C:/xampp/htdocs/moodle/lib/simpletestlib/parser.php

Go to the documentation of this file.

<?php
foreach (array('LEXER_ENTER', 'LEXER_MATCHED',
                'LEXER_UNMATCHED', 'LEXER_EXIT',
                'LEXER_SPECIAL') as $i => $constant) {
    if (! defined($constant)) {
        define($constant, $i + 1);
    }
}
class ParallelRegex {
    var $_patterns;
    var $_labels;
    var $_regex;
    var $_case;
    
    function ParallelRegex($case) {
        $this->_case = $case;
        $this->_patterns = array();
        $this->_labels = array();
        $this->_regex = null;
    }
    
    function addPattern($pattern, $label = true) {
        $count = count($this->_patterns);
        $this->_patterns[$count] = $pattern;
        $this->_labels[$count] = $label;
        $this->_regex = null;
    }
    
    function match($subject, &$match) {
        if (count($this->_patterns) == 0) {
            return false;
        }
        if (! preg_match($this->_getCompoundedRegex(), $subject, $matches)) {
            $match = '';
            return false;
        }
        $match = $matches[0];
        for ($i = 1; $i < count($matches); $i++) {
            if ($matches[$i]) {
                return $this->_labels[$i - 1];
            }
        }
        return true;
    }
    
    function _getCompoundedRegex() {
        if ($this->_regex == null) {
            for ($i = 0, $count = count($this->_patterns); $i < $count; $i++) {
                $this->_patterns[$i] = '(' . str_replace(
                        array('/', '(', ')'),
                        array('\/', '\(', '\)'),
                        $this->_patterns[$i]) . ')';
            }
            $this->_regex = "/" . implode("|", $this->_patterns) . "/" . $this->_getPerlMatchingFlags();
        }
        return $this->_regex;
    }
    
    function _getPerlMatchingFlags() {
        return ($this->_case ? "msS" : "msSi");
    }
}

class SimpleStateStack {
    var $_stack;
    
    function SimpleStateStack($start) {
        $this->_stack = array($start);
    }
    
    function getCurrent() {
        return $this->_stack[count($this->_stack) - 1];
    }
    
    function enter($state) {
        array_push($this->_stack, $state);
    }
    
    function leave() {
        if (count($this->_stack) == 1) {
            return false;
        }
        array_pop($this->_stack);
        return true;
    }
}

class SimpleLexer {
    var $_regexes;
    var $_parser;
    var $_mode;
    var $_mode_handlers;
    var $_case;
    
    function SimpleLexer(&$parser, $start = "accept", $case = false) {
        $this->_case = $case;
        $this->_regexes = array();
        $this->_parser = &$parser;
        $this->_mode = new SimpleStateStack($start);
        $this->_mode_handlers = array($start => $start);
    }
    
    function addPattern($pattern, $mode = "accept") {
        if (! isset($this->_regexes[$mode])) {
            $this->_regexes[$mode] = new ParallelRegex($this->_case);
        }
        $this->_regexes[$mode]->addPattern($pattern);
        if (! isset($this->_mode_handlers[$mode])) {
            $this->_mode_handlers[$mode] = $mode;
        }
    }
    
    function addEntryPattern($pattern, $mode, $new_mode) {
        if (! isset($this->_regexes[$mode])) {
            $this->_regexes[$mode] = new ParallelRegex($this->_case);
        }
        $this->_regexes[$mode]->addPattern($pattern, $new_mode);
        if (! isset($this->_mode_handlers[$new_mode])) {
            $this->_mode_handlers[$new_mode] = $new_mode;
        }
    }
    
    function addExitPattern($pattern, $mode) {
        if (! isset($this->_regexes[$mode])) {
            $this->_regexes[$mode] = new ParallelRegex($this->_case);
        }
        $this->_regexes[$mode]->addPattern($pattern, "__exit");
        if (! isset($this->_mode_handlers[$mode])) {
            $this->_mode_handlers[$mode] = $mode;
        }
    }
    
    function addSpecialPattern($pattern, $mode, $special) {
        if (! isset($this->_regexes[$mode])) {
            $this->_regexes[$mode] = new ParallelRegex($this->_case);
        }
        $this->_regexes[$mode]->addPattern($pattern, "_$special");
        if (! isset($this->_mode_handlers[$special])) {
            $this->_mode_handlers[$special] = $special;
        }
    }
    
