[project/print.git] / print_pdf / print_pdf.pages.inc

<?php
// $Id$

/**
 * @file
 * Generates the PDF versions of the pages
 *
 * This file is included by the print_pdf module and includes the
 * functions that interface with the PDF generation packages.
 *
 * @ingroup print
 */

require_once(DRUPAL_ROOT . '/' . drupal_get_path('module', 'print') . '/print.pages.inc');

/**
 * Generate a PDF version of the printer-friendly page
 *
 * @see print_controller()
 * @see _print_get_template()
 * @see _print_pdf_dompdf()
 * @see _print_pdf_tcpdf()
 */
function print_pdf_controller() {
  global $base_url, $language;

  // Disable caching for generated PDFs, as Drupal doesn't ouput the proper headers from the cache
  $GLOBALS['conf']['cache'] = FALSE;

  $args = func_get_args();
  $path = implode('/', $args);
  $cid = isset($_GET['comment']) ? (int)$_GET['comment'] : NULL;

  $print_pdf_pdf_tool = variable_get('print_pdf_pdf_tool', PRINT_PDF_PDF_TOOL_DEFAULT);

  $print = print_controller($path, $cid, PRINT_PDF_FORMAT);
  if ($print === FALSE) {
    return;
  }

  // Img elements must be set to absolute
  $pattern = '!<(img\s[^>]*?)>!is';
  $print['content'] = preg_replace_callback($pattern, '_print_rewrite_urls', $print['content']);
  $print['logo'] = preg_replace_callback($pattern, '_print_rewrite_urls', $print['logo']);
  $print['footer_message'] = preg_replace_callback($pattern, '_print_rewrite_urls', $print['footer_message']);
  // And converted from private to public paths
  $file_downloads = variable_get('file_downloads', FILE_DOWNLOADS_PUBLIC);
  if ($file_downloads == FILE_DOWNLOADS_PRIVATE) {
    switch (variable_get('language_negotiation', LANGUAGE_NEGOTIATION_NONE)) {
      case LANGUAGE_NEGOTIATION_PATH_DEFAULT:
      case LANGUAGE_NEGOTIATION_PATH:
        $lang = $language->language;
        break;
      default:
        $lang = '';
        break;
    }
    $pattern = "!(<img\s[^>]*?src\s*?=\s*?['\"]?)${base_url}/(?:(?:index.php)?\?q=)?(?:${lang}/)?system/files(/[^>]*?>)!is";
    $replacement = '$1file://' . realpath(file_directory_path()) . '$2';
    $print['content'] = preg_replace($pattern, $replacement, $print['content']);
    $print['logo'] = preg_replace($pattern, $replacement, $print['logo']);
    $print['footer_message'] = preg_replace($pattern, $replacement, $print['footer_message']);
  }

  $node = $print['node'];
  ob_start();
  include_once(DRUPAL_ROOT . '/' . _print_get_template(PRINT_PDF_FORMAT, $print['type']));
  $html = ob_get_contents();
  ob_end_clean();

  // Convert the a href elements, to make sure no relative links remain
  $pattern = '!<(a\s[^>]*?)>!is';
  $html = preg_replace_callback($pattern, '_print_rewrite_urls', $html);
  // And make anchor links relative again, to permit in-PDF navigation
  $html = preg_replace("!${base_url}/" . PRINTPDF_PATH . "/.*?%2523!", '#', $html);

  $pdf_filename = variable_get('print_pdf_filename', PRINT_PDF_FILENAME_DEFAULT);
  if (function_exists('token_replace') && !empty($pdf_filename)) {
    $pdf_filename = token_replace($pdf_filename, 'node', $node) . '.pdf';
  }
  else {
    $pdf_filename = str_replace('/', '_', $path) . '.pdf';
  }
  if (basename($print_pdf_pdf_tool) == 'dompdf_config.inc.php') {
    _print_pdf_dompdf($print, $html, $pdf_filename);
  }
  elseif (basename($print_pdf_pdf_tool) == 'tcpdf.php') {
    _print_pdf_tcpdf($print, $html, $pdf_filename);
  }
  elseif (basename($print_pdf_pdf_tool) == 'wkhtmltopdf') {
    _print_pdf_wkhtmltopdf($print, $html, $pdf_filename);
  }
  else {
    return drupal_not_found();
  }

