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             $start = isset($matches[1]) ? $matches[1] : 0; // first monday..
 117
 118             $use_cols = 1;
 119             foreach($days as $i=>$ddata) {
 120
 121
 122
 123                 $cols = explode(',',$cinfo['htmlcols']);
 124                 $dn = strtotime($start) + (($i * $use_cols) * 24 * 60 * 60 );
 125                 $res =  $this->$method($ddata,$cols,date('Y-m-d',$dn));
 126                 if (is_string($res)) {
 127                     $start = $res;
 128                      $use_cols =0;
 129                     continue;
 130                 }
 131                 if (!is_array($res)) {
 132                     continue;
 133                 }
 134                 $day_id = strtotime($start) + (($i *  $use_cols ) * 24 * 60 * 60 );
 135
 136                 if (empty($this->schedule[$chid][$day_id])) {
 137                     $this->schedule[$chid][$day_id] = array();
 138                 }
 139                 $this->schedule[$chid][$day_id] += $res;
 140             }
 141             return;
 142
 143
 144         }
 145
 146         $cols = explode(',',$cinfo['htmlcols']);
 147
 148         //print_r(array($data,$cols));
 149
 150         $this->schedule[$k][$this->date + ($dayoffset * 24 * 60 * 60 )]  = $this->$method($data,$cols);
 151         //print_r($this->schedule);
 152     }
 153     //tvb +???
 154     function parseTable($data,$colnames)
 155     {
 156         $rows = preg_split('/<tr[^>]*>/i', $data);
 157        // print_r($rows);exit;
 158         array_shift($rows);
 159
 160
 161         foreach($rows as $r) {
 162             //print_r($r);
 163             $cols = preg_split('/\<td[^>]*\>/i', trim($r));
 164             $rdata = array();
 165             array_shift($cols);
 166             //var_dump(count(array_values($cols)));
 167             //print_r($r);
 168             //    print_r($cols);exit;
 169
 170             foreach($cols as $i=>$c) {
 171                 if (count(array_values($cols)) != count(array_values($colnames))) {
 172                     continue;
 173                 }
 174                 if (preg_match('/<table/i', $c)) {
 175                     continue;
 176                 }
 177                 //var_dump($c);
 178                 $c = str_ireplace('<br>',' ', $c);
 179                 $c = str_ireplace('&nbsp;',' ', $c);
 180                 $c = str_replace("\n",' ', $c);
 181                 $c = str_replace("\r",' ', $c);
 182                 //var_dump($c);
 183                 $rdata[$colnames[$i]] = trim(strip_tags($c));
 184             }
 185
 186           //  print_r($rdata);
 187             if (count(array_values($rdata)) != count(array_values($colnames))) {
 188                 continue;
 189             }
 190            // print_R($rdata);
 191             $ret[] = $rdata;
 192         }
 193
 194         $this->debug(print_r($ret,true));
 195
 196         return $ret;
 197
 198
 199     }
 200
 201     function parseJade($data, $colnames)
 202     {
 203
 204         $rows = preg_split('#</li>#i', $data);
 205         //print_r($rows);
 206         $ret = array();
 207         foreach($rows as $r) {
 208             $r = str_ireplace('&nbsp;',' ', $r);
 209             $rdata = array();
 210
 211             list($time,$r) = explode('</span>', $r, 2);
 212
 213             $rdata['hour'] = trim(strip_tags($time));
 214             if (!strlen($rdata['hour'])) {
 215                 continue;
 216                }
 217             //list($title,$r) = explode('</em>', $r, 2);
 218             list($title,$r) = explode('</p>', $r, 2);
 219             $rdata['description'] = trim(strip_tags($title));
 220             $rdata['description2'] = trim(strip_tags($r));
 221             $rdata['day'] = $day;
 222             $ret[] = $rdata;
 223         }
 224         //print_R($ret);
 225        return $ret;
 226     }
 227
 228
 229     function parseatv($data,$colnames, $day)
 230     {
 231
 232         // if it's a day row..
