trunk/NP_MetaTags/NP_MetaTags.php

   1 <?php\r
   2 // vim: tabstop=2:shiftwidth=2\r
   3 \r
   4 /**\r
   5   * NP_MetaTags ($Revision: 1.118 $)\r
   6   * by hsur ( http://blog.cles.jp/np_cles )\r
   7 */\r
   8 \r
   9 /*\r
  10   * Copyright (C) 2005-2010 CLES. All rights reserved.\r
  11   *\r
  12   * This program is free software; you can redistribute it and/or\r
  13   * modify it under the terms of the GNU General Public License\r
  14   * as published by the Free Software Foundation; either version 2\r
  15   * of the License, or (at your option) any later version.\r
  16   * \r
  17   * This program is distributed in the hope that it will be useful,\r
  18   * but WITHOUT ANY WARRANTY; without even the implied warranty of\r
  19   * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\r
  20   * GNU General Public License for more details.\r
  21   * \r
  22   * You should have received a copy of the GNU General Public License\r
  23   * along with this program; if not, write to the Free Software\r
  24   * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301 USA\r
  25   * \r
  26   * In addition, as a special exception, cles( http://blog.cles.jp/np_cles ) gives\r
  27   * permission to link the code of this program with those files in the PEAR\r
  28   * library that are licensed under the PHP License (or with modified versions\r
  29   * of those files that use the same license as those files), and distribute\r
  30   * linked combinations including the two. You must obey the GNU General Public\r
  31   * License in all respects for all of the code used other than those files in\r
  32   * the PEAR library that are licensed under the PHP License. If you modify\r
  33   * this file, you may extend this exception to your version of the file,\r
  34   * but you are not obligated to do so. If you do not wish to do so, delete\r
  35   * this exception statement from your version.\r
  36 */\r
  37 \r
  38 require_once(dirname(__FILE__).'/sharedlibs/sharedlibs.php');\r
  39 require_once("cles/AsyncHTTP.php");\r
  40 \r
  41 class NP_MetaTags extends NucleusPlugin {\r
  42         function getName() {\r
  43                 return 'MetaTags';\r
  44         }\r
  45         function getAuthor() {\r
  46                 return 'hsur';\r
  47         }\r
  48         function getURL() {\r
  49                 return 'http://blog.cles.jp/np_cles/category/31/subcatid/4';\r
  50         }\r
  51         function getVersion() {\r
  52                 return '1.8';\r
  53         }\r
  54         function getDescription() {\r
  55                 return '[$Revision: 1.118 $]<br />This plug-in This plug-in inserts a &lt;META&gt; tag (robots, description, keywords), by using &lt;%MetaTags%&gt;';\r
  56         }\r
  57         function getMinNucleusVersion() {\r
  58                 return 320;\r
  59         }\r
  60         function supportsFeature($what) {\r
  61                 switch ($what) {\r
  62                         case 'SqlTablePrefix' :\r
  63                                 return 1;\r
  64                         default :\r
  65                                 return 0;\r
  66                 }\r
  67         }\r
  68 \r
  69         function hasAdminArea() {\r
  70                 return 1;\r
  71         }\r
  72         \r
  73         function getEventList() {\r
  74                 return array ('PostPluginOptionsUpdate');\r
  75         }\r
  76 \r
  77         function install() {\r
  78                 $this->createOption('yahooAppId', 'Y!API AppId', 'text', '');\r
  79                 $this->createOption('DescMaxLength', 'Description Max Length', 'text', '80');\r
  80                 $this->createOption('MaxKeywords', 'Num of Max Keywords', 'text', '6');\r
  81                 $this->createOption('isRefresh', 'Refresh data ?', 'yesno', 'no');\r
  82                 $this->createOption('isForceRefresh', 'Force Refresh ?', 'yesno', 'no');\r
  83 \r
  84                 $this->createItemOption('robots', 'META Robots', 'select', 'INDEX,FOLLOW', 'INDEX,FOLLOW|INDEX,FOLLOW|NOINDEX,FOLLOW|NOINDEX,FOLLOW|INDEX,NOFOLLOW|INDEX,NOFOLLOW|NOINDEX,NOFOLLOW|NOINDEX,NOFOLLOW');\r
  85                 $this->createItemOption('description', 'META description', 'textarea', '');\r
  86                 $this->createItemOption('keywords', 'META keywords', 'textarea', '');\r
  87         }\r
  88 \r
  89         function doSkinVar($skinType, $mode = 'tags') {\r
  90                 switch( $skinType ){\r
  91                         case 'item':\r
  92                                 global $itemid;\r
  93                                 $description = $this->getItemOption($itemid, 'description');\r
  94                                 $keywords = $this->getItemOption($itemid, 'keywords');\r
  95                                 $robots = $this->getItemOption($itemid, 'robots');\r
  96                                 \r
  97                                 switch( $mode ){\r