  $nodepath = (isset($node->path)) ? drupal_get_normal_path($node->path) : 'node/' . $path;
  db_merge('print_pdf_page_counter')
    ->key(array('path' => $nodepath))
    ->fields(array(
        'totalcount' => 1,
        'timestamp' => REQUEST_TIME,
    ))
    ->expression('totalcount', 'totalcount + :inc', array(':inc' => 1))
    ->execute();
}

/**
 * Generate the PDF file using the dompdf library
 *
 * @param $print
 *   array containing the configured data
 * @param $html
 *   contents of the post-processed template already with the node data
 * @param $filename
 *   name of the PDF file to be generated
 * @see print_pdf_controller()
 */
function _print_pdf_dompdf($print, $html, $filename) {
  $print_pdf_pdf_tool = variable_get('print_pdf_pdf_tool', PRINT_PDF_PDF_TOOL_DEFAULT);
  $print_pdf_paper_size = variable_get('print_pdf_paper_size', PRINT_PDF_PAPER_SIZE_DEFAULT);
  $print_pdf_page_orientation = variable_get('print_pdf_page_orientation', PRINT_PDF_PAGE_ORIENTATION_DEFAULT);
  $print_pdf_content_disposition = variable_get('print_pdf_content_disposition', PRINT_PDF_CONTENT_DISPOSITION_DEFAULT);
  require_once(DRUPAL_ROOT . '/' . $print_pdf_pdf_tool);

  // dompdf seems to have problems with something in system.css so let's not use it
  $html = preg_replace('!<link.*?modules/system/system.css.*?/>!', '', $html);

  $url_array  = parse_url($print['url']);

  $protocol = $url_array['scheme'] . '://';
  $host = $url_array['host'];
  $path = dirname($url_array['path']) . '/';

  $dompdf = new DOMPDF();
  $dompdf->set_base_path($path);
  $dompdf->set_host($host);
  $dompdf->set_paper(drupal_strtolower($print_pdf_paper_size), $print_pdf_page_orientation);
  $dompdf->set_protocol($protocol);

// dompdf can't handle footers cleanly, so disable the following
//  $html = theme('print_pdf_dompdf_footer', $html);

  // Convert the euro sign to an HTML entity
  $html = str_replace('€', '&#0128;', $html);

  // Convert from UTF-8 to ISO 8859-1 and then to HTML entities
  if (function_exists('utf8_decode')) {
    $html = utf8_decode($html);
  }
// iconv fails silently when it encounters something that it doesn't know, so don't use it
//  else if (function_exists('iconv')) {
//    $html = iconv('UTF-8', 'ISO-8859-1', $html);
//  }
  elseif (function_exists('mb_convert_encoding')) {
    $html = mb_convert_encoding($html, 'ISO-8859-1', 'UTF-8');
  }
  elseif (function_exists('recode_string')) {
    $html = recode_string('UTF-8..ISO_8859-1', $html);
  }
  $html = htmlspecialchars_decode(htmlentities($html, ENT_NOQUOTES, 'ISO-8859-1'), ENT_NOQUOTES);

  //must get rid of tbody (dompdf goes into recursion)
  $html = preg_replace('!<tbody[^>]*?>|</tbody>!i', '', $html);

  $dompdf->load_html($html);

  $dompdf->render();
  $dompdf->stream($filename, array('Attachment' => ($print_pdf_content_disposition == 2)));
}

/**
 * Generate the PDF file using the TCPDF library
 *
 * @param $print
 *   array containing the configured data
 * @param $html
 *   contents of the post-processed template already with the node data
 * @param $filename
 *   name of the PDF file to be generated
 * @see print_pdf_controller()
 */
function _print_pdf_tcpdf($print, $html, $filename) {
  global $base_url, $language;

