','>'), $response);
$dom = new DomDocument;
libxml_use_internal_errors(True);
- $dom->loadHTML(mb_convert_encoding($response, "UTF-8", "EUC-KR"));
- $xpath = new DomXPath($dom);
- for($i = 2; $i < 5; $i++) :
- $thisday = $day;
- $query = "//table[contains(@style,'margin-bottom:30')][".$i."]//td[contains(@colspan,'2')]/following::td[1]/table[1]//td[2]";
- $programs = $xpath->query($query);
- foreach($programs as $program) :
- $hour = $xpath->query("parent::*/parent::*/parent::*/parent::*/td[1]", $program)->item(0);
- $hour = str_replace("시", "", trim($hour->nodeValue));
- $minute = $xpath->query("preceding-sibling::td[1]", $program)->item(0);
- $hour = $hour.":".str_replace(array("[", "]"), array("",""), trim($minute->nodeValue));
- switch ($i) :
- case 2 :
- $hour = $hour." AM";
- break;
- case 3 :
- $hour = $hour." PM";
- break;
- case 4 :
- if($hour > 5 && $hour < 12) :
- $hour = $hour." PM";
- elseif($hour <5 || $hour == 12) :
+ if($dom->loadHTML($response)):
+ $xpath = new DomXPath($dom);
+ for($i = 2; $i < 5; $i++) :
+ $thisday = $day;
+ $query = "//table[contains(@style,'margin-bottom:30')][".$i."]//td[contains(@colspan,'2')]/following::td[1]/table[1]//td[2]";
+ $programs = $xpath->query($query);
+ foreach($programs as $program) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $hour = $xpath->query("parent::*/parent::*/parent::*/parent::*/td[1]", $program)->item(0);
+ $hour = str_replace("시", "", trim($hour->nodeValue));
+ $minute = $xpath->query("preceding-sibling::td[1]", $program)->item(0);
+ $minute = str_replace(array("[", "]"), array("",""), trim($minute->nodeValue));
+ $minute = substr($minute, -2);
+ $hour = $hour.":".$minute;
+ switch ($i) :
+ case 2 :
$hour = $hour." AM";
- $thisday = date("Ymd", strtotime($day." +1 days"));
- endif;
- break;
- endswitch;
- $startTime = date("YmdHis", strtotime($thisday." ".$hour));
- $programName = "";
- $subprogramName = "";
- $rating = 0;
- $episode = "";
- $rebroadcast = False;
- $pattern = '/?(.*)?(.*?)\s*(<(.*)>)?\s*(\(재\))?\s*(\(([\d,]+)회\))?()?\s*<\/td>/';
- preg_match($pattern, trim($dom->saveHTML($program)), $matches);
- if ($matches != NULL) :
- if(isset($matches[2])) $programName = trim($matches[2]) ?: "";
- if(isset($matches[4])) $subprogramName = trim($matches[4]) ?: "";
- if(isset($matches[5])) $rebroadcast = $matches[5] ? True : False;
- if(isset($matches[7])) $episode = $matches[7] ?: "";
- if(isset($matches[8])) :
- $image = $matches[8] ? $matches[8] : "";
- preg_match('/.*schedule_([\d,]+)?.*/', $image, $grade);
- if($grade != NULL) $rating = $grade[1];
+ break;
+ case 3 :
+ $hour = $hour." PM";
+ break;
+ case 4 :
+ if($hour > 5 && $hour < 12) :
+ $hour = $hour." PM";
+ elseif($hour <5 || $hour == 12) :
+ $hour = $hour." AM";
+ $thisday = date("Ymd", strtotime($day." +1 days"));
+ endif;
+ break;
+ endswitch;
+ $startTime = date("YmdHis", strtotime($thisday." ".$hour));
+ $pattern = '/^(.*?)\s*(?:<(.*)>)?\s*(?:\((재)\))?\s*(?:\(([\d,]+)회\)?)?$/';
+ $programName = trim($program->nodeValue);
+ preg_match($pattern, $programName, $matches);
+ if ($matches != NULL) :
+ if(isset($matches[1])) $programName = trim($matches[1]) ?: "";
+ if(isset($matches[2])) $subprogramName = trim($matches[2]) ?: "";
+ if(isset($matches[3])) $rebroadcast = $matches[3] ? True : False;
+ if(isset($matches[4])) $episode = $matches[4] ?: "";
endif;
- endif;
- //programName, startTime, rating, subprogramName, rebroadcast, episode
- $epginfo[] = array($programName, $startTime, $rating, $subprogramName, $rebroadcast, $episode );
- endforeach;
- endfor;
+ $images = $program->getElementsByTagName('img');
+ foreach($images as $image):
+ preg_match('/.*schedule_([\d,]+)?.*/', $image->getAttribute('src'), $grade);
+ if($grade != NULL) $rating = $grade[1];
+ endforeach;
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
+ endforeach;
+ endfor;
+ else:
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
endif;
} catch (Exception $e) {
if($GLOBALS['debug']) printError($e->getMessage());
}
endforeach;
- $zipped = array_slice(array_map(NULL, $epginfo, array_slice($epginfo,1)),0,-1);
- foreach($zipped as $epg) :
- $programName = $epg[0][0] ?: "";
- $subprogramName = $epg[0][3] ?: "";
- $startTime = $epg[0][1] ?: "";
- $endTime = $epg[1][1] ?: "";
- $desc = "";
- $actors = "";
- $producers = "";
- $category = "";
- $rebroadcast = $epg[0][4];
- $episode = $epg[0][5] ?: "";
- $rating = $epg[0][2] ?: 0;
- $programdata = array(
- 'channelId'=> $ChannelId,
- 'startTime' => $startTime,
- 'endTime' => $endTime,
- 'programName' => $programName,
- 'subprogramName'=> $subprogramName,
- 'desc' => $desc,
- 'actors' => $actors,
- 'producers' => $producers,
- 'category' => $category,
- 'episode' => $episode,
- 'rebroadcast' => $rebroadcast,
- 'rating' => $rating
- );
- writeProgram($programdata);
- endforeach;
+ epgzip($epginfo);
}
// Get EPG data from KT
@@ -537,59 +530,41 @@ function GetEPGFromKT($ChannelInfo) {
printError($ChannelName.HTTP_ERROR);
else :
$response = str_replace("charset=euc-kr", "charset=utf-8", $response);
+ $response = mb_convert_encoding($response, "UTF-8", "EUC-KR");
$dom = new DomDocument;
libxml_use_internal_errors(True);
- $dom->loadHTML(mb_convert_encoding($response, "UTF-8", "EUC-KR"));
- $xpath = new DomXPath($dom);
- $query = "//table[@id='pop_day']/tbody/tr";
- $rows = $xpath->query($query);
- foreach($rows as $row) :
- $cells = $row->getElementsByTagName('td');
