twitpic_filter.php

<?php
/**
 * when twitpic URL is found in filtered $text, download the image file
 * - use curl to get info from API on actual URL (w/ temporary auth key)
 * - then use that URL to download the file to local $dl_dir
 * @return downloaded file path or FALSE
 */
function _twitpic_filter_download_twitpic($pic_code) {
  $pic_url = 'http://twitpic.com/show/full/' . $pic_code;

  // get download directory
  $dl_dir = variable_get('twitpic_filter_images_dir', file_directory_path());  // set in filter config
  $dl_dir = trim($dl_dir, '\\/');    // clear trailing slash
  
  // does the file already exist / already downloaded?
  // PROBLEM: we don't know the file type at this point! only known from the API call...
  //  but don't want to run API call every time a pic is shown...
  // workaround: the file convention is 'twitpic_CODE.TYPE' in $dl_dir - so look for 'twitpic_CODE.*'
  // THIS ASSUMES PIC CODES ARE UNIQUE... THEY SHOULD BE, SINCE THAT'S ALL THAT ID'S THEM...
  $check_exists = glob($dl_dir . '/twitpic_'.$pic_code.'*');
    // (checking pattern from relative dir not realpath(), should be ok; otherwise have to strip prefix)
  watson_debug($check_exists, 'check ' . realpath($dl_dir) . '/twitpic_'.$pic_code.'*');
  if (!empty($check_exists)) {  // at least 1 file exists
    
    // weird scenario w/ >1 file found, shouldn't happen, but track just in case
    if (sizeof($check_exists) > 1) {
      drupal_set_message("More than one Twitpic file matching code {$pic_code} exists", 'warning');
      watchdog('twitpic_filter', 'More than one Twitpic file matching code {$pic_code} exists: ' . watson_debug($check_exists,'',false), WATCHDOG_WARNING);
    }

    // path should now be in same format as if downloaded new
    return array_shift($check_exists);
  }

  // doesn't exist yet...
  
  // GET FILE INFO (ESP TYPE), THEN DOWNLOAD
  $ch = curl_init($pic_url);

  // (set these opts thru trial and error)
  curl_setopt_array($ch, array(  
     CURLOPT_BINARYTRANSFER => 1,
     CURLOPT_RETURNTRANSFER => 0,
     CURLOPT_HEADER => 0,
     CURLOPT_FOLLOWLOCATION => 1,  // follow redirects - key!
     CURLOPT_TIMEOUT => 10,    // too short? but don't want page to freeze b/c of this
     CURLOPT_NOBODY => 1,    // don't output anywhere
    ));

  curl_exec($ch);
  $info = curl_getinfo($ch);    // xml converted to array
  //drupal_set_message("info for $pic_url:" . watson_debug($info, '', false));    // DEBUGGING
  curl_close($ch);
  
  // get file info
  // actual URL will be returned w/ auth key
  if (!isset($info['url'])) {
    watchdog('twitpic_filter', 'Unable to get file info for Twitpic #'.$pic_code, WATCHDOG_ERROR);
    drupal_set_message('Twitpic Filter unable to get file info for Twitpic #'.$pic_code, 'error');
    return FALSE;
  }
  $pic_url = $info['url'];  // actual
  
  if (! preg_match('|/photos/full/([0-9]*)\.(.{3,4})\?|i', $info['url'], $info_matches)) {
    watchdog('twitpic_filter', 'Returned URL for Twitpic #'.$pic_code.' not in expected format', WATCHDOG_ERROR);
    drupal_set_message('Returned URL for Twitpic #'.$pic_code.' not in expected format', 'error');
    return FALSE;
  }
  
  // check file type
  if (!isset($info_matches[2])) {
    watchdog('twitpic_filter', 'No file type found for Twitpic #'.$pic_code, WATCHDOG_ERROR);
    drupal_set_message('No file type found for Twitpic #'.$pic_code, 'error');
    return FALSE;
  }
  $file_type = strtolower($info_matches[2]);

  // validate file type
  // @TODO should these be set in admin form?
  if (! in_array($file_type, array('jpg', 'jpeg', 'png', 'gif'))) {
    watchdog('twitpic_filter', 'Invalid file type found for Twitpic #'.$pic_code, WATCHDOG_ERROR);
    drupal_set_message('Invalid file type found for Twitpic #'.$pic_code, 'error');
    return FALSE;
  }

  // file type is good, download it...

