034567890x456", RegionCode::NZ); |
||
| 195 | $this->doTestFindInContext("+44 2034567890x456", RegionCode::GB); |
||
| 196 | $this->doTestFindInContext("+44 2034567890 x456", RegionCode::GB); |
||
| 197 | $this->doTestFindInContext("+44 2034567890 X456", RegionCode::GB); |
||
| 198 | $this->doTestFindInContext("+44 2034567890 X 456", RegionCode::GB); |
||
| 199 | $this->doTestFindInContext("+44 2034567890 X 456", RegionCode::GB); |
||
| 200 | $this->doTestFindInContext("+44 2034567890 X 456", RegionCode::GB); |
||
| 201 | |||
| 202 | $this->doTestFindInContext("(800) 901-3355 x 7246433", RegionCode::US); |
||
| 203 | $this->doTestFindInContext("(800) 901-3355 , ext 7246433", RegionCode::US); |
||
| 204 | $this->doTestFindInContext("(800) 901-3355 ,extension 7246433", RegionCode::US); |
||
| 205 | // The next test differs from PhoneNumberUtil -> when matching we don't consider a lone comma to |
||
| 206 | // indicate an extension, although we accept it when parsing. |
||
| 207 | $this->doTestFindInContext("(800) 901-3355 ,x 7246433", RegionCode::US); |
||
| 208 | $this->doTestFindInContext("(800) 901-3355 ext: 7246433", RegionCode::US); |
||
| 209 | } |
||
| 210 | |||
| 211 | public function testFindInterspersedWithSpace() |
||
| 212 | { |
||
| 213 | $this->doTestFindInContext("0 3 3 3 1 6 0 0 5", RegionCode::NZ); |
||
| 214 | } |
||
| 215 | |||
| 216 | /** |
||
| 217 | * Test matching behaviour when starting in the middle of a phone number. |
||
| 218 | */ |
||
| 219 | public function testIntermediateParsePositions() |
||
| 220 | { |
||
| 221 | $text = "Call 033316005 or 032316005!"; |
||
| 222 | // | | | | | | |
||
| 223 | // 0 5 10 15 20 25 |
||
| 224 | |||
| 225 | // Iterate over all possible indices. |
||
| 226 | for ($i = 0; $i <= 5; $i++) { |
||
| 227 | $this->assertEqualRange($text, $i, 5, 14); |
||
| 228 | } |
||
| 229 | // 7 and 8 digits in a row are still parsed as number. |
||
| 230 | $this->assertEqualRange($text, 6, 6, 14); |
||
| 231 | $this->assertEqualRange($text, 7, 7, 14); |
||
| 232 | // Anything smaller is skipped to the second instance. |
||
| 233 | for ($i = 8; $i <= 19; $i++) { |
||
| 234 | $this->assertEqualRange($text, $i, 19, 28); |
||
| 235 | } |
||
| 236 | } |
||
| 237 | |||
| 238 | public function testFourMatchesInARow() |
||
| 239 | { |
||
| 240 | $number1 = "415-666-7777"; |
||
| 241 | $number2 = "800-443-1223"; |
||
| 242 | $number3 = "212-443-1223"; |
||
| 243 | $number4 = "650-443-1223"; |
||
| 244 | $text = $number1 . " - " . $number2 . " - " . $number3 . " - " . $number4; |
||
| 245 | |||
| 246 | $iterator = $this->phoneUtil->findNumbers($text, RegionCode::US); |
||
| 247 | |||
| 248 | $iterator->next(); |
||
| 249 | $match = $iterator->current(); |
||
| 250 | $this->assertMatchProperties($match, $text, $number1, RegionCode::US); |
||
| 251 | |||
| 252 | $iterator->next(); |
||
| 253 | $match = $iterator->current(); |
||
| 254 | $this->assertMatchProperties($match, $text, $number2, RegionCode::US); |
||
| 255 | |||
| 256 | $iterator->next(); |
||
| 257 | $match = $iterator->current(); |
||
| 258 | $this->assertMatchProperties($match, $text, $number3, RegionCode::US); |
||
| 259 | |||
| 260 | $iterator->next(); |
||
| 261 | $match = $iterator->current(); |
||
| 262 | $this->assertMatchProperties($match, $text, $number4, RegionCode::US); |
||
| 263 | } |
||
| 264 | |||
| 265 | public function testMatchesFoundWithMultipleSpaces() |
||
| 266 | { |
||
| 267 | $number1 = "(415) 666-7777"; |
||
| 268 | $number2 = "(800) 443-1223"; |
||
| 269 | $text = $number1 . " " . $number2; |
||
| 270 | |||
| 271 | $iterator = $this->phoneUtil->findNumbers($text, RegionCode::US); |
||
| 272 | |||
| 273 | $iterator->next(); |
||
| 274 | $match = $iterator->current(); |
||
| 275 | $this->assertMatchProperties($match, $text, $number1, RegionCode::US); |
||
| 276 | |||
| 277 | $iterator->next(); |
||
| 278 | $match = $iterator->current(); |
||
| 279 | $this->assertMatchProperties($match, $text, $number2, RegionCode::US); |
||
| 280 | } |
||
| 281 | |||
| 282 | public function testMatchWithSurroundingZipcodes() |
||
| 283 | { |
||
| 284 | $number = "415-666-7777"; |
||
| 285 | $zipPreceding = "My address is CA 34215 - " . $number . " is my number."; |
||
| 286 | |||
| 287 | $iterator = $this->phoneUtil->findNumbers($zipPreceding, RegionCode::US); |
||
| 288 | |||
| 289 | $iterator->next(); |
||
| 290 | $match = $iterator->current(); |
||
| 291 | $this->assertMatchProperties($match, $zipPreceding, $number, RegionCode::US); |
||
| 292 | |||
| 293 | // Now repeat, but this time the phone number has spaces in it. It should still be found. |
||
| 294 | $number = "(415) 666 7777"; |
||
| 295 | |||
| 296 | $zipFollowing = "My number is " . $number . ". 34215 is my zip-code."; |
||
| 297 | |||
| 298 | $iterator = $this->phoneUtil->findNumbers($zipFollowing, RegionCode::US); |
||
| 299 | |||
| 300 | $iterator->next(); |
||
| 301 | $match = $iterator->current(); |
||
| 302 | $this->assertMatchProperties($match, $zipFollowing, $number, RegionCode::US); |
||
| 303 | } |
||
| 304 | |||
| 305 | public function dataLatinLetters() |
||
| 306 | { |
||
| 307 | return array( |
||
| 308 | array('c', true), |
||
| 309 | array('C', true), |
||
| 310 | array("\xC3\x89", true), |
||
| 311 | array("\xCC\x81", true), // Combining acute accent |