- #programName, startTime, rating, category
- $startTime = date("YmdHis", strtotime($day." ".trim($cells->item(0)->nodeValue)));
- $rating = str_replace("all", 0, str_replace("세 이상", "", trim($cells->item(2)->nodeValue)));
- $epginfo[]= array(trim($cells->item(1)->nodeValue), $startTime, $rating, trim($cells->item(4)->nodeValue));
- endforeach;
+ if($dom->loadHTML($response)):
+ $xpath = new DomXPath($dom);
+ $query = "//table[@id='pop_day']/tbody/tr";
+ $rows = $xpath->query($query);
+ foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $cells = $row->getElementsByTagName('td');
+ //programName, startTime, rating, category
+ $startTime = date("YmdHis", strtotime($day." ".trim($cells->item(0)->nodeValue)));
+ $pattern = '/^(.*?)( <(.*)>)?$/';
+ $programName = trim($cells->item(1)->nodeValue);
+ preg_match($pattern, $programName, $matches);
+ if ($matches != NULL) :
+ if(isset($matches[1])) $programName = $matches[1] ?: "";
+ if(isset($matches[3])) $subprogramName = $matches[3] ?: "";
+ endif;
+ $category = trim($cells->item(4)->nodeValue);
+ $rating = str_replace("all", 0, str_replace("세 이상", "", trim($cells->item(2)->nodeValue)));
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
+ endforeach;
+ else :
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
endif;
} catch (Exception $e) {
if($GLOBALS['debug']) printError($e->getMessage());
}
endforeach;
- $zipped = array_slice(array_map(NULL, $epginfo, array_slice($epginfo,1)),0,-1);
- foreach($zipped as $epg) :
- $programName = "";
- $subprogramName = "";
- $pattern = '/^(.*?)( <(.*)>)?$/';
- preg_match($pattern, $epg[0][0], $matches);
- if ($matches != NULL) :
- if(isset($matches[1])) $programName = $matches[1] ?: "";
- if(isset($matches[3])) $subprogramName = $matches[3] ?: "";
- endif;
- $startTime = $epg[0][1] ?: "";
- $endTime = $epg[1][1] ?: "";
- $desc = "";
- $actors = "";
- $producers = "";
- $category = $epg[0][3] ?: "";
- $rebroadcast = False;
- $episode = "";
- $rating = $epg[0][2] ?: 0;
- $programdata = array(
- 'channelId'=> $ChannelId,
- 'startTime' => $startTime,
- 'endTime' => $endTime,
- 'programName' => $programName,
- 'subprogramName'=> $subprogramName,
- 'desc' => $desc,
- 'actors' => $actors,
- 'producers' => $producers,
- 'category' => $category,
- 'episode' => $episode,
- 'rebroadcast' => $rebroadcast,
- 'rating' => $rating
- );
- writeProgram($programdata);
- endforeach;
+ epgzip($epginfo);
}
// Get EPG data from LG
@@ -617,62 +592,40 @@ function GetEPGFromLG($ChannelInfo) {
libxml_use_internal_errors(True);
$response = mb_convert_encoding($response, "UTF-8", "EUC-KR");
$response = str_replace(array('<재>', ' [..', ' (..'), array('<재>', '', ''), $response);
- $dom->loadHTML($response);
- $xpath = new DomXPath($dom);
- $query = "//div[@class='tblType list']/table/tbody/tr";
- $rows = $xpath->query($query);
- foreach($rows as $row) :
- $cells = $row->getElementsByTagName('td');
- $programName = trim($cells->item(1)->childNodes->item(0)->nodeValue);
- $startTime = date("YmdHis", strtotime($day." ".trim($cells->item(0)->nodeValue)));
- $spans = $cells->item(1)->getElementsByTagName('span');
- $rating = trim($spans->item(1)->nodeValue)=="All" ? 0 : trim($spans->item(1)->nodeValue);
- //programName, startTime, rating, category
- $epginfo[]= array($programName, $startTime, $rating, trim($cells->item(2)->nodeValue));
- endforeach;
+ if($dom->loadHTML($response)):
+ $xpath = new DomXPath($dom);
+ $query = "//div[@class='tblType list']/table/tbody/tr";
+ $rows = $xpath->query($query);
+ foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $cells = $row->getElementsByTagName('td');
+ $startTime = date("YmdHis", strtotime($day." ".trim($cells->item(0)->nodeValue)));
+ $programName = trim($cells->item(1)->childNodes->item(0)->nodeValue);
+ $pattern = '/(<재>?)?(.*?)(?:\[(.*)\])?\s?(?:\(([\d,]+)회\))?$/';
+ preg_match($pattern, $programName, $matches);
+ if ($matches != NULL) :
+ if(isset($matches[2])) $programName = trim($matches[2]) ?: "";
+ if(isset($matches[3])) $subprogramName = trim($matches[3]) ?: "";
+ if(isset($matches[4])) $episode = trim($matches[4]) ?: "";
+ if(isset($matches[1])) $rebroadcast = trim($matches[1]) ? True: False;
+ endif;
+ $category = trim($cells->item(2)->nodeValue);
+ $spans = $cells->item(1)->getElementsByTagName('span');
+ $rating = trim($spans->item(1)->nodeValue)=="All" ? 0 : trim($spans->item(1)->nodeValue);
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
+ endforeach;
+ else :
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
endif;
} catch (Exception $e) {
if($GLOBALS['debug']) printError($e->getMessage());
}
endforeach;
- $zipped = array_slice(array_map(NULL, $epginfo, array_slice($epginfo,1)),0,-1);
- foreach($zipped as $epg) :
- $pattern = '/(<재>?)?(.*?)(\[(.*)\])?\s?(\(([\d,]+)회\))?$/';
- preg_match($pattern, $epg[0][0], $matches);
- $programName = "";
- $subprogramName = "";
- $episode = "";
- $rebroadcast = False;
- if ($matches != NULL) :
- if(isset($matches[2])) $programName = trim($matches[2]) ?: "";
- if(isset($matches[4])) $subprogramName = trim($matches[4]) ?: "";
- if(isset($matches[6])) $episode = trim($matches[6]) ?: "";
- if(isset($matches[1])) $rebroadcast = trim($matches[1]) ? True: False;
- endif;
- $startTime = $epg[0][1] ?: "";
- $endTime = $epg[1][1] ?: "";
- $desc = "";
- $actors = "";
- $producers = "";
- $category = $epg[0][3] ?: "";
-
- $rating = $epg[0][2] ?: 0;
- $programdata = array(
- 'channelId'=> $ChannelId,
