Index: WPTexturize.php
===================================================================
--- WPTexturize.php (revision 27515)
+++ WPTexturize.php (working copy)
@@ -14,7 +14,11 @@
$this->assertEquals('[a]a–b[code]---[/code]a–b[/a]', wptexturize('[a]a--b[code]---[/code]a--b[/a]'));
$this->assertEquals('
', wptexturize('
'));
- $this->assertEquals('---', wptexturize('---'));
+ $this->assertEquals( '---', wptexturize( '---' ) );
+ $this->assertEquals( '---', wptexturize( '---' ) );
+ $this->assertEquals( '', wptexturize( '' ) );
+ $this->assertEquals( '', wptexturize( '' ) );
+ $this->assertEquals( '---', wptexturize( '---' ) );
$this->assertEquals('href="baba" “baba”', wptexturize('href="baba" "baba"'));
@@ -44,7 +48,6 @@
//WP Ticket #4539
function test_basic_quotes() {
$this->assertEquals('test’s', wptexturize('test\'s'));
- $this->assertEquals('test’s', wptexturize('test\'s'));
$this->assertEquals('‘quoted’', wptexturize('\'quoted\''));
$this->assertEquals('“quoted”', wptexturize('"quoted"'));
@@ -184,14 +187,591 @@
* @ticket 23185
*/
function test_spaces_around_hyphens() {
+ $nbsp = "\xC2\xA0";
+
$this->assertEquals( ' – ', wptexturize( ' - ' ) );
$this->assertEquals( ' – ', wptexturize( ' - ' ) );
$this->assertEquals( ' – ', wptexturize( ' - ' ) );
$this->assertEquals( ' – ', wptexturize( ' - ') );
+ $this->assertEquals( "$nbsp–$nbsp", wptexturize( "$nbsp-$nbsp" ) );
+ $this->assertEquals( " –$nbsp", wptexturize( " -$nbsp" ) );
+ $this->assertEquals( "$nbsp– ", wptexturize( "$nbsp- ") );
$this->assertEquals( ' — ', wptexturize( ' -- ' ) );
$this->assertEquals( ' — ', wptexturize( ' -- ' ) );
$this->assertEquals( ' — ', wptexturize( ' -- ' ) );
$this->assertEquals( ' — ', wptexturize( ' -- ') );
+ $this->assertEquals( "$nbsp—$nbsp", wptexturize( "$nbsp--$nbsp" ) );
+ $this->assertEquals( " —$nbsp", wptexturize( " --$nbsp" ) );
+ $this->assertEquals( "$nbsp— ", wptexturize( "$nbsp-- ") );
}
+
+ /**
+ * @ticket 22692
+ */
+ function test_spaces_around_quotes() {
+ $nbsp = "\xC2\xA0";
+ $pi = "\xCE\xA0";
+
+ // This should never happen, even if the desired output changes some day.
+
+ $problem_input = "$nbsp\"A";
+ $problem_output = "$nbsp”A";
+
+ $this->assertNotEquals( $problem_output, wptexturize( $problem_input ) );
+
+ // These are desirable outputs for the current design.
