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 // seriously it's this bad..
  18 error_reporting(E_ALL & ~E_STRICT &~E_NOTICE);
  19
  20
  21
  22
  23 if (!extension_loaded('mbstring')) {
  24     dl('mbstring.so');
  25 }
  26
  27 define ('QUOTE', '"');
  28 class XML_Tv
  29 {
  30     var $config; // configuration array.
  31     var $date; // date we are fetching
  32     var $channels; // details on the channels.
  33
  34
  35     function start($ini)
  36     {
  37         // load the ini file.
  38         require_once 'JSON.php';
  39         $j = new Services_JSON();
  40         $conf = (array)($j->decode(file_get_contents($ini)));
  41         //$conf = (array)json_decode(file_get_contents($ini));
  42         //print_r($conf);
  43         foreach($conf as $k=>$v) {
  44             $conf[$k] = (array)$v;
  45         }
  46         $this->channels = $conf;
  47         $this->config = $conf['global'];
  48         unset($this->channels['global']);
  49         //print_r($this->channels);
  50
  51         $this->date = time();
  52
  53
  54         foreach($this->channels as $k=>$v) {
  55
  56             for ($i=0;$i< (isset($v['days']) ? $v['days'] : $conf['days']);$i++) {
  57                 $this->grabChannel($k,$i);
  58                    // break;
  59             }
  60
  61
  62
  63             //break;
  64         }
  65
  66         echo $this->toXML();
  67
  68     }
  69
  70     function grabChannel($k,$dayoffset)
  71     {
  72         $cinfo = $this->channels[$k];
  73
  74         //print_r($cinfo);
  75        // exit;
  76         $date = $this->date + ($dayoffset * 24 * 60 * 60 );
  77
  78         $url =  strftime( $cinfo['url'], $date);
  79         $this->debug($url );
  80         //echo "GET $url\n";
  81         $data = @file_get_contents($url);
  82         if (empty($data)) {
  83             $this->debug("NO  DATA");
  84
  85             // something went wrong..
  86             return;
  87         }
  88
  89         $odata = $data;
  90         $map = array(
  91                 array("0x22", "0x09", "0x0a"),
  92                 array('"', "\t", "\n")
  93              );
  94
  95
  96         $hs = $cinfo['htmlstart'] === false ? false  : str_replace($map[0],$map[1],$cinfo['htmlstart']);
  97         //echo $hs;
  98         $he= str_replace($map[0],$map[1],$cinfo['htmlend']);
  99         //echo $he;
 100         if ($hs !== false) {
 101             list( , $data) = explode($hs,$data);
 102         }
 103
 104         if (!empty($he)) {
 105             list($data,) =  explode($he ,$data);
 106         }
 107
 108
 109         //$this->debug("DATA:".$data);
 110
 111         $method = 'parse'.$cinfo['htmlparsetype'];
 112
 113         $chid = isset($cinfo['id']) ? $cinfo['id'] : $k;
 114         // for multiday html layout of atv:
 115         if (!empty($cinfo['htmldaysep'])) {
 116             $days = explode($cinfo['htmldaysep'], $data);
 117             // kludge. = first monday of current week..
 118             // loook for... <BR>2007-12-31 Mon
 119             $start = preg_match('/<BR>([0-9]{4}-[0-9]{2}-[0-9]{2}) Mon/i', $odata, $matches);
 120             $start = isset($matches[1]) ? $matches[1] : 0; // first monday..
