Ticket #50683: shortcode-parser.7.diff

src/wp-includes/shortcodes.php

@@ -214 +214 @@
 
         $content = do_shortcodes_in_html_tags( $content, $ignore_html, $tagnames );
 
-        $pattern = get_shortcode_regex( $tagnames );
-        $content = preg_replace_callback( "/$pattern/", 'do_shortcode_tag', $content );
+        $shortcode_parser = new Shortcode_Parser( $content, $tagnames );
+        $content = $shortcode_parser->parse();
 
         // Always restore square braces so we don't break things like <!--[if IE ]>.
         $content = unescape_invalid_shortcodes( $content );
@@ -224 +224 @@
 }
 
 /**
+ * A collection of methods for parsing and executing shortcodes in content.
+ */
+class Shortcode_Parser {
+        /**
+         * The content being parsed.
+         *
+         * @var string $content
+         */
+        private $content;
+
+        /**
+         * The current state of the parser.
+         *
+         * @var int $state One of the SHORTCODE_PARSE_SATE_* constants listed below.
+         */
+        private $state;
+
+        /**
+         * The current position of the parsing cursor in $content.
+         *
+         * @var int $cursor_position
+         */
+        private $cursor_position;
+
+        /**
+         * The stack of unprocessed shortcodes.
+         *
+         * As shortcodes are opened, they are placed on the stack, and as they're
+         * closed, they're processed and removed.
+         *
+         * @var array $stack
+         */
+        private $stack;
+
+        /**
+         * A list of all registered shortcode tag names.
+         *
+         * @var array $tagnames
+         */
+        private $tagnames;
+
+        /**
+         * The shortcode currently being parsed.
+         *
+         * As a shortcode is being parsed, it is stored here. If a new shortcode is
+         * found before parsing is complete, this shortcode is moved to the stack,
+         * and the new shortcode is stored here.
+         *
+         * @var array $current_shortcode {
+         *     @type string $full_tag The content that makes up this tag, from the opening bracket
+         *                            to the closing bracket.
+         *     @type string $extra_opening_bracket Either the character '[' or an empty string,
+         *                                         depending on whether there was an extra opening
+         *                                         bracket for this shortcode. Essentially a flag. 
+         *     @type string $tag_slug The shortcode tag slug.
+         *     @type string $atts_and_values The (unparsed) part of the shortcode tag that contains
+         *                                   attributes and their optional values.
+         *     @type string $self_closing_slash Like $extra_opening_bracket, this is either the
+         *                                      slash used to self-close the shortcode tag, or an
+         *                                      empty string.
+         *     @type int $cursor_position The cursor position where this shortcode began.
+         * }
+         */
+        private $current_shortcode;
+
+        /**
+         * The default parsing state -- the cursor is not in a shortcode tag or
+         * shortcode content or a quoted string in a shortcode attribute value.
+         *
+         * @var int $SHORTCODE_PARSE_STATE_DEFAULT
+         */
+        const SHORTCODE_PARSE_STATE_DEFAULT          = 0;
+
+        /**
+         * The cursor is inside the shortcode tag, past the shortcode tag slug.
+         *
+         * @var int $SHORTCODE_PARSE_STATE_IN_TAG
+         */
+        const SHORTCODE_PARSE_STATE_IN_TAG           = 1;
+
+        /**
+         * The cursor is in the content of the shortcode -- past the opening tag
+         * but not yet to the closing tag.
+         *
+         * @var int $SHORTCODE_PARSE_STATE_IN_CONTENT
+         */
+        const SHORTCODE_PARSE_STATE_IN_CONTENT       = 2;
+
+        /**
+         * The cursor is inside of a quoted string in the shortcode tag.
+         *
+         * @var int $SHORTCODE_PARSE_STATE_IN_QUOTED_STRING
+         */
+        const SHORTCODE_PARSE_STATE_IN_QUOTED_STRING = 3;
+
+        /**
+         * Store the content and tag names for later use.
+         *
+         * @param string $content The HTML/text content to parse for shortcodes.
+         * @param array $tagnames An array of string shortcode tag names.
+         */
+        public function __construct( $content, $tagnames ) {
+                $this->content  = $content;
+                $this->tagnames = $tagnames;
+        }
+
+        /**
+         * Parse shortcodes in content and replace them with the output that their
+         * handler functions generate.
+         *
+         * @return string The content with shortcodes replaced by their output.
+         */
+        public function parse() {
+                $this->stack = array();
+
+                /*
+                 * A regular expression that checks whether a string appears to begin
+                 * with a tag for a registered shortcode.
+                 */
+                $registered_shortcode_regex= '/^(?P<extra_opening_bracket>\\[?)(?P<opening_bracket>\\[)(?P<tag_slug>' . join( '|', array_map( 'preg_quote', $this->tagnames ) ) . ')(?![\\w-])/u';
+
+                $this->cursor_position = 0;
+
+                // Save some parsing time by starting a few characters before the first bracket.
+                $this->forward_cursor_to_next_bracket();
+
+                $this->state = self::SHORTCODE_PARSE_STATE_DEFAULT;
+
+                $is_escaped = false;
+                $delimiter = null;
+
+                while ( $this->cursor_position < strlen( $this->content ) ) {
+                        $char = substr( $this->content, $this->cursor_position, 1 );
+
+                        $found_escape_character = false;
+
+                        switch ( $this->state ) {
+                                case self::SHORTCODE_PARSE_STATE_DEFAULT:
+                                case self::SHORTCODE_PARSE_STATE_IN_CONTENT:
+                                        if (
+                                                   ! $is_escaped
+                                                && '[' === $char
+                                                && preg_match( $registered_shortcode_regex, substr( $this->content, $this->cursor_position ), $m ) ) {