  98                                         case '':\r
  99                                         case 'tags':\r
 100                                                 if (!empty ($robots))\r
 101                                                         echo "<meta name=\"robots\" content=\"".htmlspecialchars($robots, ENT_QUOTES)."\" />\n";\r
 102                                                 if (!empty ($description))\r
 103                                                         echo "<meta name=\"description\" content=\"".htmlspecialchars($description, ENT_QUOTES)."\" />\n";\r
 104                                                 if (!empty ($keywords)){\r
 105                                                         echo "<meta name=\"keywords\" content=\"".htmlspecialchars($keywords, ENT_QUOTES)."\" />\n";\r
 106                                                 }\r
 107                                                 break;\r
 108                                                 \r
 109                                         case 'robots':\r
 110                                                 if (!empty ($robots))\r
 111                                                         echo "<meta name=\"robots\" content=\"".htmlspecialchars($robots, ENT_QUOTES)."\" />\n";\r
 112                                                 break;\r
 113                                         case 'description':\r
 114                                                 if (!empty ($description))\r
 115                                                         echo "<meta name=\"description\" content=\"".htmlspecialchars($description, ENT_QUOTES)."\" />\n";\r
 116                                                 break;\r
 117                                         case 'keywords':\r
 118                                                 if (!empty ($keywords))\r
 119                                                         echo "<meta name=\"keywords\" content=\"".htmlspecialchars($keywords, ENT_QUOTES)."\" />\n";\r
 120                                                 break;\r
 121                                                 \r
 122                                         default :\r
 123                                                 break;\r
 124                                 }\r
 125                                 break;\r
 126                                 \r
 127                         case 'archive':\r
 128                                 global $archive, $archivetype;\r
 129                                 $now = time();\r
 130                                 $t = 0;\r
 131                                 if( $archivetype == _ARCHIVETYPE_DAY ){\r
 132                                         sscanf($archive, '%d-%d-%d', $y, $m, $d);\r
 133                                         $t = mktime(0, 0, 0, $m, $d, $y);\r
 134                                 } elseif ( $archivetype == _ARCHIVETYPE_MONTH ) {\r
 135                                         sscanf($archive, '%d-%d', $y, $m);\r
 136                                         $t = mktime(0, 0, 0, $m, 1, $y);\r
 137                                 }\r
 138                                 \r
 139                                 //TODO: remove hard coding.\r
 140                                 if( $t > $now || $t < mktime(0, 0, 0, 2, 1, 2004) )\r
 141                                         echo '<meta name="robots" content="NOINDEX,NOFOLLOW" />'."\n";\r
 142                                 else\r
 143                                         echo '<meta name="robots" content="INDEX,FOLLOW" />'."\n";\r
 144                                 break;\r
 145                                 \r
 146                         default:\r
 147                                 break;\r
 148                 }\r
 149         }\r
 150         \r
 151         function event_PostPluginOptionsUpdate(& $data) {\r
 152                 global $manager;\r
 153                 \r
 154                 switch($data['context']){\r
 155                         case 'global':\r
 156                                 $affected = $this->_refreshData();\r
 157                                 break;\r
 158                         case 'item':\r
 159                                 // var_dump($data);\r
 160                                 // core hack needed.\r
 161                                 if( $data['item']['draft'] ) break;\r
 162 \r
 163                                 $item = $data['item'];\r
 164                                 $item['itemid'] = $data['itemid'];\r
 165                                 $this->_setData($item);\r
 166                                 break;\r
 167                         default:\r
 168                                 // nothing      \r
 169                 }\r
 170         }\r
 171         \r
 172         function _refreshData(){\r
 173                 if( $this->getOption('isRefresh') == yes ){\r
