hphp/test/slow/ext_preg/hackarrdvarrs/ext_preg.php

   1 <?hh
   2
   3 function VS($x, $y) {
   4   var_dump($x === $y);
   5   if ($x !== $y) { echo "Failed: $y\n"; var_dump(debug_backtrace()); }
   6 }
   7 function VERIFY($x) { VS($x, true); }
   8
   9 //////////////////////////////////////////////////////////////////////
  10
  11 function test_preg_rep($a,$b,$c) {
  12   return strtoupper($c).$a;
  13 }
  14
  15 function test_preg_grep() {
  16   $array = array("foo 123.1", "fg 24bar", "123.1", "24");
  17   $fl_array = preg_grep("/^(\\d+)?\\.\\d+$/", $array);
  18   VS(count($fl_array), 1);
  19   VS($fl_array[2], "123.1");
  20
  21   VS(preg_grep("/a/", array("c", "b")), array());
  22 }
  23
  24 function test_preg_match() {
  25   // The "i" after the pattern delimiter indicates a case-insensitive search
  26   VS(preg_match("/php/i", "PHP is a scripting language."), 1);
  27
  28   // The \b in the pattern indicates a word boundary, so only the distinct
  29   // word "web" is matched, and not a word partial like "webbing" or "cobweb"
  30   VS(preg_match("/\\bweb\\b/i", "is the web scripting"), 1);
  31
  32   // get host name from URL
  33   $matches = null;
  34   preg_match_with_matches(
  35     "@^(?:http://)?([^/]+)@i",
  36     "http://www.php.net/index.html",
  37     inout $matches,
  38   );
  39   $host = $matches[1];
  40   VS($host, "www.php.net");
  41
  42   // get last two segments of host name
  43   preg_match_with_matches("/[^.]+\\.[^.]+$/", $host, inout $matches);
  44   VS($matches[0], "php.net");
  45
  46   $str = "foobar: 2008";
  47   preg_match_with_matches("/(?<name>\\w+): (?<digit>\\d+)/", $str, inout $matches);
  48   VS(print_r($matches, true),
  49      "Dict\n".
  50      "(\n".
  51      "    [0] => foobar: 2008\n".
  52      "    [name] => foobar\n".
  53      "    [1] => foobar\n".
  54      "    [digit] => 2008\n".
  55      "    [2] => 2008\n".
  56      ")\n");
  57 }
  58
  59
  60 function test_preg_match_all() {
  61   $matches = null;
  62   preg_match_all_with_matches(
  63     "/\\(?  (\\d{3})?  \\)?  (?(1)  [\\-\\s] ) \\d{3}-\\d{4}/x",
  64     "Call 555-1212 or 1-800-555-1212",
  65     inout $matches,
  66   );
  67   VS(print_r($matches, true),
  68      "Dict\n".
  69      "(\n".
  70      "    [0] => Dict\n".
  71      "        (\n".
  72      "            [0] => 555-1212\n".
  73      "            [1] => 800-555-1212\n".
  74      "        )\n".
  75      "\n".
  76      "    [1] => Dict\n".
  77      "        (\n".
  78      "            [0] => \n".
  79      "            [1] => 800\n".
  80      "        )\n".
  81      "\n".
  82      ")\n");
  83
  84   // The \\2 is an example of backreferencing. This tells pcre that
  85   // it must match the second set of parentheses in the regular expression
  86   // itself, which would be the ([\w]+) in this case. The extra backslash is
  87   // required because the string is in double quotes.
  88   $html = "<b>bold text</b><a href=howdy.html>click me</a>";
  89   preg_match_all_with_matches(
  90     "/(<([\\w]+)[^>]*>)(.*)(<\\/\\2>)/",
  91     $html,
  92     inout $matches,
  93     PREG_SET_ORDER,
  94   );
  95   VS(print_r($matches, true),
  96      "Dict\n".
  97      "(\n".
  98      "    [0] => Dict\n".