+                                                if ( $this->current_shortcode ) {
+                                                        $this->stack[] = $this->current_shortcode;
+                                                }
+
+                                                // We have found the beginning of a shortcode.
+                                                $this->current_shortcode = array(
+                                                        'full_tag' => $m[0],
+                                                        'extra_opening_bracket' => $m['extra_opening_bracket'],
+                                                        'tag_slug' => $m['tag_slug'],
+                                                        'atts_and_values' => '',
+                                                        'self_closing_slash' => '',
+                                                        'inner_content' => '',
+                                                        'extra_closing_bracket' => '',
+                                                        'cursor_position' => $this->cursor_position,
+                                                );
+
+                                                $this->cursor_position += strlen( $m[0] );
+
+                                                // Move back one so it's as if we just processed the last character of the shortcode slug.
+                                                $this->cursor_position--;
+
+                                                $this->state = self::SHORTCODE_PARSE_STATE_IN_TAG;
+                                        } elseif ( self::SHORTCODE_PARSE_STATE_IN_CONTENT === $this->state ) {
+                                                $this->current_shortcode['full_tag'] .= $char;
+
+                                                if ( '[' === $char ) {
+                                                        // Check whether it's a closing tag of any currently open shortcode.
+                                                        $rest_of_closing_tag = '/' . $this->current_shortcode['tag_slug'] . ']';
+
+                                                        if ( $rest_of_closing_tag === substr( $this->content, $this->cursor_position + 1, strlen( $rest_of_closing_tag ) ) ) {
+                                                                // The end of this shortcode.
+
+                                                                $this->current_shortcode['full_tag'] .= $rest_of_closing_tag;
+
+                                                                // Move the cursor to the end of the closing tag.
+                                                                $this->cursor_position += strlen( $rest_of_closing_tag );
+
+                                                                if ( $this->current_shortcode['extra_opening_bracket'] ) {
+                                                                        if ( ']' === substr( $this->content, $this->cursor_position + 1, 1 ) ) {
+                                                                                $this->current_shortcode['full_tag'] .= ']';
+                                                                                $this->current_shortcode['extra_closing_bracket'] = ']';
+                                                                                $this->cursor_position++;
+                                                                        } else {
+                                                                                // If there was an extra opening bracket but not an extra closing bracket, ignore the extra opening bracket.
+
+                                                                                $this->current_shortcode['full_tag'] = substr( $this->current_shortcode['full_tag'], 1 );
+                                                                                $this->current_shortcode['extra_opening_bracket'] = '';
+
+                                                                                // We initially thought it had an extra opening bracket, but it doesn't so it started one character later than we thought.
+                                                                                $this->current_shortcode['cursor_position'] += 1;
+                                                                        }
+                                                                }
+
+                                                                $this->process_current_shortcode();
+                                                        } else {
+                                                                $found_matching_shortcode = false;
+
+                                                                for ( $stack_index = count( $this->stack ) - 1; $stack_index >= 0; $stack_index-- ) {
+                                                                        $rest_of_closing_tag = '/' . $this->stack[ $stack_index ]['tag_slug'] . ']';
+
+                                                                        if ( $rest_of_closing_tag === substr( $this->content, $this->cursor_position + 1, strlen( $rest_of_closing_tag ) ) ) {
+                                                                                // Yes, it closes this one.
+                                                                                $found_matching_shortcode = true;
+
+                                                                                /*
+                                                                                 * We already saved the bracket as part of the full tag, expecting that the
+                                                                                 * closing tag would be for the current shortcode. It's not, so remove it.
+                                                                                 */
+                                                                                $this->current_shortcode['full_tag'] = substr( $this->current_shortcode['full_tag'], 0, -1 );
+
+                                                                                // This means that the "current" shortcode and any others above this one on the stack need to be closed out, because they are self-closing.
+                                                                                do {