 174                         ACTIONLOG :: add(INFO, 'MetaTags: Invoked refreshData()');\r
 175 \r
 176                         $isForce = ($this->getOption('isForceRefresh') == 'yes')? true : false;                 \r
 177                         $this->setOption('isRefresh','no');\r
 178                         $this->setOption('isForceRefresh','no');\r
 179                         \r
 180                         $affected = 0;\r
 181                         $query = 'SELECT inumber, ibody, imore FROM '.sql_table('item').' order by inumber';\r
 182                         $res = sql_query($query);\r
 183                         if (! @mysql_num_rows($res) )\r
 184                                 return;\r
 185                         while ($assoc = mysql_fetch_assoc($res)) {\r
 186                                 // description\r
 187                                 if( !$isForce ){\r
 188                                         $description = $this->getItemOption($assoc['inumber'], 'description');\r
 189                                         if (empty($description)) {\r
 190                                                 $description = $this->_makeDescription($assoc['ibody'].$assoc['imore']);\r
 191                                                 $this->setItemOption($assoc['inumber'], 'description', $description);\r
 192                                         }\r
 193                                 } else {\r
 194                                         $description = $this->_makeDescription($assoc['ibody'].$assoc['imore']);\r
 195                                         $this->setItemOption($assoc['inumber'], 'description', $description);\r
 196                                 }\r
 197                                 // keywords\r
 198                                 if( !$isForce ){\r
 199                                         $keywords = $this->getItemOption($assoc['inumber'], 'keywords');\r
 200                                         if (empty($keywords)) {\r
 201                                                 $keywords = $this->_getKeywords($assoc['ibody'].$assoc['imore']);\r
 202                                                 $this->setItemOption($assoc['inumber'], 'keywords', $keywords);\r
 203                                         }\r
 204                                 } else {\r
 205                                         $keywords = $this->_getKeywords($assoc['ibody'].$assoc['imore']);\r
 206                                         $this->setItemOption($assoc['inumber'], 'keywords', $keywords);\r
 207                                 }\r
 208                                 $affected += 1;\r
 209                         }\r
 210                         mysql_free_result($res);\r
 211                         ACTIONLOG :: add(INFO, "MetaTags: Finished refreshData(): affected items [$affected]");\r
 212                         return $affected;\r
 213                 }               \r
 214         }\r
 215 \r
 216         function _setData(& $data) {\r
 217                 $itemid = intval($data['itemid']);\r
 218                 \r
 219                 $this->setItemOption($itemid, 'lastupdate', time());\r
 220                 \r
 221                 $description = $this->getItemOption($itemid, 'description');\r
 222                 if (empty ($description)) {\r
 223                         $description = $this->_makeDescription($data['body'].$data['more']);\r
 224                         $this->setItemOption($itemid, 'description', $description);\r
 225                 }\r
 226 \r
 227                 $keywords = $this->getItemOption($itemid, 'keywords');\r
 228                 if( strlen($keywords) > 100 ) $keywords = null;\r
 229                 if (empty ($keywords)) {\r
 230                         $keywords = $this->_getKeywords($data['body'].$data['more']);\r
 231                         $this->setItemOption($itemid, 'keywords', $keywords);\r
 232                 }\r
 233         }\r
 234 \r
 235         function _getKeywords($text) {\r
 236                 if ( ! $this->getOption('yahooAppId') )\r
 237                         return '';\r
 238                 $maxKeywords = $this->getOption('MaxKeywords');\r
 239                 $appid = $this->getOption('yahooAppId');\r
 240                 $tfidf = array();\r
 241                 \r
 242                 if( _CHARSET != 'UTF-8' )\r
 243                         $string = mb_convert_encoding($string, 'UTF-8', _CHARSET);\r
 244                 \r
 245                 $text = strip_tags($text);\r
 246                 $text = str_replace("\n", "", $text);\r
 247                 $text = str_replace("\r", "", $text);\r
 248                 \r
 249                 $postData = array();\r
 250                 $postData['appid'] = $appid;\r
 251                 $postData['results'] = 'uniq';\r
 252                 $postData['filter'] = '9';\r
 253                 $postData['response'] = 'surface';\r
