Ticket #22692: WPTexturize.php.diff

WPTexturize.php

@@ -194 +194 @@
                 $this->assertEquals( ' — ', wptexturize( ' -- ' ) );
                 $this->assertEquals( ' — ', wptexturize( ' -- ') );
         }
+
+        /**
+         * @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_current_behaviors() {
+                $input  = array();
+                $output = array();
+
+                // Apostrophe before a number always becomes &#8217 (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 &#8216; (opening_single_quote)
+
+                $input[]  = "word 'word word";
+                $output[] = "word &#8216;word word";
+
+                $input[]  = "word ('word word";
+                $output[] = "word (&#8216;word word";
+
+                $input[]  = "word ['word word";
+                $output[] = "word [&#8216;word word";
+
+                $input[]  = "word <'word word"; // Invalid HTML input?
+                $output[] = "word <&#8216;word word";
+
+                $input[]  = "word &lt;'word word"; // Valid HTML input triggers the apos in a word pattern
+                $output[] = "word &lt;&#8217;word word";
+
+                $input[]  = "word {'word word";
+                $output[] = "word {&#8216;word word";
+
+                $input[]  = "word \"'word word";
+                $output[] = "word &#8220;&#8216;word word"; // Two opening quotes
+
+                $input[]  = "'word word";
+                $output[] = "&#8216;word word";
+
+                $input[]  = "word('word word";
+                $output[] = "word(&#8216;word word";
+
+                $input[]  = "word['word word";
+                $output[] = "word[&#8216;word word";
+
+                $input[]  = "word<'word word";
+                $output[] = "word<&#8216;word word";
+
+                $input[]  = "word&lt;'word word";
+                $output[] = "word&lt;&#8217;word word";
+
+                $input[]  = "word{'word word";
+                $output[] = "word{&#8216;word word";
+
+                $input[]  = "word\"'word word";
+                $output[] = "word&#8221;&#8216;word word"; // Closing quote, then opening quote
+
+                $input[]  = "word ' word word";
+                $output[] = "word &#8216; word word";
+
+                $input[]  = "word (' word word";
+                $output[] = "word (&#8216; word word";
+
+                $input[]  = "word [' word word";
+                $output[] = "word [&#8216; word word";
+
+                $input[]  = "word <' word word"; // Invalid HTML input?
+                $output[] = "word <&#8216; word word";
+
+                $input[]  = "word &lt;' word word"; // Valid HTML input triggers the closing single quote here
+                $output[] = "word &lt;&#8217; word word";
+
+                $input[]  = "word {' word word";
+                $output[] = "word {&#8216; word word";
+
+                $input[]  = "word \"' word word";
+                $output[] = "word &#8220;&#8216; word word"; // Two opening quotes
+
+                $input[]  = "' word word";
+                $output[] = "&#8216; word word";
+
+                $input[]  = "word(' word word";
+                $output[] = "word(&#8216; word word";
+
+                $input[]  = "word[' word word";
+                $output[] = "word[&#8216; word word";
+
+                $input[]  = "word<' word word";
+                $output[] = "word<&#8216; word word";
+
+                $input[]  = "word&lt;' word word";
+                $output[] = "word&lt;&#8217; word word";
+
+                $input[]  = "word{' word word";
+                $output[] = "word{&#8216; word word";
+
+                $input[]  = "word\"' word word";
+                $output[] = "word&#8221;&#8216; word word"; // Closing quote, then opening quote
+
+                // Double quote after a number becomes &#8243; (double_prime)
+
+                $input[]  = 'word 99" word';
+                $output[] = 'word 99&#8243; word';
+
+                $input[]  = 'word 99"word';
+                $output[] = 'word 99&#8243;word';
+
+                $input[]  = 'word99" word';
+                $output[] = 'word99&#8243; word';
+
+                $input[]  = 'word99"word';
+                $output[] = 'word99&#8243;word';
+
+                // Apostrophe after a number becomes &#8242; (prime)
+
+                $input[]  = "word 99' word";
+                $output[] = "word 99&#8242; word";
+
+                $input[]  = "word 99'word";
+                $output[] = "word 99&#8242;word";
+
+                $input[]  = "word99' word";
+                $output[] = "word99&#8242; word";
+
+                $input[]  = "word99'word";
+                $output[] = "word99&#8242;word";
+
+                // Apostrophe "in a word" becomes &#8217; (apos)
+
+                $input[]  = "word word's word";
+                $output[] = "word word&#8217;s word";
+
+                $input[]  = "word ]'. word"; // Quotes wrapped in braces could end with apostrophes instead of closing quotes (may affect i18n)
+                $output[] = "word ]&#8217;. word";
+
+                $input[]  = "word )'. word";
+                $output[] = "word )&#8217;. word";
+
+                $input[]  = "word }'. word";
+                $output[] = "word }&#8217;. word";
+
+                $input[]  = "word >'. word"; // Not tested
+                $output[] = "word >&#8217;. word";
+
+                $input[]  = "word &gt;'. word";
+                $output[] = "word &gt;&#8217;. word";
+
+                // Double quote after a space or ([{< becomes &#8220; (opening_quote) if not followed by spaces
+
+                $input[]  = 'word "word word';
+                $output[] = 'word &#8220;word word';
+
+                $input[]  = 'word ("word word';
+                $output[] = 'word (&#8220;word word';
+
+                $input[]  = 'word ["word word';
+                $output[] = 'word [&#8220;word word';
+
+                $input[]  = 'word <"word word'; // Invalid HTML input?