+                                                                                        $this->current_shortcode['full_tag'] = substr( $this->current_shortcode['full_tag'], 0, -1 * strlen( $this->current_shortcode['inner_content'] ) );
+
+                                                                                        // And there is no inner content.
+                                                                                        $this->current_shortcode['inner_content'] = '';
+
+                                                                                        $this->process_current_shortcode(); // This sets $current_shortcode using the top stack item, so we don't need to do it.
+                                                                                } while ( count( $this->stack ) > $stack_index + 1 );
+
+                                                                                /*
+                                                                                 * At this point, the shortcode that is being closed right now is $this->current_shortcode.
+                                                                                 * The easiest way to process this without duplicating code is to reprocess the current
+                                                                                 * character with the new stack and current shortcode, so the section above will get
+                                                                                 * triggered, since the closing tag will be for the current shortcode.
+                                                                                 */
+
+                                                                                continue 3;
+                                                                        }
+                                                                }
+
+                                                                if ( ! $found_matching_shortcode ) {
+                                                                        $this->current_shortcode['inner_content'] .= $char;
+                                                                }
+                                                        }
+                                                } else {
+                                                        $this->current_shortcode['inner_content'] .= $char;
+                                                }
+                                        }
+
+                                        break;
+                                case self::SHORTCODE_PARSE_STATE_IN_TAG:
+                                        $this->current_shortcode['full_tag'] .= $char;
+
+                                        if ( '/' === $char && substr( $this->content, $this->cursor_position + 1, 1 ) === ']' ) {
+                                                // The shortcode is over.
+                                                $this->current_shortcode['self_closing_slash'] = '/';
+                                                $this->current_shortcode['full_tag'] .= ']';
+                                                $this->cursor_position++;
+
+                                                // If the shortcode had an extra opening bracket but doesn't have an extra closing bracket, ignore the extra opening bracket.
+
+                                                if ( $this->current_shortcode['extra_opening_bracket'] ) {
+                                                        if ( ']' === substr( $this->content, $this->cursor_position + 1, 1 ) ) {
+                                                                $this->current_shortcode['extra_closing_bracket'] = ']';
+                                                                $this->current_shortcode['full_tag'] .= ']';
+                                                                $this->cursor_position++;
+                                                        } else {
+                                                                $this->current_shortcode['full_tag'] = substr( $this->current_shortcode['full_tag'], 1 );
+                                                                $this->current_shortcode['extra_opening_bracket'] = '';
+
+                                                                /*
+                                                                 * We initially thought it had an extra opening bracket, but it doesn't,
+                                                                 * so it started one character later than we thought.
+                                                                 */
+                                                                $this->current_shortcode['cursor_position'] += 1;
+                                                        }
+                                                }
+
+                                                $this->process_current_shortcode();
+
+                                                break;
+                                        } elseif ( ']' === $char ) {
+                                                if ( $this->current_shortcode['extra_opening_bracket'] ) {
+                                                        /*
+                                                         * This makes the assumption that this shortcode is closed as soon as the double brackets are found:
+                                                         *
+                                                         * [[my-shortcode]][/my-shortcode]]
+                                                         *
+                                                         * But in theory, this could just be a shortcode with the content "]".
+                                                         */
+
+                                                        if ( ']' === substr( $this->content, $this->cursor_position + 1, 1 ) ) {
+                                                                $this->current_shortcode['extra_closing_bracket'] = ']';
+                                                                $this->current_shortcode['full_tag'] .= ']';
+                                                                $this->cursor_position++;
+
+                                                                $this->process_current_shortcode();
+                                                                break;
+                                                        } else {