||
| 312 | // Punctuation, digits and white-space are not considered "latin letters". |
||
| 313 | array(':', false), |
||
| 314 | array('5', false), |
||
| 315 | array('-', false), |
||
| 316 | array('.', false), |
||
| 317 | array(' ', false), |
||
| 318 | array("\xE6\x88\x91", false), // Chinese character |
||
| 319 | array("\xE3\x81\xAE", false), // Hiragana letter no |
||
| 320 | ); |
||
| 321 | } |
||
| 322 | |||
| 323 | /** |
||
| 324 | * @dataProvider dataLatinLetters |
||
| 325 | * @param string $letter |
||
| 326 | * @param string $expectedResult |
||
| 327 | */ |
||
| 328 | public function testIsLatinLetter($letter, $expectedResult) |
||
| 329 | { |
||
| 330 | $this->assertEquals($expectedResult, PhoneNumberMatcher::isLatinLetter($letter), |
||
| 331 | "{$letter} should return {$expectedResult}"); |
||
| 332 | } |
||
| 333 | |||
| 334 | public function testMatchesWithSurroundingLatinChars() |
||
| 335 | { |
||
| 336 | $possibleOnlyContexts = array(); |
||
| 337 | $possibleOnlyContexts[] = array("abc", "def"); |
||
| 338 | $possibleOnlyContexts[] = array("abc", ""); |
||
| 339 | $possibleOnlyContexts[] = array("", "def"); |
||
| 340 | // Latin capital letter e with an acute accent. |
||
| 341 | $possibleOnlyContexts[] = array("\xC3\x89", ""); |
||
| 342 | // e with an acute accent decomposed (with combining mark). |
||
| 343 | $possibleOnlyContexts[] = array("e\xCC\x81", ""); |
||
| 344 | |||
| 345 | // Numbers should not be considered valid, if they are surrounded by Latin characters, but |
||
| 346 | // should be considered possible. |
||
| 347 | $this->findMatchesInContexts($possibleOnlyContexts, false, true); |
||
| 348 | } |
||
| 349 | |||
| 350 | public function testMoneyNotSeenAsPhoneNumber() |
||
| 351 | { |
||
| 352 | $possibleOnlyContexts = array(); |
||
| 353 | $possibleOnlyContexts[] = array("$", ""); |
||
| 354 | $possibleOnlyContexts[] = array("", "$"); |
||
| 355 | $possibleOnlyContexts[] = array("\xC2\xA3", ""); // Pound sign |
||
| 356 | $possibleOnlyContexts[] = array("\xC2\xA5", ""); // Yen sign |
||
| 357 | |||
| 358 | $this->findMatchesInContexts($possibleOnlyContexts, false, true); |
||
| 359 | } |
||
| 360 | |||
| 361 | public function testPercentageNotSeenAsPhoneNumber() |
||
| 362 | { |
||
| 363 | $possibleOnlyContexts = array(); |
||
| 364 | $possibleOnlyContexts[] = array("", "%"); |
||
| 365 | // Numbers followed by % should be dropped |
||
| 366 | $this->findMatchesInContexts($possibleOnlyContexts, false, true); |
||
| 367 | } |
||
| 368 | |||
| 369 | public function testPhoneNumberWithLeadingOrTrailingMoneyMatches() |
||
| 370 | { |
||
| 371 | // Because of the space after the 20 (or before the 100) these dollar amounts should not stop |
||
| 372 | // the actual number from being found. |
||
| 373 | $contexts = array(); |
||
| 374 | $contexts[] = array('$20 ', ''); |
||
| 375 | $contexts[] = array('', ' 100$'); |
||
| 376 | |||
| 377 | $this->findMatchesInContexts($contexts, true, true); |
||
| 378 | } |
||
| 379 | |||
| 380 | public function testMatchesWithSurroundingLatinCharsAndLeadingPunctuation() |
||
| 381 | { |
||
| 382 | // Contexts with trailing characters. Leading characters are okay here since the numbers we will |
||
| 383 | // insert start with punctuation, but trailing characters are still not allowed. |
||
| 384 | $possibleOnlyContexts = array(); |
||
| 385 | $possibleOnlyContexts[] = array("abc", "def"); |
||
| 386 | $possibleOnlyContexts[] = array("", "def"); |
||
| 387 | $possibleOnlyContexts[] = array("", "\xC3\x89"); |
||
| 388 | |||
| 389 | // Numbers should not be considered valid, if they have trailing Latin characters, but should be |
||
| 390 | // considered possible. |
||
| 391 | $numberWithPlus = "+14156667777"; |
||
| 392 | $numberWithBrackets = "(415)6667777"; |
||
| 393 | $this->findMatchesInContexts($possibleOnlyContexts, false, true, RegionCode::US, $numberWithPlus); |
||
| 394 | $this->findMatchesInContexts($possibleOnlyContexts, false, true, RegionCode::US, $numberWithBrackets); |
||
| 395 | |||
| 396 | $validContexts = array(); |
||
| 397 | $validContexts[] = array("abc", ""); |
||
| 398 | $validContexts[] = array("\xC3\x89", ""); |
||
| 399 | $validContexts[] = array("\xC3\x89", "."); // Trailing punctuation. |
||
| 400 | $validContexts[] = array("\xC3\x89", " def"); // Trailing white space. |
||
| 401 | |||
| 402 | // Numbers should be considered valid, since they start with punctuation. |
||
| 403 | $this->findMatchesInContexts($validContexts, true, true, RegionCode::US, $numberWithPlus); |
||
| 404 | $this->findMatchesInContexts($validContexts, true, true, RegionCode::US, $numberWithBrackets); |
||
| 405 | } |
||
| 406 | |||
| 407 | public function testMatchesWithSurroundingChineseChars() |
||
| 408 | { |
||
| 409 | $validContexts = array(); |
||
| 410 | $validContexts[] = array("我的电话号码是", ""); |
||
| 411 | $validContexts[] = array("", "是我的电话号码"); |
||
| 412 | $validContexts[] = array("请拨打", "我在明天"); |
||
| 413 | |||
| 414 | // Numbers should be considered valid, since they are surrounded by Chinese. |
||
| 415 | $this->findMatchesInContexts($validContexts, true, true); |
||
| 416 | } |
||
| 417 | |||
| 418 | public function testMatchesWithSurroundingPunctuation() |
||
| 419 | { |
||
| 420 | $validContexts = array(); |