 121
 122             $use_cols = 1;
 123             foreach($days as $i=>$ddata) {
 124
 125
 126
 127                 $cols = explode(',',$cinfo['htmlcols']);
 128                 $dn = strtotime($start) + (($i * $use_cols) * 24 * 60 * 60 );
 129                 $res =  $this->$method($ddata,$cols,date('Y-m-d',$dn));
 130                 if (is_string($res)) {
 131                     $start = $res;
 132                     $use_cols =0;
 133                     continue;
 134                 }
 135                 if (!is_array($res)) {
 136                     continue;
 137                 }
 138                 $day_id = strtotime($start) + (($i *  $use_cols ) * 24 * 60 * 60 );
 139
 140                 //print_r($res);
 141
 142                 if (empty($this->schedule[$chid][$day_id])) {
 143                     $this->schedule[$chid][$day_id] = array();
 144                 }
 145                 $this->schedule[$chid][$day_id] = array_merge($this->schedule[$chid][$day_id],  $res);
 146             }
 147             return;
 148
 149
 150         }
 151
 152         $cols = explode(',',$cinfo['htmlcols']);
 153
 154         //print_r(array($data,$cols));
 155
 156         $this->schedule[$k][$this->date + ($dayoffset * 24 * 60 * 60 )]  = $this->$method($data,$cols);
 157         //print_r($this->schedule);
 158     }
 159     //tvb +???
 160     function parseTable($data,$colnames)
 161     {
 162         $rows = preg_split('/<tr[^>]*>/i', $data);
 163        // print_r($rows);exit;
 164         array_shift($rows);
 165
 166
 167         foreach($rows as $r) {
 168             //print_r($r);
 169             $cols = preg_split('/\<td[^>]*\>/i', trim($r));
 170             $rdata = array();
 171             array_shift($cols);
 172             //var_dump(count(array_values($cols)));
 173             //print_r($r);
 174             //    print_r($cols);exit;
 175
 176             foreach($cols as $i=>$c) {
 177                 if (count(array_values($cols)) != count(array_values($colnames))) {
 178                     continue;
 179                 }
 180                 if (preg_match('/<table/i', $c)) {
 181                     continue;
 182                 }
 183                 //var_dump($c);
 184                 $c = str_ireplace('<br>',' ', $c);
 185                 $c = str_ireplace('&nbsp;',' ', $c);
 186                 $c = str_replace("\n",' ', $c);
 187                 $c = str_replace("\r",' ', $c);
 188                 //var_dump($c);
 189                 $rdata[$colnames[$i]] = trim(strip_tags($c));
 190             }
 191
 192           //  print_r($rdata);
 193             if (count(array_values($rdata)) != count(array_values($colnames))) {
 194                 continue;
 195             }
 196            // print_R($rdata);
 197             $ret[] = $rdata;
 198         }
 199
 200         $this->debug(print_r($ret,true));
 201
 202         return $ret;
 203
 204
 205     }
 206
 207     function parseJade($data, $colnames, $day=0)
 208     {
 209
 210         $rows = preg_split('#</li>#i', $data);
 211         //print_r($rows);
 212         $ret = array();
 213         foreach($rows as $r) {
 214             $r = str_ireplace('&nbsp;',' ', $r);
 215             $rdata = array();
 216
 217             @list($time,$r) = explode('</span>', $r, 2);
 218
 219             $rdata['hour'] = trim(strip_tags($time));
 220             if (!strlen($rdata['hour'])) {
 221                 continue;
 222                }
 223             //list($title,$r) = explode('</em>', $r, 2);
 224             list($title,$r) = explode('</p>', $r, 2);
 225             $rdata['description'] = trim(strip_tags($title));
 226             $rdata['description2'] = trim(strip_tags($r));
 227             $rdata['day'] = $day;
 228             $ret[] = $rdata;
 229         }
 230         //print_R($ret);
 231        return $ret;
 232     }
 233
 234
 235     function parseatv($data,$colnames, $day)
 236     {
 237
 238         // if it's a day row..
