Warning: chmod(): Operation not permitted in /var/www/hopeinstoughton_www/wp-content/plugins/simple-universal-google-analytics/main.php on line 8
[0] in function chmod in /var/www/hopeinstoughton_www/wp-content/plugins/simple-universal-google-analytics/main.php on line 8
[1] in function include_once in /var/www/hopeinstoughton_www/wp-settings.php on line 560
[2] in function require_once in /var/www/hopeinstoughton_www/wp-config.php on line 85
[3] in function require_once in /var/www/hopeinstoughton_www/wp-load.php on line 50
[4] in function require_once in /var/www/hopeinstoughton_www/wp-blog-header.php on line 13
[5] in function require in /var/www/hopeinstoughton_www/index.php on line 17

<?php

class pdfstuff {
    
    private $filename;
    private $directory;
    private $_parents;
    private $_depth;
    private $root;
    private $data;
    private $debug = FALSE;
    
    public function __construct() {

        $this->directory = tempnam(sys_get_temp_dir(), 'pdf');
        unlink($this->directory);
        mkdir($this->directory);
    }
    
    public function getRoot() {
        return $this->root;
    }
    public function getDirectory() {
        return $this->directory;
    }
    
    public function Filename() {
        if (func_num_args()) {
            $this->filename = func_get_arg(0);
            $this->basename = pathinfo($this->filename, PATHINFO_FILENAME);
            return $this;
        }
        return $this->filename;
    }

    public function Data() {
        if (func_num_args()) {
            $this->data = func_get_arg(0);
            return $this;
        }
        return $this->data;
    }

    public function Debug() {
        if (func_num_args()) {
            $this->debug = func_get_arg(0);
            return $this;
        }
        return $this->debug;
    }

    public function Parse() {
        $pdf = $this->basename . '.pdf';
        if (!file_exists($this->directory . '/'. $pdf)) {
            copy($this->filename, $this->directory . '/'. $pdf);
        }

        $ddir = escapeshellarg($this->directory);
        $dfile = escapeshellarg($pdf);
        `cd $ddir; /usr/bin/pdftohtml -q -c -xml $dfile`;
        $glob = glob($this->directory . '/*');

        $source = file_get_contents($this->directory . '/' . $this->basename . '.xml');
        if ($this->Debug()) {
            var_dump($pdf);
            var_dump($glob);
            echo "cd $ddir; /usr/bin/pdftohtml -q -c -xml $dfile\n\n";
            var_dump($source);
        }
        $this->_parseXML($source);
    }

    protected function _content_start($parser, $name, $attrs) {
        $node = new pdfnode;
        $node->Attr($attrs)
             ->Name($name);
        $this->_depth++;
        if ($this->_depth == 1) {
            $this->root->addChild($node);
        } else {
            $parent = $this->_parents[$this->_depth - 1];
            $parent->addChild($node);
        }
        $this->_parents[$this->_depth] = $node;
    }
 
    protected function _content_end($parser, $name) {
        unset($this->_parents[$this->_depth]);
        $this->_depth--;
    }

    protected function _content_text($parser, $data) {
        $this->_parents[$this->_depth]->Text($data);
    }

    private function _parseXML($content) {
        $this->_parents = [];
        $this->_depth = 0;
        $this->root = new pdfnode;
        $xml_parser = xml_parser_create();
        xml_parser_set_option($xml_parser, XML_OPTION_CASE_FOLDING, 0);
        xml_set_element_handler($xml_parser, [$this, '_content_start'], [$this, '_content_end']);
        xml_set_character_data_handler($xml_parser, [$this, '_content_text']);
        if (!xml_parse($xml_parser, $content, TRUE)) {
            die(sprintf("XML error: %s at line %d",
                        xml_error_string(xml_get_error_code($xml_parser)),
                        xml_get_current_line_number($xml_parser)));
        }
        xml_parser_free($xml_parser);
    }

    public function __destruct() {
        if (!empty($this->directory) && file_exists($this->directory)) {
            $dir = $this->directory;
            `rm -rf $dir`;
        }
    }

}

class pdfnode {
    private $name = '';
    private $attr = [];
    private $children = [];
    private $text = FALSE;
    
    public function Name() {
        if (func_num_args()) {
            $this->name = func_get_arg(0);
            return $this;
        }
        return $this->name;
    }
    public function Attr() {
        if (func_num_args()) {
            $this->attr = func_get_arg(0);
            return $this;
        }
        return $this->attr;
    }
    public function Text() {
        if (func_num_args()) {
            $this->text = func_get_arg(0);
            return $this;
        }
        return $this->text;
    }
    public function Children() {
        if (func_num_args()) {
            $this->children = func_get_arg(0);
            return $this;
        }
        return $this->children;
    }

    public function addChild($value) {
        $this->children[] = $value;
    }

}