||
| 421 | $validContexts[] = array("My number-", ""); // At end of text |
||
| 422 | $validContexts[] = array("", ".Nice day."); // At start of text |
||
| 423 | $validContexts[] = array("Tel:", "."); // Punctuation surrounds number. |
||
| 424 | $validContexts[] = array("Tel: ", " on Saturdays."); // White-space is also fine. |
||
| 425 | |||
| 426 | // Numbers should be considered valid, since they are surrounded by punctuation. |
||
| 427 | $this->findMatchesInContexts($validContexts, true, true); |
||
| 428 | } |
||
| 429 | |||
| 430 | public function testMatchesMultiplePhoneNumbersSeparatedByPhoneNumberPunctuation() |
||
| 431 | { |
||
| 432 | $text = "Call 650-253-4561 -- 455-234-3451"; |
||
| 433 | $region = RegionCode::US; |
||
| 434 | |||
| 435 | $number1 = new PhoneNumber(); |
||
| 436 | $number1->setCountryCode($this->phoneUtil->getCountryCodeForRegion($region)); |
||
| 437 | $number1->setNationalNumber(6502534561); |
||
| 438 | $match1 = new PhoneNumberMatch(5, "650-253-4561", $number1); |
||
| 439 | |||
| 440 | $number2 = new PhoneNumber(); |
||
| 441 | $number2->setCountryCode($this->phoneUtil->getCountryCodeForRegion($region)); |
||
| 442 | $number2->setNationalNumber(4552343451); |
||
| 443 | $match2 = new PhoneNumberMatch(21, "455-234-3451", $number2); |
||
| 444 | |||
| 445 | $matches = $this->phoneUtil->findNumbers($text, $region); |
||
| 446 | |||
| 447 | $matches->next(); |
||
| 448 | $this->assertEquals($match1, $matches->current()); |
||
| 449 | |||
| 450 | $matches->next(); |
||
| 451 | $this->assertEquals($match2, $matches->current()); |
||
| 452 | } |
||
| 453 | |||
| 454 | public function testDoesNotMatchMultiplePhoneNumbersSeparatedWithNoWhiteSpace() |
||
| 455 | { |
||
| 456 | // No white-space found between numbers - neither is found. |
||
| 457 | $text = "Call 650-253-4561--455-234-3451"; |
||
| 458 | $region = RegionCode::US; |
||
| 459 | |||
| 460 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers($text, $region))); |
||
| 461 | } |
||
| 462 | |||
| 463 | /** |
||
| 464 | * Strings with number-like things that shouldn't be found under any level. |
||
| 465 | * @return array |
||
| 466 | */ |
||
| 467 | public function dataImpossibleCases() |
||
| 468 | { |
||
| 469 | return array( |
||
| 470 | array("12345", RegionCode::US), |
||
| 471 | array("23456789", RegionCode::US), |
||
| 472 | array("234567890112", RegionCode::US), |
||
| 473 | array("650+253+1234", RegionCode::US), |
||
| 474 | array("3/10/1984", RegionCode::CA), |
||
| 475 | array("03/27/2011", RegionCode::US), |
||
| 476 | array("31/8/2011", RegionCode::US), |
||
| 477 | array("1/12/2011", RegionCode::US), |
||
| 478 | array("10/12/82", RegionCode::DE), |
||
| 479 | array("650x2531234", RegionCode::US), |
||
| 480 | array("2012-01-02 08:00", RegionCode::US), |
||
| 481 | array("2012/01/02 08:00", RegionCode::US), |
||
| 482 | array("20120102 08:00", RegionCode::US), |
||
| 483 | array("2014-04-12 04:04 PM", RegionCode::US), |
||
| 484 | array("2014-04-12 04:04 PM", RegionCode::US), |
||
| 485 | array("2014-04-12 04:04 PM", RegionCode::US), |
||
| 486 | array("2014-04-12 04:04 PM", RegionCode::US), |
||
| 487 | ); |
||
| 488 | } |
||
| 489 | |||
| 490 | /** |
||
| 491 | * Strings with number-like things that should only be found under "possible". |
||
| 492 | * @return array |
||
| 493 | */ |
||
| 494 | public function dataPossibleOnlyCases() |
||
| 495 | { |
||
| 496 | return array( |
||
| 497 | // US numbers cannot start with 7 in the test metadata to be valid. |
||
| 498 | array("7121115678", RegionCode::US), |
||
| 499 | // 'X' should not be found in numbers at leniencies stricter than POSSIBLE, unless it represents |
||
| 500 | // a carrier code or extension. |
||
| 501 | array("1650 x 253 - 1234", RegionCode::US), |
||
| 502 | array("650 x 253 - 1234", RegionCode::US), |
||
| 503 | array("6502531x234", RegionCode::US), |
||
| 504 | array("(20) 3346 1234", RegionCode::GB), // Non-optional NP omitted |
||
| 505 | ); |
||
| 506 | } |
||
| 507 | |||
| 508 | /** |
||
| 509 | * Strings with number-like things that should only be found up to and including the "valid" |
||
| 510 | * leniency level. |
||
| 511 | * @return array |
||
| 512 | */ |
||
| 513 | public function dataValidCases() |
||
| 514 | { |
||
| 515 | return array( |
||
| 516 | array("65 02 53 00 00", RegionCode::US), |
||
| 517 | array("6502 538365", RegionCode::US), |
||
| 518 | array("650//253-1234", RegionCode::US), // 2 slashes are illegal at higher levels |
||
| 519 | array("650/253/1234", RegionCode::US), |
||
| 520 | array("9002309. 158", RegionCode::US), |
||
| 521 | array("12 7/8 - 14 12/34 - 5", RegionCode::US), |
||
| 522 | array("12.1 - 23.71 - 23.45", RegionCode::US), |
||
| 523 | array("800 234 1 111x1111", RegionCode::US), |
||
| 524 | array("1979-2011 100", RegionCode::US), |
||
| 525 | array("+494949-4-94", RegionCode::DE), // National number in wrong format |
||
| 526 | array("4156666-777", RegionCode::US), |
||
| 527 | array("2012-0102 08", RegionCode::US), // Very strange formatting. |
||
| 528 | array("2012-01-02 08", RegionCode::US), |
||
| 529 | // Breakdown assistance number with unexpected formatting. |
||
| 530 | array("1800-1-0-10 22", RegionCode::AU), |
||
| 531 | array("030-3-2 23 12 34", RegionCode::DE), |