    function mapHandler($mode, $handler) {
        $this->_mode_handlers[$mode] = $handler;
    }
    
    function parse($raw) {
        if (! isset($this->_parser)) {
            return false;
        }
        $length = strlen($raw);
        while (is_array($parsed = $this->_reduce($raw))) {
            list($raw, $unmatched, $matched, $mode) = $parsed;
            if (! $this->_dispatchTokens($unmatched, $matched, $mode)) {
                return false;
            }
            if ($raw === '') {
                return true;
            }
            if (strlen($raw) == $length) {
                return false;
            }
            $length = strlen($raw);
        }
        if (! $parsed) {
            return false;
        }
        return $this->_invokeParser($raw, LEXER_UNMATCHED);
    }
    
    function _dispatchTokens($unmatched, $matched, $mode = false) {
        if (! $this->_invokeParser($unmatched, LEXER_UNMATCHED)) {
            return false;
        }
        if (is_bool($mode)) {
            return $this->_invokeParser($matched, LEXER_MATCHED);
        }
        if ($this->_isModeEnd($mode)) {
            if (! $this->_invokeParser($matched, LEXER_EXIT)) {
                return false;
            }
            return $this->_mode->leave();
        }
        if ($this->_isSpecialMode($mode)) {
            $this->_mode->enter($this->_decodeSpecial($mode));
            if (! $this->_invokeParser($matched, LEXER_SPECIAL)) {
                return false;
            }
            return $this->_mode->leave();
        }
        $this->_mode->enter($mode);
        return $this->_invokeParser($matched, LEXER_ENTER);
    }
    
    function _isModeEnd($mode) {
        return ($mode === "__exit");
    }
    
    function _isSpecialMode($mode) {
        return (strncmp($mode, "_", 1) == 0);
    }
    
    function _decodeSpecial($mode) {
        return substr($mode, 1);
    }
    
    function _invokeParser($content, $is_match) {
        if (($content === '') || ($content === false)) {
            return true;
        }
        $handler = $this->_mode_handlers[$this->_mode->getCurrent()];
        return $this->_parser->$handler($content, $is_match);
    }
    
    function _reduce($raw) {
        if ($action = $this->_regexes[$this->_mode->getCurrent()]->match($raw, $match)) {
            $unparsed_character_count = strpos($raw, $match);
            $unparsed = substr($raw, 0, $unparsed_character_count);
            $raw = substr($raw, $unparsed_character_count + strlen($match));
            return array($raw, $unparsed, $match, $action);
        }
        return true;
    }
}

class SimpleHtmlLexer extends SimpleLexer {
    
    function SimpleHtmlLexer(&$parser) {
        $this->SimpleLexer($parser, 'text');
        $this->mapHandler('text', 'acceptTextToken');
        $this->_addSkipping();
        foreach ($this->_getParsedTags() as $tag) {
            $this->_addTag($tag);
        }
        $this->_addInTagTokens();
    }
    
    function _getParsedTags() {
        return array('a', 'base', 'title', 'form', 'input', 'button', 'textarea', 'select',
                'option', 'frameset', 'frame', 'label');
    }
    
    function _addSkipping() {
        $this->mapHandler('css', 'ignore');
        $this->addEntryPattern('<style', 'text', 'css');
        $this->addExitPattern('</style>', 'css');
        $this->mapHandler('js', 'ignore');
        $this->addEntryPattern('<script', 'text', 'js');
        $this->addExitPattern('</script>', 'js');
        $this->mapHandler('comment', 'ignore');
        $this->addEntryPattern('<!--', 'text', 'comment');
        $this->addExitPattern('-->', 'comment');
    }
    
    function _addTag($tag) {
        $this->addSpecialPattern("</$tag>", 'text', 'acceptEndToken');
        $this->addEntryPattern("<$tag", 'text', 'tag');
    }
    
    function _addInTagTokens() {
        $this->mapHandler('tag', 'acceptStartToken');
        $this->addSpecialPattern('\s+', 'tag', 'ignore');
        $this->_addAttributeTokens();
        $this->addExitPattern('/>', 'tag');
        $this->addExitPattern('>', 'tag');
    }
    
