diff --git a/src/wp-includes/formatting.php b/src/wp-includes/formatting.php
index ee3bcd0a8d..e1a0ee3104 100644
--- a/src/wp-includes/formatting.php
+++ b/src/wp-includes/formatting.php
@@ -73,27 +73,50 @@ function wptexturize($text) {
$static_characters = array_merge( array( '---', ' -- ', '--', ' - ', 'xn–', '...', '``', '\'\'', ' (tm)' ), $cockney );
$static_replacements = array_merge( array( $em_dash, ' ' . $em_dash . ' ', $en_dash, ' ' . $en_dash . ' ', 'xn--', '…', $opening_quote, $closing_quote, ' ™' ), $cockneyreplace );
- $dynamic = array();
- if ( "'" != $apos ) {
- $dynamic[ '/\'(\d\d(?:’|\')?s)/' ] = $apos . '$1'; // '99's
- $dynamic[ '/\'(\d)/' ] = $apos . '$1'; // '99
- }
- if ( "'" != $opening_single_quote )
- $dynamic[ '/(\s|\A|[([{<]|")\'/' ] = '$1' . $opening_single_quote; // opening single quote, even after (, {, <, [
- if ( '"' != $double_prime )
- $dynamic[ '/(\d)"/' ] = '$1' . $double_prime; // 9" (double prime)
- if ( "'" != $prime )
- $dynamic[ '/(\d)\'/' ] = '$1' . $prime; // 9' (prime)
- if ( "'" != $apos )
- $dynamic[ '/(\S)\'([^\'\s])/' ] = '$1' . $apos . '$2'; // apostrophe in a word
- if ( '"' != $opening_quote )
- $dynamic[ '/(\s|\A|[([{<])"(?!\s)/' ] = '$1' . $opening_quote . '$2'; // opening double quote, even after (, {, <, [
- if ( '"' != $closing_quote )
- $dynamic[ '/"(\s|\S|\Z)/' ] = $closing_quote . '$1'; // closing double quote
- if ( "'" != $closing_single_quote )
- $dynamic[ '/\'([\s.]|\Z)/' ] = $closing_single_quote . '$1'; // closing single quote
+ /*
+ * Regex for common whitespace characters.
+ *
+ * By default, spaces include new lines, tabs, nbsp entities, and the UTF-8 nbsp.
+ * This is designed to replace the PCRE \s sequence. In #WP22692, that sequence
+ * was found to be unreliable due to random inclusion of the A0 byte.
+ */
+ $spaces = '[\r\n\t ]|\xC2\xA0| ';
- $dynamic[ '/\b(\d+)x(\d+)\b/' ] = '$1×$2'; // 9x9 (times)
+
+ // Pattern-based replacements of characters.
+ $dynamic = array();
+
+ // '99 '99s '99's (apostrophe)
+ if ( "'" != $apos )
+ $dynamic[ '/\'(?=\d)/' ] = $apos;
+
+ // Single quote at start, or preceded by (, {, <, [, ", or spaces.
+ if ( "'" != $opening_single_quote )
+ $dynamic[ '/(?<=\A|[([{<"]|' . $spaces . ')\'/' ] = $opening_single_quote;
+
+ // 9" (double prime)
+ if ( '"' != $double_prime )
+ $dynamic[ '/(?<=\d)"/' ] = $double_prime;
+
+ // 9' (prime)
+ if ( "'" != $prime )
+ $dynamic[ '/(?<=\d)\'/' ] = $prime;
+
+ // Apostrophe in a word. No spaces or double primes.
+ if ( "'" != $apos )
+ $dynamic[ '/(?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 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
//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"'));
@@ -194,4 +197,801 @@ class Tests_Formatting_WPTexturize extends WP_UnitTestCase {
$this->assertEquals( ' — ', wptexturize( ' -- ' ) );
$this->assertEquals( ' — ', wptexturize( ' -- ') );
}
+
+ /**
+ * Test spaces around quotes.