||
| 532 | array("03 0 -3 2 23 12 34", RegionCode::DE), |
||
| 533 | array("(0)3 0 -3 2 23 12 34", RegionCode::DE), |
||
| 534 | array("0 3 0 -3 2 23 12 34", RegionCode::DE), |
||
| 535 | ); |
||
| 536 | } |
||
| 537 | |||
| 538 | /** |
||
| 539 | * Strings with number-like things that should only be found up to and including the |
||
| 540 | * "strict_grouping" leniency level. |
||
| 541 | * @return array |
||
| 542 | */ |
||
| 543 | public function dataStrictGroupingCases() |
||
| 544 | { |
||
| 545 | return array( |
||
| 546 | array("(415) 6667777", RegionCode::US), |
||
| 547 | array("415-6667777", RegionCode::US), |
||
| 548 | // Should be found by strict grouping but not exact grouping, as the last two groups are |
||
| 549 | // formatted together as a block. |
||
| 550 | array("0800-2491234", RegionCode::DE), |
||
| 551 | // Doesn't match any formatting in the test file, but almost matches an alternate format (the |
||
| 552 | // last two groups have been squashed together here). |
||
| 553 | array("0900-1 123123", RegionCode::DE), |
||
| 554 | array("(0)900-1 123123", RegionCode::DE), |
||
| 555 | array("0 900-1 123123", RegionCode::DE), |
||
| 556 | // NDC also found as part of the country calling code; this shouldn't ruin the grouping |
||
| 557 | // expectations. |
||
| 558 | array("+33 3 34 2312", RegionCode::FR), |
||
| 559 | ); |
||
| 560 | } |
||
| 561 | |||
| 562 | /** |
||
| 563 | * Strings with number-like things that should be found at all levels. |
||
| 564 | * @return array |
||
| 565 | */ |
||
| 566 | public function dataExactGroupingCases() |
||
| 567 | { |
||
| 568 | return array( |
||
| 569 | array("4156667777", RegionCode::US), |
||
| 570 | array("415-666-7777", RegionCode::US), |
||
| 571 | array("4156667777", RegionCode::US), |
||
| 572 | array("4156667777 x 123", RegionCode::US), |
||
| 573 | array("415-666-7777", RegionCode::US), |
||
| 574 | array("415/666-7777", RegionCode::US), |
||
| 575 | array("415-666-7777 ext. 503", RegionCode::US), |
||
| 576 | array("1 415 666 7777 x 123", RegionCode::US), |
||
| 577 | array("+1 415-666-7777", RegionCode::US), |
||
| 578 | array("+494949 49", RegionCode::DE), |
||
| 579 | array("+49-49-34", RegionCode::DE), |
||
| 580 | array("+49-4931-49", RegionCode::DE), |
||
| 581 | array("04931-49", RegionCode::DE), // With National Prefix |
||
| 582 | array("+49-494949", RegionCode::DE), // One group with country code |
||
| 583 | array("+49-494949 ext. 49", RegionCode::DE), |
||
| 584 | array("+49494949 ext. 49", RegionCode::DE), |
||
| 585 | array("0494949", RegionCode::DE), |
||
| 586 | array("0494949 ext. 49", RegionCode::DE), |
||
| 587 | array("01 (33) 3461 2234", RegionCode::MX), // Optional NP present |
||
| 588 | array("(33) 3461 2234", RegionCode::MX), // Optional NP omitted |
||
| 589 | array("1800-10-10 22", RegionCode::AU), // Breakdown assistance number. |
||
| 590 | // Doesn't match any formatting in the test file, but matches an alternate format exactly. |
||
| 591 | array("0900-1 123 123", RegionCode::DE), |
||
| 592 | array("(0)900-1 123 123", RegionCode::DE), |
||
| 593 | array("0 900-1 123 123", RegionCode::DE), |
||
| 594 | array("+33 3 34 23 12", RegionCode::FR), |
||
| 595 | ); |
||
| 596 | } |
||
| 597 | |||
| 598 | public function data_testMatchesWithPossibleLeniency() |
||
| 599 | { |
||
| 600 | return $this->dataStrictGroupingCases() |
||
| 601 | + $this->dataExactGroupingCases() |
||
| 602 | + $this->dataValidCases() |
||
| 603 | + $this->dataPossibleOnlyCases(); |
||
| 604 | } |
||
| 605 | |||
| 606 | /** |
||
| 607 | * @param string $rawString |
||
| 608 | * @param string $region |
||
| 609 | * @dataProvider data_testMatchesWithPossibleLeniency |
||
| 610 | */ |
||
| 611 | public function testMatchesWithPossibleLeniency($rawString, $region) |
||
| 612 | { |
||
| 613 | $this->doTestNumberMatchesForLeniency($rawString, $region, Leniency::POSSIBLE()); |
||
| 614 | } |
||
| 615 | |||
| 616 | /** |
||
| 617 | * @param string $rawString |
||
| 618 | * @param string $region |
||
| 619 | * @dataProvider dataImpossibleCases |
||
| 620 | */ |
||
| 621 | public function testNonMatchesWithPossibleLeniency($rawString, $region) |
||
| 622 | { |
||
| 623 | $this->doTestNumberNonMatchesForLeniency($rawString, $region, Leniency::POSSIBLE()); |
||
| 624 | } |
||
| 625 | |||
| 626 | public function data_testMatchesWithValidLeniency() |
||
| 627 | { |
||
| 628 | return $this->dataStrictGroupingCases() |
||
| 629 | + $this->dataExactGroupingCases() |
||
| 630 | + $this->dataValidCases(); |
||
| 631 | } |
||
| 632 | |||
| 633 | /** |
||
| 634 | * @param string $rawString |
||
| 635 | * @param string $region |
||
| 636 | * @dataProvider data_testMatchesWithValidLeniency |
||
| 637 | */ |
||
| 638 | public function testMatchesWithValidLeniency($rawString, $region) |
||
| 639 | { |
||
| 640 | $this->doTestNumberMatchesForLeniency($rawString, $region, Leniency::VALID()); |
||
| 641 | } |
||
| 642 | |||
| 643 | public function data_testNonMatchesWithValidLeniency() |
||
| 644 | { |
||
| 645 | return $this->dataImpossibleCases() |
||
| 646 | + $this->dataPossibleOnlyCases(); |
||
| 647 | } |
||
| 648 | |||
| 649 | /** |
||
| 650 | * @param $rawString |
||
| 651 | * @param $region |
||
| 652 | * @dataProvider data_testNonMatchesWithValidLeniency |