 254                 $postData['sentence'] = $text;\r
 255 \r
 256                 $ahttp = new cles_AsyncHTTP();\r
 257                 $ahttp->asyncMode = false;\r
 258                 $ahttp->userAgent = "NP_MetaTags/".$this->getVersion();\r
 259                 $ahttp->setRequest('http://jlp.yahooapis.jp/MAService/V1/parse', 'POST', '', $postData);\r
 260                 list($data) = $ahttp->getResponses();\r
 261                 if( !$data )\r
 262                         ACTIONLOG :: add(WARNING, 'NP_MetaTags: AsyncHTTP Error['.$ahttp->getErrorNo(0).']'.$ahttp->getError(0));\r
 263                 \r
 264                 if( $data ){\r
 265                         $p =& new NP_MetaTags_MA_XMLParser();\r
 266                         $words = $p->parse($data);\r
 267                         if( $p->isError ){\r
 268                                 ACTIONLOG :: add(WARNING, 'NP_MetaTags: Y!API Error( '. (isset($rawtokens[0]) ? $rawtokens[0] : 'Unknown Error -> '.$data) . ' )');\r
 269                                 $words = array();\r
 270                         }\r
 271                         $p->free();\r
 272                         $p = null;\r
 273                         \r
 274                         if( $words ){\r
 275                                 arsort($words);\r
 276                                 $words = array_slice($words, 0, 20);\r
 277                                 \r
 278                                 $postData = array();\r
 279                                 $postData['appid'] = $appid;\r
 280                                 $postData['results'] = '1';\r
 281                                 $postData['adult_ok'] = '1';\r
 282                                 $postData['similar_ok'] = '1';  \r
 283                                 $ahttp = new cles_AsyncHTTP();\r
 284                                 $ahttp->userAgent = "NP_MetaTags/".$this->getVersion();\r
 285                                 \r
 286                                 $requests = array();\r
 287                                 foreach ($words as $word => $count) {\r
 288                                         $postData['query'] = $word;\r
 289                                         \r
 290                                         $qs = array();\r
 291                                         foreach($postData as $k => $v){\r
 292                                                 $qs[] = $k."=".urlencode($v);   \r
 293                                         }\r
 294                                         $u = 'http://search.yahooapis.jp/WebSearchService/V2/webSearch?'.implode("&", $qs);\r
 295                                         $id = $ahttp->setRequest($u, 'GET');\r
 296                                         \r
 297                                         $requests[$id] = $word;\r
 298                                 }\r
 299                                 \r
 300                                 $responses = $ahttp->getResponses();\r
 301                                 \r
 302                                 foreach( $requests as $id => $word ){\r
 303                                         if( $respXml = $responses[$id] ){\r
 304                                                 $p =& new NP_MetaTags_WS_XMLParser();\r
 305                                                 list($totalResultsAvailable) = $p->parse($respXml);\r
 306                                                 if( $p->isError ){\r
 307                                                         ACTIONLOG :: add(WARNING, 'NP_MetaTags: Y!API Error( '. (isset($totalResultsAvailable) ? $totalResultsAvailable : 'Unknown Error') . ' )');\r
 308                                                         $totalResultsAvailable = 0;\r
 309                                                 }\r
 310                                                 $p->free();\r
 311                                                 $p = null;\r
 312                                                 \r
 313                                                 if( $totalResultsAvailable ){\r
 314                                                         $tfidf[$word] = $words[$word] * log10(30000000000/$totalResultsAvailable);\r
 315                                                 }\r
 316                                         } else {\r
 317                                                 ACTIONLOG :: add(WARNING, $this->getName().': AsyncHTTP Error['.$ahttp->getErrorNo($id).']'.$ahttp->getError($id));\r
 318                                         }\r
 319                                 }\r
 320                                 //var_dump($ahttp);\r
 321                         }\r
 322                 }\r
 323                 //var_dump($tfidf);     \r
 324 \r
 325                 arsort($tfidf);\r
 326                 $tfidf = array_slice($tfidf, 0, $maxKeywords);\r
 327                 $result = "";\r
 328                 foreach( $tfidf as $word => $score ){\r