+
+ $input = array();
+ $output = array();
+
+ $input[] = "stop. $nbsp\"A quote after 2 spaces.\"";
+ $output[] = "stop. $nbsp“A quote after 2 spaces.”";
+
+ $input[] = "stop.$nbsp$nbsp\"A quote after 2 spaces.\"";
+ $output[] = "stop.$nbsp$nbsp“A quote after 2 spaces.”";
+
+ $input[] = "stop. $nbsp'A quote after 2 spaces.'";
+ $output[] = "stop. $nbsp‘A quote after 2 spaces.’";
+
+ $input[] = "stop.$nbsp$nbsp'A quote after 2 spaces.'";
+ $output[] = "stop.$nbsp$nbsp‘A quote after 2 spaces.’";
+
+ $input[] = "stop. \"A quote after 2 spaces.\"";
+ $output[] = "stop. “A quote after 2 spaces.”";
+
+ $input[] = "stop. \"A quote after 2 spaces.\"";
+ $output[] = "stop. “A quote after 2 spaces.”";
+
+ $input[] = "stop. 'A quote after 2 spaces.'";
+ $output[] = "stop. ‘A quote after 2 spaces.’";
+
+ $input[] = "stop. 'A quote after 2 spaces.'";
+ $output[] = "stop. ‘A quote after 2 spaces.’";
+
+ $input[] = "Contraction: $pi's";
+ $output[] = "Contraction: $pi’s";
+
+ foreach($input as $key => $in) {
+ $this->assertEquals( $output[$key], wptexturize( $in ) );
+ }
+ }
+
+ /**
+ * Check all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ */
+ function test_all_regexp_behaviors() {
+ $input = array();
+ $output = array();
+
+ // Apostrophe before a number always becomes ’ (apos);
+
+ $input[] = "word '99 word";
+ $output[] = "word ’99 word";
+
+ $input[] = "word'99 word";
+ $output[] = "word’99 word";
+
+ $input[] = "word '99word";
+ $output[] = "word ’99word";
+
+ $input[] = "word'99word";
+ $output[] = "word’99word";
+
+ $input[] = "word '99’s word"; // Appears as a separate but logically superfluous pattern in 3.8.
+ $output[] = "word ’99’s word";
+
+ $input[] = "word '99's word"; // Due to the logic error, second apos becomes a prime. See ticket #22823
+ $output[] = "word ’99′s word";
+
+ $input[] = "word '99'samsonite";
+ $output[] = "word ’99′samsonite";
+
+ $input[] = "according to our source, '33% of all students scored less than 50' on the test."; // Apostrophes and primes have priority over quotes
+ $output[] = "according to our source, ’33% of all students scored less than 50′ on the test.";
+
+ $input[] = "word '99' word"; // See ticket #8775
+ $output[] = "word ’99′ word";
+
+ // Apostrophe after a space or ([{<" becomes ‘ (opening_single_quote)
+
+ $input[] = "word 'word word";
+ $output[] = "word ‘word word";
+
+ $input[] = "word ('word word";
+ $output[] = "word (‘word word";
+
+ $input[] = "word ['word word";
+ $output[] = "word [‘word word";
+
+ $input[] = "word <'word word"; // Invalid HTML input?
+ $output[] = "word <‘word word";
+
+ $input[] = "word <'word word"; // Valid HTML input triggers the apos in a word pattern
+ $output[] = "word <’word word";
+
+ $input[] = "word {'word word";
+ $output[] = "word {‘word word";
+
+ $input[] = "word \"'word word";
+ $output[] = "word “‘word word"; // Two opening quotes
+
+ $input[] = "'word word";
+ $output[] = "‘word word";
+
+ $input[] = "word('word word";
+ $output[] = "word(‘word word";
+
+ $input[] = "word['word word";
+ $output[] = "word[‘word word";
+
+ $input[] = "word<'word word";
+ $output[] = "word<‘word word";
+
+ $input[] = "word<'word word";
+ $output[] = "word<’word word";
+
+ $input[] = "word{'word word";
+ $output[] = "word{‘word word";
+
+ $input[] = "word\"'word word";
+ $output[] = "word”‘word word"; // Closing quote, then opening quote
+
+ $input[] = "word ' word word";
+ $output[] = "word ‘ word word";
+
+ $input[] = "word (' word word";
+ $output[] = "word (‘ word word";
+
+ $input[] = "word [' word word";
+ $output[] = "word [‘ word word";
+
+ $input[] = "word <' word word"; // Invalid HTML input?