 233
 234         $lines = explode("\n",  trim($data));
 235         //var_dump($lines[1]);
 236
 237
 238         if (isset($lines[1]) && preg_match('/<div/', trim($lines[1]))) {
 239
 240             preg_match('/<BR>([0-9]{4}-[0-9]{2}-[0-9]{2})/i', $lines[1], $matches);
 241             var_dump($matches[1]);
 242             return $matches[1];
 243
 244         }
 245         $rows = preg_split('/<tr[^>]*>/i', $data);
 246         if ($day == '2011-10-29') {
 247             //$this->debug(print_r($rows,true));
 248         }
 249         if ($day < date('Y-m-d')) {
 250            // $this->debug("OLD DATA  $day");
 251             return array();
 252         }
 253         //$this->debug($day);
 254         //$this->debug(print_r($rows,true));
 255     //return;
 256         array_shift($rows);
 257
 258
 259         //$day = false;
 260         foreach($rows as $r) {
 261             //print_r($r);
 262             $cols = preg_split('/<td[^>]*>/i', $r);
 263             $rdata = array();
 264
 265             if (!isset($cols[2])) {
 266                 continue;
 267             }
 268             //PRINT_r($cols);
 269             $c= $cols[2];
 270              //var_dump($c);
 271             // look for time..
 272             if (!preg_match('/^[0-9]+:[0-9]+/', $c)) {
 273                 continue;
 274             }
 275             //$this->debug("GOT HOUR: $c");
 276              $rdata['hour'] = trim(array_shift(explode('<', $c)));
 277             $c =  $cols[3];
 278
 279             $kv = preg_split('/<br>/',$c);
 280
 281             $c = $kv[0];
 282             $c = str_ireplace('<br>',' ', $c);
 283             $c = str_ireplace('&nbsp;',' ', $c);
 284             $c = str_replace("\n",' ', $c);
 285             $c = str_replace("\r",' ', $c);
 286             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 287             $rdata['description2']  = trim($c);
 288
 289
 290
 291             $c = $kv[1];
 292             $c = str_ireplace('<br>',' ', $c);
 293             $c = str_ireplace('&nbsp;',' ', $c);
 294             $c = str_replace("\n",' ', $c);
 295             $c = str_replace("\r",' ', $c);
 296             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 297             $rdata['description']  = trim($c);
 298
 299
 300             $rdata['day'] = $day;
 301
 302
 303             //print_R($rdata);
 304             $ret[] = $rdata;
 305         }
 306         //print_r($ret); exit;
 307         return $ret;
 308
 309
 310     }
 311
 312
 313 /*
 314     function parseTableCells($data,$colnames, $day)
 315     {
 316         $rows = preg_split('/<tr[^>]*>/i', $data);
 317         //$this->debug(print_r($rows,true));
 318         //exit;
 319         //exit;
 320         array_shift($rows);
 321
 322
 323         $day = false;
 324         foreach($rows as $r) {
 325             //print_r($r);
 326             $cols = preg_split('/<td[^>]*>/i', $r);
 327             $rdata = array();
 328             $c= $cols[1];
 329              //var_dump($c);
 330             // look for time..
 331             if (!preg_match('/^[0-9]+:[0-9]+\s/', $c)) {
 332                 continue;
 333             }
 334
 335             $c = str_ireplace('<br>',' ', $c);
 336             $c = str_ireplace('&nbsp;',' ', $c);
 337             $c = str_replace("\n",' ', $c);
 338             $c = str_replace("\r",' ', $c);
 339
 340
 341             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 342             $c = trim($c);
 343             $kv = preg_split("/\s+/", $c, 2);
 344
 345             $rdata[$colnames[0]] = trim($kv[0]);
 346             $rdata[$colnames[1]] = trim($kv[1]);
 347
 348
 349             //print_r($kv);
 350             if (count(array_values($rdata)) != count(array_values($colnames))) {
 351                 continue;
 352             }
 353             $rdata['day'] = $day;
 354             $ret[] = $rdata;
 355         }
 356         //print_r($ret);
 357         return $ret;
 358
 359
 360     }
 361 */
 362     /*
 363
 364        <tv generator-info-name="tv_grab_uk">
 365           <channel id="bbc2.bbc.co.uk">
 366             <display-name lang="en">BBC2</display-name>
 367           </channel>
 368           <channel id="channel4.com">
 369             <display-name lang="en">Channel 4</display-name>
 370           </channel>
 371
 372           <programme channel="bbc2.bbc.co.uk" start="20010829000500 +0100">
 373             <title lang="en">The Phil Silvers Show</title>
 374             <desc lang="en">
 375               Bilko claims he's had a close encounter with an alien in order
 376               to be given some compassionate leave so he can visit an old
 377               flame in New York.