- 'startTime' => $startTime,
- 'endTime' => $endTime,
- 'programName' => $programName,
- 'subprogramName'=> $subprogramName,
- 'desc' => $desc,
- 'actors' => $actors,
- 'producers' => $producers,
- 'category' => $category,
- 'episode' => $episode,
- 'rebroadcast' => $rebroadcast,
- 'rating' => $rating
- );
- writeProgram($programdata);
- endforeach;
+ epgzip($epginfo);
}
// Get EPG data from SK
@@ -704,10 +657,9 @@ function GetEPGFromSK($ChannelInfo) {
else :
$programs = $data['channel']['programs'];
foreach ($programs as $program) :
- $programName = "";
- $subprogramName = "";
- $episode = "";
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
$rebroadcast = False;
+ $rating = 0;
$pattern = '/^(.*?)(?:\s*[\(<]([\d,회]+)[\)>])?(?:\s*<([^<]*?)>)?(\((재)\))?$/';
preg_match($pattern, str_replace('...', '>', $program['programName']), $matches);
if ($matches != NULL) :
@@ -760,12 +712,11 @@ function GetEPGFromSKB($ChannelInfo) {
$ServiceId = $ChannelInfo[3];
$epginfo = array();
foreach(range(1, $GLOBALS['period']) as $k) :
- $url = "http://www.skbroadband.com/content/realtime/Channel_List.do";
+ $url = "http://m.skbroadband.com/content/realtime/Channel_List.do";
$day = date("Ymd", strtotime("+".($k - 1)." days"));
$params = array(
'key_depth2' => $ServiceId,
- 'key_depth3' => $day,
- 'tab_gubun' => 'lst'
+ 'key_depth3' => $day
);
$params = http_build_query($params);
$method = "POST";
@@ -778,61 +729,44 @@ function GetEPGFromSKB($ChannelInfo) {
$dom = new DomDocument;
libxml_use_internal_errors(True);
$response = mb_convert_encoding($response, "UTF-8", "EUC-KR");
- $dom->loadHTML($response);
- $xpath = new DomXPath($dom);
- $query = "//tr[@class='".$day."']";
- $rows = $xpath->query($query);
- foreach($rows as $row) :
- $cells = $row->getElementsByTagName('td');
- $pattern = '/^(.*?)(\(([\d,]+)회\))?(<(.*)>)?(\((재)\))?$/';
- preg_match($pattern, trim($cells->item(0)->nodeValue), $matches);
- if ($matches != NULL) :
- if(isset($matches[1])) $programName = trim($matches[1]) ?: "";
- if(isset($matches[5])) $subprogramName = trim($matches[5]) ?: "";
- if(isset($matches[3])) $episode = $matches[3] ?: "";
- if(isset($matches[7])) $rebroadcast = $matches[7] ? True : False;
- endif;
- preg_match('/.*\s*([\d,]+)\s*.*/', $cells->item(1)->nodeValue, $rating);
- $startTime = $row->getElementsByTagName('th')->item(0)->nodeValue;
- $startTime = date("YmdHis", strtotime($day." ".$startTime));
- $rating = $rating[1];
- //programName, startTime, rating, subprogramName, rebroadcast, episode
- $epginfo[]= array($programName, $startTime, $rating, $subprogramName, $rebroadcast, $episode);
- endforeach;
+ if($dom->loadHTML($response)):
+ $xpath = new DomXPath($dom);
+ $query = "//span[@class='caption' or @class='explan' or @class='fullHD' or @class='UHD' or @class='nowon']";
+ $spans = $xpath->query($query);
+ foreach($spans as $span) :
+ $span->parentNode->removeChild( $span);
+ endforeach;
+ $query = "//div[@id='dawn']/ul/li";
+ $rows = $xpath->query($query);
+ foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $cells = $row->getElementsByTagName('span');
+ $startTime = $cells->item(0)->nodeValue ?: "";
+ $startTime = date("YmdHis", strtotime($day." ".$startTime));
+ $programName = trim($cells->item(2)->nodeValue) ?: "";
+ $pattern = '/^(.*?)(\(([\d,]+)회\))?(<(.*)>)?(\((재)\))?$/';
+ preg_match($pattern, $programName, $matches);
+ if ($matches != NULL) :
+ if(isset($matches[1])) $programName = trim($matches[1]) ?: "";
+ if(isset($matches[5])) $subprogramName = trim($matches[5]) ?: "";
+ if(isset($matches[3])) $episode = $matches[3] ?: "";
+ if(isset($matches[7])) $rebroadcast = $matches[7] ? True : False;
+ endif;
+ if($cells->length > 3) $rating = str_replace('세', '', $cells->item(3)->nodeValue) ?: 0;
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
+ endforeach;
+ epgzip($epginfo);
+ else :
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
endif;
} catch (Exception $e) {
if($GLOBALS['debug']) printError($e->getMessage());
}
endforeach;
- $zipped = array_slice(array_map(NULL, $epginfo, array_slice($epginfo,1)),0,-1);
- foreach($zipped as $epg) :
- $programName = trim($epg[0][0]) ?: "";
- $subprogramName = trim($epg[0][3]) ?: "";
- $episode = $epg[0][5] ?: "";
- $rebroadcast = $epg[0][4] ? True: False;
- $startTime = $epg[0][1] ?: "";
- $endTime = $epg[1][1] ?: "";
- $desc = "";
- $actors = "";
- $producers = "";
- $category = "";
- $rating = $epg[0][2] ?: 0;
- $programdata = array(
- 'channelId'=> $ChannelId,
- 'startTime' => $startTime,
- 'endTime' => $endTime,
- 'programName' => $programName,
- 'subprogramName'=> $subprogramName,
- 'desc' => $desc,
- 'actors' => $actors,
- 'producers' => $producers,
- 'category' => $category,
- 'episode' => $episode,
- 'rebroadcast' => $rebroadcast,
- 'rating' => $rating
- );
- writeProgram($programdata);
- endforeach;
}
// Get EPG data from SKY
@@ -866,6 +800,9 @@ function GetEPGFromSKY($ChannelInfo) {
else :
$programs = $data['scheduleListIn'];
foreach($programs as $program) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
$programName = htmlspecialchars_decode($program['program_name']) ?: "";
$subprogramName = str_replace(array('lt;', 'gt;', 'amp;'), array('<', '>', '&'),$program['program_subname']) ?: "";
$startTime = $program['starttime'];
@@ -875,8 +812,12 @@ function GetEPGFromSKY($ChannelInfo) {
$description = str_replace(array('lt;', 'gt;', 'amp;'), array('<', '>', '&'),$program['description']) ?: "";