+ *
+ * These should never happen, even if the desired output changes some day.
+ *
+ * @ticket 22692
+ */
+ function test_spaces_around_quotes_never() {
+ $nbsp = "\xC2\xA0";
+
+ $problem_input = "$nbsp\"A";
+ $problem_output = "$nbsp”A";
+
+ $this->assertNotEquals( $problem_output, wptexturize( $problem_input ) );
+ }
+
+ /**
+ * Test spaces around quotes.
+ *
+ * These are desirable outputs for the current design.
+ *
+ * @ticket 22692
+ * @dataProvider data_spaces_around_quotes
+ */
+ function test_spaces_around_quotes( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_spaces_around_quotes() {
+ $nbsp = "\xC2\xA0";
+ $pi = "\xCE\xA0";
+
+ return array(
+ array(
+ "stop. $nbsp\"A quote after 2 spaces.\"",
+ "stop. $nbsp“A quote after 2 spaces.”",
+ ),
+ array(
+ "stop.$nbsp$nbsp\"A quote after 2 spaces.\"",
+ "stop.$nbsp$nbsp“A quote after 2 spaces.”",
+ ),
+ array(
+ "stop. $nbsp'A quote after 2 spaces.'",
+ "stop. $nbsp‘A quote after 2 spaces.’",
+ ),
+ array(
+ "stop.$nbsp$nbsp'A quote after 2 spaces.'",
+ "stop.$nbsp$nbsp‘A quote after 2 spaces.’",
+ ),
+ array(
+ "stop. \"A quote after 2 spaces.\"",
+ "stop. “A quote after 2 spaces.”",
+ ),
+ array(
+ "stop. \"A quote after 2 spaces.\"",
+ "stop. “A quote after 2 spaces.”",
+ ),
+ array(
+ "stop. 'A quote after 2 spaces.'",
+ "stop. ‘A quote after 2 spaces.’",
+ ),
+ array(
+ "stop. 'A quote after 2 spaces.'",
+ "stop. ‘A quote after 2 spaces.’",
+ ),
+ array(
+ "Contraction: $pi's",
+ "Contraction: $pi’s",
+ ),
+ );
+ }
+
+ /**
+ * Apostrophe before a number always becomes ’ (apos);
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_apos_before_digits
+ */
+ function test_apos_before_digits( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_apos_before_digits() {
+ return array(
+ array(
+ "word '99 word",
+ "word ’99 word",
+ ),
+ array(
+ "word'99 word",
+ "word’99 word",
+ ),
+ array(
+ "word '99word",
+ "word ’99word",
+ ),
+ array(
+ "word'99word",
+ "word’99word",
+ ),
+ array(
+ "word '99’s word", // Appears as a separate but logically superfluous pattern in 3.8.
+ "word ’99’s word",
+ ),
+ array(
+ "word '99's word", // Due to the logic error, second apos becomes a prime. See ticket #22823
+ "word ’99′s word",
+ ),
+ array(
+ "word '99'samsonite",
+ "word ’99′samsonite",
+ ),
+ array(
+ "according to our source, '33% of all students scored less than 50' on the test.", // Apostrophes and primes have priority over quotes
+ "according to our source, ’33% of all students scored less than 50′ on the test.",
+ ),
+ array(
+ "word '99' word", // See ticket #8775
+ "word ’99′ word",
+ ),
+ );
+ }
+
+ /**
+ * Apostrophe after a space or ([{<" becomes ‘ (opening_single_quote)
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_opening_single_quote
+ */
+ function test_opening_single_quote( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_opening_single_quote() {
+ return array(
+ array(
+ "word 'word word",
+ "word ‘word word",
+ ),
+ array(
+ "word ('word word",
+ "word (‘word word",
+ ),
+ array(
+ "word ['word word",
+ "word [‘word word",
+ ),
+ array(
+ "word <'word word", // Invalid HTML input?
