pdf_file.inc

<?php
require_once SQ_CORE_PACKAGE_PATH.'/files/file/file.inc';
require_once SQ_DATA_PATH.'/private/conf/tools.inc';


class PDF_File extends File
{


    public $allowed_extensions = Array('pdf');


    function __construct($assetid=0)
    {
        parent::__construct($assetid);

    }//end constructor


    function getContent()
    {
        $file_info = $this->getExistingFile();
        if (empty($file_info)) return '';
        $file = $file_info['path'];

        $indexer_path = dirname(__FILE__).'/files';

        $content = '';

        if (SQ_TOOL_PDFTOHTML_ENABLED) {
            $pwd = $this->attr('password');
            $pwd = (empty($pwd)) ? '' : $pwd = ' -opw '.$pwd.' ';

            // Run a program that converts a PDF to HTML, set environment variable of bin_path
            $cmd = SQ_TOOL_PDFTOHTML_PATH." -i -nomerge -noframes -stdout $pwd $file";
            $cmd = escapeshellcmd($cmd);
            // Run the command
            $log = Array(
                        'pipe',
                        'w',
                   );

            $status = FALSE;
            $spec = Array(
                        0   => Array(
                                    'pipe',
                                    'r',
                               ),                   // stdin is a pipe that the child will read from
                        1   => $log,                // stdout is a pipe that the child will write to
                        2   => $log,                // stderr is a file to write to
                    );

            $cwd = NULL;
            $env = Array();
            $process = proc_open($cmd, $spec, $pipes, $cwd, $env);

            if (is_resource($process)) {
                fclose($pipes[0]);

                // Get the output
                $content = stream_get_contents($pipes[1]);
                fclose($pipes[1]);

                // Find the errors ... do NOT throw to output
                $errors = stream_get_contents($pipes[2]);
                if (!empty($errors)) {
                    $errors = array_unique(explode(PHP_EOL, $errors));
                    trigger_error('Error occured while getting contents for pdf File Asset #'.$this->id.' : '.trim(implode(', ', $errors), ', '), E_USER_WARNING);
                    return FALSE;
                }//end if
                fclose($pipes[2]);

                // Finally, get the status of this command
                $status = proc_close($process);

                // check to see if the command was executed correctly return value is 0,
                // all good or else log a warning depending upon the error code returned
                if ($status != '0') {
                    switch ($status) {
                        case 1:
                            $error_str = 'Invalid Argument.';
                        break;

                        case 2:
                            $error_str = 'Invalid Encoded Output.';
                        break;

                        case 3:
                            $error_str = 'Copying text from this Document is not allowed.';
                        break;

                        case 4:
                            $error_str = 'Error Opening File (e.g. corrupted file).';
                        break;

                        default :
                            $error_str = 'Unknown Error.';

                    }//end switch

                    trigger_error('Error occured while getting contents for pdf File Asset #'.$this->id.' : '.$error_str, E_USER_WARNING);
                    return '';
                }//end if
            }//end if
        }//end if
        $file = str_replace('"','\\\"',$file);
        if (empty($content)) return '';

        // strip out the tags
        $content = trim(strip_tags($content));
        if (empty($content)) return '';

        // in some cases file path ends up as the first string in the file
        // this is the way pdftohtml treats files without the title, it puts filename into html title
        if (($this->status & SQ_SC_STATUS_SAFE_EDITING) && ($GLOBALS['SQ_SYSTEM']->user instanceof Public_User)) {
            // strip file versioning suffix from the filename, e.g. 'file.pdf,ffv5' becomes 'file.pdf'
            $file = preg_replace('/,ffv\d+$/', '', $file);
        }
        // if file extension is .pdf, the .pdf part is omitted by pdftohtml from the first string (strange)
        // here we prepare the file name for that case
        if (substr($file, -4) == '.pdf') {
            $file = substr($file, 0, strlen($file)-4);
        }

        // if we suspect that the first string is the file path, remove it
        // NOTE: this will only work on *nix type filesystems. This is
        if (isset($content[0]) && $content[0] == '/') {
            $content = substr($content, strlen($file));
        }

        // strip out the unnecessary whitespace and html entities like &amp;
        $search = Array('/[\s,]+/', '/\&[^;\s]+;/');
        $replace = Array(' ', '');

        $content = strtolower(trim(preg_replace($search, $replace, $content)));

        return $content;

    }//end getContent()


}//end class
?>