$summary = str_replace(array('lt;', 'gt;', 'amp;'), array('<', '>', '&'),$program['summary']) ?: "";
$desc = $description ?: "";
- if($summary) :
- $desc = $desc."\n".$summary;
+ if($desc) :
+ if($summary):
+ $desc = $desc."\n".$summary;
+ endif;
+ else :
+ $desc = $summary;
endif;
$category = $program['program_category1'];
$episode = $program['episode_id'] ?: "";
@@ -956,41 +897,20 @@ function GetEPGFromNaver($ChannelInfo) {
for($i = 0; $i < count($data['displayDates']); $i++) :
for($j = 0; $j < 24; $j++) :
foreach($data['schedules'][$j][$i] as $program) :
- //programName, startTime, episode, rebroadcast, rating
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
$startTime = date("YmdHis", strtotime($data['displayDates'][$i]['date']." ".$program['startTime']));
- $epginfo[] = array($program['title'], $startTime, str_replace("회","", $program['episode']), $program['isRerun'], $program['grade']);
+ $programName = htmlspecialchars_decode(trim($program['title']), ENT_XML1);
+ $episode = str_replace("회","", $program['episode']);
+ $rebroadcast = $program['isRerun'] ? True : False;
+ $rating = $program['grade'];
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
endforeach;
endfor;
endfor;
- $zipped = array_slice(array_map(NULL, $epginfo, array_slice($epginfo,1)),0,-1);
- foreach($zipped as $epg) :
- $programName = htmlspecialchars_decode($epg[0][0], ENT_XML1) ?: "";
- $subprogramName = "";
- $startTime = $epg[0][1] ?: "";
- $endTime = $epg[1][1] ?: "";
- $desc = "";
- $actors = "";
- $producers = "";
- $category = "";
- $rebroadcast = $epg[0][3] ? True: False;
- $episode = $epg[0][2] ?: "";
- $rating = $epg[0][4] ?: 0;
- $programdata = array(
- 'channelId'=> $ChannelId,
- 'startTime' => $startTime,
- 'endTime' => $endTime,
- 'programName' => $programName,
- 'subprogramName'=> $subprogramName,
- 'desc' => $desc,
- 'actors' => $actors,
- 'producers' => $producers,
- 'category' => $category,
- 'episode' => $episode,
- 'rebroadcast' => $rebroadcast,
- 'rating' => $rating
- );
- writeProgram($programdata);
- endforeach;
+ epgzip($epginfo);
endif;
} catch(Exception $e) {
if($GLOBALS['debug']) printError($e->getMessage());
@@ -1001,19 +921,218 @@ function GetEPGFromNaver($ChannelInfo) {
}
}
-// Get EPG data from Tbroad
-function GetEPGFromTbroad($ChannelInfo) {
- $url='https://www.tbroad.com/chplan/selectRealTimeListForNormal.tb';
-}
-
// Get EPG data from Iscs
function GetEPGFromIscs($ChannelInfo) {
- $url='http://service.iscs.co.kr/sub/channel_view.asp';
+ $ChannelId = $ChannelInfo[0];
+ $ChannelName = $ChannelInfo[1];
+ $ServiceId = $ChannelInfo[3];
+ $epginfo = array();
+ foreach(range(1, $GLOBALS['period']) as $k) :
+ $url = "https://www.iscs.co.kr/service/sub/ajax_channel_view.asp";
+ $day = date("Y-m-d", strtotime("+".($k - 1)." days"));
+ $params = array(
+ 's_idx' => $ServiceId,
+ 'C_date' => $day
+ );
+ $params = http_build_query($params);
+ $method = "POST";
+ try {
+ $response = getWeb($url, $params, $method);
+ if ($response === False && $GLOBALS['debug']) :
+ printError($ChannelName.HTTP_ERROR);
+ else :
+ try {
+ $data = json_decode($response, TRUE);
+ if(json_last_error() != JSON_ERROR_NONE) throw new Exception(JSON_SYNTAX_ERROR);
+ if(count($data['html']) == 0) :
+ if($GLOBALS['debug']) :
+ printError($ChannelName.CHANNEL_ERROR);
+ endif;
+ else :
+ $html = $data['html'];
+ $pattern = '/| (.*)<\/td>/';
+ $html = preg_replace_callback($pattern, function($matches) { return ' | '.htmlspecialchars($matches[1]).' | ';}, $html);
+ $dom = new DomDocument;
+ libxml_use_internal_errors(True);
+ $html = mb_convert_encoding($html, "HTML-ENTITIES", "UTF-8");
+ if($dom->loadHTML($html)):
+ $xpath = new DomXPath($dom);
+ $query = "//div[@class='pp_tbl']/table/tbody/tr";
+ $rows = $xpath->query($query);
+ foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $cells = $row->getElementsByTagName('td');
+ $startTime = $cells->item(0)->nodeValue ?: "";
+ $startTime = date("YmdHis", strtotime($day." ".$startTime));
+ $programName = trim($cells->item(1)->nodeValue) ?: "";
+ $pattern = '/^(.*?)(?:\(([\d,]+)회\))?(?:\((재)\))?$/';
+ preg_match($pattern, $programName, $matches);
+ if ($matches != NULL) :
+ if(isset($matches[1])) $programName = trim($matches[1]) ?: "";
+ if(isset($matches[2])) $episode = $matches[2] ?: "";
+ if(isset($matches[3])) $rebroadcast = $matches[3] ? True : False;
+ endif;
+ $rating = $cells->item(2)->nodeValue=='전체관람' ? 0 : str_replace('세이상', '', $cells->item(2)->nodeValue);
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
+ endforeach;
+ epgzip($epginfo);
+ else :
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
+ endif;
+ } catch(Exception $e) {
+ if($GLOBALS['debug']) printError($e->getMessage());
+ }
+ endif;
+ } catch (Exception $e) {
+ if($GLOBALS['debug']) printError($e->getMessage());
+ }
+ endforeach;
+}
+
+// Get EPG data from Hcn
+function GetEPGFromHcn($ChannelInfo) {
+ $ChannelId = $ChannelInfo[0];
+ $ChannelName = $ChannelInfo[1];
+ $ServiceId = $ChannelInfo[3];
+ $epginfo = array();
+ foreach(range(1, $GLOBALS['period']) as $k) :
+ $url = "https://www.hcn.co.kr/ur/bs/ch/channelInfo.hcn";
+ $day = date("Y-m-d", strtotime("+".($k - 1)." days"));
+ $params = array(
+ 'method' => 'ajax_00',
+ 'pageType' => 'sheetList',
+ 'ch_id' => $ServiceId,
+ 'onairdate' => $day
+ );
+ $params = http_build_query($params);
+ $method = "POST";
+ try {
+ $response = getWeb($url, $params, $method);