  $print_pdf_pdf_tool = variable_get('print_pdf_pdf_tool', PRINT_PDF_PDF_TOOL_DEFAULT);
  $print_pdf_paper_size = variable_get('print_pdf_paper_size', PRINT_PDF_PAPER_SIZE_DEFAULT);
  $print_pdf_page_orientation = variable_get('print_pdf_page_orientation', PRINT_PDF_PAGE_ORIENTATION_DEFAULT);
  $print_pdf_content_disposition = variable_get('print_pdf_content_disposition', PRINT_PDF_CONTENT_DISPOSITION_DEFAULT);

  $pdf_tool_path = realpath(dirname($print_pdf_pdf_tool));

  if (variable_get('print_pdf_autoconfig', PRINT_PDF_AUTOCONFIG_DEFAULT)) {
    define('K_TCPDF_EXTERNAL_CONFIG', TRUE);
    define('K_PATH_MAIN', dirname($_SERVER['SCRIPT_FILENAME']));
    define('K_PATH_URL', $base_url);
    define('K_PATH_FONTS', $pdf_tool_path . '/fonts/');
    define('K_PATH_CACHE', $pdf_tool_path . '/cache/');
    define('K_PATH_IMAGES', '');
    define('K_BLANK_IMAGE', $pdf_tool_path . '/images/_blank.png');
    define('K_CELL_HEIGHT_RATIO', 1.25);
    define('K_SMALL_RATIO', 2/3);
  }

  // Decode HTML entities in image filenames
  $pattern = "!<img\s[^>]*?src\s*?=\s*?['\"]?${base_url}[^>]*?>!is";
  $html = preg_replace_callback($pattern, create_function('$matches', 'return html_entity_decode($matches[0], ENT_QUOTES);'), $html);
  // Remove queries from the image URL
  $pattern = "!(<img\s[^>]*?src\s*?=\s*?['\"]?${base_url}.*?)(?:%3F|\?)[\w=&]+([^>]*?>)!is";
  $html = preg_replace($pattern, '$1$2', $html);

  require_once(DRUPAL_ROOT . '/' . $print_pdf_pdf_tool);
  if (strpos(PDF_PRODUCER, 'PHP4') === FALSE) {
    require_once(DRUPAL_ROOT . '/' . drupal_get_path('module', 'print_pdf') . '/print_pdf.class.inc');
  }
  else {
    drupal_set_message(t("The PHP4 version of TCPDF is not supported. Please upgrade it."), 'error', FALSE);
    drupal_goto($_GET['q']);
  }

  $font = Array(
    check_plain(variable_get('print_pdf_font_family', PRINT_PDF_FONT_FAMILY_DEFAULT)),
    '',
    check_plain(variable_get('print_pdf_font_size', PRINT_PDF_FONT_SIZE_DEFAULT)),
  );
  $orientation = drupal_strtoupper($print_pdf_page_orientation[0]);

  // create new PDF document
  $pdf = new PrintTCPDF($orientation , 'mm', $print_pdf_paper_size, TRUE);

  // set document information
  $pdf->SetAuthor(strip_tags($print['submitted']));
  $pdf->SetCreator(variable_get('site_name', 'Drupal'));
  $pdf->SetTitle(html_entity_decode($print['title'], ENT_QUOTES, 'UTF-8'));
  $keys = implode(' ', explode("\n", trim(strip_tags($print['taxonomy']))));
  $pdf->SetKeywords($keys);
  $pdf->setPDFVersion('1.6');

  if (defined('LANGUAGE_RTL') && $language->direction == LANGUAGE_RTL) {
    $pdf->setRTL(TRUE);
  }

  $pdf = theme('print_pdf_tcpdf_header', $pdf, $html, $font);
  $pdf = theme('print_pdf_tcpdf_footer', $pdf, $html, $font);
  $pdf = theme('print_pdf_tcpdf_page', $pdf);

  //initialize document
  $pdf->AliasNbPages();

  // add a page
  $pdf->AddPage();

