nucleus/convert/bloggercaif.php

   1 <?php
   2 /*
   3  * Nucleus: PHP/MySQL Weblog CMS (http://nucleuscms.org/)
   4  * Copyright (C) 2002-2006 The Nucleus Group
   5  *
   6  * This program is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU General Public License
   8  * as published by the Free Software Foundation; either version 2
   9  * of the License, or (at your option) any later version.
  10  * (see nucleus/documentation/index.html#license for more info)
  11  */
  12 /**
  13  * This script will import a Blogger blog into a Nucleus blog, using
  14  * a easy to use wizard.
  15  *
  16  * Notes:
  17  *   - Templates are not converted
  18  *   - Nucleus should already be installed
  19  *       - Members should exist for all teammembers
  20  *
  21  * @license http://nucleuscms.org/license.txt GNU General Public License
  22  * @copyright Copyright (C) 2002-2006 The Nucleus Group
  23  * @version $Id: bloggercaif.php 1889 2012-06-17 08:46:45Z sakamocchi $
  24  */
  25
  26 include("../../config.php");
  27 include("functions.inc.php");
  28 include($DIR_LIBS . "ADMIN.php");
  29 include($DIR_LIBS . "MEDIA.php");
  30
  31 if (!$member->isLoggedIn()) {
  32         convert_showLogin('bloggercaif.php');
  33 }
  34
  35 if (!$member->isAdmin()) {
  36         convert_doError('Only Super-Admins are allowed to perform blog conversions');
  37 }
  38
  39 $ver = convert_getNucleusVersion();
  40 if ($ver > 210)
  41         convert_doError("You should check the Nucleus website for updates to this convert tool. This one might not work with your current Nucleus installation.");
  42
  43 // include PRAX lib (to read XML files easily)
  44 include ('PRAX.php');
  45
  46 switch($action) {
  47         case "assignMembers":
  48                 bc_assignMembers(); break;
  49         case "showOverview":
  50                 bc_showOverview(); break;
  51         case "doConversion":
  52                 bc_doConversion(); break;
  53         case "login": // drop through
  54         default:
  55                 bc_getBloggerBlogID();
  56 }
  57
  58 // step 1: get the Blogger Blog ID
  59 function bc_getBloggerBlogID() {
  60         global $HTTP_SERVER_VARS, $PHP_SELF;
  61
  62         convert_head();
  63
  64         ?>
  65                 <div class="note">
  66                 <b>Note:</b> This conversion tool was written for Free Blogger blogs. If you use it for BloggerPro blogs, you'll lose the titles of your posts (as I currently have no information on the templating system of BloggerPro)
  67                 </div>
  68
  69                 <h1>Step 1: Exporting to a file</h1>
  70
  71                 <p>
  72                 The first step in the conversion is to export all your Blogger entries into one single file. This is done by logging in in Blogger and by temporary changing your settings and templates.
  73                 <br />The full procedure is explained below:
  74                 </p>
  75
  76                 <h2>Exporting from Blogger</h2>
  77
  78                 <div class="note"><b>Note:</b> If you intend to keep using your weblog afterwards, write down the changes you made, so they can be undone afterwards. For the templates, copy paste the old ones in a textfile.</div>
  79
  80                 <ol>
  81                         <li>
  82                                 Log into <a href="http://www.blogger.com/">Blogger</a> and go to the blog you want to export.
  83                         </li>
  84                         <li>
  85                                 Change the template of your blog to the following:
  86
  87                                 <pre>
  88 &lt;?xml version="1.0"?&gt;
  89
  90 &lt;bloggerblog&gt;
  91
  92  &lt;Blogger&gt;
  93
  94   &lt;blogentry&gt;
  95    &lt;bloggerid&gt;&lt;$BlogItemNumber$&gt;&lt;/bloggerid&gt;
  96    &lt;body&gt;&lt;![CDATA[&lt;$BlogItemBody$&gt;]]&gt;&lt;/body&gt;
  97    &lt;date&gt;&lt;$BlogItemDateTime$&gt;&lt;/date&gt;
  98    &lt;author&gt;&lt;$BlogItemAuthor$&gt;&lt;/author&gt;
  99   &lt;/blogentry&gt;
 100
 101  &lt;/Blogger&gt;
 102
 103 &lt;/bloggerblog&gt;