 239
 240         $lines = explode("\n",  trim($data));
 241         //var_dump($lines[1]);
 242
 243
 244         if (isset($lines[1]) && preg_match('/<div/', trim($lines[1]))) {
 245
 246             preg_match('/<BR>([0-9]{4}-[0-9]{2}-[0-9]{2})/i', $lines[1], $matches);
 247             //var_dump($matches[1]);
 248             return $matches[1];
 249
 250         }
 251         $rows = preg_split('/<tr[^>]*>/i', $data);
 252         if ($day == '2011-10-29') {
 253             //$this->debug(print_r($rows,true));
 254         }
 255         if ($day < date('Y-m-d')) {
 256            // $this->debug("OLD DATA  $day");
 257             return array();
 258         }
 259         //$this->debug($day);
 260         //$this->debug(print_r($rows,true));
 261     //return;
 262         array_shift($rows);
 263
 264         $ret = array();
 265         //$day = false;
 266         foreach($rows as $r) {
 267             //print_r($r);
 268             $cols = preg_split('/<td[^>]*>/i', $r);
 269             $rdata = array();
 270
 271             if (!isset($cols[2])) {
 272                 continue;
 273             }
 274             //PRINT_r($cols);
 275             $c= $cols[2];
 276              //var_dump($c);
 277             // look for time..
 278             if (!preg_match('/^[0-9]+:[0-9]+/', $c)) {
 279                 continue;
 280             }
 281             //$this->debug("GOT HOUR: $c");
 282              $rdata['hour'] = trim(array_shift(explode('<', $c)));
 283             $c =  $cols[3];
 284
 285             $kv = preg_split('/<br>/',$c);
 286
 287             $c = $kv[0];
 288             $c = str_ireplace('<br>',' ', $c);
 289             $c = str_ireplace('&nbsp;',' ', $c);
 290             $c = str_replace("\n",' ', $c);
 291             $c = str_replace("\r",' ', $c);
 292             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 293             $rdata['description2']  = trim($c);
 294
 295
 296
 297             $c = isset($kv[1]) ? $kv[1] : '';
 298             $c = str_ireplace('<br>',' ', $c);
 299             $c = str_ireplace('&nbsp;',' ', $c);
 300             $c = str_replace("\n",' ', $c);
 301             $c = str_replace("\r",' ', $c);
 302             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 303             $rdata['description']  = trim($c);
 304
 305
 306             $rdata['day'] = $day;
 307
 308
 309             //print_R($rdata);
 310             $ret[] = $rdata;
 311         }
 312         //print_r($ret); exit;
 313         return $ret;
 314
 315
 316     }
 317
 318
 319 /*
 320     function parseTableCells($data,$colnames, $day)
 321     {
 322         $rows = preg_split('/<tr[^>]*>/i', $data);
 323         //$this->debug(print_r($rows,true));
 324         //exit;
 325         //exit;
 326         array_shift($rows);
 327
 328
 329         $day = false;
 330         foreach($rows as $r) {
 331             //print_r($r);
 332             $cols = preg_split('/<td[^>]*>/i', $r);
 333             $rdata = array();
 334             $c= $cols[1];
 335              //var_dump($c);
 336             // look for time..
 337             if (!preg_match('/^[0-9]+:[0-9]+\s/', $c)) {
 338                 continue;
 339             }
 340
 341             $c = str_ireplace('<br>',' ', $c);
 342             $c = str_ireplace('&nbsp;',' ', $c);
 343             $c = str_replace("\n",' ', $c);
 344             $c = str_replace("\r",' ', $c);
 345
 346
 347             $c = preg_replace('/\<[^>]+\>/', ' ', $c);
 348             $c = trim($c);
 349             $kv = preg_split("/\s+/", $c, 2);
 350
 351             $rdata[$colnames[0]] = trim($kv[0]);
 352             $rdata[$colnames[1]] = trim($kv[1]);
 353
 354
 355             //print_r($kv);
 356             if (count(array_values($rdata)) != count(array_values($colnames))) {
 357                 continue;
 358             }
 359             $rdata['day'] = $day;
 360             $ret[] = $rdata;
 361         }
 362         //print_r($ret);
 363         return $ret;
 364
 365
 366     }
 367 */
 368     /*
 369
 370        <tv generator-info-name="tv_grab_uk">
 371           <channel id="bbc2.bbc.co.uk">
 372             <display-name lang="en">BBC2</display-name>
 373           </channel>
 374           <channel id="channel4.com">
 375             <display-name lang="en">Channel 4</display-name>
 376           </channel>
 377
 378           <programme channel="bbc2.bbc.co.uk" start="20010829000500 +0100">
 379             <title lang="en">The Phil Silvers Show</title>
 380             <desc lang="en">
 381               Bilko claims he's had a close encounter with an alien in order
 382               to be given some compassionate leave so he can visit an old
 383               flame in New York.