||
| 653 | */ |
||
| 654 | public function testNonMatchesWithValidLeniency($rawString, $region) |
||
| 655 | { |
||
| 656 | $this->doTestNumberNonMatchesForLeniency($rawString, $region, Leniency::VALID()); |
||
| 657 | } |
||
| 658 | |||
| 659 | public function data_testMatchesWithStrictGroupingLeniency() |
||
| 660 | { |
||
| 661 | return $this->dataStrictGroupingCases() |
||
| 662 | + $this->dataExactGroupingCases(); |
||
| 663 | } |
||
| 664 | |||
| 665 | /** |
||
| 666 | * @param string $rawString |
||
| 667 | * @param string $region |
||
| 668 | * @dataProvider data_testMatchesWithStrictGroupingLeniency |
||
| 669 | */ |
||
| 670 | public function testMatchesWithStrictGroupingLeniency($rawString, $region) |
||
| 671 | { |
||
| 672 | $this->doTestNumberMatchesForLeniency($rawString, $region, Leniency::STRICT_GROUPING()); |
||
| 673 | } |
||
| 674 | |||
| 675 | public function data_testNonMatchesWithStrictGroupLeniency() |
||
| 676 | { |
||
| 677 | return $this->dataImpossibleCases() |
||
| 678 | + $this->dataPossibleOnlyCases() |
||
| 679 | + $this->dataValidCases(); |
||
| 680 | } |
||
| 681 | |||
| 682 | /** |
||
| 683 | * @param string $rawString |
||
| 684 | * @param string $region |
||
| 685 | * @dataProvider data_testNonMatchesWithStrictGroupLeniency |
||
| 686 | */ |
||
| 687 | public function testNonMatchesWithStrictGroupLeniency($rawString, $region) |
||
| 688 | { |
||
| 689 | $this->doTestNumberNonMatchesForLeniency($rawString, $region, Leniency::STRICT_GROUPING()); |
||
| 690 | } |
||
| 691 | |||
| 692 | /** |
||
| 693 | * @param string $rawString |
||
| 694 | * @param string $region |
||
| 695 | * @dataProvider dataExactGroupingCases |
||
| 696 | */ |
||
| 697 | public function testMatchesWithExactGroupingLeniency($rawString, $region) |
||
| 698 | { |
||
| 699 | $this->doTestNumberMatchesForLeniency($rawString, $region, Leniency::EXACT_GROUPING()); |
||
| 700 | } |
||
| 701 | |||
| 702 | public function data_testNonMatchesExactGroupLeniency() |
||
| 703 | { |
||
| 704 | return $this->dataImpossibleCases() |
||
| 705 | + $this->dataPossibleOnlyCases() |
||
| 706 | + $this->dataValidCases() |
||
| 707 | + $this->dataStrictGroupingCases(); |
||
| 708 | } |
||
| 709 | |||
| 710 | /** |
||
| 711 | * @param string $rawString |
||
| 712 | * @param string $region |
||
| 713 | * @dataProvider data_testNonMatchesExactGroupLeniency |
||
| 714 | */ |
||
| 715 | public function testNonMatchesExactGroupLeniency($rawString, $region) |
||
| 716 | { |
||
| 717 | $this->doTestNumberNonMatchesForLeniency($rawString, $region, Leniency::EXACT_GROUPING()); |
||
| 718 | } |
||
| 719 | |||
| 720 | protected function doTestNumberMatchesForLeniency($string, $region, Leniency\AbstractLeniency $leniency) |
||
| 721 | { |
||
| 722 | $iterator = $this->findNumbersForLeniency($string, $region, $leniency); |
||
| 723 | |||
| 724 | $iterator->next(); |
||
| 725 | $match = $iterator->current(); |
||
| 726 | |||
| 727 | $this->assertNotNull($match, "No match found in {$string} ({$region}) for leniency {$leniency}"); |
||
| 728 | $this->assertEquals($string, $match->rawString(), "Found wrong match in test {$string} ({$region}). Found {$match->rawString()}"); |
||
| 729 | } |
||
| 730 | |||
| 731 | protected function doTestNumberNonMatchesForLeniency($string, $region, Leniency\AbstractLeniency $leniency) |
||
| 732 | { |
||
| 733 | $iterator = $this->findNumbersForLeniency($string, $region, $leniency); |
||
| 734 | |||
| 735 | $iterator->next(); |
||
| 736 | $match = $iterator->current(); |
||
| 737 | |||
| 738 | $this->assertNull($match, "Match found in {$string} ({$region}) for leniency {$leniency}"); |
||
| 739 | } |
||
| 740 | |||
| 741 | /** |
||
| 742 | * Helper method which tests the contexts provided and ensures that: |
||
| 743 | * -- if isValid is true, they all find a test number inserted in the middle when leniency of |
||
| 744 | * matching is set to VALID; else no test number should be extracted at that leniency level |
||
| 745 | * -- if isPossible is true, they all find a test number inserted in the middle when leniency of |
||
| 746 | * matching is set to POSSIBLE; else no test number should be extracted at that leniency level |
||
| 747 | * |
||
| 748 | * |
||
| 749 | * @param array $contexts |
||
| 750 | * @param bool $isValid |
||
| 751 | * @param bool $isPossible |
||
| 752 | * @param string $region |
||
| 753 | * @param string $number |
||
| 754 | */ |
||
| 755 | protected function findMatchesInContexts($contexts, $isValid, $isPossible, $region = RegionCode::US, $number = "415-666-7777") |
||
| 756 | { |
||
| 757 | if ($isValid) { |
||
| 758 | $this->doTestInContext($number, $region, $contexts, Leniency::VALID()); |
||
| 759 | } else { |
||
| 760 | foreach ($contexts as $context) { |
||
| 761 | $text = $context[0] . $number . $context[1]; |
||
| 762 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers($text, $region)), |
||
| 763 | "Should not have found a number in {$text}"); |
||
| 764 | } |
||
| 765 | } |
||
| 766 | |||
| 767 | if ($isPossible) { |
||
| 768 | $this->doTestInContext($number, $region, $contexts, Leniency::POSSIBLE()); |
||
| 769 | } else { |
||
| 770 | foreach ($contexts as $context) { |
||
| 771 | $text = $context[0] . $number . $context[1]; |
||