 104                                 </pre>
 105                                 Don't forget to save changes!
 106                         </li>
 107                         <li>
 108                                 Go to the settings and change the following options:
 109                                 <ul>
 110                                         <li>
 111                                                 Blog filename: <b>blogger.xml</b>
 112                                         </li>
 113                                         <li>
 114                                                 Show <b>900</b> <b>day's posts</b> on main page
 115                                         </li>
 116                                         <li>
 117                                                 Date/Time Format: <b>MM/DD/YYYY HH:MM:SS AM|PM</b> (that option is not listed as such, instead it has the current date filled in)
 118                                         </li>
 119                                         <li>
 120                                                 Archive Frequency: <b>No Archive</b>
 121                                         </li>
 122                                 </ul>
 123                         </li>
 124                         <li>
 125                                 Save the settings and publish your blog. You'll end up with a blogger.xml file on your server containing all of your blog items.
 126                         </li>
 127                         <li>
 128                                 If you're running blogspot, you'll need to edit this file and take out the advertising banner code.
 129                         </li>
 130                 </ol>
 131
 132                 <h2>Exporting comments to CAIF (Instructions for YACCS)</h2>
 133
 134                 <ol>
 135                         <li>Log into YACCS</li>
 136                         <li>Export using the CAIF format. This will result in an XML file, which you should save as <tt>caif.xml</tt> on your computer.</li>
 137                 </ol>
 138
 139                 <h2>Importing</h2>
 140
 141                 <p>
 142                 Now you have two files (<b>blogger.xml</b> and <b>caif.xml</b>). Upload both to the same directory as the convert files (/nucleus/convert) and continue to the next step.
 143                 </p>
 144
 145                 <p>
 146                 <form method="post" action="bloggercaif.php">
 147                 <input type="submit" value="Next Step: Assign Members to Authors" />
 148                 <input type="hidden" name="action" value="assignMembers" />
 149                 </form>
 150                 </p>
 151
 152         <?php
 153         convert_foot();
 154 }
 155
 156
 157 function bc_assignMembers() {
 158         global $HTTP_POST_VARS, $CONF;
 159
 160         // some checks
 161         if (!file_exists('blogger.xml'))
 162                 convert_doError("blogger.xml not found. Make sure it is in the correct directory");
 163         if (!is_readable('blogger.xml'))
 164                 convert_doError("The blogger.xml file is not readable. Make sure the file permissions are set correctly so PHP can access it.");
 165         if (!file_exists('caif.xml'))
 166                 convert_doError("caif.xml not found. Make sure it is in the correct directory");
 167         if (!is_readable('caif.xml'))
 168                 convert_doError("The caif.xml file is not readable. Make sure the file permissions are set correctly so PHP can access it.");
 169
 170         convert_head();
 171
 172         ?>
 173                 <form method="post" action="bloggercaif.php">
 174
 175
 176                 <h1>Step 2: Assign Members to Authors</h1>
 177
 178                 <p>
 179                 Below is a list of all the authors that Nucleus could discover (only authors that have posted at least one entry are listed). Please assign a Nucleus Member to all of these authors.
 180                 </p>
 181
 182
 183                 <table>
 184                 <tr>
 185                         <th>Blogger Author</th>
 186                         <th>Nucleus Member</th>
 187                         <th>Blog Admin?</th>
 188                 </tr>
 189
 190                 <?php
 191                 $blog = new RAX();
 192                 $blog->openfile('blogger.xml');
 193                 $blog->record_delim = 'blogentry';
 194                 $blog->parse();
 195
 196
 197                 $authors = array();
 198
 199
 200                 while ( $entry = $blog->readRecord() )  {
 201                         $row = $entry->getRow();
 202
 203                         // handle one item
 204                         if (!in_array($row['author'],$authors))
 205                                 array_push($authors,$row['author']);
 206
 207                 }
 208
 209                 $blog->close(); // close the file
 210                 $idx = 0;
 211
 212 while ($a_name = array_pop($authors)) {
 213         ?>
 214                 <tr>
 215                         <td>
 216                                 <b><?php echo $a_name?></b>
 217                                 <input name="author[<?php echo $idx?>]" value="<?php echo Entity::hsc($a_name)?>" type="hidden" />
 218                         </td>
 219                         <td>
 220                 <?php                   $query =  'SELECT mname as text, mnumber as value FROM '.sql_table('member');
 221
 222                         $template['name'] = 'memberid[' . $idx . ']';
 223                         echo showlist($query,'select',$template);
 224                 ?>
 225                         </td>
 226                         <td>
 227                                 <input name="admin[<?php echo $idx?>]" type="checkbox" value="1" id="admin<?php echo $idx?>" /><label for="admin<?php echo $idx?>">Blog Admin</label>
 228                         </td>
 229                 </tr>
 230         <?php   $idx++;
 231 } // while
 232
 233
 234                 ?>
 235                 </table>
 236
 237
 238                 <h1>Choose Destination Weblog</h1>
 239
 240                 <p>
 241                 There are two options: you can either choose an existing blog to add the blogger entries into, or you can choose to create a new weblog.