+ if ($response === False && $GLOBALS['debug']) :
+ printError($ChannelName.HTTP_ERROR);
+ else :
+ $dom = new DomDocument;
+ libxml_use_internal_errors(True);
+ $response = mb_convert_encoding($response, "HTML-ENTITIES", "EUC-KR");
+ if($dom->loadHTML($response)):
+ $xpath = new DomXPath($dom);
+ $query = "//tr[@class='']";
+ $rows = $xpath->query($query);
+ foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $cells = $row->getElementsByTagName('td');
+ $startTime = $cells->item(0)->nodeValue ?: "";
+ $startTime = date("YmdHis", strtotime($day." ".$startTime));
+ $programName = trim($cells->item(1)->nodeValue) ?: "";
+ $images = $row->getElementsByTagName('img');
+ foreach($images as $image):
+ preg_match('/re\.png/', $image->getAttribute('src'), $rebroadcast);
+ if($rebroadcast != NULL) $rebroadcast = True;
+ preg_match('/.*plus([\d,]+)\.png/', $image->getAttribute('src'), $grade);
+ if($grade != NULL) $rating = $grade[1];
+ endforeach;
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
+ endforeach;
+ epgzip($epginfo);
+ else :
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
+ endif;
+ } catch (Exception $e) {
+ if($GLOBALS['debug']) printError($e->getMessage());
+ }
+ endforeach;
+}
+
+// Get EPG data from POOQ
+function GetEPGFromPooq($ChannelInfo) {
+ $ChannelId = $ChannelInfo[0];
+ $ChannelName = $ChannelInfo[1];
+ $ServiceId = $ChannelInfo[3];
+ $today = date("Ymd");
+ $lastday = date("Ymd", strtotime("+".($GLOBALS['period'] - 1)." days"));
+ $url = "https://wapie.pooq.co.kr/v1/epgs30/".$ServiceId."/";
$params = array(
- 'chan_idx'=>'242',
- 'source_id'=>'203',
- 'Chan_Date'=>'2017-04-18'
+ 'deviceTypeId'=> 'pc',
+ 'marketTypeId'=> 'generic',
+ 'apiAccessCredential'=> 'EEBE901F80B3A4C4E5322D58110BE95C',
+ 'offset'=> '0',
+ 'limit'=> '1000',
+ 'startTime'=> date("Y/m/d", strtotime($today)).' 00:00',
+ 'endTime'=> date("Y/m/d", strtotime($lastday)).' 00:00'
);
+ foreach(range(1, $GLOBALS['period']) as $k) :
+ $day = date("Y-m-d", strtotime("+".($k - 1)." days"));
+ $date_list[] = $day;
+ endforeach;
+ $params = http_build_query($params);
+ $method = "GET";
+ try {
+ $response = getWeb($url, $params, $method);
+ if ($response === False && $GLOBALS['debug']) :
+ printError($ChannelName.HTTP_ERROR);
+ else :
+ try {
+ $data = json_decode($response, TRUE);
+ if(json_last_error() != JSON_ERROR_NONE) throw new Exception(JSON_SYNTAX_ERROR);
+ if($data['result']['count'] == 0) :
+ if($GLOBALS['debug']) :
+ printError($ChannelName.CHANNEL_ERROR);
+ endif;
+ else :
+ $programs = $data['result']['list'];
+ foreach ($programs as $program) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ if(in_array($program['startDate'] , $date_list)) :
+ $startTime = $program['startDate']." ".$program['startTime'];
+ $startTime = date("YmdHis", strtotime($startTime));
+ $endTime = $program['startDate']." ".$program['endTime'];
+ $endTime = date("YmdHis", strtotime($endTime));
+ $pattern = '/^(.*?)(?:([\d,]+)회)?(?:\((재)\))?$/';
+ $programName = $program['programTitle'];
+ preg_match($pattern, $programName, $matches);
+ if($matches !== NULL) :
+ if(isset($matches[1])) $programName = trim($matches[1]) ?: "";
+ if(isset($matches[2])) $episode = trim($matches[2]) ?: "";
+ if(isset($matches[3])) $rebroadcast = $matches[3] ? True : False;
+ endif;
+ if($program['programStaring']) $actors = trim($program['programStaring'], ',');
+ if($program['programSummary']) $desc = trim($program['programSummary']);
+ $rating = $program['age'];
+ $programdata = array(
+ 'channelId'=> $ChannelId,
+ 'startTime' => $startTime,
+ 'endTime' => $endTime,
+ 'programName' => $programName,
+ 'subprogramName'=> $subprogramName,
+ 'desc' => $desc,
+ 'actors' => $actors,
+ 'producers' => $producers,
+ 'category' => $category,
+ 'episode' => $episode,
+ 'rebroadcast' => $rebroadcast,
+ 'rating' => $rating
+ );
+ writeProgram($programdata);
+ endif;
+ endforeach;
+ endif;
+ } catch(Exception $e) {
+ if($GLOBALS['debug']) printError($e->getMessage());
+ }
+ endif;
+ } catch (Exception $e) {
+ if($GLOBALS['debug']) printError($e->getMessage());
+ }
}
// Get EPG data from MBC
@@ -1046,24 +1165,19 @@ function GetEPGFromMbc($ChannelInfo) {
$programs = $data['Programs'];
foreach($programs as $program) :
if($program['Channel'] == "CHAM" && $program['LiveDays'] == $dayofweek[date("w", strtotime($day))]) :
- $programName = "";
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
$rebroadcast = False;
+ $rating = 0;
$pattern = '/^(.*?)(\(재\))?$/';
preg_match($pattern, htmlspecialchars_decode($program['ProgramTitle']), $matches);
if ($matches != NULL) :
$programName = $matches[1];
- $rebroadcast = $matches[2] ? True : False;
+ if(isset($matches[2])) $rebroadcast = $matches[2] ? True : False;
endif;
- $subprogramName = "";
$startTime = $day." ".$program['StartTime'];
$startTime = date("YmdHis", strtotime($startTime));
$endTime = date("YmdHis", strtotime("+".$program['RunningTime']." minutes", strtotime($startTime)));
- $desc = "";
- $actors = "";
- $producers = "";
$category = "음악";
- $episode = "";
- $rating = 0;
$programdata = array(
'channelId'=> $ChannelId,
'startTime' => $startTime,
@@ -1120,25 +1234,22 @@ function GetEPGFromMil($ChannelInfo) {
else :
$programs = $data['resultList'];
foreach($programs as $program) :
- $programName = "";
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
$rebroadcast = False;
+ $rating = 0;
$pattern = '/^(.*?)(\(재\))?$/';