+ "word <‘word word",
+ ),
+ array(
+ "word <'word word", // Valid HTML input triggers the apos in a word pattern
+ "word <’word word",
+ ),
+ array(
+ "word {'word word",
+ "word {‘word word",
+ ),
+ array(
+ "word \"'word word",
+ "word “‘word word", // Two opening quotes
+ ),
+ array(
+ "'word word",
+ "‘word word",
+ ),
+ array(
+ "word('word word",
+ "word(‘word word",
+ ),
+ array(
+ "word['word word",
+ "word[‘word word",
+ ),
+ array(
+ "word<'word word",
+ "word<‘word word",
+ ),
+ array(
+ "word<'word word",
+ "word<’word word",
+ ),
+ array(
+ "word{'word word",
+ "word{‘word word",
+ ),
+ array(
+ "word\"'word word",
+ "word”‘word word", // Closing quote, then opening quote
+ ),
+ array(
+ "word ' word word",
+ "word ‘ word word",
+ ),
+ array(
+ "word (' word word",
+ "word (‘ word word",
+ ),
+ array(
+ "word [' word word",
+ "word [‘ word word",
+ ),
+ array(
+ "word <' word word", // Invalid HTML input?
+ "word <‘ word word",
+ ),
+ array(
+ "word <' word word", // Valid HTML input triggers the closing single quote here
+ "word <’ word word",
+ ),
+ array(
+ "word {' word word",
+ "word {‘ word word",
+ ),
+ array(
+ "word \"' word word",
+ "word “‘ word word", // Two opening quotes
+ ),
+ array(
+ "' word word",
+ "‘ word word",
+ ),
+ array(
+ "word(' word word",
+ "word(‘ word word",
+ ),
+ array(
+ "word[' word word",
+ "word[‘ word word",
+ ),
+ array(
+ "word<' word word",
+ "word<‘ word word",
+ ),
+ array(
+ "word<' word word",
+ "word<’ word word",
+ ),
+ array(
+ "word{' word word",
+ "word{‘ word word",
+ ),
+ array(
+ "word\"' word word",
+ "word”‘ word word", // Closing quote, then opening quote
+ ),
+ );
+ }
+
+ /**
+ * Double quote after a number becomes ″ (double_prime)
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_double_prime
+ */
+ function test_double_prime( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_double_prime() {
+ return array(
+ array(
+ 'word 99" word',
+ 'word 99″ word',
+ ),
+ array(
+ 'word 99"word',
+ 'word 99″word',
+ ),
+ array(
+ 'word99" word',
+ 'word99″ word',
+ ),
+ array(
+ 'word99"word',
+ 'word99″word',
+ ),
+ );
+ }
+
+ /**
+ * Apostrophe after a number becomes ′ (prime)
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_single_prime
+ */
+ function test_single_prime( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_single_prime() {
+ return array(
+ array(
+ "word 99' word",
+ "word 99′ word",
+ ),
+ array(
+ "word 99'word",
+ "word 99′word",
+ ),
+ array(
+ "word99' word",
+ "word99′ word",
+ ),
+ array(
+ "word99'word",
+ "word99′word",
+ ),
+ );
+ }
+
+ /**
+ * Apostrophe "in a word" becomes ’ (apos)
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_contractions
+ */
+ function test_contractions( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_contractions() {
+ return array(
+ array(
+ "word word's word",
+ "word word’s word",
+ ),
+ array(
+ "word word'. word", // Quotes with outside punctuation could end with apostrophes instead of closing quotes (may affect i18n)
+ "word word’. word",
+ ),
+ array(
+ "word ]'. word",
+ "word ]’. word",
+ ),
+ array(
+ "word )'. word",
+ "word )’. word",
+ ),
+ array(
+ "word }'. word",
+ "word }’. word",
+ ),
+ array(
+ "word >'. word", // Not tested
+ "word >’. word",
+ ),
+ array(
+ "word >'. word",
+ "word >’. word",
+ ),
+ );
+ }
+
+ /**
+ * Double quote after a space or ([{< becomes “ (opening_quote) if not followed by spaces
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_opening_quote
+ */
+ function test_opening_quote( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_opening_quote() {
+ return array(
+ array(
+ 'word "word word',
+ 'word “word word',
+ ),
+ array(
+ 'word ("word word',
+ 'word (“word word',
+ ),
+ array(
+ 'word ["word word',
+ 'word [“word word',
+ ),
+ array(
+ 'word <"word word', // Invalid HTML input?