  99      "        (\n".
 100      "            [0] => <b>bold text</b>\n".
 101      "            [1] => <b>\n".
 102      "            [2] => b\n".
 103      "            [3] => bold text\n".
 104      "            [4] => </b>\n".
 105      "        )\n".
 106      "\n".
 107      "    [1] => Dict\n".
 108      "        (\n".
 109      "            [0] => <a href=howdy.html>click me</a>\n".
 110      "            [1] => <a href=howdy.html>\n".
 111      "            [2] => a\n".
 112      "            [3] => click me\n".
 113      "            [4] => </a>\n".
 114      "        )\n".
 115      "\n".
 116      ")\n");
 117
 118   $str = "a: 1\nb: 2\nc: 3\n";
 119   preg_match_all_with_matches(
 120     "/(?<name>\\w+): (?<digit>\\d+)/",
 121     $str,
 122     inout $matches,
 123   );
 124   VS(print_r($matches, true),
 125      "Dict\n".
 126      "(\n".
 127      "    [0] => Dict\n".
 128      "        (\n".
 129      "            [0] => a: 1\n".
 130      "            [1] => b: 2\n".
 131      "            [2] => c: 3\n".
 132      "        )\n".
 133      "\n".
 134      "    [name] => Dict\n".
 135      "        (\n".
 136      "            [0] => a\n".
 137      "            [1] => b\n".
 138      "            [2] => c\n".
 139      "        )\n".
 140      "\n".
 141      "    [1] => Dict\n".
 142      "        (\n".
 143      "            [0] => a\n".
 144      "            [1] => b\n".
 145      "            [2] => c\n".
 146      "        )\n".
 147      "\n".
 148      "    [digit] => Dict\n".
 149      "        (\n".
 150      "            [0] => 1\n".
 151      "            [1] => 2\n".
 152      "            [2] => 3\n".
 153      "        )\n".
 154      "\n".
 155      "    [2] => Dict\n".
 156      "        (\n".
 157      "            [0] => 1\n".
 158      "            [1] => 2\n".
 159      "            [2] => 3\n".
 160      "        )\n".
 161      "\n".
 162      ")\n");
 163 }
 164
 165 function test_preg_replace() {
 166   $str = "April 15, 2003";
 167   $pattern = "/(\\w+) (\\d+), (\\d+)/i";
 168   $replacement = "\${1}1,\$3";
 169   VS(preg_replace($pattern, $replacement, $str), "April1,2003");
 170
 171   $str = "The quick brown fox jumped over the lazy dog.";
 172   $patterns = array();
 173   $replacements = array();
 174   $patterns[0] = "/quick/";
 175   $patterns[1] = "/brown/";
 176   $patterns[2] = "/fox/";
 177   $replacements[2] = "bear";
 178   $replacements[1] = "black";
 179   $replacements[0] = "slow";
 180   VS(preg_replace($patterns, $replacements, $str),
 181      "The bear black slow jumped over the lazy dog.");
 182
 183   ksort(inout $patterns);
 184   ksort(inout $replacements);
 185   VS(preg_replace($patterns, $replacements, $str),
 186      "The slow black bear jumped over the lazy dog.");
 187
 188   $foos = array();
 189   $foos[0] = "foo";
 190   $foos[1] = "Foo";
 191   $foos[2] = "FOO";
 192   $expFoo = dict[];
 193   $expFoo[0] = "FOO";
 194   $expFoo[1] = "FOO";
 195   $expFoo[2] = "FOO";
 196   VS(preg_replace("/some pattern/", "", array()), dict[]);
 197   VS(preg_replace("/foo/i", "FOO", $foos), $expFoo);
 198