+                                                                // There was not an extra closing bracket.
+                                                        }
+                                                }
+
+                                                if ( false === strpos( substr( $this->content, $this->cursor_position ), '[/' . $this->current_shortcode['tag_slug'] . ']' ) ) {
+                                                        // If there's no closing tag, it's a self-enclosed shortcode, and we're done with it.
+                                                        $this->process_current_shortcode();
+                                                } else {
+                                                        $this->state = self::SHORTCODE_PARSE_STATE_IN_CONTENT;
+                                                        
+                                                        $current_cursor_position = $this->cursor_position;
+                                                        $this->forward_cursor_to_next_bracket();
+                        
+                                                        if ( $this->cursor_position != $current_cursor_position ) {
+                                                                // The +1 is because the character at $current_cursor_position has already been recorded.
+                                                                $skipped_content = substr( $this->content, $current_cursor_position + 1, $this->cursor_position - $current_cursor_position );
+                                
+                                                                $this->current_shortcode['inner_content'] .= $skipped_content;
+                                                                $this->current_shortcode['full_tag'] .= $skipped_content;
+                                                        }
+                                                }
+                                        } else {
+                                                $this->current_shortcode['atts_and_values'] .= $char;
+
+                                                if ( '"' === $char || "'" === $char ) {
+                                                        $this->state = self::SHORTCODE_PARSE_STATE_IN_QUOTED_STRING;
+                                                        $delimiter = $char;
+                                                } else {
+                                                        // Nothing to do.
+                                                }
+                                        }
+
+                                        break;
+                                case self::SHORTCODE_PARSE_STATE_IN_QUOTED_STRING:
+                                        $this->current_shortcode['full_tag'] .= $char;
+                                        $this->current_shortcode['atts_and_values'] .= $char;
+
+                                        if ( $is_escaped ) {
+                                                // Nothing to do. This is just an escaped character to be taken literally.
+                                        } else {
+                                                // Not escaped.
+                                                if ( '\\' === $char ) {
+                                                        // The next character is escaped.
+                                                        $found_escape_character = true;
+                                                } elseif ( $char === $delimiter ) {
+                                                        $this->state = self::SHORTCODE_PARSE_STATE_IN_TAG;
+                                                        $delimiter = null;
+                                                }
+                                        }
+
+                                        break;
+                        }
+
+                        // Is the next character escaped?
+                        if ( $found_escape_character ) {
+                                $is_escaped = true;
+                        } else {
+                                // If we didn't find an escape character here, then no.
+                                $is_escaped = false;
+                        }
+
+                        $this->cursor_position++;
+                }
+
+                if ( self::SHORTCODE_PARSE_STATE_IN_QUOTED_STRING === $this->state ) {
+                        /*
+                         * example: This is my content [footag foo=" [bartag]
+                         * Should it be reprocessed in order to convert [bartag] or is this considered malformed?
+                         */
+                }
+
+                if ( self::SHORTCODE_PARSE_STATE_IN_TAG === $this->state ) {
+                        /*
+                         * example: This is my content [footag foo="abc" bar="def" [bartag]
+                         * Should it be reprocessed in order to convert [bartag] or is this considered malformed?
+                         */
+                }
+
+                if ( $this->current_shortcode ) {
+                        /*
+                         * If we end with shortcodes still on the stack, then there was a situation like this:
+                         *
+                         * [footag] [bartag] [baztag] [footag]content[/footag]
+                         *
+                         * i.e., a scenario where the parser was unsure whether the first [footag] was self-closing or not.
+                         *
+                         * By this point, $content will be in this format:
+                         *
+                         * [footag] bartag-output baztag-output footag-content-output
+                         *
+                         * so we need to back up and process the still-stored shortcodes as unclosed.
+                         *
+                         * An extreme version of this would look like:
+                         *
+                         * [footag] [footag] [footag] [footag] [footag] [footag] [footag] [footag] ... [footag][/footag]
+                         *
+                         * where the last tag would be the only one processed normally above and there would be n-1 [footag]s still on the stack.
+                         */
+                        while ( $this->current_shortcode ) {
+                                // What we thought was part of this tag was just regular content.