preg_match($pattern, htmlspecialchars_decode($program['program_title']), $matches);
if ($matches != NULL) :
$programName = $matches[1];
- $rebroadcast = $matches[2] ? True : False;
+ if(isset($matches[2])) $rebroadcast = $matches[2] ? True : False;
endif;
$subprogramName = htmlspecialchars_decode($program['program_subtitle']);
$startTime = $day." ".$program['program_time'];
$startTime = date("YmdHis", strtotime($startTime));
$endTime = $day." ".$program['program_end_time'];
$endTime = date("YmdHis", strtotime($endTime));
- $desc = "";
$actors = htmlspecialchars_decode($program['movie_actor']);
$producers = htmlspecialchars_decode($program['movie_director']);
- $category = "";
- $episode = "";
- $rating = 0;
$programdata = array(
'channelId'=> $ChannelId,
'startTime' => $startTime,
@@ -1196,19 +1307,16 @@ function GetEPGFromIfm($ChannelInfo) {
else :
$programs = $data['hybMusicInfoList'];
foreach($programs as $program) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
$programName = htmlspecialchars_decode($program['pgmTitle']) ?: "";
- $subprogramName = "";
$startTime = $day." ".$program['pgmStime'];
$startTime = date("YmdHis", strtotime($startTime));
$endTime = $day." ".$program['pgmEtime'];
$endTime = date("YmdHis", strtotime($endTime));
- $desc = "";
$actors = htmlspecialchars_decode($program['pgmDj']);
$producers = htmlspecialchars_decode($program['pgmPd']);
- $category = "";
- $episode = "";
- $rebroadcast = False;
- $rating = 0;
$programdata = array(
'channelId'=> $ChannelId,
'startTime' => $startTime,
@@ -1270,13 +1378,18 @@ function GetEPGFromKbs($ChannelInfo) {
$query = "//li";
$rows = $xpath->query($query);
foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
$cells = $row->getElementsByTagName('span');
+ $startTime = $day." ".trim($cells->item(0)->childNodes->item(0)->nodeValue);
+ $startTime = date("YmdHis", strtotime($startTime));
$programName = trim($cells->item(2)->childNodes->item(0)->nodeValue);
$programName = str_replace(array("[","]", " Broadcast"), array("", "", ""), $programName);
- $startTime = $day." ".trim($cells->item(0)->childNodes->item(0)->nodeValue);
- //programName, startTime, rating, category
- $epginfo[]= array($programName, $startTime);
+ //ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
+ $epginfo[] = array($ChannelId, $startTime, $programName, $subprogramName, $desc, $actors, $producers, $category, $episode, $rebroadcast, $rating);
endforeach;
+ epgzip($epginfo);
endif;
} catch(Exception $e) {
if($GLOBALS['debug']) printError($e->getMessage());
@@ -1286,21 +1399,97 @@ function GetEPGFromKbs($ChannelInfo) {
if($GLOBALS['debug']) printError($e->getMessage());
}
endforeach;
+}
+
+function GetEPGFromArirang($ChannelInfo) {
+ $ChannelId = $ChannelInfo[0];
+ $ChannelName = $ChannelInfo[1];
+ $ServiceId = $ChannelInfo[3];
+ $epginfo = array();
+ foreach(range(1, $GLOBALS['period']) as $k) :
+ $url = "http://www.arirang.com/Radio/Radio_Index.asp";
+ $day = date("Y-m-d", strtotime("+".($k - 1)." days"));
+ $params = array();
+ $params = http_build_query($params);
+ $method = "GET";
+ try {
+ $response = getWeb($url, $params, $method);
+ if ($response === False && $GLOBALS['debug']) :
+ printError($ChannelName.HTTP_ERROR);
+ else :
+ $dom = new DomDocument;
+ libxml_use_internal_errors(True);
+ //echo $response;
+
+ $response = mb_convert_encoding($response, "HTML-ENTITIES", "EUC-KR");
+ if($dom->loadHTML($response)):
+ $xpath = new DomXPath($dom);
+ $dayofweek = date("w", strtotime($day));
+ if($dayofweek == 0):
+ $query = "//table[@id='aIRSW_sun']/tr";
+ elseif($dayofweek == 6):
+ $query = "//table[@id='aIRSW_sat']/tr";
+ else :
+ $query = "//table[@id='aIRSW_week']/tr";
+ endif;
+ $rows = $xpath->query($query);
+ foreach($rows as $row) :
+ $startTime = $endTime = $programName = $subprogramName = $desc = $actors = $producers = $category = $episode = "";
+ $rebroadcast = False;
+ $rating = 0;
+ $time = $row->getElementsByTagName('th');
+ $times = explode('~', trim($time->item(0)->nodeValue));
+ $startTime = date("YmdHis", strtotime($day." ".$times[0]));
+ $endTime = date("YmdHis", strtotime($day." ".$times[1]));
+ $program = $row->getElementsByTagName('td');
+ $pattern = '/^(.*?)(?:\((Re)\))?$/';
+ preg_match($pattern, trim($program->item(0)->nodeValue), $matches);
+ if ($matches != NULL) :
+ $programName = $matches[1];
+ if(isset($matches[2])) $rebroadcast = $matches[2] ? True : False;
+ endif;
+ $programdata = array(
+ 'channelId'=> $ChannelId,
+ 'startTime' => $startTime,
+ 'endTime' => $endTime,
+ 'programName' => $programName,
+ 'subprogramName'=> $subprogramName,
+ 'desc' => $desc,
+ 'actors' => $actors,
+ 'producers' => $producers,
+ 'category' => $category,
+ 'episode' => $episode,
+ 'rebroadcast' => $rebroadcast,
+ 'rating' => $rating
+ );
+ writeProgram($programdata);
+ endforeach;
+ else :
+ if($GLOBALS['debug']) printError($ChannelName.CONTENT_ERROR);
+ endif;
+ endif;
+ } catch (Exception $e) {
+ if($GLOBALS['debug']) printError($e->getMessage());
+ }
+ endforeach;
+}
+# Zip epginfo
+function epgzip($epginfo) {
+ #ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
$zipped = array_slice(array_map(NULL, $epginfo, array_slice($epginfo,1)),0,-1);
foreach($zipped as $epg) :
- $programName = $epg[0][0] ?: "";
- $subprogramName = "";
+ $ChannelId = $epg[0][0] ?: "";
$startTime = $epg[0][1] ?: "";
- $startTime = date("YmdHis", strtotime($startTime));
$endTime = $epg[1][1] ?: "";