 242                 </p>
 243
 244                 <div>
 245                         <input name="createnew" value="0" type="radio" checked='checked' id="createnew_no" /><label for="createnew_no">Choose existing weblog to add to:</label>
 246
 247                         <?php                                   $query =  'SELECT bname as text, bnumber as value FROM '.sql_table('blog');
 248                                         $template['name'] = 'blogid';
 249                                         $template['selected'] = $CONF['DefaultBlog'];
 250                                         echo showlist($query,'select',$template);
 251                         ?>
 252                 </div>
 253                 <div>
 254                         <input name="createnew" value="1" type="radio" id="createnew_yes" /><label for="createnew_yes">Create new weblog</label>
 255                         <ul>
 256                                 <li>New blog name: <input name="newblogname" /></li>
 257                                 <li>Blog owner:
 258                                 <?php                                   $query =  'SELECT mname as text, mnumber as value FROM '.sql_table('member');
 259
 260                                         $template['name'] = 'newowner';
 261                                         echo showlist($query,'select',$template);
 262                                 ?>
 263                                 </li>
 264                         </ul>
 265                 </div>
 266
 267                 <h1>Do the conversion!</h1>
 268                 <p>
 269                 <input type="hidden" name="authorcount" value="<?php echo $idx?>" />
 270                 <input type="submit" value="Step 3: Do the conversion!" />
 271                 <input type="hidden" name="action" value="doConversion" />
 272                 </p>
 273
 274                 <div class="note">
 275                 <b>Note:</b> Clicking the button once is enough, even if it takes a while to complete.
 276                 </div>
 277
 278                 </form>
 279         <?php
 280         convert_foot();
 281
 282 }
 283
 284
 285 function bc_doConversion() {
 286         global $HTTP_POST_VARS, $manager;
 287
 288         // 1. get all data
 289         $authorcount = intval($HTTP_POST_VARS['authorcount']);
 290         $author = $HTTP_POST_VARS['author'];
 291
 292         for ($i=0;$i<$authorcount;$i++) {
 293                 $key = $author[$i];
 294                 $memberid[$key] = intval($HTTP_POST_VARS['memberid'][$i]);
 295                 $isadmin[$key] = intval($HTTP_POST_VARS['admin'][$i]);
 296         }
 297
 298         $createnew = intval($HTTP_POST_VARS['createnew']);
 299         $newblogname = stripslashes($HTTP_POST_VARS['newblogname']);
 300         $newowner = intval($HTTP_POST_VARS['newowner']);
 301
 302         global $nucleus_blogid;
 303         $nucleus_blogid = intval($HTTP_POST_VARS['blogid']);
 304
 305
 306         // 2. check data
 307
 308         convert_head();