+ 'word <“word word',
+ ),
+ array(
+ 'word <"word word', // Valid HTML input triggers the closing quote pattern
+ 'word <”word word',
+ ),
+ array(
+ 'word {"word word',
+ 'word {“word word',
+ ),
+ array(
+ '"word word',
+ '“word word',
+ ),
+ array(
+ 'word("word word',
+ 'word(“word word',
+ ),
+ array(
+ 'word["word word',
+ 'word[“word word',
+ ),
+ array(
+ 'word<"word word', // Invalid HTML input?
+ 'word<“word word',
+ ),
+ array(
+ 'word<"word word', // Valid HTML input triggers the closing quote pattern
+ 'word<”word word',
+ ),
+ array(
+ 'word{"word word',
+ 'word{“word word',
+ ),
+ array(
+ 'word "99 word',
+ 'word “99 word',
+ ),
+ );
+ }
+
+ /**
+ * Double quote becomes ” (closing_quote) unless it is already converted to double_prime or opening_quote.
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_closing_quote
+ */
+ function test_closing_quote( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_closing_quote() {
+ return array(
+ array(
+ 'word word" word',
+ 'word word” word',
+ ),
+ array(
+ 'word word") word',
+ 'word word”) word',
+ ),
+ array(
+ 'word word"] word',
+ 'word word”] word',
+ ),
+ array(
+ 'word word"} word',
+ 'word word”} word',
+ ),
+ array(
+ 'word word"> word', // Invalid HTML input?
+ 'word word”> word',
+ ),
+ array(
+ 'word word"> word', // Valid HTML should work
+ 'word word”> word',
+ ),
+ array(
+ 'word word"',
+ 'word word”',
+ ),
+ array(
+ 'word word"word',
+ 'word word”word',
+ ),
+ array(
+ 'word"word"word',
+ 'word”word”word',
+ ),
+ array(
+ 'test sentence".',
+ 'test sentence”.',
+ ),
+ array(
+ 'test sentence."',
+ 'test sentence.”',
+ ),
+ array(
+ 'test sentence". word',
+ 'test sentence”. word',
+ ),
+ array(
+ 'test sentence." word',
+ 'test sentence.” word',
+ ),
+ );
+ }
+
+ /**
+ * Test that single quotes followed by a space or a period become ’ (closing_single_quote)
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_closing_single_quote
+ */
+ function test_closing_single_quote( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_closing_single_quote() {
+ return array(
+ array(
+ "word word' word",
+ "word word’ word",
+ ),
+ array(
+ "word word'. word",
+ "word word’. word",
+ ),
+ array(
+ "word word'.word",
+ "word word’.word",
+ ),
+ array(
+ "word word'",
+ "word word’",
+ ),
+ array(
+ "test sentence'.",
+ "test sentence’.",
+ ),
+ array(
+ "test sentence.'",
+ "test sentence.’",
+ ),
+ array(
+ "test sentence'. word",
+ "test sentence’. word",
+ ),
+ array(
+ "test sentence.' word",
+ "test sentence.’ word",
+ ),
+ );
+ }
+
+ /**
+ * Tests multiplication.
