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 *  $use_cols ) * 24 * 60 * 60 );
 135
 136                 //print_r($res);
 137
 138                 if (empty($this->schedule[$chid][$day_id])) {
 139                     $this->schedule[$chid][$day_id] = array();
 140                 }
 141                 $this->schedule[$chid][$day_id] = array_merge($this->schedule[$chid][$day_id],  $res);
 142             }
 143             return;
 144
 145
 146         }
 147
 148         $cols = explode(',',$cinfo['htmlcols']);
 149
 150         //print_r(array($data,$cols));
 151
 152         $this->schedule[$k][$this->date + ($dayoffset * 24 * 60 * 60 )]  = $this->$method($data,$cols);
 153         //print_r($this->schedule);
 154     }
 155     //tvb +???
 156     function parseTable($data,$colnames)
 157     {
 158         $rows = preg_split('/<tr[^>]*>/i', $data);
 159        // print_r($rows);exit;
 160         array_shift($rows);
 161
 162
 163         foreach($rows as $r) {
 164             //print_r($r);
 165             $cols = preg_split('/\<td[^>]*\>/i', trim($r));
 166             $rdata = array();
 167             array_shift($cols);
 168             //var_dump(count(array_values($cols)));
 169             //print_r($r);
 170             //    print_r($cols);exit;
 171
 172             foreach($cols as $i=>$c) {
 173                 if (count(array_values($cols)) != count(array_values($colnames))) {
 174                     continue;
 175                 }
 176                 if (preg_match('/<table/i', $c)) {
 177                     continue;
 178                 }
 179                 //var_dump($c);
 180                 $c = str_ireplace('<br>',' ', $c);
 181                 $c = str_ireplace('&nbsp;',' ', $c);
 182                 $c = str_replace("\n",' ', $c);
 183                 $c = str_replace("\r",' ', $c);
 184                 //var_dump($c);
 185                 $rdata[$colnames[$i]] = trim(strip_tags($c));
 186             }
 187
 188           //  print_r($rdata);
 189             if (count(array_values($rdata)) != count(array_values($colnames))) {
 190                 continue;
 191             }
 192            // print_R($rdata);
 193             $ret[] = $rdata;
 194         }
 195
 196         $this->debug(print_r($ret,true));
 197
 198         return $ret;
 199
 200
 201     }
 202
 203     function parseJade($data, $colnames, $day)
 204     {
 205
 206         $rows = preg_split('#</li>#i', $data);
 207         //print_r($rows);
 208         $ret = array();
 209         foreach($rows as $r) {
 210             $r = str_ireplace('&nbsp;',' ', $r);
 211             $rdata = array();
 212
 213             list($time,$r) = explode('</span>', $r, 2);
 214
 215             $rdata['hour'] = trim(strip_tags($time));
 216             if (!strlen($rdata['hour'])) {
 217                 continue;
 218                }
 219             //list($title,$r) = explode('</em>', $r, 2);
 220             list($title,$r) = explode('</p>', $r, 2);
 221             $rdata['description'] = trim(strip_tags($title));
 222             $rdata['description2'] = trim(strip_tags($r));
 223             $rdata['day'] = $day;
 224             $ret[] = $rdata;
 225         }
 226         //print_R($ret);
 227        return $ret;
 228     }
 229
 230
 231     function parseatv($data,$colnames, $day)
 232     {
 233
 234         // if it's a day row..
 235
 236         $lines = explode("\n",  trim($data));
 237         //var_dump($lines[1]);
 238
 239
 240         if (isset($lines[1]) && preg_match('/<div/', trim($lines[1]))) {
 241
 242             preg_match('/<BR>([0-9]{4}-[0-9]{2}-[0-9]{2})/i', $lines[1], $matches);
 243             //var_dump($matches[1]);
 244             return $matches[1];
 245
 246         }
 247         $rows = preg_split('/<tr[^>]*>/i', $data);
 248         if ($day == '2011-10-29') {
 249             //$this->debug(print_r($rows,true));
 250         }
 251         if ($day < date('Y-m-d')) {
 252            // $this->debug("OLD DATA  $day");
 253             return array();
 254         }
 255         //$this->debug($day);
 256         //$this->debug(print_r($rows,true));
 257     //return;
 258         array_shift($rows);
 259
 260         $ret = array();
 261         //$day = false;
 262         foreach($rows as $r) {
 263             //print_r($r);
 264             $cols = preg_split('/<td[^>]*>/i', $r);
 265             $rdata = array();
 266
 267             if (!isset($cols[2])) {
 268                 continue;
 269             }
 270             //PRINT_r($cols);
 271             $c= $cols[2];
 272              //var_dump($c);
 273             // look for time..