| 772 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers($text, $region)), |
||
| 773 | "Should not have found a number in {$text}"); |
||
| 774 | } |
||
| 775 | } |
||
| 776 | } |
||
| 777 | |||
| 778 | public function testNonMatchingBracketsAreInvalid() |
||
| 779 | { |
||
| 780 | // The digits up to the ", " form a valid US number, but it shouldn't be matched as one since |
||
| 781 | // there was a non-matching bracket present. |
||
| 782 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers( |
||
| 783 | "80.585 [79.964, 81.191]", RegionCode::US))); |
||
| 784 | |||
| 785 | // The trailing "]" is thrown away before parsing, so the resultant number, while a valid US |
||
| 786 | // number, does not have matching brackets. |
||
| 787 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers( |
||
| 788 | "80.585 [79.964]", RegionCode::US))); |
||
| 789 | |||
| 790 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers( |
||
| 791 | "80.585 ((79.964)", RegionCode::US))); |
||
| 792 | |||
| 793 | // This case has too many sets of brackets to be valid. |
||
| 794 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers( |
||
| 795 | "(80).(585) (79).(9)64", RegionCode::US))); |
||
| 796 | } |
||
| 797 | |||
| 798 | public function testNoMatchIfRegionIsNull() |
||
| 799 | { |
||
| 800 | // Fail on non-international prefix if region code is null. |
||
| 801 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers( |
||
| 802 | "Random text body - number is 0331 6005, see you there", null))); |
||
| 803 | } |
||
| 804 | |||
| 805 | public function testNoMatchInEmptyString() |
||
| 806 | { |
||
| 807 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers("", RegionCode::US))); |
||
| 808 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers(" ", RegionCode::US))); |
||
| 809 | } |
||
| 810 | |||
| 811 | public function testNoMatchIfNoNumber() |
||
| 812 | { |
||
| 813 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers( |
||
| 814 | "Random text body - number is foobar, see you there", RegionCode::US))); |
||
| 815 | } |
||
| 816 | |||
| 817 | public function testSequences() |
||
| 818 | { |
||
| 819 | // Test multiple occurrences. |
||
| 820 | $text = "Call 033316005 or 032316005!"; |
||
| 821 | $region = RegionCode::NZ; |
||
| 822 | |||
| 823 | $number1 = new PhoneNumber(); |
||
| 824 | $number1->setCountryCode($this->phoneUtil->getCountryCodeForRegion($region)); |
||
| 825 | $number1->setNationalNumber(33316005); |
||
| 826 | $match1 = new PhoneNumberMatch(5, "033316005", $number1); |
||
| 827 | |||
| 828 | $number2 = new PhoneNumber(); |
||
| 829 | $number2->setCountryCode($this->phoneUtil->getCountryCodeForRegion($region)); |
||
| 830 | $number2->setNationalNumber(32316005); |
||
| 831 | $match2 = new PhoneNumberMatch(19, "032316005", $number2); |
||
| 832 | |||
| 833 | $matches = $this->phoneUtil->findNumbers($text, $region, Leniency::POSSIBLE()); |
||
| 834 | |||
| 835 | $matches->next(); |
||
| 836 | $this->assertEquals($match1, $matches->current()); |
||
| 837 | |||
| 838 | $matches->next(); |
||
| 839 | $this->assertEquals($match2, $matches->current()); |
||
| 840 | } |
||
| 841 | |||
| 842 | public function testNullInput() |
||
| 843 | { |
||
| 844 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers(null, RegionCode::US))); |
||
| 845 | $this->assertTrue($this->hasNoMatches($this->phoneUtil->findNumbers(null, null))); |
||
| 846 | } |
||
| 847 | |||
| 848 | public function testMaxMatches() |
||
| 849 | { |
||
| 850 | // Set up text with 100 valid phone numbers. |
||
| 851 | $numbers = str_repeat("My info: 415-666-7777,", 100); |
||
| 852 | |||
| 853 | // Matches all 100. Max only applies to failed cases. |
||
| 854 | $number = $this->phoneUtil->parse("+14156667777", null); |
||
| 855 | $expected = array_fill(0, 100, $number); |
||
| 856 | |||
| 857 | $iterable = $this->phoneUtil->findNumbers($numbers, RegionCode::US, Leniency::VALID(), 10); |
||
| 858 | $actual = array(); |
||
| 859 | foreach ($iterable as $match) { |
||
| 860 | $actual[] = $match->number(); |
||
| 861 | } |
||
| 862 | |||
| 863 | $this->assertEquals($expected, $actual); |
||
| 864 | } |
||
| 865 | |||
| 866 | public function testMaxMatchesInvalid() |
||
| 867 | { |
||
| 868 | // Set up text with 10 invalid phone numbers followed by 100 valid. |
||
| 869 | $numbers = ""; |
||
| 870 | for ($i = 0; $i < 10; $i++) { |
||
| 871 | $numbers .= "My address is 949-8945-0"; |
||
| 872 | } |
||
| 873 | for ($i = 0; $i < 100; $i++) { |
||
| 874 | $numbers .= "My info: 415-666-7777,"; |
||
| 875 | } |
||
| 876 | |||
| 877 | $iterable = $this->phoneUtil->findNumbers($numbers, RegionCode::US, Leniency::VALID(), 10); |
||
| 878 | $iterable->next(); |
||
| 879 | $this->assertNull($iterable->current()); |
||
| 880 | } |
||
| 881 | |||
| 882 | public function testMaxMatchesMixed() |
||
| 883 | { |
||
| 884 | // Set up text with 100 valid numbers inside an invalid number. |
||
| 885 | $numbers = ""; |
||
| 886 | for ($i = 0; $i < 100; $i++) { |
||
| 887 | $numbers .= "My info: 415-666-7777 123 fake street"; |
||
| 888 | } |
||
| 889 | |||
| 890 | // Only matches the first 10 despite there being 100 numbers due to max matches |
||
| 891 | $number = $this->phoneUtil->parse("+14156667777", null); |
||
| 892 | $expected = array_fill(0, 10, $number); |
||
| 893 | |||