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_multiplication
+ */
+ function test_multiplication( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_multiplication() {
+ return array(
+ array(
+ "9x9",
+ "9×9",
+ ),
+ array(
+ "12x34",
+ "12×34",
+ ),
+ array(
+ "9 x 9",
+ "9 x 9",
+ ),
+ );
+ }
+
+ /**
+ * Test ampersands. & always becomes & unless it is followed by # or ;
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_ampersand
+ */
+ function test_ampersand( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_ampersand() {
+ return array(
+ array(
+ "word & word",
+ "word & word",
+ ),
+ array(
+ "word&word",
+ "word&word",
+ ),
+ array(
+ "word word",
+ "word word",
+ ),
+ array(
+ "word & word",
+ "word & word",
+ ),
+ array(
+ "word &# word",
+ "word &# word", // invalid output?
+ ),
+ array(
+ "word &44; word",
+ "word &44; word",
+ ),
+ array(
+ "word && word",
+ "word && word",
+ ),
+ array(
+ "word &!amp; word",
+ "word &!amp; word",
+ ),
+ );
+ }
+
+ /**
+ * Test "cockney" phrases, which begin with an apostrophe instead of an opening single quote.
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_cockney
+ */
+ function test_cockney( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_cockney() {
+ return array(
+ array(
+ "word 'tain't word",
+ "word ’tain’t word",
+ ),
+ array(
+ "word 'twere word",
+ "word ’twere word",
+ ),
+ array(
+ "word 'twas word",
+ "word ’twas word",
+ ),
+ array(
+ "word 'tis word",
+ "word ’tis word",
+ ),
+ array(
+ "word 'twill word",
+ "word ’twill word",
+ ),
+ array(
+ "word 'til word",
+ "word ’til word",
+ ),
+ array(
+ "word 'bout word",
+ "word ’bout word",
+ ),
+ array(
+ "word 'nuff word",
+ "word ’nuff word",
+ ),
+ array(
+ "word 'round word",
+ "word ’round word",
+ ),
+ array(
+ "word 'cause word",
+ "word ’cause word",
+ ),
+ array(
+ "word 'em word",
+ "word ‘em word",
+ ),
+ );
+ }
+
+ /**
+ * Test smart dashes.
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_smart_dashes
+ */
+ function test_smart_dashes( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_smart_dashes() {
+ return array(
+ array(
+ "word --- word",
+ "word — word",
+ ),
+ array(
+ "word---word",
+ "word—word",
+ ),
+ array(
+ "word -- word",
+ "word — word",
+ ),
+ array(
+ "word--word",
+ "word–word",
+ ),
+ array(
+ "word - word",
+ "word – word",
+ ),
+ array(
+ "word-word",
+ "word-word",
+ ),
+ array(
+ "word xn– word",
+ "word xn-- word",
+ ),
+ array(
+ "wordxn–word",
+ "wordxn--word",
+ ),
+ );
+ }
+
+ /**
+ * Test miscellaneous static replacements.
+ *
+ * Checks all baseline patterns. If anything ever changes in wptexturize(), these tests may fail.
+ *
+ * @ticket 22692
+ * @dataProvider data_misc_static_replacements
+ */
+ function test_misc_static_replacements( $input, $output ) {
+ return $this->assertEquals( $output, wptexturize( $input ) );
+ }
+
+ function data_misc_static_replacements() {
+ return array(
+ array(
+ "word ... word",
+ "word … word",
+ ),
+ array(
+ "word...word",
+ "word…word",
+ ),
+ array(
+ "word `` word",
+ "word “ word",
+ ),
+ array(
+ "word``word",
+ "word“word",
+ ),
+ array(
+ "word '' word",
+ "word ” word",
+ ),
+ array(
+ "word''word",
+ "word”word",
+ ),
+ array(
+ "word (tm) word",
+ "word ™ word",
+ ),
+ array(
+ "word (tm)word",
+ "word ™word",
+ ),
+ array(
+ "word(tm) word",
+ "word(tm) word",
+ ),
+ array(
+ "word(tm)word",
+ "word(tm)word",
+ ),
+ );
+ }
}