git.roojs.org Git - lib.XML_Tv/blob - Tv.php

   1 <?php
   2 /*
   3 Usage:
   4
   5 make a weekly cronjob (/etc/cron.weekly/filltv)
   6 #!/bin/sh
   7 /usr/bin/php /path/to/application/Tv.php /path/to/application/hongkong.ini > /tmp/hongkong.xml
   8 /usr/bin/mythfilldatabase --file 1 1 /tmp/hongkong.xml
   9
  10 make sure the filltv file is chmod +x
  11
  12
  13
  14 Quick hack to generate xmltv listings from TVxb style .ini files.
  15
  16 */
  17
  18
  19 if (!extension_loaded('mbstring')) {
  20     dl('mbstring.so');
  21 }
  22
  23 define ('QUOTE', '"');
  24 class XML_Tv
  25 {
  26     var $config; // configuration array.
  27     var $date; // date we are fetching
  28     var $channels; // details on the channels.
  29
  30
  31     function start($ini)
  32     {
  33         // load the ini file.
  34         require_once 'JSON.php';
  35         $j = new Services_JSON();
  36         $conf = (array)($j->decode(file_get_contents($ini)));
  37         //$conf = (array)json_decode(file_get_contents($ini));
  38         //print_r($conf);
  39         foreach($conf as $k=>$v) {
  40             $conf[$k] = (array)$v;
  41         }
  42         $this->channels = $conf;
  43         $this->config = $conf['global'];
  44         unset($this->channels['global']);
  45         //print_r($this->channels);
  46
  47         $this->date = time();
  48
  49
  50         foreach($this->channels as $k=>$v) {
  51
  52             for ($i=0;$i< (isset($v['days']) ? $v['days'] : $conf['days']);$i++) {
  53                 $this->grabChannel($k,$i);
  54                    // break;
  55             }
  56
  57
  58
  59             //break;
  60         }
  61
  62         echo $this->toXML();
  63
  64     }
  65
  66     function grabChannel($k,$dayoffset)
  67     {
  68         $cinfo = $this->channels[$k];
  69
  70         //print_r($cinfo);
  71        // exit;
  72         $date = $this->date + ($dayoffset * 24 * 60 * 60 );
  73
  74         $url =  strftime( $cinfo['url'], $date);
  75         $this->debug($url );
  76         //echo "GET $url\n";
  77         $data = @file_get_contents($url);
  78         if (empty($data)) {
  79             $this->debug("NO  DATA");
  80
  81             // something went wrong..
  82             return;
  83         }
  84
  85         $odata = $data;
  86         $map = array(
  87                 array("0x22", "0x09", "0x0a"),
  88                 array('"', "\t", "\n")
  89              );
  90
  91
  92         $hs = $cinfo['htmlstart'] === false ? false  : str_replace($map[0],$map[1],$cinfo['htmlstart']);
  93         //echo $hs;
  94         $he= str_replace($map[0],$map[1],$cinfo['htmlend']);
  95         //echo $he;
  96         if ($hs !== false) {
  97             list( , $data) = explode($hs,$data);
  98         }
  99
 100         if (!empty($he)) {
 101             list($data,) =  explode($he ,$data);
 102         }
 103
 104
 105         //$this->debug("DATA:".$data);
 106
 107         $method = 'parse'.$cinfo['htmlparsetype'];
 108
 109         $chid = isset($cinfo['id']) ? $cinfo['id'] : $k;
 110         // for multiday html layout of atv:
 111         if (!empty($cinfo['htmldaysep'])) {
 112             $days = explode($cinfo['htmldaysep'], $data);
 113             // kludge. = first monday of current week..
 114             // loook for... <BR>2007-12-31 Mon
 115             $start = preg_match('/<BR>([0-9]{4}-[0-9]{2}-[0-9]{2}) Mon/i', $odata, $matches);
 116             $start = $matches[1]; // first monday..
 117             foreach($days as $i=>$ddata) {
 118                 $cols = explode(',',$cinfo['htmlcols']);
 119                 $dn = strtotime($start) + (($i ) * 24 * 60 * 60 );
 120                 $this->schedule[$chid][strtotime($start) + (($i ) * 24 * 60 * 60 )]  = $this->$method($ddata,$cols,date('Y-m-d',$dn));
 121             }
 122             return;
 123
 124
 125         }
 126
 127         $cols = explode(',',$cinfo['htmlcols']);
 128
 129         //print_r(array($data,$cols));
 130
 131         $this->schedule[$k][$this->date + ($dayoffset * 24 * 60 * 60 )]  = $this->$method($data,$cols);
 132         //print_r($this->schedule);
 133     }
 134     //tvb +???