 309
 310         ?>
 311                 <h1>Step 3: Converting...</h1>
 312
 313                 <p>
 314                 Please be patient. Don't hit reload! The conversion progress should be showing below.
 315                 </p>
 316         <?php
 317         // try to extend time limit
 318         // surpress error messages when not allowed to do so
 319         @set_time_limit(1200);
 320
 321         echo "<pre>";
 322
 323         echo "Authors: <br/>";
 324         for ($i=0;$i<$authorcount;$i++) {
 325                 echo  "\tAuthor=" . $author[$i];
 326                 echo " ID=" . $memberid[$author[$i]];
 327                 echo " ADMIN=" . $isadmin[$author[$i]];
 328                 echo "<br />";
 329         }
 330         echo "Create New Weblog = $createnew (name='$newblogname', owner=$newowner, dest=$nucleus_blogid)<br />";
 331
 332         echo "</pre>";
 333
 334         // create blog
 335         if ($createnew == 1) {
 336                 // choose unique name
 337                 $shortname = 'blogger';
 338                 if (Blog::exists($shortname)) {
 339                         $idx = 1;
 340                         while (Blog::exists($shortname . $idx))
 341                                 $idx++;
 342                         $shortname = $shortname . $idx;
 343                 }
 344
 345                 $nucleus_blogid = convert_addToBlog($newblogname, $shortname, $newowner);
 346                 echo "<pre>New blog created</pre>";
 347         }
 348
 349         // add authors to blog team
 350         $b =& $manager->getBlog($nucleus_blogid);
 351         global $catid;
 352         $catid = $b->getDefaultCategory();
 353
 354         for ($i=0;$i<$authorcount;$i++)
 355                 $b->addTeamMember($memberid[$author[$i]],$isadmin[$author[$i]]);
 356
 357
 358         global $mapping;
 359         $mapping = array();
 360
 361         // 3. go through blogger.xml file
 362
 363         $blog = new RAX();
 364         $blog->openfile('blogger.xml');
 365         $blog->record_delim = 'blogentry';
 366         $blog->parse();
 367
 368         while ( $entry = $blog->readRecord() )  {
 369                 $row = $entry->getRow();
 370
 371                 bc_convertOneItem($row, $memberid, $nucleus_blogid);
 372         }
 373
 374         $blog->close(); // close the file
 375
 376         // 4. convert comments
 377         bc_convertComments($mapping);
 378
 379         echo "<pre>All done!</pre>";
 380
 381         convert_foot();
 382
 383 }
 384
 385 function bc_convertOneItem($row, $memberid, $nucleus_blogid) {
 386         global $catid, $comments;
 387
 388         $nucl_id = $memberid[$row['author']];
 389         $bloggerid = $row['bloggerid'];
 390
 391         $timestamp = date("Y-m-d H:i:s",bc_transformDate($row['date']));
 392
 393         echo "<ul>";
 394         echo "<li>Blogger ID: $bloggerid </li>";
 395         echo "<li>Body:" .  Entity::hsc(substr($row['body'],0,20)) . "...(time: " . $timestamp . ") </li>";
 396         echo "<li>author: " . $row['author'] ;
 397         echo " (nucleus-id: " . $nucl_id . ")</li>";
 398
 399         $nucleus_itemid = convert_addToItem($row['title'], $row['body'], '', $nucleus_blogid, $nucl_id, $timestamp, 0, $catid, 0, 0);
 400
 401         echo "<li>New itemid= $nucleus_itemid</li>";
 402
 403         // save mapping
 404         global $mapping;
 405         $mapping[$bloggerid] = $nucleus_itemid;
 406
 407         echo "</ul>";
 408 }
 409
 410 function bc_transformDate($date) {
 411         // 7/24/2000 11:27:13 AM
 412         if (eregi("(.*)/(.*)/(.*) (.*):(.*):(.*) (.*)",$date,$regs) != false) {
 413                 if ($regs[7] == "PM")
 414                         $regs[4] += 12;
 415                 return mktime($regs[4],$regs[5],$regs[6],$regs[1],$regs[2],$regs[3]);
 416
 417         } else {
 418                 return 0;
 419         }
 420
 421 }
 422
 423 // mapping is an array mapping blogger IDs to nucleus itemids
 424 function bc_convertComments($mapping) {
 425         $reader = new ReadCAIF('caif.xml',$mapping);
 426 }
 427
 428 function bc_addComment($nucleus_itemid, $data) {
 429         global $nucleus_blogid;
 430
 431         if (!$nucleus_itemid) {
 432                 echo "<div>No matching itemid found: ";
 433                 print_r($data);
 434                 echo "</div>";
 435                 return;
 436         }
 437
 438         $c_datetime = $data['datetime'];
 439         $c_name = $data['name'];
 440         $c_email = $data['email'];
 441         $c_uri = $data['uri'];