 199   $patterns = array("/(19|20)(\\d{2})-(\\d{1,2})-(\\d{1,2})/",
 200                                   "/^\\s*{(\\w+)}\\s*=/");
 201   $replace = array("\\3/\\4/\\1\\2", "$\\1 =");
 202   VS(preg_replace($patterns, $replace, "{startDate} = 1999-5-27"),
 203      "\$startDate = 5/27/1999");
 204
 205   $str = "foo   o";
 206   $str = preg_replace("/\\s\\s+/", " ", $str);
 207   VS($str, "foo o");
 208
 209   $count = 0;
 210   preg_replace_with_count(array("/\\d/", "/\\s/"), "*", "xp 4 to", -1, inout $count);
 211   VS($count, 3);
 212
 213   VS(preg_replace("/xxx", "w", "xxxx"), NULL);
 214   VS(preg_replace("/xxx/", "w", "xxxx"), "wx");
 215   VS(preg_replace("/xxy/", "w", "xxxx"), "xxxx");
 216
 217   VS(preg_replace("/xxx", "w", array("xxxx")), dict[]);
 218   VS(preg_replace("/xxx/", "w", array("xxxx")), dict(array("wx")));
 219   VS(preg_replace("/xxx/", "w", array("xxxx", "yyyy")), dict(array("wx", "yyyy")));
 220   VS(preg_replace(array("/xxx/", "/xxx"), "w", array("xxxx")), dict[]);
 221   VS(preg_replace(array("/xxx/", "/xxx/"), "w", array("xxxx")), dict(array("wx")));
 222
 223   VS(preg_replace("/xxx", array("w"), array("xxxx")), false);
 224   VS(preg_replace(array("/xxx"), array("w"), array("xxxx")), dict[]);
 225   VS(preg_replace(array("/xxx/"), array("w"), array("xxxx")), dict(array("wx")));
 226 }
 227
 228 function next_year($m) {
 229   return $m[1].((int)$m[2] + 1);
 230 }
 231
 232 function test_preg_replace_callback() {
 233   $text = "April fools day is 04/01/2002\n".
 234     "Last christmas was 12/24/2001\n";
 235   $count = -1;
 236   $text = preg_replace_callback("|(\\d{2}/\\d{2}/)(\\d{4})|", fun("next_year"),
 237                                 $text, -1, inout $count);
 238   VS($text, "April fools day is 04/01/2003\nLast christmas was 12/24/2002\n");
 239 }
 240
 241 function test_preg_split() {
 242   $keywords = preg_split("/[\\s,]+/",
 243                          "hypertext language, programming");
 244   VS(count($keywords), 3);
 245   VS($keywords[0], "hypertext");
 246   VS($keywords[1], "language");
 247   VS($keywords[2], "programming");
 248
 249   $str = "string";
 250   $chars = preg_split("//", $str, -1, PREG_SPLIT_NO_EMPTY);
 251   VS(count($chars), 6);
 252   VS($chars[0], "s");
 253   VS($chars[1], "t");
 254   VS($chars[2], "r");
 255   VS($chars[3], "i");
 256   VS($chars[4], "n");
 257   VS($chars[5], "g");
 258
 259   $chars = preg_split("//", $str, null, PREG_SPLIT_NO_EMPTY);
 260   VS(count($chars), 6);
 261   VS($chars[0], "s");
 262   VS($chars[1], "t");
 263   VS($chars[2], "r");
 264   VS($chars[3], "i");
 265   VS($chars[4], "n");
 266   VS($chars[5], "g");
 267
 268   $chars = preg_split("//", $str, 3, PREG_SPLIT_NO_EMPTY);
 269   VS(count($chars), 3);
 270   VS($chars[0], "s");
 271   VS($chars[1], "t");
 272   VS($chars[2], "ring");
 273
 274
 275   $str = "hypertext language programming";
 276   $chars = preg_split("/ /", $str, -1, PREG_SPLIT_OFFSET_CAPTURE);
 277   VS(print_r($chars, true),
 278      "Array\n".
 279      "(\n".
 280      "    [0] => Array\n".
 281      "        (\n".
 282      "            [0] => hypertext\n".
 283      "            [1] => 0\n".