+ $output[] = "word <‘ word word";
+
+ $input[] = "word <' word word"; // Valid HTML input triggers the closing single quote here
+ $output[] = "word <’ word word";
+
+ $input[] = "word {' word word";
+ $output[] = "word {‘ word word";
+
+ $input[] = "word \"' word word";
+ $output[] = "word “‘ word word"; // Two opening quotes
+
+ $input[] = "' word word";
+ $output[] = "‘ word word";
+
+ $input[] = "word(' word word";
+ $output[] = "word(‘ word word";
+
+ $input[] = "word[' word word";
+ $output[] = "word[‘ word word";
+
+ $input[] = "word<' word word";
+ $output[] = "word<‘ word word";
+
+ $input[] = "word<' word word";
+ $output[] = "word<’ word word";
+
+ $input[] = "word{' word word";
+ $output[] = "word{‘ word word";
+
+ $input[] = "word\"' word word";
+ $output[] = "word”‘ word word"; // Closing quote, then opening quote
+
+ // Double quote after a number becomes ″ (double_prime)
+
+ $input[] = 'word 99" word';
+ $output[] = 'word 99″ word';
+
+ $input[] = 'word 99"word';
+ $output[] = 'word 99″word';
+
+ $input[] = 'word99" word';
+ $output[] = 'word99″ word';
+
+ $input[] = 'word99"word';
+ $output[] = 'word99″word';
+
+ // Apostrophe after a number becomes ′ (prime)
+
+ $input[] = "word 99' word";
+ $output[] = "word 99′ word";
+
+ $input[] = "word 99'word";
+ $output[] = "word 99′word";
+
+ $input[] = "word99' word";
+ $output[] = "word99′ word";
+
+ $input[] = "word99'word";
+ $output[] = "word99′word";
+
+ // Apostrophe "in a word" becomes ’ (apos)
+
+ $input[] = "word word's word";
+ $output[] = "word word’s word";
+
+ $input[] = "word word'. word"; // Quotes with outside punctuation could end with apostrophes instead of closing quotes (may affect i18n)
+ $output[] = "word word’. word";
+
+ $input[] = "word ]'. word";
+ $output[] = "word ]’. word";
+
+ $input[] = "word )'. word";
+ $output[] = "word )’. word";
+
+ $input[] = "word }'. word";
+ $output[] = "word }’. word";
+
+ $input[] = "word >'. word"; // Not tested
+ $output[] = "word >’. word";
+
+ $input[] = "word >'. word";
+ $output[] = "word >’. word";
+
+ // Double quote after a space or ([{< becomes “ (opening_quote) if not followed by spaces
+
+ $input[] = 'word "word word';
+ $output[] = 'word “word word';
+
+ $input[] = 'word ("word word';
+ $output[] = 'word (“word word';
+
+ $input[] = 'word ["word word';
+ $output[] = 'word [“word word';
+
+ $input[] = 'word <"word word'; // Invalid HTML input?
+ $output[] = 'word <“word word';
+
+ $input[] = 'word <"word word'; // Valid HTML input triggers the closing quote pattern
+ $output[] = 'word <”word word';
+
+ $input[] = 'word {"word word';
+ $output[] = 'word {“word word';
+
+ $input[] = '"word word';
+ $output[] = '“word word';
+
+ $input[] = 'word("word word';
+ $output[] = 'word(“word word';
+
+ $input[] = 'word["word word';
+ $output[] = 'word[“word word';
+
+ $input[] = 'word<"word word'; // Invalid HTML input?
+ $output[] = 'word<“word word';
+
+ $input[] = 'word<"word word'; // Valid HTML input triggers the closing quote pattern
+ $output[] = 'word<”word word';
+
+ $input[] = 'word{"word word';
+ $output[] = 'word{“word word';
+
+ $input[] = 'word "99 word';
+ $output[] = 'word “99 word';
+
+ // Double quote becomes ” (closing_quote) unless it is already converted to double_prime or opening_quote.
+
+ $input[] = 'word word" word';
+ $output[] = 'word word” word';
+
+ $input[] = 'word word") word';
+ $output[] = 'word word”) word';
+
+ $input[] = 'word word"] word';
+ $output[] = 'word word”] word';
+
+ $input[] = 'word word"} word';
+ $output[] = 'word word”} word';
+
+ $input[] = 'word word"> word'; // Invalid HTML input?