 135     function parseTable($data,$colnames)
 136     {
 137         $rows = preg_split('/<tr[^>]*>/i', $data);
 138        // print_r($rows);exit;
 139         array_shift($rows);
 140
 141
 142         foreach($rows as $r) {
 143             //print_r($r);
 144             $cols = preg_split('/\<td[^>]*\>/i', trim($r));
 145             $rdata = array();
 146             array_shift($cols);
 147             //var_dump(count(array_values($cols)));
 148             //print_r($r);
 149             //    print_r($cols);exit;
 150
 151             foreach($cols as $i=>$c) {
 152                 if (count(array_values($cols)) != count(array_values($colnames))) {
 153                     continue;
 154                 }
 155                 if (preg_match('/<table/i', $c)) {
 156                     continue;
 157                 }
 158                 //var_dump($c);
 159                 $c = str_ireplace('<br>',' ', $c);
 160                 $c = str_ireplace('&nbsp;',' ', $c);
 161                 $c = str_replace("\n",' ', $c);
 162                 $c = str_replace("\r",' ', $c);
 163                 //var_dump($c);
 164                 $rdata[$colnames[$i]] = trim(strip_tags($c));
 165             }
 166
 167           //  print_r($rdata);
 168             if (count(array_values($rdata)) != count(array_values($colnames))) {
 169                 continue;
 170             }
 171            // print_R($rdata);
 172             $ret[] = $rdata;
 173         }
 174
 175         $this->debug(print_r($ret,true));
 176
 177         return $ret;
 178
 179
 180     }
 181
 182     function parseJade($data, $colnames)
 183     {
 184
 185         $rows = preg_split('#</li>#i', $data);
 186         //print_r($rows);
 187         $ret = array();
 188         foreach($rows as $r) {
 189             $r = str_ireplace('&nbsp;',' ', $r);
 190             $rdata = array();
 191
 192             list($time,$r) = explode('</span>', $r, 2);
 193
 194             $rdata['hour'] = trim(strip_tags($time));
 195             if (!strlen($rdata['hour'])) {
 196                 continue;
 197                }
 198             //list($title,$r) = explode('</em>', $r, 2);
 199             list($title,$r) = explode('</p>', $r, 2);
 200             $rdata['description'] = trim(strip_tags($title));
 201             $rdata['description2'] = trim(strip_tags($r));
 202             $rdata['day'] = $day;
 203             $ret[] = $rdata;
 204         }
 205         //print_R($ret);
 206        return $ret;
 207     }
 208
 209
 210     function parseatv($data,$colnames, $day)
 211     {
 212
 213
 214         $rows = preg_split('/<tr[^>]*>/i', $data);
 215         //$this->debug(print_r($rows,true));
 216         //exit;
 217         array_shift($rows);
 218
 219
 220         $day = false;
 221         foreach($rows as $r) {
 222             //print_r($r);
 223             $cols = preg_split('/<td[^>]*>/i', $r);
 224             $rdata = array();
 225
 226
 227             PRINT_r($cols);
 228             $c= $cols[2];
 229              //var_dump($c);
 230             // look for time..
 231             if (!preg_match('/^[0-9]+:[0-9]+/', $c)) {
 232                 continue;
 233             }
 234
 235             $rdata['time'] = trim(preg_replace('/\<.*$/', '', $c));
 236             $c =  $cols[3];
 237
 238             $kv = preg_split('/<br>/',$c);
 239
 240             $c = $kv[0];
 241             $c = str_ireplace('<br>',' ', $c);
 242             $c = str_ireplace('&nbsp;',' ', $c);
 243             $c = str_replace("\n",' ', $c);
 244             $c = str_replace("\r",' ', $c);
 245             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 246             $rdata['description2']  = trim($c);
 247
 248
 249
 250             $c = $kv[1];
 251             $c = str_ireplace('<br>',' ', $c);
 252             $c = str_ireplace('&nbsp;',' ', $c);
 253             $c = str_replace("\n",' ', $c);
 254             $c = str_replace("\r",' ', $c);
 255             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 256             $rdata['description']  = trim($c);
 257
 258
 259             $rdata['day'] = $day;
 260
 261
 262             print_R($rdata);
 263             $ret[] = $rdata;
 264         }
 265         //print_r($ret); exit;
 266         return $ret;
 267
 268
 269     }
 270
 271
 272 /*
 273     function parseTableCells($data,$colnames, $day)
 274     {
 275         $rows = preg_split('/<tr[^>]*>/i', $data);
 276         //$this->debug(print_r($rows,true));
 277         //exit;
 278         //exit;
 279         array_shift($rows);
 280
 281
 282         $day = false;
 283         foreach($rows as $r) {
 284             //print_r($r);
 285             $cols = preg_split('/<td[^>]*>/i', $r);
 286             $rdata = array();
 287             $c= $cols[1];
 288              //var_dump($c);
 289             // look for time..