+                                $this->current_shortcode['full_tag'] = substr( $this->current_shortcode['full_tag'], 0, -1 * strlen( $this->current_shortcode['inner_content'] ) );
+
+                                // And there is no inner content.
+                                $this->current_shortcode['inner_content'] = '';
+
+                                $this->process_current_shortcode(); // This sets $current_shortcode, so we don't need to do it.
+                        }
+                }
+
+                return $this->content;
+        }
+
+        /**
+         * Create an argument to pass to do_shortcode_tag.
+         *
+         * The format of this argument was determined by the capture groups of the
+         * regular expression that used to be used to parse shortcodes out of content.
+         *
+         * @param array $shortcode An associative array comprising data about a shortcode in the text.
+         * @return array A numerically-indexed array of the shortcode data ready for do_shortcode_tag().
+         */
+        private function shortcode_argument( $shortcode ) {
+                return array(
+                        $shortcode['full_tag'],
+                        $shortcode['extra_opening_bracket'],
+                        $shortcode['tag_slug'],
+                        $shortcode['atts_and_values'],
+                        $shortcode['self_closing_slash'],
+                        $shortcode['inner_content'],
+                        $shortcode['extra_closing_bracket'],
+                );
+        }
+
+        /**
+         * Process the shortcode at the top of the stack.
+         *
+         * The shortcode at the top of the stack is complete and can be processed.
+         * Process it and modify the enclosing shortcode as if the content was passed in
+         * with this shortcode already converted into HTML.
+         */
+        private function process_current_shortcode() {
+                $argument_for_do_shortcode_tag = $this->shortcode_argument( $this->current_shortcode );
+
+                $shortcode_output = do_shortcode_tag( $argument_for_do_shortcode_tag );
+
+                /*
+                 * Replace based on position rather than find and replace, since this content is possible:
+                 *
+                 * Test 123 [some-shortcode] To use my shortcode, type [[some-shortcode]].
+                 */
+                $this->content =
+                          substr( $this->content, 0, $this->current_shortcode['cursor_position'] )
+                        . $shortcode_output
+                        . substr( $this->content, $this->current_shortcode['cursor_position'] + strlen( $this->current_shortcode['full_tag'] ) )
+                        ;
+
+                /*
+                 * Update the cursor position to the end of this shortcode's output.
+                 * The -1 is because the position is incremented after this gets called to move it to the next character.
+                 */
+                $this->cursor_position = $this->current_shortcode['cursor_position'] + strlen( $shortcode_output ) - 1;
+
+                // For any enclosing shortcode, its inner content needs to include the full output of this shortcode.
+                if ( ! empty( $this->stack ) ) {
+                        $this->current_shortcode = array_pop( $this->stack );
+
+                        $this->current_shortcode['inner_content'] .= $shortcode_output;
+                        $this->current_shortcode['full_tag'] .= $shortcode_output;
+
+                        $this->state = self::SHORTCODE_PARSE_STATE_IN_CONTENT;
+                        
+                        $current_cursor_position = $this->cursor_position;
+                        $this->forward_cursor_to_next_bracket();
+                        
+                        if ( $this->cursor_position != $current_cursor_position ) {
+                                /*
+                                 * The +1 is because the character at $current_cursor_position has already been recorded.
+                                 */
+                                $skipped_content = substr( $this->content, $current_cursor_position + 1, $this->cursor_position - $current_cursor_position );
+                                
+                                $this->current_shortcode['inner_content'] .= $skipped_content;
+                                $this->current_shortcode['full_tag'] .= $skipped_content;
+                        }
+                } else {
+                        $this->current_shortcode = null;
+
+                        $this->state = self::SHORTCODE_PARSE_STATE_DEFAULT;
+
+                        // In the default state, we can skip over any content that couldn't be a shortcode, so let's move forward near the next bracket.
+                        $this->forward_cursor_to_next_bracket();
+                }
+        }
+
+        /**
+         * Moves the parsing cursor to the next possible location that might
+         * include a shortcode.
+         *
+         * The specific location is directly before the next bracket or the end
+         * of the content if there is no next bracket.
+         */
+        private function forward_cursor_to_next_bracket() {
+                /*
+                 * The max() here is because $cursor_position can be -1 if a shortcode
+                 * at the beginning of the content didn't have any output and reset the
+                 * cursor back to the beginning. It's -1 instead of zero because it will
+                 * be incremented later in the loop to set it to zero for the next iteration.
+                 */
+                $next_bracket_location = strpos( $this->content, '[', max( 0, $this->cursor_position ) );
+
+                if ( false === $next_bracket_location ) {
+                        // There is no next bracket, so fast-forward to the end.
+                        $next_bracket_location = strlen( $this->content );
+                }
+
+                /*
+                 * Again, the -1 is because this will be incremented before it is used,
+                 * and we really want it to have a minimum value of zero.
+                 */
+                $this->cursor_position = max( -1, $next_bracket_location - 1 );
+        }
+}
+
+/**
  * Retrieve the shortcode regular expression for searching.
  *
  * The regular expression combines the shortcode tags in the regular expression