- $endTime = date("YmdHis", strtotime($endTime));
- $desc = "";
- $actors = "";
- $producers = "";
- $category = "";
- $rebroadcast = False;
- $episode = "";
- $rating = 0;
+ $programName = $epg[0][2] ?: "";
+ $subprogramName = $epg[0][3] ?: "";
+ $desc = $epg[0][4] ?: "";
+ $actors = $epg[0][5] ?: "";
+ $producers = $epg[0][6] ?: "";
+ $category = $epg[0][7] ?: "";
+ $episode = $epg[0][8] ?: "";
+ $rebroadcast = $rebroadcast = $epg[0][9] ? True: False;
+ $rating = $epg[0][10] ?: 0;
$programdata = array(
'channelId'=> $ChannelId,
'startTime' => $startTime,
@@ -1318,7 +1507,6 @@ function GetEPGFromKbs($ChannelInfo) {
writeProgram($programdata);
endforeach;
}
-
function writeProgram($programdata) {
$fp = $GLOBALS['fp'];
$ChannelId = $programdata['channelId'];
@@ -1349,6 +1537,7 @@ function writeProgram($programdata) {
if($GLOBALS['addverbose'] == 'y') :
$desc = htmlspecialchars($programdata['programName'], ENT_XML1);
if($subprogramName) $desc = $desc."\n부제 : ".$subprogramName;
+ if($rebroadcast == True && $GLOBALS['addrebroadcast'] == 'y') $desc = $desc."\n방송 : 재방송";
if($episode) $desc = $desc."\n회차 : ".$episode."회";
if($category) $desc = $desc."\n장르 : ".$category;
if($actors) $desc = $desc."\n출연 : ".$actors;
@@ -1358,6 +1547,7 @@ function writeProgram($programdata) {
$desc = "";
endif;
if($programdata['desc']) $desc = $desc."\n".htmlspecialchars($programdata['desc'], ENT_XML1);
+ $desc = preg_replace('/ +/', ' ', $desc);
$contentTypeDict = array(
'교양' => 'Arts / Culture (without music)',
'만화' => 'Cartoons / Puppets',
@@ -1422,10 +1612,10 @@ function getWeb($url, $params, $method) {
curl_setopt ($ch, CURLOPT_POSTFIELDS, $params);
endif;
curl_setopt($ch, CURLOPT_URL, $url);
- curl_setopt($ch, CURLOPT_RETURNTRANSFER,True);
- curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 3);
+ curl_setopt($ch, CURLOPT_RETURNTRANSFER, True);
+ curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $GLOBALS['timeout']);
curl_setopt($ch, CURLOPT_HEADER, False);
- curl_setopt($ch, CURLOPT_FAILONERROR,True);
+ curl_setopt($ch, CURLOPT_FAILONERROR, True);
curl_setopt($ch, CURLOPT_USERAGENT, $GLOBALS['ua']);
$response = curl_exec($ch);
if(curl_error($ch) && $GLOBALS['debug']) printError($url." ".curl_error($ch));
@@ -1439,5 +1629,4 @@ function printLog($string) {
function printError($string) {
fwrite(STDERR, "Error : ".$string."\n");
}
-?>
-
+?>
\ No newline at end of file
diff --git a/epg2xml.py b/epg2xml.py
index 6bb40b8..300af80 100644
--- a/epg2xml.py
+++ b/epg2xml.py
@@ -13,6 +13,8 @@ import re
from xml.sax.saxutils import escape, unescape
import argparse
import pprint
+from functools import partial
+
try:
imp.find_module('bs4')
from bs4 import BeautifulSoup, SoupStrainer
@@ -21,6 +23,7 @@ except ImportError:
sys.exit()
try:
imp.find_module('lxml')
+ from lxml import html
except ImportError:
print("Error : ", "lxml 모듈이 설치되지 않았습니다.", file=sys.stderr)
sys.exit()
@@ -30,6 +33,7 @@ try:
except ImportError:
print("Error : ", "requests 모듈이 설치되지 않았습니다.", file=sys.stderr)
sys.exit()
+
reload(sys)
sys.setdefaultencoding('utf-8')
@@ -158,6 +162,8 @@ def GetEPGFromEPG(ChannelInfo):
response.raise_for_status()
html_data = response.content
data = unicode(html_data, 'euc-kr', 'ignore').encode('utf-8', 'ignore')
+ pattern = '(.*)<\/td>'
+ data = re.sub(pattern, partial(replacement,txt='title'), data)
strainer = SoupStrainer('table', {'style':'margin-bottom:30'})
soup = BeautifulSoup(data, 'lxml', parse_only=strainer, from_encoding='utf-8')
html = soup.find_all('table', {'style':'margin-bottom:30'})
@@ -174,6 +180,8 @@ def GetEPGFromEPG(ChannelInfo):
hour = 'AM ' + str(hour)
thisday = day + datetime.timedelta(days=1)
for celldata in cell.parent.find_all('tr'):
+ celldata = str(celldata).replace('</b></a>','').replace('</b>','')
+ celldata = celldata.replace('<img','').replace('align=absmiddle>','>').replace('</a>','')
startTime = endTime = programName = subprogramName = desc = actors = producers = category = episode = ''
rebroadcast = False
rating = 0
@@ -188,19 +196,19 @@ def GetEPGFromEPG(ChannelInfo):
grade = re.match('.*schedule_([\d,]+)?.*',image)
if not (grade is None): rating = int(grade.group(1))
programName = matches.group(2).strip() if matches.group(2) else ''
+ programName = unescape(programName)
subprogramName = matches.group(4).strip() if matches.group(4) else ''
+ subprogramName = unescape(subprogramName)
rebroadcast = True if matches.group(5) else False;
episode = matches.group(7) if matches.group(7) else ''
#ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
epginfo.append([ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating])
- epgzip(epginfo)
else:
if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
-
except (requests.exceptions.RequestException) as e:
if(debug): printError(ChannelName + str(e))
- else: pass
+ epgzip(epginfo)
# Get EPG data from KT
def GetEPGFromKT(ChannelInfo):
@@ -239,13 +247,13 @@ def GetEPGFromKT(ChannelInfo):
if not(matches is None): rating = int(matches.group())
#ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
epginfo.append([ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating])
- epgzip(epginfo)
else:
if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except (requests.exceptions.RequestException) as e:
if(debug): printError(ChannelName + str(e))
else: pass
+ epgzip(epginfo)
# Get EPG data from LG
def GetEPGFromLG(ChannelInfo):
@@ -277,23 +285,23 @@ def GetEPGFromLG(ChannelInfo):