  $pdf = theme('print_pdf_tcpdf_content', $pdf, $html, $font);

  // reset pointer to the last page
  $pdf->lastPage();

  // try to recover from any warning/error
  ob_clean();

  //Close and output PDF document
  $output_dest = ($print_pdf_content_disposition == 2) ? 'D' : 'I';
  $pdf->Output($filename, $output_dest);
}

/**
 * Generate the PDF file using wkhtmltopdf
 *
 * @param $print
 *   array containing the configured data
 * @param $html
 *   contents of the post-processed template already with the node data
 * @param $filename
 *   name of the PDF file to be generated
 * @see print_pdf_controller()
 */
function _print_pdf_wkhtmltopdf($print, $html, $filename) {
  $print_pdf_pdf_tool = variable_get('print_pdf_pdf_tool', PRINT_PDF_PDF_TOOL_DEFAULT);
  $print_pdf_paper_size = drupal_strtolower(variable_get('print_pdf_paper_size', PRINT_PDF_PAPER_SIZE_DEFAULT));
  $print_pdf_page_orientation = variable_get('print_pdf_page_orientation', PRINT_PDF_PAGE_ORIENTATION_DEFAULT);
  $print_pdf_content_disposition = variable_get('print_pdf_content_disposition', PRINT_PDF_CONTENT_DISPOSITION_DEFAULT);
  $print_pdf_wkhtmltopdf_options = variable_get('print_pdf_wkhtmltopdf_options', PRINT_PDF_WKHTMLTOPDF_OPTIONS);
  $print_pdf_xvfb_options = variable_get('print_pdf_xvfb_options', PRINT_PDF_XVFB_OPTIONS);

  $xvfb_binary = key(file_scan_directory(drupal_get_path('module', 'print'), '!^Xvfb$!'));
  $dpi = 96;

  // If available, launch a temporary X server
  if (isset($xvfb_binary)) {
    $xdisplay = variable_get('print_pdf_wkhtmltopdf_xdisplay', 10);
    variable_set('print_pdf_wkhtmltopdf_xdisplay', ($xdisplay == 4990) ? 10 : $xdisplay + 10);
    $xdisplay += mt_rand(0, 9);

    $xcmd = realpath($xvfb_binary) . " :$xdisplay -screen 0 320x200x24 -dpi $dpi -terminate -nolisten tcp -tst $print_pdf_xvfb_options";

    $xdescriptor = array(0 => array('pipe', 'r'), 1 => array('pipe', 'w'), 2 => array('pipe', 'w'));
    $xprocess = proc_open($xcmd, $xdescriptor, $xpipes, NULL, NULL);
  }

  if (function_exists('token_replace') && !empty($print_pdf_wkhtmltopdf_options)) {
    $print_pdf_wkhtmltopdf_options = token_replace($print_pdf_wkhtmltopdf_options, 'node', $print['node']);
  }

  $descriptor = array(0 => array('pipe', 'r'), 1 => array('pipe', 'w'), 2 => array('pipe', 'w'));
  $cmd = realpath($print_pdf_pdf_tool) . " --page-size $print_pdf_paper_size --orientation $print_pdf_page_orientation --dpi $dpi $print_pdf_wkhtmltopdf_options - -";

  $process = proc_open($cmd, $descriptor, $pipes, NULL, isset($xvfb_binary) ? array('DISPLAY' => ':' . $xdisplay) : NULL);

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

    $pdf = stream_get_contents($pipes[1]);
    fclose($pipes[1]);

    stream_set_blocking($pipes[2], 0);
    $error = stream_get_contents($pipes[2]);
    if (!empty($error)) {
      watchdog('print_pdf', 'wkhtmltopdf: ' . $error);
    }
    fclose($pipes[2]);