+                $output[] = 'word <&#8220;word word';
+
+                $input[]  = 'word &lt;"word word'; // Valid HTML input triggers the closing quote pattern
+                $output[] = 'word &lt;&#8221;word word';
+
+                $input[]  = 'word {"word word';
+                $output[] = 'word {&#8220;word word';
+
+                $input[]  = '"word word';
+                $output[] = '&#8220;word word';
+
+                $input[]  = 'word("word word';
+                $output[] = 'word(&#8220;word word';
+
+                $input[]  = 'word["word word';
+                $output[] = 'word[&#8220;word word';
+
+                $input[]  = 'word<"word word'; // Invalid HTML input?
+                $output[] = 'word<&#8220;word word';
+
+                $input[]  = 'word&lt;"word word'; // Valid HTML input triggers the closing quote pattern
+                $output[] = 'word&lt;&#8221;word word';
+
+                $input[]  = 'word{"word word';
+                $output[] = 'word{&#8220;word word';
+
+                $input[]  = 'word "99 word';
+                $output[] = 'word &#8220;99 word';
+
+                // Double quote becomes &#8221; (closing_quote) unless it is already converted to double_prime or opening_quote.
+
+                $input[]  = 'word word" word';
+                $output[] = 'word word&#8221; word';
+
+                $input[]  = 'word word") word';
+                $output[] = 'word word&#8221;) word';
+
+                $input[]  = 'word word"] word';
+                $output[] = 'word word&#8221;] word';
+
+                $input[]  = 'word word"} word';
+                $output[] = 'word word&#8221;} word';
+
+                $input[]  = 'word word"> word'; // Invalid HTML input?
+                $output[] = 'word word&#8221;> word';
+
+                $input[]  = 'word word"&gt; word'; // Valid HTML should work
+                $output[] = 'word word&#8221;&gt; word';
+
+                $input[]  = 'word word"';
+                $output[] = 'word word&#8221;';
+
+                $input[]  = 'word word"word';
+                $output[] = 'word word&#8221;word';
+
+                $input[]  = 'word"word"word';
+                $output[] = 'word&#8221;word&#8221;word';
+
+                $input[]  = 'test sentence".';
+                $output[] = 'test sentence&#8221;.';
+
+                $input[]  = 'test sentence."';
+                $output[] = 'test sentence.&#8221;';
+
+                $input[]  = 'test sentence". word';
+                $output[] = 'test sentence&#8221;. word';
+
+                $input[]  = 'test sentence." word';
+                $output[] = 'test sentence.&#8221; word';
+
+                // Single quote followed by a space or a period becomes &#8217; (closing_single_quote)
+
+                $input[]  = "word word' word";
+                $output[] = "word word&#8217; word";
+
+                $input[]  = "word word'. word";
+                $output[] = "word word&#8217;. word";
+
+                $input[]  = "word word'.word";
+                $output[] = "word word&#8217;.word";
+
+                $input[]  = "word word'";
+                $output[] = "word word&#8217;";
+
+                $input[]  = "test sentence'.";
+                $output[] = "test sentence&#8217;.";
+
+                $input[]  = "test sentence.'";
+                $output[] = "test sentence.&#8217;";
+
+                $input[]  = "test sentence'. word";
+                $output[] = "test sentence&#8217;. word";
+
+                $input[]  = "test sentence.' word";
+                $output[] = "test sentence.&#8217; word";
+
+                // 9x9 becomes 9&#215;9
+
+                $input[]  = "9x9";
+                $output[] = "9&#215;9";
+
+                $input[]  = "12x34";
+                $output[] = "12&#215;34";
+
+                $input[]  = "9 x 9";
+                $output[] = "9 x 9";
+
+                foreach($input as $key => $in) {
+                        $this->assertEquals( $output[$key], wptexturize( $in ) );
+                }
+        }
 }