 290             if (!preg_match('/^[0-9]+:[0-9]+\s/', $c)) {
 291                 continue;
 292             }
 293
 294             $c = str_ireplace('<br>',' ', $c);
 295             $c = str_ireplace('&nbsp;',' ', $c);
 296             $c = str_replace("\n",' ', $c);
 297             $c = str_replace("\r",' ', $c);
 298
 299
 300             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 301             $c = trim($c);
 302             $kv = preg_split("/\s+/", $c, 2);
 303
 304             $rdata[$colnames[0]] = trim($kv[0]);
 305             $rdata[$colnames[1]] = trim($kv[1]);
 306
 307
 308             //print_r($kv);
 309             if (count(array_values($rdata)) != count(array_values($colnames))) {
 310                 continue;
 311             }
 312             $rdata['day'] = $day;
 313             $ret[] = $rdata;
 314         }
 315         //print_r($ret);
 316         return $ret;
 317
 318
 319     }
 320 */
 321     /*
 322
 323        <tv generator-info-name="tv_grab_uk">
 324           <channel id="bbc2.bbc.co.uk">
 325             <display-name lang="en">BBC2</display-name>
 326           </channel>
 327           <channel id="channel4.com">
 328             <display-name lang="en">Channel 4</display-name>
 329           </channel>
 330
 331           <programme channel="bbc2.bbc.co.uk" start="20010829000500 +0100">
 332             <title lang="en">The Phil Silvers Show</title>
 333             <desc lang="en">
 334               Bilko claims he's had a close encounter with an alien in order
 335               to be given some compassionate leave so he can visit an old
 336               flame in New York.
 337             </desc>
 338           </programme>
 339
 340           <programme channel="channel4.com" start="20010829095500 +0100">
 341             <title lang="en">King of the Hill</title>
 342             <sub-title lang="en">Meet the Propaniacs</sub-title>
 343             <desc lang="en">
 344                Bobby tours with a comedy troupe who specialize in
 345                propane-related mirth.
 346             </desc>
 347             <credits>
 348               <actor>Mike Judge</actor>
 349               <actor>Lane Smith</actor>
 350             </credits>
 351             <category lang="en">animation</category>
 352           </programme>
 353         </tv>
 354     */
 355
 356     function toXml()
 357     {
 358
 359         //print_r($this->schedule);
 360         $doc = new DomDocument('1.0', 'UTF-8');
 361         $tv = $doc->createElement('tv');
 362         $tv->setAttribute( 'generator-info-name','akpear_xml_tv');
 363         $doc->appendChild($tv);
 364
 365
 366         //$out = '<'.'?xml version="1.0" encoding="UTF-8"?.'>'."\n" .
 367          //       '<!DOCTYPE tv SYSTEM "xmltv.dtd">'."\n" ."\n" .
 368           ///      '<tv generator-info-name="akpear_xml_tv">'."\n";
 369           $donec = array();
 370         foreach($this->channels as $k => $v) {
 371             // dont dupe?!
 372             $chid = isset($v['id']) ? $v['id'] : $k;
 373             if (isset($donec[$chid])) {
 374                 continue;
 375             }
 376             $donec[$chid] = true;
 377             $ch = $doc->createElement('channel');
 378             $ch->setAttribute('id', $chid);
 379             $disp = $doc->createElement('display-name');
 380             $disp->setAttribute('lang', 'en');
 381             $disp->appendChild($doc->createTextNode($v['name']));
 382             $ch->appendChild($disp);
 383             $tv->appendChild($ch);
 384             //$out .=
 385             //    '<channel id="'. $k .'">
 386             //        <display-name lang="en">'. $v['name'] .'</display-name>
 387             //    </channel>'."\n";
 388         }
 389         //print_r($this->schedule);
 390         foreach($this->schedule as $chan => $scheds) {
 391             //print_r($sched);
 392             foreach($scheds as $day => $sched) {
 393                 $hoffset = 0;
 394                 $last = -1;
 395
 396                 if (empty($sched)) {
 397                     continue;
 398                 }
 399                 foreach($sched as $item) {
 400                     //print_r($item);
 401                     $bits = explode(':', $item['hour']);
 402                     if ($bits[0] < $last) {
 403                         $hoffset +=12;
 404                     }
 405                     $last = $bits[0];
 406                     //var_dump($bits[0] + $hoffset);
 407                     $start = mktime(/*hmsmdy  */
 408                             $bits[0] + $hoffset,
 409                             $bits[1],
 410                             0,
 411                             date('m', $day),
 412                             date('d', $day),
 413                             date('Y', $day)
 414                             );
 415
 416                     $start_str = date('YmdHis',$start) . ' ' . $this->config['gmtoffset'];
 417                     //var_dump($start_str);
 418                     //var_dump($this->channels);
 419                     $description =   iconv($this->channels[$chan]['encoding'], 'UTF-8',$item['description'] .