tests/phpunit/tests/shortcode.php

@@ -972 +972 @@
                 );
                 $this->assertSame( 'test-shortcode-tag', $this->tagname );
         }
-}
+        
+        function test_bracket_in_shortcode_attribute() {
+                do_shortcode( '[test-shortcode-tag subject="[This is my subject]" /]' );
+                $expected_attrs = array(
+                        "subject" => "[This is my subject]",
+                );
+                $this->assertEquals( $expected_attrs, $this->atts );
+        }
+        
+        function test_self_closing_shortcode_with_quoted_end_tag() {
+                $out = do_shortcode( '[test-shortcode-tag]Test 123[footag foo="[/test-shortcode-tag]"/] [baztag]bazcontent[/baztag]' );
+                
+                $this->assertEquals( 'Test 123foo = [/test-shortcode-tag] content = bazcontent', $out );
+        }
+        
+        function test_nested_shortcodes() {
+                do_shortcode( '[test-shortcode-tag]Some content [footag foo="foo content"/] some other content[/test-shortcode-tag]' );
+                
+                $this->assertEquals( 'Some content foo = foo content some other content', $this->content );
+
+                $out = do_shortcode( '[footag foo="1"][footag foo="2"][footag foo="3"][footag foo="4"][/footag][footag foo="4a"][/footag][/footag][/footag][/footag]' );
+                
+                $this->assertEquals( 'foo = 1', $out );
+
+                $out = do_shortcode( '[footag foo="1"] abc [bartag foo="2"] def [/footag] something else [test-shortcode-tag attr="[/footag]" attr2="[/bartag]"][/test-shortcode-tag]' );
+                
+                $this->assertEquals( 'foo = 1something else ', $out );
+        }
+        
+        /**
+         * @ticket 49955
+         */
+        function test_ticket_49955() {
+                add_shortcode( 'ucase', function ( $atts, $content ) {
+                        return strtoupper( $content );
+                } );
+                        
+                $out = do_shortcode( 'This [[ucase]] shortcode [ucase]demonstrates[/ucase] the usage of enclosing shortcodes.' );
+                
+                $this->assertEquals( 'This [ucase] shortcode DEMONSTRATES the usage of enclosing shortcodes.', $out );
+        }
+        
+        /**
+         * @ticket 43725
+         */
+        public function test_same_tag_multiple_formats_open_closed_one() {
+                $in = <<<EOT
+This post uses URL multiple times.
+
+[url]Now this is wrapped[/url]
+
+[url] This one is standalone
+
+[url]Now this is wrapped too[/url]
+EOT;
+                $expected = <<<EOT
+This post uses URL multiple times.
+
+http://www.wordpress.org/
+
+http://www.wordpress.org/ This one is standalone
+
+http://www.wordpress.org/
+EOT;
+                $out      = do_shortcode( $in );
+                $this->assertEquals( strip_ws( $expected ), strip_ws( $out ) );
+        }
+
+        /**
+         * @ticket 43725
+         */
+        public function test_same_tag_multiple_formats_open_closed_two() {
+                $in = <<<EOT
+This post uses URL multiple times.
+
+[url]Now this is wrapped[/url]
+
+[url/] This one is standalone
+
+[url]Now this is wrapped too[/url]
+EOT;
+                $expected = <<<EOT
+This post uses URL multiple times.
+
+http://www.wordpress.org/
+
+http://www.wordpress.org/ This one is standalone
+
+http://www.wordpress.org/
+EOT;
+                $out      = do_shortcode( $in );
+                $this->assertEquals( strip_ws( $expected ), strip_ws( $out ) );
+        }
+
+        /**
+         * Not really a test suite, but the easiest way I could find to test the speed of shortcode parsing.
+         */
+        public function test_speed() {
+                $total_time = 0;
+                
+                // Load a list of example shortcodes.
+                $example_shortcodes = glob( "./tests/phpunit/tests/shortcode_examples/*.txt" );
+                