 378             </desc>
 379           </programme>
 380
 381           <programme channel="channel4.com" start="20010829095500 +0100">
 382             <title lang="en">King of the Hill</title>
 383             <sub-title lang="en">Meet the Propaniacs</sub-title>
 384             <desc lang="en">
 385                Bobby tours with a comedy troupe who specialize in
 386                propane-related mirth.
 387             </desc>
 388             <credits>
 389               <actor>Mike Judge</actor>
 390               <actor>Lane Smith</actor>
 391             </credits>
 392             <category lang="en">animation</category>
 393           </programme>
 394         </tv>
 395     */
 396
 397     function toXml()
 398     {
 399
 400         //print_r($this->schedule);
 401         $doc = new DomDocument('1.0', 'UTF-8');
 402         $tv = $doc->createElement('tv');
 403         $tv->setAttribute( 'generator-info-name','akpear_xml_tv');
 404         $doc->appendChild($tv);
 405
 406
 407         //$out = '<'.'?xml version="1.0" encoding="UTF-8"?.'>'."\n" .
 408          //       '<!DOCTYPE tv SYSTEM "xmltv.dtd">'."\n" ."\n" .
 409           ///      '<tv generator-info-name="akpear_xml_tv">'."\n";
 410           $donec = array();
 411         foreach($this->channels as $k => $v) {
 412             // dont dupe?!
 413             $chid = isset($v['id']) ? $v['id'] : $k;
 414             if (isset($donec[$chid])) {
 415                 continue;
 416             }
 417             $donec[$chid] = true;
 418             $ch = $doc->createElement('channel');
 419             $ch->setAttribute('id', $chid);
 420             $disp = $doc->createElement('display-name');
 421             $disp->setAttribute('lang', 'en');
 422             $disp->appendChild($doc->createTextNode($v['name']));
 423             $ch->appendChild($disp);
 424             $tv->appendChild($ch);
 425             //$out .=
 426             //    '<channel id="'. $k .'">
 427             //        <display-name lang="en">'. $v['name'] .'</display-name>
 428             //    </channel>'."\n";
 429         }
 430         //print_r($this->schedule);
 431         foreach($this->schedule as $chan => $scheds) {
 432             //print_r($sched);
 433             foreach($scheds as $day => $sched) {
 434                 $hoffset = 0;
 435                 $last = -1;
 436
 437                 if (empty($sched)) {
 438                     continue;
 439                 }
 440                 foreach($sched as $item) {
 441                     $item['day'] = $day;
 442
 443                     $bits = explode(':', $item['hour']);
 444                     if ($bits[0] < $last) {
 445                         $hoffset +=12;
 446                     }
 447                     $last = $bits[0];
 448
 449                     //var_dump($bits[0] + $hoffset);
 450                     $start = mktime(/*hmsmdy  */
 451                             $bits[0] + $hoffset,
 452                             $bits[1],
 453                             0,
 454                             date('m', $day),
 455                             date('d', $day),
 456                             date('Y', $day)
 457                             );
 458                     if ($start < strtotime(date("Y-m-d 00:00:00", strtotime('NOW - 1 DAY')))) {
 459                         continue;
 460                     }
 461                     $item['hoffset'] = $hoffset;
 462                     $item['hoffset_ar'] = $bits;
 463
 464                     //$this->debug(print_r($item, true));
 465                     $start_str = date('YmdHis',$start) . ' ' . $this->config['gmtoffset'];
 466                     //var_dump($start_str);
 467                     //var_dump($this->channels);
 468                     $description =   iconv($this->channels[$chan]['encoding'], 'UTF-8',$item['description'] .