 442         $c_ip = $data['ip'];
 443         $c_text = $data['text'];
 444
 445         // to be sure the CDATA part is gone
 446         $c_text = str_replace('<![CDATA[','',$c_text);
 447         $c_text = str_replace(']]>','',$c_text);
 448         $c_text = trim($c_text);
 449
 450         // get unix timestamp out of datetime thing (for yaccs, this contains an erroneous comma)
 451         $c_timestamp = strtotime(str_replace(',','',$c_datetime));
 452
 453         // choose url or email to pass as userid
 454         if ((stristr($c_uri,'http://') != false) && ($c_uri != 'http://'))
 455                 $c_userid = $c_uri;
 456         else
 457                 $c_userid = $c_email;
 458
 459         echo '<div>',Entity::hsc(shorten($c_text,50,'...')),'</div><ul>';
 460 //      echo '<li>Date: ',strftime('%x %X',$c_timestamp),'</li>';
 461         echo "<li>Name: $c_name</li>";
 462 //      echo "<li>Email: $c_email</li>";
 463 //      echo "<li>URI: $c_uri</li>";
 464         echo "<li>UserID: $c_userid</li>";
 465 //      echo "<li>ip: $c_ip</li>";
 466         echo "<li>nucleus itemid: $nucleus_itemid</li>";
 467 //      echo "<li>nucleus blogid: $nucleus_blogid</li>";
 468         echo '</ul>';
 469
 470         // prepare for MySQL
 471         $c_timestamp = date("Y-m-d H:i:s",$c_timestamp);
 472
 473         // add to comments
 474         convert_addToComments($c_name, $c_userid, $c_text, $nucleus_blogid, $nucleus_itemid, 0, $c_timestamp, $c_ip);
 475
 476
 477 }
 478
 479 /* by Wouter Demuynck, slightly based on PRAX.php */
 480 class ReadCAIF {
 481         function ReadCAIF($filename, $mapping) {
 482                 $this->fp = fopen($filename, 'r');
 483                 $this->mapping =& $mapping;
 484                 $this->currentItem = 0;
 485                 $this->tag_stack = array();
 486                 $this->in_rec = 0;
 487                 $this->in_field = 0;
 488                 $this->field_data = '';
 489                 $this->fields = array();
 490
 491                 $this->parser = xml_parser_create();
 492                 xml_set_object($this->parser,$this);
 493                 xml_set_element_handler($this->parser,  'startElement',  'endElement');
 494                 xml_set_character_data_handler($this->parser,  'characterData');
 495                 xml_parser_set_option($this->parser, XML_OPTION_CASE_FOLDING, 0);
 496                 while ($buffer = fread($this->fp, 4096))
 497                         xml_parse( $this->parser, $buffer, feof($this->fp) );
 498         }
 499
 500
 501         function startElement($parser, $name, $attrs) {
 502                 array_push($this->tag_stack, $name);
 503
 504 //              echo "<br />BEGIN: ", implode($this->tag_stack,'-&gt;');
 505                 if ( !$this->in_rec and !strcmp($name, 'comment') ) {
 506                         $this->in_rec = 1;
 507                         $this->rec_lvl = sizeof($this->tag_stack);
 508                         $this->field_lvl = $this->rec_lvl + 1;
 509                 }
 510                 else if ( $this->in_rec and sizeof($this->tag_stack) == $this->field_lvl ) {
 511                         $this->in_field = 1;
 512                 } else if ($name == 'thread') {
 513                         $this->currentItem = $this->mapping[$attrs['id']];
 514 //                      echo '<b>THREAD START (',$this->currentItem,')</b>';
 515
 516                 }
 517
 518         }
 519
 520
 521         function endElement($parser, $name) {
 522                 array_pop($this->tag_stack);
 523
 524 //              echo "<br />END: ", implode($this->tag_stack,'-&gt;');
 525
 526                 if ( $this->in_rec ) {
 527
 528                         if ( sizeof($this->tag_stack) < $this->rec_lvl ) {
 529                                 $this->in_rec = 0;
 530                                 bc_addComment($this->currentItem, $this->fields);
 531                                 $this->fields = array();
 532
 533                         }
 534                         else if ( sizeof($this->tag_stack) < $this->field_lvl ) {
 535                                 $this->in_field = 0;
 536                                 $this->fields[$name] = $this->field_data;
 537                                 $this->field_data = '';
 538                         }
 539
 540                 }
 541
 542         }
 543
 544         function characterData ($parser, $data) {
 545                 if ( $this->in_field )
 546                         $this->field_data .= $data;
 547
 548         }
 549
 550
 551
 552 }
 553
 554
 555
 556 ?>