+ $output[] = 'word word”> word';
+
+ $input[] = 'word word"> word'; // Valid HTML should work
+ $output[] = 'word word”> word';
+
+ $input[] = 'word word"';
+ $output[] = 'word word”';
+
+ $input[] = 'word word"word';
+ $output[] = 'word word”word';
+
+ $input[] = 'word"word"word';
+ $output[] = 'word”word”word';
+
+ $input[] = 'test sentence".';
+ $output[] = 'test sentence”.';
+
+ $input[] = 'test sentence."';
+ $output[] = 'test sentence.”';
+
+ $input[] = 'test sentence". word';
+ $output[] = 'test sentence”. word';
+
+ $input[] = 'test sentence." word';
+ $output[] = 'test sentence.” word';
+
+ // Single quote followed by a space or a period becomes ’ (closing_single_quote)
+
+ $input[] = "word word' word";
+ $output[] = "word word’ word";
+
+ $input[] = "word word'. word";
+ $output[] = "word word’. word";
+
+ $input[] = "word word'.word";
+ $output[] = "word word’.word";
+
+ $input[] = "word word'";
+ $output[] = "word word’";
+
+ $input[] = "test sentence'.";
+ $output[] = "test sentence’.";
+
+ $input[] = "test sentence.'";
+ $output[] = "test sentence.’";
+
+ $input[] = "test sentence'. word";
+ $output[] = "test sentence’. word";
+
+ $input[] = "test sentence.' word";
+ $output[] = "test sentence.’ word";
+
+ // 9x9 becomes 9×9
+
+ $input[] = "9x9";
+ $output[] = "9×9";
+
+ $input[] = "12x34";
+ $output[] = "12×34";
+
+ $input[] = "9 x 9";
+ $output[] = "9 x 9";
+
+ // & always becomes & unless it is followed by # or ;
+
+ $input[] = "word & word";
+ $output[] = "word & word";
+
+ $input[] = "word&word";
+ $output[] = "word&word";
+
+ $input[] = "word word";
+ $output[] = "word word";
+
+ $input[] = "word & word";
+ $output[] = "word & word";
+
+ $input[] = "word &# word";
+ $output[] = "word &# word"; // invalid output?
+
+ $input[] = "word &44; word";
+ $output[] = "word &44; word";
+
+
+ foreach($input as $key => $in) {
+ $this->assertEquals( $output[$key], wptexturize( $in ) );
+ }
+ }
+
+
+
+ /**
+ * Check all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ */
+ function test_all_static_behaviors() {
+ $input = array();
+ $output = array();
+
+ // "cockney" phrases begin with apos instead of opening_single_quote.
+
+ $input[] = "word 'tain't word";
+ $output[] = "word ’tain’t word";
+
+ $input[] = "word 'twere word";
+ $output[] = "word ’twere word";
+
+ $input[] = "word 'twas word";
+ $output[] = "word ’twas word";
+
+ $input[] = "word 'tis word";
+ $output[] = "word ’tis word";
+
+ $input[] = "word 'twill word";
+ $output[] = "word ’twill word";
+
+ $input[] = "word 'til word";
+ $output[] = "word ’til word";
+
+ $input[] = "word 'bout word";
+ $output[] = "word ’bout word";
+
+ $input[] = "word 'nuff word";
+ $output[] = "word ’nuff word";
+
+ $input[] = "word 'round word";
+ $output[] = "word ’round word";
+
+ $input[] = "word 'cause word";
+ $output[] = "word ’cause word";
+
+ $input[] = "word 'em word";
+ $output[] = "word ‘em word";
+
+ // Smart dashes, etc
+
+ $input[] = "word --- word";
+ $output[] = "word — word";
+
+ $input[] = "word---word";
+ $output[] = "word—word";
+
+ $input[] = "word -- word";
+ $output[] = "word — word";
+
+ $input[] = "word--word";
+ $output[] = "word–word";
+
+ $input[] = "word - word";
+ $output[] = "word – word";
+
+ $input[] = "word-word";
+ $output[] = "word-word";
+
+ $input[] = "word xn– word";
+ $output[] = "word xn-- word";
+
+ $input[] = "wordxn–word";
+ $output[] = "wordxn--word";
+
+ $input[] = "word ... word";
+ $output[] = "word … word";
+
+ $input[] = "word...word";
+ $output[] = "word…word";
+
+ $input[] = "word `` word";
+ $output[] = "word “ word";
+
+ $input[] = "word``word";
+ $output[] = "word“word";
+
+ $input[] = "word '' word";
+ $output[] = "word ” word";
+
+ $input[] = "word''word";
+ $output[] = "word”word";
+
+ $input[] = "word (tm) word";
+ $output[] = "word ™ word";
+
+ $input[] = "word (tm)word";
+ $output[] = "word ™word";
+
+ $input[] = "word(tm) word";
+ $output[] = "word(tm) word";
+
+ $input[] = "word(tm)word";
+ $output[] = "word(tm)word";
+
+ foreach($input as $key => $in) {
+ $this->assertEquals( $output[$key], wptexturize( $in ) );
+ }
+ }
+
+
+ /**
+ * All of these tests fail against 3.8 but were not yet listed above.