+                $alphabet = "`1234567890-=qwertyuiop\asdfghjkl;'zxcvbnm,./~!@#$%^&*()_+QWERTYUIOP{}|ASDFGHJKL:\"ZXCVBNM<>?          \n\n\n\n\n\n\n\n\t\t\t\t\t\t\t\t";
+                
+                // Add random strings to all of the example shortcodes.
+                // This lets us test speed on content that is different lengths, and it also
+                // works as a fuzzer, letting us find any situation where parsing fails because of the content.
+
+                $short_content = '';
+                $medium_content = '';
+                $long_content = '';
+                
+                for ( $i = 0; $i < 10; $i++ ) {
+                        $short_content .= $alphabet[ rand( 0, strlen( $alphabet ) - 1 ) ];
+                }
+
+                for ( $i = 0; $i < 100; $i++ ) {
+                        $medium_content .= $alphabet[ rand( 0, strlen( $alphabet ) - 1 ) ];
+                }
+
+                for ( $i = 0; $i < 1000; $i++ ) {
+                        $long_content .= $alphabet[ rand( 0, strlen( $alphabet ) - 1 ) ];
+                }
+                
+                foreach ( $example_shortcodes as $example_shortcode ) {
+                        $shortcode = file_get_contents( $example_shortcode );
+                
+                        $variations = array(
+                                "shortcode only" => $shortcode,
+                                "short content suffix" => $short_content . $shortcode,
+                                "medium content suffix" => $medium_content . $shortcode,
+                                "long content suffix" => $long_content . $shortcode,
+                                "short content prefix" => $shortcode . $short_content,
+                                "medium content prefix" => $shortcode . $medium_content,
+                                "long content prefix" => $shortcode . $long_content,
+                                "short content prefix and suffix" => $shortcode . $short_content . $shortcode,
+                                "medium content prefix and suffix" => $shortcode . $medium_content . $shortcode,
+                                "long content prefix and suffix" => $shortcode . $long_content . $shortcode,
+                                "short content inner" => mb_substr( $short_content, 0, round( mb_strlen( $short_content ) ) ) . $shortcode . mb_substr( $short_content, round( mb_strlen( $short_content ) / 2 ) ),
+                                "medium content inner" => mb_substr( $medium_content, 0, round( mb_strlen( $medium_content ) ) ) . $shortcode . mb_substr( $medium_content, round( mb_strlen( $medium_content ) / 2 ) ),
+                                "long content inner" => mb_substr( $long_content, 0, round( mb_strlen( $long_content ) ) ) . $shortcode . mb_substr( $long_content, round( mb_strlen( $long_content ) / 2 ) ),
+                        );
+                        
+                        foreach ( $variations as $variation => $content ) {
+                                $variation_key = basename( $example_shortcode ) . ":" . $variation;
+                                error_log( "Running variation " . $variation_key );
+                                
+                                $start = microtime( true );
+                                
+                                do_shortcode( $content );
+                                
+                                $end = microtime( true );
+                                
+                                $total_time += ( $end - $start );
+                                
+                                error_log( $variation_key . ": " . ( $end - $start ) . " seconds\n" );
+                        }
+                }
+                
+                error_log( "\nTotal time: " . $total_time . " seconds\n" );
+        }
+}
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/bracket_but_no_shortcode.txt