 384             </desc>
 385           </programme>
 386
 387           <programme channel="channel4.com" start="20010829095500 +0100">
 388             <title lang="en">King of the Hill</title>
 389             <sub-title lang="en">Meet the Propaniacs</sub-title>
 390             <desc lang="en">
 391                Bobby tours with a comedy troupe who specialize in
 392                propane-related mirth.
 393             </desc>
 394             <credits>
 395               <actor>Mike Judge</actor>
 396               <actor>Lane Smith</actor>
 397             </credits>
 398             <category lang="en">animation</category>
 399           </programme>
 400         </tv>
 401     */
 402
 403     function toXml()
 404     {
 405
 406         //print_r($this->schedule);
 407         $doc = new DomDocument('1.0', 'UTF-8');
 408         $tv = $doc->createElement('tv');
 409         $tv->setAttribute( 'generator-info-name','akpear_xml_tv');
 410         $doc->appendChild($tv);
 411
 412
 413         //$out = '<'.'?xml version="1.0" encoding="UTF-8"?.'>'."\n" .
 414          //       '<!DOCTYPE tv SYSTEM "xmltv.dtd">'."\n" ."\n" .
 415           ///      '<tv generator-info-name="akpear_xml_tv">'."\n";
 416           $donec = array();
 417         foreach($this->channels as $k => $v) {
 418             // dont dupe?!
 419             $chid = isset($v['id']) ? $v['id'] : $k;
 420             if (isset($donec[$chid])) {
 421                 continue;
 422             }
 423             $donec[$chid] = true;
 424             $ch = $doc->createElement('channel');
 425             $ch->setAttribute('id', $chid);
 426             $disp = $doc->createElement('display-name');
 427             $disp->setAttribute('lang', 'en');
 428             $disp->appendChild($doc->createTextNode($v['name']));
 429             $ch->appendChild($disp);
 430             $tv->appendChild($ch);
 431             //$out .=
 432             //    '<channel id="'. $k .'">
 433             //        <display-name lang="en">'. $v['name'] .'</display-name>
 434             //    </channel>'."\n";
 435         }
 436         //print_r($this->schedule);
 437         foreach($this->schedule as $chan => $scheds) {
 438             //print_r($sched);
 439             foreach($scheds as $day => $sched) {
 440                 $hoffset = 0;
 441                 $last = -1;
 442
 443                 if (empty($sched)) {
 444                     continue;
 445                 }
 446                 foreach($sched as $item) {
 447                     $item['day'] = $day;
 448
 449                     $bits = explode(':', $item['hour']);
 450                     if ($bits[0] < $last) {
 451                         $hoffset +=12;
 452                     }
 453                     $last = $bits[0];
 454
 455                     //var_dump($bits[0] + $hoffset);
 456                     $start = mktime(/*hmsmdy  */
 457                             $bits[0] + $hoffset,
 458                             $bits[1],
 459                             0,
 460                             date('m', $day),
 461                             date('d', $day),
 462                             date('Y', $day)
 463                             );
 464                     if ($start < strtotime(date("Y-m-d 00:00:00", strtotime('NOW - 1 DAY')))) {
 465                         continue;
 466                     }
 467                     $item['hoffset'] = $hoffset;
 468                     $item['hoffset_ar'] = $bits;
 469
 470                     //$this->debug(print_r($item, true));
 471                     $start_str = date('YmdHis',$start) . ' ' . $this->config['gmtoffset'];
 472                     //var_dump($start_str);
 473                     //var_dump($this->channels);
 474                     $description =   iconv($this->channels[$chan]['encoding'], 'UTF-8',$item['description'] .