| 894 | $iterable = $this->phoneUtil->findNumbers($numbers, RegionCode::US, Leniency::VALID(), 10); |
||
| 895 | |||
| 896 | $actual = array(); |
||
| 897 | foreach ($iterable as $match) { |
||
| 898 | $actual[] = $match->number(); |
||
| 899 | } |
||
| 900 | |||
| 901 | $this->assertEquals($expected, $actual); |
||
| 902 | } |
||
| 903 | |||
| 904 | public function testNonPlusPrefixedNumbersNotFoundForInvalidRegion() |
||
| 905 | { |
||
| 906 | // Does not start with a "+", we won't match it. |
||
| 907 | $iterable = $this->phoneUtil->findNumbers("1 456 764 156", RegionCode::ZZ); |
||
| 908 | |||
| 909 | $iterable->next(); |
||
| 910 | $this->assertFalse($iterable->valid()); |
||
| 911 | } |
||
| 912 | |||
| 913 | public function testEmptyIteration() |
||
| 914 | { |
||
| 915 | $iterable = $this->phoneUtil->findNumbers("", RegionCode::ZZ); |
||
| 916 | |||
| 917 | $iterable->next(); |
||
| 918 | $this->assertFalse($iterable->valid()); |
||
| 919 | } |
||
| 920 | |||
| 921 | public function testSingleIteration() |
||
| 922 | { |
||
| 923 | $iterable = $this->phoneUtil->findNumbers("+14156667777", RegionCode::ZZ); |
||
| 924 | |||
| 925 | $iterable->next(); |
||
| 926 | $this->assertTrue($iterable->valid()); |
||
| 927 | $this->assertNotNull($iterable->current()); |
||
| 928 | |||
| 929 | $iterable->next(); |
||
| 930 | $this->assertFalse($iterable->valid()); |
||
| 931 | $this->assertNull($iterable->current()); |
||
| 932 | } |
||
| 933 | |||
| 934 | public function testDoubleIteration() |
||
| 935 | { |
||
| 936 | $iterable = $this->phoneUtil->findNumbers("+14156667777 foobar +14156667777 ", RegionCode::ZZ); |
||
| 937 | |||
| 938 | $iterable->next(); |
||
| 939 | $this->assertTrue($iterable->valid()); |
||
| 940 | $this->assertNotNull($iterable->current()); |
||
| 941 | |||
| 942 | $iterable->next(); |
||
| 943 | $this->assertTrue($iterable->valid()); |
||
| 944 | $this->assertNotNull($iterable->current()); |
||
| 945 | |||
| 946 | $iterable->next(); |
||
| 947 | $this->assertFalse($iterable->valid()); |
||
| 948 | $this->assertNull($iterable->current()); |
||
| 949 | } |
||
| 950 | |||
| 951 | /** |
||
| 952 | * Asserts that the expected match is non-null, and that the raw string and expected |
||
| 953 | * proto buffer are set appropriately |
||
| 954 | * |
||
| 955 | * @param PhoneNumberMatch $match |
||
| 956 | * @param string $text |
||
| 957 | * @param string $number |
||
| 958 | * @param string $region |
||
| 959 | */ |
||
| 960 | protected function assertMatchProperties(PhoneNumberMatch $match, $text, $number, $region) |
||
| 961 | { |
||
| 962 | $expectedResult = $this->phoneUtil->parse($number, $region); |
||
| 963 | $this->assertNotNull($match, "Did not find a number in {$text}; expected {$number}"); |
||
| 964 | $this->assertEquals($expectedResult, $match->number()); |
||
| 965 | $this->assertEquals($number, $match->rawString()); |
||
| 966 | } |
||
| 967 | |||
| 968 | /** |
||
| 969 | * Asserts that another number can be found in $text starting at $index, and that its |
||
| 970 | * corresponding range in $start to $end |
||
| 971 | * |
||
| 972 | * @param string $text |
||
| 973 | * @param int $index |
||
| 974 | * @param int $start |
||
| 975 | * @param int $end |
||
| 976 | */ |
||
| 977 | protected function assertEqualRange($text, $index, $start, $end) |
||
| 978 | { |
||
| 979 | $sub = mb_substr($text, $index, mb_strlen($text) - $index); |
||
| 980 | $matches = $this->phoneUtil->findNumbers($sub, RegionCode::NZ, Leniency::POSSIBLE(), PHP_INT_MAX); |
||
| 981 | |||
| 982 | $matches->next(); |
||
| 983 | $this->assertTrue($matches->valid()); |
||
| 984 | |||
| 985 | $match = $matches->current(); |
||
| 986 | |||
| 987 | $this->assertEquals($start - $index, $match->start()); |
||
| 988 | $this->assertEquals($end - $index, $match->end()); |
||
| 989 | $this->assertEquals(mb_substr($sub, $match->start(), $match->end() - $match->start()), $match->rawString()); |
||
| 990 | } |
||
| 991 | |||
| 992 | /** |
||
| 993 | * Tests numbers found by PhoneNumberUtil->findNumbers in various |
||
| 994 | * textual contexts |
||
| 995 | * |
||
| 996 | * @param string $number The number to test |
||
| 997 | * @param string $defaultCountry The corresponding region code |
||
| 998 | */ |
||
| 999 | protected function doTestFindInContext($number, $defaultCountry) |
||
| 1000 | { |
||
| 1001 | $this->findPossibleInContext($number, $defaultCountry); |
||
| 1002 | |||
| 1003 | $parsed = $this->phoneUtil->parse($number, $defaultCountry); |
||
| 1004 | if ($this->phoneUtil->isValidNumber($parsed)) { |
||
| 1005 | $this->findValidInContext($number, $defaultCountry); |
||
| 1006 | } |
||
| 1007 | } |
||
| 1008 | |||
| 1009 | /** |
||
| 1010 | * Tests valid numbers in contexts that should pass for Leniency::POSSIBLE() |
||
| 1011 | * @param $number |
||
| 1012 | * @param $defaultCountry |
||
| 1013 | */ |
||
| 1014 | protected function findPossibleInContext($number, $defaultCountry) |
||
| 1015 | { |
||
| 1016 | $contextPairs = array(); |
||
| 1017 | $contextPairs[] = array("", ""); // no content |
||
| 1018 | $contextPairs[] = array(" ", "\t"); // whitespace only |
||
| 1019 | $contextPairs[] = array("Hello ", ""); // no context at end |
||
| 1020 | $contextPairs[] = array("", " to call me!"); // no context at start |
||