 469                                 (isset($item['description2']) ? ('   ' . $item['description2']) : ''));
 470
 471                     $this->debug(date("Y-m-d H:i - ", $start). $description);
 472
 473
 474                     $pg = $doc->createElement('programme');
 475
 476                     $pg->setAttribute('channel', $chan);
 477                     $pg->setAttribute('start', $start_str);
 478
 479                     $title = $doc->createElement('title');
 480                     $title->setAttribute('lang', 'zh');
 481                     $title->appendChild($doc->createTextNode($this->toTitle($description,$chan)));
 482                     $pg->appendChild($title);
 483
 484                     $title = $doc->createElement('desc');
 485                     $title->setAttribute('lang', 'zh');
 486                     $title->appendChild($doc->createTextNode($description));
 487                     $pg->appendChild($title);
 488
 489                     $tv->appendChild($pg);
 490                     //$out.= '<programme channel="'.$chan. '" start="'.$start_str. '">
 491                     //    <title lang="zh">'. $this->toTitle($description) .'</title>
 492                     //      <desc lang="zh">'. $description .'</desc>
 493                     //    </programme>'."\n";
 494                 }
 495             }
 496         }
 497
 498         $doc->formatOutput = true;
 499
 500         //$out .= "</tv>\n";
 501
 502         return $doc->saveXML();
 503
 504     }
 505
 506     function toTitle($description, $chan)
 507     {
 508         // remove sponsor message.
 509         $title_pre = '';
 510         @list($title, $fuldesc) = explode("&gt;&gt;",$description);
 511
 512
 513
 514         if (preg_match('/Followed\s*By/i', $title)) {
 515             $bits = preg_split('/Followed\s*By/i', $title);
 516             $title_pre  = $bits[0]  . ' Followed By ';
 517             $title = $bits[1];
 518         }
 519         $title = preg_replace('#countdown to[a-z0-9 ]+#i', '' , $title); # NICAM Language
 520         $title = preg_replace('#^(solar x|Samsung Digital)\s*#i', '' , $title); # known sponsors..
 521
 522         $title = preg_replace('#[a-z0-9 ]+(presents|special|blockbuster|movie of the month|showtime)\s*:\s*#i', '', $title);
 523
 524         $title = preg_replace('#\([a-z]+/[a-z]+\s*(|bilingual)\)#i', '' , $title); # NICAM Language
 525         $title = preg_replace('#\(live\)#i', '' , $title);            # live
 526         $title = preg_replace('#\((s|c|l|e|cs|es|ecs|can|ce)[*]*\)#i', '' , $title);            # Subtitle
 527         $title = preg_replace('#\((pg\d*\w*)\)#i', '' , $title);         # Adult
 528         $title = preg_replace('#\(r\)#i', '' , $title);         # Repeated
 529         $title = trim($title, '/');
 530         $title = trim($title);
 531
 532         $ret = $title_pre . $title;
 533
 534       //  if (!strlen($ret)) {
 535        //     die("got $description : nothing to return");
 536         //   }
 537         if (in_array($chan, array('tvbpearl.hk', 'english.atvworld.hk'))) {
 538             $enonly = preg_replace('#^[^a-z0-9]+#i', '', $title);
 539             if (strlen($enonly) > 10) {
 540                 $title = $enonly;
 541             }
 542         }
 543
 544
 545         return $title_pre . $title;
 546     }
 547
 548
 549
 550     function debug($str)
 551     {
 552         if (empty($this->config['debug'])) {
 553              return;
 554            }
 555         echo $str."\n";
 556        }
 557
 558
 559 }
 560
 561 $x = new XML_Tv;
 562 //print_r($_SERVER);
 563
 564 $x->start($_SERVER['argv'][1]);