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