  // same convention as above -- see comment about file type
  // (only now is file type known for sure)
  $dl_file = $dl_dir . '/' . 'twitpic_'.$pic_code . '.' . $file_type;    // relative to webroot
  $dl_realpath = realpath($dl_file);                    // absolute
  
  // running all these now w/ errors suppressed
  // @TODO curl handle errors more gracefully?
  @$ch = curl_init($pic_url);      // reinitialize
  @$fp = fopen($dl_realpath, "wb");
  @curl_setopt_array($ch, array(
      CURLOPT_FILE => $fp,
      CURLOPT_HEADER => 0,
      CURLOPT_FOLLOWLOCATION => 1,
      CURLOPT_TIMEOUT => 60,
     ));
          
  @curl_exec($ch);
  @curl_close($ch);
  @fclose($fp);

  // @TODO handle permission error?
  
  // make sure file not empty
  if (!file_exists($dl_realpath) or !filesize($dl_realpath)) {
    watchdog('twitpic_filter', "Attempted to download twitpic '$pic_code' to $dl_realpath, but file is missing or empty.", WATCHDOG_ERROR);
    drupal_set_message("Attempted to download twitpic '$pic_code' to $dl_file, but file is missing or empty.", 'error');
    return FALSE;
  }

  drupal_set_message("Downloaded Twitpic '$pic_code' to $dl_file");

  return $dl_file;
}


/**
 * passed $text from the filter 'processing', find Twitpic URLs and replace with images
 * @param $text passed from filter 'process' $op
 * @param $format passed from filter, for checking per-filter configs (imagecache preset, etc)
 * @return $text for output (changed or unchanged)
 */
function _twitpic_filter_put_pics($text, $format) {  // called by 'process' $op of twitpic_filter_filter()
  
  /**
   * pattern: twitpic url w/ 5-digit code, not in html attribute (src or href),
   *  not w/ alphanumeric, slash, period, or quotations before or after
   * (see testpattern.php)
   */
  $twitpic_pattern = '#(?<![\'".a-zA-Z0-9/])(http://twitpic.com/([a-zA-Z0-9]{5}))(?![\'".a-zA-Z0-9/])#im';

  if ($matchres===FALSE) {  // regex ERROR
    watchdog('twitpic_filter', 'Regex error', WATCHDOG_ERROR);
    drupal_set_message('Regex error in Twitpic Filter.', 'error');
    return $text;
  }
  
  // technically there can only be 1 twitpic per tweet (preg_match), but allow for many (preg_match_all) for other uses
  elseif (preg_match_all($twitpic_pattern, $text, $matches, PREG_SET_ORDER)) {

    // watson_debug($matches, 'all matches');

    // $matches will be 2-D, each element having [0]=full match, [1]=URL, [2]=code  (0=1 effectively)
    foreach($matches as $match) {

      // not sure if/when this scenario would occur, but check just in case
      if (!isset($match[2])) {
        watchdog('twitpic_filter', 'Twitpic match apparently found but missing code: '
            . '<pre>'.print_r($match,true).'</pre>', WATCHDOG_WARNING);
        drupal_set_message('Twitpic Filter unable to find an image code.', 'error', false);
      }
      else {  // code found
        $replace = $match[1];  // URL to replace w/ image
        $pic_code = $match[2];  // twitpic image code to look for

        // Download file or use existing if already downloaded
        $pic_file = _twitpic_filter_download_twitpic($pic_code);

        if (!empty($pic_file)) {  // found/downloaded
          $pic_url = url($pic_file, array('absolute'=>true));

          // substitute plain URL w/ image
          
          // @TODO pass this thru a theme hook, handle w/ imagecache if requested (by $format)

          // temporary...
          $replace_with =  "<img src='{$pic_url}' width='300' />";
          //$text .= watson_debug($match, '',false);
          
          // just do a simple str_replace... case-sensitive should be fine
          //  (maybe even necessary - if twitpic codes are case-sensitive)
          // - no $count limit here, will replace all, is that ok?
          $text = str_replace($replace, $replace_with, $text);
          
          // @TODO it's coming out as a link, maybe b/c turned into links by prior filters? need to resolve
          
          // @TODO remove all the watson calls, message debugging
        }
      }

    }

    return $text;   // breaks loop

  } // for

  return $text; // unchanged
}

?>