@@ +1 @@
+Square brackets can be used like this: [square brackets]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/closing_tag_and_content.txt

@@ +1 @@
+[footag foo="this is the foo attribute"]this is the content[/footag]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/closing_tag_and_no_content.txt

@@ +1 @@
+[footag foo="this is the foo attribute"][/footag]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/flat_ten_across.txt

@@ +1 @@
+[footag]footag content[/footag]
+[baztag]baztag content[/footag]
+[bartag]bartag content[/footag]
+[test-shortcode-tag]test-shortcode-tag content[/footag]
+[dumptag att1="att1"]dumptag content[/footag]
+[url]url content[/footag]
+[hyphen]hyphen content[/footag]
+[img att1="att1"]img content[/footag]
+[hyphen-foo]hyphen-foo content[/footag]
+[hyphen-foo-bar]hyphen-foo-bar content[/footag]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/long_content.txt

@@ +1 @@
+[footag]
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+I am the very model of a modern major general.
+[/footag]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/many_attributes.txt

@@ +1 @@
+[footag attr1="This is attr1." attr2="This is attr2" attr3="This is attr3" attr4="This is attr4" attr5="This is attr5" attr6="This is attr6" attr7="This is attr7" attr8="This is attr8" attr9="This is attr9" attr10="This is attr10"]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/nested_ten_deep.txt

@@ +1 @@
+[footag]
+        [baztag]
+                [bartag]
+                        [test-shortcode-tag]
+                                [dumptag att1="att1"]
+                                        [url]
+                                                [hyphen]
+                                                        [img att1="att1"]
+                                                                [hyphen-foo]
+                                                                        [hyphen-foo-bar]
+                                                                        [/hyphen-foo-bar]
+                                                                [/hyphen-foo]
+                                                        [/img]
+                                                [/hyphen]
+                                        [/url]
+                                [/dumptag]
+                        [/test-shortcode-tag]
+                [/bartag]
+        [/baztag]
+[/footag]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/nested_two_deep.txt

@@ +1 @@
+[footag foo="This is the foo attribute"]
+        [baztag foo="This is the baztag foo attribute." bar="This is the baztag bar attribute"]
+                This is the baztag content.
+        [/baztag]
+[/footag]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/no_closing_tag.txt

@@ +1 @@
+[footag foo="this is the foo attribute"]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/not_actually_a_shortcode.txt

@@ +1 @@
+[footag This is not a shortcode because there is no closing bracket.
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/positional_attribute.txt

@@ +1 @@
+[test-shortcode-tag http://www.youtube.com/watch?v=eBGIQ7ZuuiU]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/positional_attributes.txt

@@ +1 @@
+[test-shortcode-tag 123 https://wordpress.org/ 0 "foo" bar]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/self_closing_tag.txt

@@ +1 @@
+[footag foo="this is the foo attribute" /]
+ No newline at end of file

tests/phpunit/tests/shortcode_examples/self_closing_tag_no_space.txt

@@ +1 @@
+[footag foo="this is the foo attribute"/]
+ No newline at end of file