    $retval = proc_terminate($process);
  }
  if (isset($xvfb_binary)) {
    fclose($xpipes[0]);
    fclose($xpipes[1]);
    stream_set_blocking($xpipes[2], 0);
    $xerror = stream_get_contents($xpipes[2]);
    if (!empty($xerror)) {
      watchdog('print_pdf', 'wkhtmltopdf Xvfb: ' . $xerror);
    }
    fclose($xpipes[2]);

    proc_terminate($xprocess);
  }

  if (!empty($pdf)) {
    if (headers_sent()) {
      die("Unable to stream pdf: headers already sent");
    }
    header("Cache-Control: private");
    header("Content-Type: application/pdf");

    $attachment =  ($print_pdf_content_disposition == 2) ?  "attachment" :  "inline";

    header("Content-Disposition: $attachment; filename=\"$filename\"");

    echo $pdf;
    flush();
  }
  else {
    drupal_goto($print['url']);
    exit;
  }
}

/**
 * Format the dompdf footer contents
 *
 * @param $html
 *   contents of the body of the HTML from the original node
 * @see theme_print_pdf_tcpdf_footer()
 */
function theme_print_pdf_dompdf_footer(&$html) {
  preg_match('!<div class="print-footer">(.*?)</div>!si', $html, $tpl_footer);
  $html = str_replace($tpl_footer[0], '', $html);

  $text = '<script type="text/php">
    if (isset($pdf)) {
      $font = Font_Metrics::get_font("verdana");;
      $size = 10;
      $color = array(0,0,0);
      $text_height = Font_Metrics::get_font_height($font, $size);

      $w = $pdf->get_width();
      $h = $pdf->get_height();

      $footer = $pdf->open_object();

      // Draw a line along the bottom
      $y = $h - 25;
      $pdf->line(15, $y, $w - 15, $y, $color, 1);

      $y += $text_height / 2;
      $pdf->page_text(15, $y, \'' . addslashes(strip_tags($tpl_footer[1])) . '\', $font, $size, $color);

      $pdf->close_object();
      $pdf->add_object($footer, "all");

      // Center the text
      $width = Font_Metrics::get_text_width("Page 1 of 2", $font, $size);
      $pagenumtxt = t("Page !n of !total", array("!n" => "{PAGE_NUM}", "!total" => "{PAGE_COUNT}"));
      $pdf->page_text($w - 15 - $width, $y, $pagenumtxt, $font, $size, $color);
    }
  </script>';

  return str_replace("<body>", "<body>" . $text, $html);
}

/**
 * Format the TCPDF header
 *
 * @param $pdf
 *   current TCPDF object
 * @param $html
 *   contents of the body of the HTML from the original node
 * @param $font
 *   array with the font definition (font name, styles and size)
 * @see theme_print_pdf_tcpdf_header()
 */
function theme_print_pdf_tcpdf_header(&$pdf, &$html, $font) {
  preg_match('!<div class="print-logo">(.*?)</div>!si', $html, $tpl_logo);
  preg_match('!<h1 class="print-title">(.*?)</h1>!si', $html, $tpl_title);
  preg_match('!<div class="print-site_name">(.*?)</div>!si', $html, $tpl_site_name);

  $ratio = 0;
  $logo = '';
  $logo_ret = preg_match('!src\s*=\s*(\'.*?\'|".*?"|[^\s]*)!i', $tpl_logo[1], $matches);
  if ($logo_ret) {
    $logo = trim($matches[1], '\'"');
    $size = getimagesize($logo);
    $ratio = $size ? ($size[0] / $size[1]) : 0;
  }

  // set header font
  $pdf->setHeaderFont($font);
  // set header margin
  $pdf->SetHeaderMargin(5);
  // set header data
  $pdf->SetHeaderData($logo, 10 * $ratio, html_entity_decode($tpl_title[1], ENT_QUOTES, 'UTF-8'), strip_tags($tpl_site_name[1]));

  return $pdf;
}

/**
 * Format the TCPDF page settings (margins, etc)
 *
 * @param $pdf
 *   current TCPDF object
 * @see theme_print_pdf_tcpdf_page()
 */
function theme_print_pdf_tcpdf_page(&$pdf) {
  // set margins
  $pdf->SetMargins(15, 20, 15);
  // set auto page breaks
  $pdf->SetAutoPageBreak(TRUE, 15);
  // set image scale factor
  sscanf(PDF_PRODUCER, "TCPDF %d.%d.%d", $major, $minor, $build);
  $imagescale = (($major >= 4) && ($minor >= 6) && ($build >= 2)) ? 1 : 4;
  $pdf->setImageScale($imagescale);
  // set image compression quality
  $pdf->setJPEGQuality(100);