 274             if (!preg_match('/^[0-9]+:[0-9]+/', $c)) {
 275                 continue;
 276             }
 277             //$this->debug("GOT HOUR: $c");
 278              $rdata['hour'] = trim(array_shift(explode('<', $c)));
 279             $c =  $cols[3];
 280
 281             $kv = preg_split('/<br>/',$c);
 282
 283             $c = $kv[0];
 284             $c = str_ireplace('<br>',' ', $c);
 285             $c = str_ireplace('&nbsp;',' ', $c);
 286             $c = str_replace("\n",' ', $c);
 287             $c = str_replace("\r",' ', $c);
 288             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 289             $rdata['description2']  = trim($c);
 290
 291
 292
 293             $c = $kv[1];
 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             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 299             $rdata['description']  = trim($c);
 300
 301
 302             $rdata['day'] = $day;
 303
 304
 305             //print_R($rdata);
 306             $ret[] = $rdata;
 307         }
 308         //print_r($ret); exit;
 309         return $ret;
 310
 311
 312     }
 313
 314
 315 /*
 316     function parseTableCells($data,$colnames, $day)
 317     {
 318         $rows = preg_split('/<tr[^>]*>/i', $data);
 319         //$this->debug(print_r($rows,true));
 320         //exit;
 321         //exit;
 322         array_shift($rows);
 323
 324
 325         $day = false;
 326         foreach($rows as $r) {
 327             //print_r($r);
 328             $cols = preg_split('/<td[^>]*>/i', $r);
 329             $rdata = array();
 330             $c= $cols[1];
 331              //var_dump($c);
 332             // look for time..
 333             if (!preg_match('/^[0-9]+:[0-9]+\s/', $c)) {
 334                 continue;
 335             }
 336
 337             $c = str_ireplace('<br>',' ', $c);
 338             $c = str_ireplace('&nbsp;',' ', $c);
 339             $c = str_replace("\n",' ', $c);
 340             $c = str_replace("\r",' ', $c);
 341
 342
 343             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 344             $c = trim($c);
 345             $kv = preg_split("/\s+/", $c, 2);
 346
 347             $rdata[$colnames[0]] = trim($kv[0]);
 348             $rdata[$colnames[1]] = trim($kv[1]);
 349
 350
 351             //print_r($kv);
 352             if (count(array_values($rdata)) != count(array_values($colnames))) {
 353                 continue;
 354             }
 355             $rdata['day'] = $day;
 356             $ret[] = $rdata;
 357         }
 358         //print_r($ret);
 359         return $ret;
 360
 361
 362     }
 363 */
 364     /*
 365
 366        <tv generator-info-name="tv_grab_uk">
 367           <channel id="bbc2.bbc.co.uk">
 368             <display-name lang="en">BBC2</display-name>
 369           </channel>
 370           <channel id="channel4.com">
 371             <display-name lang="en">Channel 4</display-name>
 372           </channel>
 373
 374           <programme channel="bbc2.bbc.co.uk" start="20010829000500 +0100">
 375             <title lang="en">The Phil Silvers Show</title>
 376             <desc lang="en">
 377               Bilko claims he's had a close encounter with an alien in order
 378               to be given some compassionate leave so he can visit an old
 379               flame in New York.
 380             </desc>
 381           </programme>
 382
 383           <programme channel="channel4.com" start="20010829095500 +0100">
 384             <title lang="en">King of the Hill</title>
 385             <sub-title lang="en">Meet the Propaniacs</sub-title>
 386             <desc lang="en">
 387                Bobby tours with a comedy troupe who specialize in
 388                propane-related mirth.
 389             </desc>
 390             <credits>
 391               <actor>Mike Judge</actor>
 392               <actor>Lane Smith</actor>
 393             </credits>
 394             <category lang="en">animation</category>
 395           </programme>
 396         </tv>
 397     */
 398
 399     function toXml()
 400     {
 401
 402         //print_r($this->schedule);
 403         $doc = new DomDocument('1.0', 'UTF-8');
 404         $tv = $doc->createElement('tv');
 405         $tv->setAttribute( 'generator-info-name','akpear_xml_tv');
 406         $doc->appendChild($tv);
 407
 408
 409         //$out = '<'.'?xml version="1.0" encoding="UTF-8"?.'>'."\n" .
 410          //       '<!DOCTYPE tv SYSTEM "xmltv.dtd">'."\n" ."\n" .
 411           ///      '<tv generator-info-name="akpear_xml_tv">'."\n";
 412           $donec = array();
 413         foreach($this->channels as $k => $v) {
 414             // dont dupe?!