+ *
+ * @ticket 22692
+ */
+ function test_more_bugs() {
+ $input = array();
+ $output = array();
+
+ // Tests derived from, and contrary to, the baseline patterns above, where output seemed unusual
+
+ $input[] = "according to our source, '33% of all students scored less than 50' on the test.";
+ $output[] = "according to our source, ‘33% of all students scored less than 50’ on the test.";
+
+ $input[] = 'according to our source, "33% of all students scored less than 50" on the test.';
+ $output[] = 'according to our source, “33% of all students scored less than 50” on the test.';
+
+ $input[] = "word <'test quote'&rt; word";
+ $output[] = "word <‘test quote’&rt; word";
+
+ $input[] = 'word <"test quote"&rt; word';
+ $output[] = 'word <“test quote”&rt; word';
+
+ $input[] = "word 'em word";
+ $output[] = "word ’em word";
+
+ $input[] = "word xn– word";
+ $output[] = "word xn– word";
+
+ $input[] = "wordxn–word";
+ $output[] = "wordxn–word";
+
+ // ticket #4116
+
+ $input[] = "www.4x3.com";
+ $output[] = "www.4x3.com";
+
+ $input[] = "4x3.domain.com";
+ $output[] = "4x3.domain.com";
+
+ $input[] = "http://www.24x7.com/";
+ $output[] = "http://www.24x7.com/";
+
+ $input[] = "http://www.domain.com/24x7/";
+ $output[] = "http://www.domain.com/24x7/";
+
+ // ticket #6969 -- Some of these appear to be fixed already.
+
+ $input[] = '[caption]
The Great Wave[/caption]';
+ $output[] = '[caption] The Great Wave[/caption]';
+
+ $input[] = '[code lang="php"]$foo = \'bar\';[/code]';
+ $output[] = '[code lang="php"]$foo = \'bar\';[/code]';
+
+ $input[] = '[foobar]This is how you "quote" things.[/foobar]';
+ $output[] = '[foobar]This is how you "quote" things.[/foobar]';
+
+ $input[] = "\n[[php foo=\"bar\"]\necho \"foo\" . 'bar';\n[/php]]\n
";
+ $output[] = "\n[[php foo=\"bar\"]\necho \"foo\" . 'bar';\n[/php]]\n
";
+
+ $input[] = "foo[shortcode]\n\"bar\"\n[/shortcode]foo";
+ $output[] = "foo[shortcode]\n\"bar\"\n[/shortcode]foo";
+
+ // ticket #20342
+
+ $input[] = 'word---"quote"';
+ $output[] = 'word—“quote”';
+
+ $input[] = 'word--"quote"';
+ $output[] = 'word–“quote”';
+
+ $input[] = 'word-"quote"';
+ $output[] = 'word-“quote”';
+
+ // ticket #26850
+
+ $input[] = "in '4 years, 3 months,' Obama cut the deficit";
+ $output[] = "in ‘4 years, 3 months,’ Obama cut the deficit";
+
+ $input[] = "in '40 years, 3 months,' Obama cut the deficit";
+ $output[] = "in ‘40 years, 3 months,’ Obama cut the deficit";
+
+ $input[] = "in '400 years, 3 months,' Obama cut the deficit";
+ $output[] = "in ‘400 years, 3 months,’ Obama cut the deficit";
+
+ foreach($input as $key => $in) {
+ $this->assertEquals( $output[$key], wptexturize( $in ) );
+ }
+ }
}