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