| 1021 | $contextPairs[] = array("Hi there, call ", " to reach me!"); // no context at start |
||
| 1022 | $contextPairs[] = array("Hi here, call ", " , or don't"); // with commas |
||
| 1023 | // Three examples without whitespace around the number |
||
| 1024 | $contextPairs[] = array("Hi call", ""); |
||
| 1025 | $contextPairs[] = array("", "forme"); |
||
| 1026 | $contextPairs[] = array("Hi call", "forme"); |
||
| 1027 | // With other small numbers. |
||
| 1028 | $contextPairs[] = array("It's cheap! Call ", " before 6:30"); |
||
| 1029 | // With a second number later. |
||
| 1030 | $contextPairs[] = array("Call ", " or +1800-123-4567!"); |
||
| 1031 | $contextPairs[] = array("Call me on June 2 at", ""); // with a Month-Day date |
||
| 1032 | // With publication pages |
||
| 1033 | $contextPairs[] = array("As quoted by Alfonso 12-15 (2009), you may call me at ", ""); |
||
| 1034 | $contextPairs[] = array("As quoted by Alfonso et al. 12-15 (2009), you may call me at ", ""); |
||
| 1035 | // With dates, written in the American style. |
||
| 1036 | $contextPairs[] = array("As I said on 03/10/2011, you may call be at ", ""); |
||
| 1037 | // With trailing numbers after a comma. The 45 should not be considered an extension |
||
| 1038 | $contextPairs[] = array("", ", 45 days a year"); |
||
| 1039 | // When matching we don't consider semicolon along with legitimate extension symbol to indicate |
||
| 1040 | // an extension. The 7246433 should not be considered an extension. |
||
| 1041 | $contextPairs[] = array("", ";x 7246433"); |
||
| 1042 | // With a postfix stripped off as it looks like the start of another number. |
||
| 1043 | $contextPairs[] = array("Call ", "/x12 more"); |
||
| 1044 | |||
| 1045 | $this->doTestInContext($number, $defaultCountry, $contextPairs, Leniency::POSSIBLE()); |
||
| 1046 | } |
||
| 1047 | |||
| 1048 | /** |
||
| 1049 | * Tests valid numbers in contexts that fail for Leniency::POSSIBLE() but are valid for |
||
| 1050 | * Leniency::VALID() |
||
| 1051 | * |
||
| 1052 | * @param string $number |
||
| 1053 | * @param string $defaultCountry |
||
| 1054 | */ |
||
| 1055 | protected function findValidInContext($number, $defaultCountry) |
||
| 1056 | { |
||
| 1057 | $contextPairs = array(); |
||
| 1058 | // With other small numbers |
||
| 1059 | $contextPairs[] = array("It's only 9.99! Call ", " to buy"); |
||
| 1060 | // with a number Day.Month.Year date. |
||
| 1061 | $contextPairs[] = array("Call me on 21.6.1984 at ", ""); |
||
| 1062 | // With a number Month/Day date. |
||
| 1063 | $contextPairs[] = array("Call me on 06/21 at ", ""); |
||
| 1064 | // With a number Day.Month date. |
||
| 1065 | $contextPairs[] = array("Call me on 21.6. at ", ""); |
||
| 1066 | // With a number Month/Day/Year date. |
||
| 1067 | $contextPairs[] = array("Call me on 06/21/84 at ", ""); |
||
| 1068 | |||
| 1069 | $this->doTestInContext($number, $defaultCountry, $contextPairs, Leniency::VALID()); |
||
| 1070 | } |
||
| 1071 | |||
| 1072 | protected function doTestInContext($number, $defaultCountry, $contextPairs, Leniency\AbstractLeniency $leniency) |
||
| 1073 | { |
||
| 1074 | foreach ($contextPairs as $context) { |
||
| 1075 | $prefix = $context[0]; |
||
| 1076 | $text = $prefix . $number . $context[1]; |
||
| 1077 | |||
| 1078 | $start = mb_strlen($prefix); |
||
| 1079 | $end = $start + mb_strlen($number); |
||
| 1080 | |||
| 1081 | $iterator = $this->phoneUtil->findNumbers($text, $defaultCountry, $leniency, PHP_INT_MAX); |
||
| 1082 | |||
| 1083 | $iterator->next(); |
||
| 1084 | $match = $iterator->current(); |
||
| 1085 | $this->assertNotNull($match, "Did not find number in '{$text}'; expected '{$number}'"); |
||
| 1086 | |||
| 1087 | $extracted = mb_substr($text, $match->start(), $match->end() - $match->start()); |
||
| 1088 | $this->assertEquals($start, $match->start(), "Unexpected phone region in '{$text}'; extracted '{$extracted}'"); |
||
| 1089 | $this->assertEquals($end, $match->end(), "Unexpected phone region in '{$text}'; extracted '{$extracted}'"); |
||
| 1090 | $this->assertEquals($number, $extracted); |
||
| 1091 | $this->assertEquals($extracted, $match->rawString()); |
||
| 1092 | |||
| 1093 | $this->ensureTermination($text, $defaultCountry, $leniency); |
||
| 1094 | } |
||
| 1095 | } |
||
| 1096 | |||
| 1097 | protected function ensureTermination($text, $defaultCountry, Leniency\AbstractLeniency $leniency) |
||
| 1098 | { |
||
| 1099 | $textLength = mb_strlen($text); |
||
| 1100 | for ($index = 0; $index <= $textLength; $index++) { |
||
| 1101 | $sub = mb_substr($text, $index); |
||
| 1102 | $matches = ""; |
||
| 1103 | // Iterates over all matches. |
||
| 1104 | foreach ($this->phoneUtil->findNumbers($sub, $defaultCountry, $leniency, PHP_INT_MAX) as $match) { |
||
| 1105 | $matches .= ", " . $match; |
||
| 1106 | } |
||
| 1107 | } |
||
| 1108 | } |
||
| 1109 | |||
| 1110 | protected function findNumbersForLeniency($text, $defaultCountry, Leniency\AbstractLeniency $leniency) |
||
| 1111 | { |
||
| 1112 | return $this->phoneUtil->findNumbers($text, $defaultCountry, $leniency, PHP_INT_MAX); |
||
| 1113 | } |
||
| 1114 | |||
| 1115 | protected function hasNoMatches(PhoneNumberMatcher $match) |
||
| 1116 | { |
||
| 1117 | return !$match->valid(); |
||
| 1118 | } |
||
| 1119 | } |
||
| 1120 |