 329                         $result .= $word.',';\r
 330                 }\r
 331                 \r
 332                 if( _CHARSET != 'UTF-8' )\r
 333                         $result = mb_convert_encoding($result, _CHARSET, 'UTF-8');\r
 334                 \r
 335                 return mb_substr($result, 0, -1);\r
 336         }\r
 337 \r
 338         function _makeDescription($description) {\r
 339                 $maxLength = $this->getOption('DescMaxLength');\r
 340                 $description = strip_tags($description);\r
 341                 $description = Str_Replace("\n", "", $description);\r
 342                 $description = Str_Replace("\r", "", $description);\r
 343                 return htmlspecialchars(shorten($description, $maxLength, ''));\r
 344         }\r
 345 }\r
 346 \r
 347 class NP_MetaTags_Base_XMLParser {\r
 348         function init(){\r
 349                 $this->parser = xml_parser_create('UTF-8');\r
 350                 xml_set_object($this->parser, $this);\r
 351                 xml_set_element_handler($this->parser, "_open", "_close");\r
 352                 xml_set_character_data_handler($this->parser, "_cdata");\r
 353                 \r
 354                 $this->isError = false;\r
 355                 $this->inTarget = false;\r
 356         }\r
 357 \r
 358         function parse($data){\r
 359                 $this->words = array();\r
 360                 xml_parse($this->parser, $data);\r
 361                 $errcode = xml_get_error_code($this->parser);\r
 362             if ( $errcode != XML_ERROR_NONE ) {\r
 363                 $this->isError = true;\r
 364                 $this->words = array();\r
 365                         $this->words[] = 'XML Parse Error: ' . xml_error_string($errcode) . ' in '. xml_get_current_line_number($this->parser);\r
 366             }\r
 367                 return $this->words;\r
 368         }\r
 369 \r
 370         function free(){\r
 371                 xml_parser_free($this->parser);\r
 372                 $this->words = null;\r
 373         }\r
 374 }\r
 375 \r
 376 class NP_MetaTags_MA_XMLParser extends NP_MetaTags_Base_XMLParser {\r
 377         function NP_MetaTags_MA_XMLParser(){\r
 378                 $this->init();\r
 379         }\r
 380         \r
 381         function _open($parser, $name, $attribute){\r
 382                 switch( $name ){\r
 383                         case 'WORD':\r
 384                                 $this->tmp = array();\r
 385                                 break;\r
 386                         case 'SURFACE':\r
 387                                 $this->inTarget = 'SURFACE';\r
 388                                 break;\r
 389                         case 'COUNT':\r
 390                                 $this->inTarget = 'COUNT';\r
 391                                 break;                  \r
 392                         case 'MESSAGE':\r
 393                                 $this->inTarget = 'MESSAGE';\r
 394                                 break;\r
 395                         case 'ERROR':\r
 396                                 $this->isError = true;\r
 397                                 break;\r
 398                 }\r
 399         }\r
 400 \r
 401         function _close($parser, $name){\r
 402                 switch( $name ){\r
 403                         case 'WORD':\r
 404                                 if( $this->tmp['SURFACE'] && $this->tmp['COUNT'] && !is_numeric($this->tmp['SURFACE']) && mb_strlen($this->tmp['SURFACE']) > 1 ){\r
 405                                         $this->words[$this->tmp['SURFACE']] = $this->tmp['COUNT'];\r
 406                                 }\r
 407                                 break;\r
 408                         case 'MESSAGE':\r
 409                                 $this->words = array();\r
 410                                 $this->words[] = $this->tmp['MESSAGE'];\r
 411                                 break;\r
 412                 }\r
 413                 if( $name == $this->inTarget ) $this->inTarget = false;\r
 414         }\r
 415 \r
 416         function _cdata($parser, $data){\r
 417                 if( $this->inTarget ){\r
 418                         $this->tmp[$this->inTarget] = trim($data);\r
 419                 }\r
 420         }\r
 421 }\r
 422 \r
 423 class NP_MetaTags_WS_XMLParser extends NP_MetaTags_Base_XMLParser {\r
 424         function NP_MetaTags_WS_XMLParser(){\r
 425                 $this->init();\r
 426         }\r
 427         \r
 428         function _open($parser, $name, $attribute){\r
 429                 switch( $name ){\r
 430                         case 'RESULTSET':\r
 431                                 $this->words[] = $attribute['TOTALRESULTSAVAILABLE'];\r
 432                                 break;\r
 433                 }\r
 434         }\r
 435         \r
 436         function _close($parser, $name){}\r
 437         function _cdata($parser, $data){}\r
 438 }\r