 420                                 (isset($item['description2']) ? ('   ' . $item['description2']) : ''));
 421
 422                     $pg = $doc->createElement('programme');
 423
 424                     $pg->setAttribute('channel', $chan);
 425                     $pg->setAttribute('start', $start_str);
 426
 427                     $title = $doc->createElement('title');
 428                     $title->setAttribute('lang', 'zh');
 429                     $title->appendChild($doc->createTextNode($this->toTitle($description,$chan)));
 430                     $pg->appendChild($title);
 431
 432                     $title = $doc->createElement('desc');
 433                     $title->setAttribute('lang', 'zh');
 434                     $title->appendChild($doc->createTextNode($description));
 435                     $pg->appendChild($title);
 436
 437                     $tv->appendChild($pg);
 438                     //$out.= '<programme channel="'.$chan. '" start="'.$start_str. '">
 439                     //    <title lang="zh">'. $this->toTitle($description) .'</title>
 440                     //      <desc lang="zh">'. $description .'</desc>
 441                     //    </programme>'."\n";
 442                 }
 443             }
 444         }
 445
 446         $doc->formatOutput = true;
 447
 448         //$out .= "</tv>\n";
 449
 450         return $doc->saveXML();
 451
 452     }
 453
 454     function toTitle($description, $chan)
 455     {
 456         // remove sponsor message.
 457         $title_pre = '';
 458         list($title, $fuldesc) = explode("&gt;&gt;",$description);
 459
 460
 461
 462         if (preg_match('/Followed\s*By/i', $title)) {
 463             $bits = preg_split('/Followed\s*By/i', $title);
 464             $title_pre  = $bits[0]  . ' Followed By ';
 465             $title = $bits[1];
 466         }
 467         $title = preg_replace('#countdown to[a-z0-9 ]+#i', '' , $title); # NICAM Language
 468         $title = preg_replace('#^(solar x|Samsung Digital)\s*#i', '' , $title); # known sponsors..
 469
 470         $title = preg_replace('#[a-z0-9 ]+(presents|special|blockbuster|movie of the month|showtime)\s*:\s*#i', '', $title);
 471
 472         $title = preg_replace('#\([a-z]+/[a-z]+\s*(|bilingual)\)#i', '' , $title); # NICAM Language
 473         $title = preg_replace('#\(live\)#i', '' , $title);            # live
 474         $title = preg_replace('#\((s|c|l|e|cs|es|ecs|can|ce)[*]*\)#i', '' , $title);            # Subtitle
 475         $title = preg_replace('#\((pg\d*\w*)\)#i', '' , $title);         # Adult
 476         $title = preg_replace('#\(r\)#i', '' , $title);         # Repeated
 477         $title = trim($title, '/');
 478         $title = trim($title);
 479
 480         $ret = $title_pre . $title;
 481
 482       //  if (!strlen($ret)) {
 483        //     die("got $description : nothing to return");
 484         //   }
 485         if (in_array($chan, array('tvbpearl.hk', 'english.atvworld.hk'))) {
 486             $enonly = preg_replace('#^[^a-z0-9]+#i', '', $title);
 487             if (strlen($enonly) > 10) {
 488                 $title = $enonly;
 489             }
 490         }
 491
 492
 493         return $title_pre . $title;
 494     }
 495
 496
 497
 498     function debug($str)
 499     {
 500         if (empty($this->config['debug'])) {
 501              return;
 502            }
 503         echo $str."\n";
 504        }
 505
 506
 507 }
 508
 509 $x = new XML_Tv;
 510 //print_r($_SERVER);
 511
 512 $x->start($_SERVER['argv'][1]);