 475                                 (isset($item['description2']) ? ('   ' . $item['description2']) : ''));
 476
 477                     $this->debug(date("Y-m-d H:i - ", $start). $description);
 478
 479
 480                     $pg = $doc->createElement('programme');
 481
 482                     $pg->setAttribute('channel', $chan);
 483                     $pg->setAttribute('start', $start_str);
 484
 485                     $title = $doc->createElement('title');
 486                     $title->setAttribute('lang', 'zh');
 487                     $title->appendChild($doc->createTextNode($this->toTitle($description,$chan)));
 488                     $pg->appendChild($title);
 489
 490                     $title = $doc->createElement('desc');
 491                     $title->setAttribute('lang', 'zh');
 492                     $title->appendChild($doc->createTextNode($description));
 493                     $pg->appendChild($title);
 494
 495                     $tv->appendChild($pg);
 496                     //$out.= '<programme channel="'.$chan. '" start="'.$start_str. '">
 497                     //    <title lang="zh">'. $this->toTitle($description) .'</title>
 498                     //      <desc lang="zh">'. $description .'</desc>
 499                     //    </programme>'."\n";
 500                 }
 501             }
 502         }
 503
 504         $doc->formatOutput = true;
 505
 506         //$out .= "</tv>\n";
 507
 508         return $doc->saveXML();
 509
 510     }
 511
 512     function toTitle($description, $chan)
 513     {
 514         // remove sponsor message.
 515         $title_pre = '';
 516         @list($title, $fuldesc) = explode("&gt;&gt;",$description);
 517
 518
 519
 520         if (preg_match('/Followed\s*By/i', $title)) {
 521             $bits = preg_split('/Followed\s*By/i', $title);
 522             $title_pre  = $bits[0]  . ' Followed By ';
 523             $title = $bits[1];
 524         }
 525         $title = preg_replace('#countdown to[a-z0-9 ]+#i', '' , $title); # NICAM Language
 526         $title = preg_replace('#^(solar x|Samsung Digital)\s*#i', '' , $title); # known sponsors..
 527
 528         $title = preg_replace('#[a-z0-9 ]+(presents|special|blockbuster|movie of the month|showtime)\s*:\s*#i', '', $title);
 529
 530         $title = preg_replace('#\([a-z]+/[a-z]+\s*(|bilingual)\)#i', '' , $title); # NICAM Language
 531         $title = preg_replace('#\(live\)#i', '' , $title);            # live
 532         $title = preg_replace('#\((s|c|l|e|cs|es|ecs|can|ce)[*]*\)#i', '' , $title);            # Subtitle
 533         $title = preg_replace('#\((pg\d*\w*)\)#i', '' , $title);         # Adult
 534         $title = preg_replace('#\(r\)#i', '' , $title);         # Repeated
 535         $title = trim($title, '/');
 536         $title = trim($title);
 537
 538         $ret = $title_pre . $title;
 539
 540       //  if (!strlen($ret)) {
 541        //     die("got $description : nothing to return");
 542         //   }
 543         if (in_array($chan, array('tvbpearl.hk', 'english.atvworld.hk'))) {
 544             $enonly = preg_replace('#^[^a-z0-9]+#i', '', $title);
 545             if (strlen($enonly) > 10) {
 546                 $title = $enonly;
 547             }
 548         }
 549
 550
 551         return $title_pre . $title;
 552     }
 553
 554
 555
 556     function debug($str)
 557     {
 558         if (empty($this->config['debug'])) {
 559              return;
 560            }
 561         echo $str."\n";
 562        }
 563
 564
 565 }
 566
 567 $x = new XML_Tv;
 568 //print_r($_SERVER);
 569
 570 $x->start($_SERVER['argv'][1]);