 415             $chid = isset($v['id']) ? $v['id'] : $k;
 416             if (isset($donec[$chid])) {
 417                 continue;
 418             }
 419             $donec[$chid] = true;
 420             $ch = $doc->createElement('channel');
 421             $ch->setAttribute('id', $chid);
 422             $disp = $doc->createElement('display-name');
 423             $disp->setAttribute('lang', 'en');
 424             $disp->appendChild($doc->createTextNode($v['name']));
 425             $ch->appendChild($disp);
 426             $tv->appendChild($ch);
 427             //$out .=
 428             //    '<channel id="'. $k .'">
 429             //        <display-name lang="en">'. $v['name'] .'</display-name>
 430             //    </channel>'."\n";
 431         }
 432         //print_r($this->schedule);
 433         foreach($this->schedule as $chan => $scheds) {
 434             //print_r($sched);
 435             foreach($scheds as $day => $sched) {
 436                 $hoffset = 0;
 437                 $last = -1;
 438
 439                 if (empty($sched)) {
 440                     continue;
 441                 }
 442                 foreach($sched as $item) {
 443                     $item['day'] = $day;
 444
 445                     $bits = explode(':', $item['hour']);
 446                     if ($bits[0] < $last) {
 447                         $hoffset +=12;
 448                     }
 449                     $last = $bits[0];
 450
 451                     //var_dump($bits[0] + $hoffset);
 452                     $start = mktime(/*hmsmdy  */
 453                             $bits[0] + $hoffset,
 454                             $bits[1],
 455                             0,
 456                             date('m', $day),
 457                             date('d', $day),
 458                             date('Y', $day)
 459                             );
 460                     if ($start < strtotime(date("Y-m-d 00:00:00", strtotime('NOW - 1 DAY')))) {
 461                         continue;
 462                     }
 463                     $item['hoffset'] = $hoffset;
 464                     $item['hoffset_ar'] = $bits;
 465
 466                     //$this->debug(print_r($item, true));
 467                     $start_str = date('YmdHis',$start) . ' ' . $this->config['gmtoffset'];
 468                     //var_dump($start_str);
 469                     //var_dump($this->channels);
 470                     $description =   iconv($this->channels[$chan]['encoding'], 'UTF-8',$item['description'] .
 471                                 (isset($item['description2']) ? ('   ' . $item['description2']) : ''));
 472
 473                     $this->debug(date("Y-m-d H:i - ", $start). $description);
 474
 475
 476                     $pg = $doc->createElement('programme');
 477
 478                     $pg->setAttribute('channel', $chan);
 479                     $pg->setAttribute('start', $start_str);
 480
 481                     $title = $doc->createElement('title');
 482                     $title->setAttribute('lang', 'zh');
 483                     $title->appendChild($doc->createTextNode($this->toTitle($description,$chan)));
 484                     $pg->appendChild($title);
 485
 486                     $title = $doc->createElement('desc');
 487                     $title->setAttribute('lang', 'zh');
 488                     $title->appendChild($doc->createTextNode($description));
 489                     $pg->appendChild($title);
 490
 491                     $tv->appendChild($pg);
 492                     //$out.= '<programme channel="'.$chan. '" start="'.$start_str. '">
 493                     //    <title lang="zh">'. $this->toTitle($description) .'</title>
 494                     //      <desc lang="zh">'. $description .'</desc>
 495                     //    </programme>'."\n";
 496                 }
 497             }
 498         }
 499
 500         $doc->formatOutput = true;
 501
 502         //$out .= "</tv>\n";
 503
 504         return $doc->saveXML();
 505
 506     }
 507
 508     function toTitle($description, $chan)
 509     {
 510         // remove sponsor message.
 511         $title_pre = '';
 512         @list($title, $fuldesc) = explode("&gt;&gt;",$description);
 513
 514
 515
 516         if (preg_match('/Followed\s*By/i', $title)) {
 517             $bits = preg_split('/Followed\s*By/i', $title);
 518             $title_pre  = $bits[0]  . ' Followed By ';
 519             $title = $bits[1];
 520         }
 521         $title = preg_replace('#countdown to[a-z0-9 ]+#i', '' , $title); # NICAM Language
 522         $title = preg_replace('#^(solar x|Samsung Digital)\s*#i', '' , $title); # known sponsors..
 523
 524         $title = preg_replace('#[a-z0-9 ]+(presents|special|blockbuster|movie of the month|showtime)\s*:\s*#i', '', $title);
 525
 526         $title = preg_replace('#\([a-z]+/[a-z]+\s*(|bilingual)\)#i', '' , $title); # NICAM Language
 527         $title = preg_replace('#\(live\)#i', '' , $title);            # live
 528         $title = preg_replace('#\((s|c|l|e|cs|es|ecs|can|ce)[*]*\)#i', '' , $title);            # Subtitle
 529         $title = preg_replace('#\((pg\d*\w*)\)#i', '' , $title);         # Adult
 530         $title = preg_replace('#\(r\)#i', '' , $title);         # Repeated
 531         $title = trim($title, '/');
 532         $title = trim($title);
 533
 534         $ret = $title_pre . $title;
 535
 536       //  if (!strlen($ret)) {
 537        //     die("got $description : nothing to return");
 538         //   }
 539         if (in_array($chan, array('tvbpearl.hk', 'english.atvworld.hk'))) {
 540             $enonly = preg_replace('#^[^a-z0-9]+#i', '', $title);
 541             if (strlen($enonly) > 10) {
 542                 $title = $enonly;
 543             }
 544         }
 545
 546
 547         return $title_pre . $title;
 548     }
 549
 550
 551
 552     function debug($str)
 553     {
 554         if (empty($this->config['debug'])) {
 555              return;
 556            }
 557         echo $str."\n";
 558        }
 559
 560
 561 }
 562
 563 $x = new XML_Tv;
 564 //print_r($_SERVER);
 565
 566 $x->start($_SERVER['argv'][1]);