    function _addAttributeTokens() {
        $this->mapHandler('dq_attribute', 'acceptAttributeToken');
        $this->addEntryPattern('=\s*"', 'tag', 'dq_attribute');
        $this->addPattern("\\\\\"", 'dq_attribute');
        $this->addExitPattern('"', 'dq_attribute');
        $this->mapHandler('sq_attribute', 'acceptAttributeToken');
        $this->addEntryPattern("=\s*'", 'tag', 'sq_attribute');
        $this->addPattern("\\\\'", 'sq_attribute');
        $this->addExitPattern("'", 'sq_attribute');
        $this->mapHandler('uq_attribute', 'acceptAttributeToken');
        $this->addSpecialPattern('=\s*[^>\s]*', 'tag', 'uq_attribute');
    }
}

class SimpleHtmlSaxParser {
    var $_lexer;
    var $_listener;
    var $_tag;
    var $_attributes;
    var $_current_attribute;
    
    function SimpleHtmlSaxParser(&$listener) {
        $this->_listener = &$listener;
        $this->_lexer = &$this->createLexer($this);
        $this->_tag = '';
        $this->_attributes = array();
        $this->_current_attribute = '';
    }
    
    function parse($raw) {
        return $this->_lexer->parse($raw);
    }
    
    function &createLexer(&$parser) {
        $lexer = new SimpleHtmlLexer($parser);
        return $lexer;
    }
    
    function acceptStartToken($token, $event) {
        if ($event == LEXER_ENTER) {
            $this->_tag = strtolower(substr($token, 1));
            return true;
        }
        if ($event == LEXER_EXIT) {
            $success = $this->_listener->startElement(
                    $this->_tag,
                    $this->_attributes);
            $this->_tag = '';
            $this->_attributes = array();
            return $success;
        }
        if ($token != '=') {
            $this->_current_attribute = strtolower(SimpleHtmlSaxParser::decodeHtml($token));
            $this->_attributes[$this->_current_attribute] = '';
        }
        return true;
    }
    
    function acceptEndToken($token, $event) {
        if (! preg_match('/<\/(.*)>/', $token, $matches)) {
            return false;
        }
        return $this->_listener->endElement(strtolower($matches[1]));
    }
    
    function acceptAttributeToken($token, $event) {
        if ($this->_current_attribute) {
            if ($event == LEXER_UNMATCHED) {
                $this->_attributes[$this->_current_attribute] .=
                        SimpleHtmlSaxParser::decodeHtml($token);
            }
            if ($event == LEXER_SPECIAL) {
                $this->_attributes[$this->_current_attribute] .=
                        preg_replace('/^=\s*/' , '', SimpleHtmlSaxParser::decodeHtml($token));
            }
        }
        return true;
    }
    
    function acceptEntityToken($token, $event) {
    }
    
    function acceptTextToken($token, $event) {
        return $this->_listener->addContent($token);
    }
    
    function ignore($token, $event) {
        return true;
    }
    
    function decodeHtml($html) {
        return html_entity_decode($html, ENT_QUOTES);
    }
    
    function normalise($html) {
        $text = preg_replace('|<!--.*?-->|', '', $html);
        $text = preg_replace('|<script[^>]*>.*?</script>|', '', $text);
        $text = preg_replace('|<img[^>]*alt\s*=\s*"([^"]*)"[^>]*>|', ' \1 ', $text);
        $text = preg_replace('|<img[^>]*alt\s*=\s*\'([^\']*)\'[^>]*>|', ' \1 ', $text);
        $text = preg_replace('|<img[^>]*alt\s*=\s*([a-zA-Z_]+)[^>]*>|', ' \1 ', $text);
        $text = preg_replace('|<[^>]*>|', '', $text);
        $text = SimpleHtmlSaxParser::decodeHtml($text);
        $text = preg_replace('|\s+|', ' ', $text);
        return trim(trim($text), "\xA0");        // TODO: The \xAO is a &nbsp;. Add a test for this.
    }
}

class SimpleSaxListener {
    
    function SimpleSaxListener() {
    }
    
    function startElement($name, $attributes) {
    }
    
    function endElement($name) {
    }
    
    function addContent($text) {
    }
}
?>