 284      "        )\n".
 285      "\n".
 286      "    [1] => Array\n".
 287      "        (\n".
 288      "            [0] => language\n".
 289      "            [1] => 10\n".
 290      "        )\n".
 291      "\n".
 292      "    [2] => Array\n".
 293      "        (\n".
 294      "            [0] => programming\n".
 295      "            [1] => 19\n".
 296      "        )\n".
 297      "\n".
 298      ")\n");
 299 }
 300
 301 function test_preg_quote() {
 302   $keywords = "$40 for a g3/400";
 303   $keywords = preg_quote($keywords, "/");
 304   VS($keywords, "\\$40 for a g3\\/400");
 305
 306   // In this example, preg_quote($word) is used to keep the
 307   // asterisks from having special meaning to the regular
 308   // expression.
 309   $textbody = "This book is *very* difficult to find.";
 310   $word = "*very*";
 311   $textbody = preg_replace("/".preg_quote($word)."/",
 312                             "<i>". $word ."</i>",
 313                             $textbody);
 314   VS($textbody, "This book is <i>*very*</i> difficult to find.");
 315 }
 316
 317 function test_ereg_replace() {
 318   $str = "This is a test";
 319   VS(str_replace(" is", " was", $str), "This was a test");
 320   VS(ereg_replace("( )is", "\\1was", $str), "This was a test");
 321   VS(ereg_replace("(( )is)", "\\2was", $str), "This was a test");
 322
 323   $num = '4';
 324   $str = "This string has four words.";
 325   $str = ereg_replace("four", $num, $str);
 326   VS($str, "This string has 4 words.");
 327
 328   $test = "http://test.com/test";
 329   $test = ereg_replace("[[:alpha:]]+://[^<>[:space:]]+[[:alnum:]/]",
 330                         "<a href=\"\\0\">\\0</a>", $test);
 331   VS($test, "<a href=\"http://test.com/test\">http://test.com/test</a>");
 332 }
 333
 334 function test_eregi_replace() {
 335   $pattern = "(>[^<]*)(suffix)";
 336   $replacement = "\\1<span class=\"search\">\\2</span>";
 337   $body = ">whateversuffix";
 338   $body = eregi_replace($pattern, $replacement, $body);
 339   VS($body, ">whatever<span class=\"search\">suffix</span>");
 340 }
 341
 342 function test_split() {
 343   $mb = "\xe4\xbf\xa1\xe6\x81\xaf\x01  2366797";
 344   $ret = split("\x01", $mb);
 345   VS($ret[0], "\xe4\xbf\xa1\xe6\x81\xaf");
 346   VS($ret[1], "  2366797");
 347
 348   $date = "04/30/1973";
 349   $ret = split("[/.-]", $date);
 350   VS($ret[0], "04");
 351   VS($ret[1], "30");
 352   VS($ret[2], "1973");
 353 }
 354
 355 function test_spliti() {
 356   $str = "aBBBaCCCADDDaEEEaGGGA";
 357   $chunks = spliti("a", $str, 5);
 358   VS($chunks[0], "");
 359   VS($chunks[1], "BBB");
 360   VS($chunks[2], "CCC");
 361   VS($chunks[3], "DDD");
 362   VS($chunks[4], "EEEaGGGA");
 363 }
 364
 365 function test_sql_regcase() {
 366   VS(sql_regcase("Foo - bar."), "[Ff][Oo][Oo] - [Bb][Aa][Rr].");
 367 }
 368
 369
 370 <<__EntryPoint>>
 371 function main_ext_preg() {
 372 test_preg_grep();
 373 test_preg_match();
 374 test_preg_match_all();
 375 test_preg_replace();
 376 test_preg_replace_callback();
 377 test_preg_split();
 378 test_preg_quote();
 379 test_ereg_replace();
 380 test_eregi_replace();
 381 test_split();
 382 test_spliti();
 383 test_sql_regcase();
 384 }