[Summary view] [Print] [Text view]

   1  <?php
   2  //Long live zem!
   3  
   4  //Time for 45KB: 0.9648-1.3698sec.
   5  function doImportBLOGGER($file, $section, $status, $invite) {
   6  
   7      # Parse a file in the MT Import Format, as described here:
   8      # http://www.movabletype.org/docs/mtimport.html
   9      # This doesn't interpret the data at all, just parse it into
  10      # a structure.
  11  
  12      $fp = fopen($file, 'r');
  13      if (!$fp)
  14          return false;
  15  
  16      //Keep some response on some part
  17      $results = array();
  18      $multiline_type = '';
  19      $multiline_data = '';
  20      $state = 'metadata';
  21      $item = array();
  22  
  23      while (!feof($fp)) {
  24          $line = rtrim(fgets($fp, 8192));
  25  
  26          # The states suggested by the spec are inconsisent, but
  27          # we'll do our best to fake it
  28  
  29          if ($line == '--------') {
  30              # End of an item, so we can process it
  31              $results[]=import_blogger_item($item, $section, $status, $invite);
  32              $item = array();
  33              $state = 'metadata';
  34          }
  35          elseif ($line == '-----' and $state == 'metadata') {
  36              $state = 'multiline';
  37              $multiline_type = '';
  38          }
  39          elseif ($line == '-----' and $state == 'multiline') {
  40              if (!empty($multiline_type))
  41                  $item[$multiline_type][] = $multiline_data;
  42              $state = 'multiline';
  43              $multiline_type = '';
  44          }
  45          elseif ($state == 'metadata') {
  46              if (preg_match('/^([A-Z ]+):\s*(.*)$/', $line, $match))
  47                  $item[$match[1]] = $match[2];
  48          }
  49          elseif ($state == 'multiline' and empty($multiline_type)) {
  50              if (preg_match('/^([A-Z ]+):\s*$/', $line, $match)) {
  51                  $multiline_type = $match[1];
  52                  $multiline_data = array();
  53              }
  54          }
  55          elseif ($state == 'multiline') {
  56              # Here's where things get hinky.  Rather than put the
  57              # multiline metadata before the field name, it goes
  58              # after, with no clear separation between metadata
  59              # and data.  And either the metadata or data might be
  60              # missing.
  61  
  62              if (empty($multiline_data['content']) and preg_match('/^([A-Z ]+):\s*(.*)$/', $line, $match)) {
  63                  # Metadata within the multiline field
  64                  $multiline_data[$match[1]] = $match[2];
  65              }
  66              elseif (empty($multiline_data['content'])) {
  67                  $multiline_data['content'] = ($line . "\n");
  68              }
  69              else {
  70                  $multiline_data['content'] .= ($line . "\n");
  71              }
  72          }
  73      }
  74  
  75      # Catch the last item in the file, if it doesn't end with a separator
  76      if (!empty($item))
  77          $results[]= import_blogger_item($item, $section, $status, $invite, $blogid);
  78  
  79      fclose($fp);
  80      return join('<br />', $results);
  81  }
  82  //Some \n chars on empty fields should be removed from body_extended and excerpt
  83  //What about the new title_html field?
  84  
  85  //Fields that are on MT format, but have not sense for blogger are commented
  86  function import_blogger_item($item, $section, $status, $invite) {
  87  
  88      # Untested import code follows
  89  
  90      if (empty($item)) return;
  91  
  92      include_once txpath.'/lib/classTextile.php';
  93      $textile = new Textile();
  94  
  95      $title = $textile->TextileThis($item['TITLE'], 1);
  96      $url_title = stripSpace($title,1);
  97  
  98      $body = $item['BODY'][0]['content'];
  99      $body_html = $textile->textileThis($body, 1);
 100  
 101      $date = strtotime($item['DATE']);
 102      $date = date('Y-m-d H:i:s', $date);
 103  
 104      if (isset($item['STATUS']))
 105          $post_status = ($item['STATUS'] == 'Draft' ? 1 : 4);
 106      else
 107          $post_status = $status;
 108  
 109      //Bogger can use special chars on author names. Strip them and check for realname
 110      $authorid = safe_field('user_id', 'txp_users', "RealName = '".doSlash($item['AUTHOR'])."'");
 111      if (!$authorid)
 112  //        $authorid = safe_field('user_id', 'txp_users', 'order by user_id asc limit 1');
 113          //Add new authors
 114          safe_insert('txp_users', "name='".doSlash(stripSpace($textile->TextileThis($item['AUTHOR'],1)))."', RealName='".doSlash($item['AUTHOR'])."'");
 115  
 116  
 117      if (!safe_field("ID", "textpattern", "Title = '".doSlash($title)."' AND Posted = '".doSlash($date)."'")) {
 118          safe_insert('textpattern',
 119              "Posted='".doSlash($date)."',".
 120              "LastMod='".doSlash($date)."',".
 121              "AuthorID='".doSlash($item['AUTHOR'])."',".
 122              "LastModID='".doSlash($item['AUTHOR'])."',".
 123              "Title='".doSlash($title)."',".
 124              "Body='".doSlash($body)."',".
 125              "Body_html='".doSlash($body_html)."',".
 126              "AnnotateInvite='".doSlash($invite)."',".
 127              "Status='".doSlash($post_status)."',".
 128              "Section='".doSlash($section)."',".
 129              "uid='".md5(uniqid(rand(),true))."',".
 130              "feed_time='".substr($date,0,10)."',".
 131              "url_title='".doSlash($url_title)."'");
 132  
 133  
 134          $parentid = mysql_insert_id();
 135  
 136          if (!empty($item['COMMENT'])) {
 137              foreach ($item['COMMENT'] as $comment) {
 138                  $comment_date = date('Y-m-d H:i:s', strtotime(@$comment['DATE']));
 139                  $comment_content = $textile->TextileThis(nl2br(@$comment['content']),1);
 140                  //Check for Comments authors
 141                  if (preg_match('/<a href="(.*)">(.*)<\/a>/', @$comment['AUTHOR'], $match)) {
 142                      @$comment['URL'] = $match[1];
 143                      @$comment['AUTHOR'] = $match[2];
 144                  }
 145                  if (!safe_field("discussid","txp_discuss","posted = '".doSlash($comment_date)."' AND message = '".doSlash($comment_content)."'")) {
 146                      safe_insert('txp_discuss',
 147                          "parentid='".doSlash($parentid)."',".
 148                          //blogger places the link to user profile page as comment author
 149                          "name='".doSlash(strip_tags(@$comment['AUTHOR']))."',".
 150  //                        "email='".doSlash(@$item['EMAIL'])."',".
 151                          "web='".doSlash(@$comment['URL'])."',".
 152  //                        "ip='".doSlash(@$item['IP'])."',".
 153                          "posted='".doSlash($comment_date)."',".
 154                          "message='".doSlash($comment_content)."',".
 155                          "visible='1'");
 156                  }
 157              }
 158          }
 159          return $title;
 160      }
 161      return $title.' already imported';
 162  }
 163  
 164  ?>