  return $pdf;
}

/**
 * Format the TCPDF page content
 *
 * @param $pdf
 *   current TCPDF object
 * @param $html
 *   contents of the body of the HTML from the original node
 * @param $font
 *   array with the font definition (font name, styles and size)
 * @see theme_print_pdf_tcpdf_content()
 */
function theme_print_pdf_tcpdf_content(&$pdf, &$html, $font) {
  // set content font
  $pdf->setFont($font[0], $font[1], $font[2]);

  preg_match('!<body.*?>(.*)</body>!sim', $html, $matches);
  $pattern = '!(?:<div class="print-(?:logo|site_name|breadcrumb|footer)">.*?</div>|<hr class="print-hr" />)!si';
  $matches[1] = preg_replace($pattern, '', $matches[1]);

  // Make CCK fields look better
  $matches[1] = preg_replace('!(<div class="field.*?>)\s*!sm', '$1', $matches[1]);
  $matches[1] = preg_replace('!(<div class="field.*?>.*?</div>)\s*!sm', '$1', $matches[1]);
  $matches[1] = preg_replace('!<div( class="field-label.*?>.*?)</div>!sm', '<strong$1</strong>', $matches[1]);

  // Since TCPDF's writeHTML is so bad with <p>, do everything possible to make it look nice
  $matches[1] = preg_replace('!<(?:p(|\s+.*?)/?|/p)>!i', '<br$1 />', $matches[1]);
  $matches[1] = str_replace(array('<div', 'div>'), array('<span', 'span><br />'), $matches[1]);
  do {
    $prev = $matches[1];
    $matches[1] = preg_replace('!(</span>)<br />(\s*?</span><br />)!s', '$1$2', $matches[1]);
  } while ($prev != $matches[1]);

  @$pdf->writeHTML($matches[1]);

  return $pdf;
}

/**
 * Format the TCPDF footer contents
 *
 * @param $pdf
 *   current TCPDF object
 * @param $html
 *   contents of the body of the HTML from the original node
 * @param $font
 *   array with the font definition (font name, styles and size)
 * @see theme_print_pdf_tcpdf_footer()
 */
function theme_print_pdf_tcpdf_footer(&$pdf, &$html, $font) {
  preg_match('!<div class="print-footer">(.*?)</div>!si', $html, $tpl_footer);
  $footer = trim(preg_replace('!</?div[^>]*?>!i', '', $tpl_footer[1]));

  // set footer font
  $font[2] *= 0.8;
  $pdf->setFooterFont($font);
  // set footer margin
  $pdf->SetFooterMargin(10);
  // set footer data
  $pdf->SetFooterData($footer);

  return $pdf;
}

/**
 * Format the TCPDF footer layout
 *
 * @param $pdf
 *   current TCPDF object
 * @see theme_print_pdf_tcpdf_footer2()
 */
function theme_print_pdf_tcpdf_footer2(&$pdf) {
  //Position at 1.5 cm from bottom
  $pdf->writeHTMLCell(0, 15, 15, -10, $pdf->footer, 0, 0, 0, TRUE, '');

  $ormargins = $pdf->getOriginalMargins();
  $pagenumtxt = t('Page !n of !total', array('!n' => $pdf->PageNo(), '!total' => $pdf->getAliasNbPages()));
  //Print page number
  if ($pdf->getRTL()) {
    $pdf->SetX($ormargins['right']);
    $pdf->Cell(0, 10, $pagenumtxt, 'T', 0, 'L');
  }
  else {
    $pdf->SetX($ormargins['left']);
    $pdf->Cell(0, 10, $pagenumtxt, 'T', 0, 'R');
  }

  return $pdf;
}