startTime = startTime.strftime('%Y%m%d%H%M%S')
rating = 0 if cell[1].find('span', {'class': 'tag cte_all'}).text.strip()=="All" else int(cell[1].find('span', {'class': 'tag cte_all'}).text.strip())
cell[1].find('span', {'class': 'tagGroup'}).decompose()
- pattern = '(<재>?)?(.*?)(\[(.*)\])?\s?(\(([\d,]+)회\))?$'
+ pattern = '(<재>?)?(.*?)(?;\[(.*)\])?\s?(?:\(([\d,]+)회\))?$'
matches = re.match(pattern, cell[1].text.strip().decode('string_escape'))
if not (matches is None):
programName = matches.group(2).strip() if matches.group(2) else ''
- subprogramName = matches.group(4).strip() if matches.group(4) else ''
- episode = matches.group(6) if matches.group(6) else ''
+ subprogramName = matches.group(3).strip() if matches.group(3) else ''
+ episode = matches.group(4) if matches.group(4) else ''
rebroadcast = True if matches.group(1) else False
category = cell[2].text.strip()
#ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating
epginfo.append([ChannelId, startTime, programName, subprogramName, desc, actors, producers, category, episode, rebroadcast, rating])
- epgzip(epginfo)
else:
if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except (requests.exceptions.RequestException) as e:
if(debug): printError(ChannelName + str(e))
else: pass
+ epgzip(epginfo)
# Get EPG data from SK
def GetEPGFromSK(ChannelInfo):
@@ -371,7 +379,8 @@ def GetEPGFromSKB(ChannelInfo):
startTime = str(day) + ' ' + row.find('span', {'class':'time'}).text
startTime = datetime.datetime.strptime(startTime, '%Y-%m-%d %H:%M')
startTime = startTime.strftime('%Y%m%d%H%M%S')
- row.find('span', {'class':['fullHD', 'UHD']}).decompose()
+ if row.find('span', {'class':['caption', 'explan', 'fullHD', 'UHD', 'nowon']}) :
+ row.find('span', {'class':['caption', 'explan', 'fullHD', 'UHD', 'nowon']}).decompose()
cell = row.find('span', {'class':None}).text.decode('string_escape').strip()
pattern = "^(.*?)(\(([\d,]+)회\))?(<(.*)>)?(\((재)\))?$"
matches = re.match(pattern, cell)
@@ -491,7 +500,7 @@ def GetEPGFromIscs(ChannelInfo):
ChannelName = ChannelInfo[1]
ServiceId = ChannelInfo[3]
epginfo = []
- url='https://www.iscs.co.kr/service/sub/ajax_channel_view.asp'
+ url='http://www.iscs.co.kr/service/sub/ajax_channel_view.asp'
for k in range(period):
day = today + datetime.timedelta(days=k)
params = {'s_idx': ServiceId, 'C_date': day}
@@ -500,6 +509,8 @@ def GetEPGFromIscs(ChannelInfo):
json_data = response.text
try:
data = json.loads(json_data, encoding='utf-8')
+ pattern = '(.*)<\/td>'
+ data['html'] = re.sub(pattern, partial(replacement, txt='name'), data['html'])
strainer = SoupStrainer('tbody')
soup = BeautifulSoup(data['html'], 'lxml', parse_only=strainer)
html = soup.find_all('tr') if soup.find_all('tr') else ''
@@ -578,8 +589,8 @@ def GetEPGFromPooq(ChannelInfo):
ChannelId = ChannelInfo[0]
ChannelName = ChannelInfo[1]
ServiceId = ChannelInfo[3]
- url = 'https://wapie.pooq.co.kr/v1/epgs30/' + str(ServiceId) + '/'
lastday = today + datetime.timedelta(days=period-1)
+ url = 'https://wapie.pooq.co.kr/v1/epgs30/' + str(ServiceId) + '/'
params = {'deviceTypeId': 'pc', 'marketTypeId': 'generic', 'apiAccessCredential': 'EEBE901F80B3A4C4E5322D58110BE95C', 'offset': '0', 'limit': '1000', 'startTime': today.strftime('%Y/%m/%d') + ' 00:00', 'endTime': lastday.strftime('%Y/%m/%d') + ' 00:00'}
date_list = [(today + datetime.timedelta(days=x)).strftime('%Y-%m-%d') for x in range(0, period)]
try:
@@ -807,7 +818,6 @@ def GetEPGFromArirang(ChannelInfo):
response = requests.get(url, params=params, headers=ua, timeout=timeout)
response.raise_for_status()
data = response.content
- #data = html_data.encode('utf-8', 'ignore')
if day.weekday() < 5 :
strainer = SoupStrainer('table', {'id':'aIRSW_week'})
elif day.weekday() == 5:
@@ -876,7 +886,6 @@ def writeProgram(programdata):
endTime = programdata['endTime']
programName = escape(programdata['programName']).strip()
subprogramName = escape(programdata['subprogramName']).strip()
-
matches = re.match('(.*) \(?(\d+부)\)?', unescape(programName.encode('utf-8', 'ignore')))
if not(matches is None):
programName = escape(matches.group(1));
@@ -884,7 +893,6 @@ def writeProgram(programdata):
subprogramName = subprogramName.strip()
if programName is None:
programName = subprogramName
-
actors = escape(programdata['actors'])
producers = escape(programdata['producers'])
category = escape(programdata['category'])
@@ -899,7 +907,7 @@ def writeProgram(programdata):
if addverbose == 'y':
desc = escape(programdata['programName'])
if subprogramName : desc = desc + '\n부제 : ' + subprogramName
- if rebroadcast == True and addrebroadcast == 'y' : desc = desc + '\n방송 : 재방송'
+ if rebroadcast == True and addrebroadcast == 'y' : desc = desc + '\n방송 : 재방송'
if episode : desc = desc + '\n회차 : ' + str(episode) + '회'
if category : desc = desc + '\n장르 : ' + category
if actors : desc = desc + '\n출연 : ' + actors
@@ -908,6 +916,8 @@ def writeProgram(programdata):
else:
desc =''
if programdata['desc'] : desc = desc + '\n' + escape(programdata['desc'])
+ desc = re.sub(' +',' ', desc)
+ #desc = re.sub('\s+','\s', desc)
contentTypeDict={'교양':'Arts / Culture (without music)', '만화':'Cartoons / Puppets', '교육':'Education / Science / Factual topics', '취미':'Leisure hobbies', '드라마':'Movie / Drama', '영화':'Movie / Drama', '음악':'Music / Ballet / Dance', '뉴스':'News / Current affairs', '다큐':'Documentary', '라이프':'Documentary', '시사/다큐':'Documentary', '연예':'Show / Game show', '스포츠':'Sports', '홈쇼핑':'Advertisement / Shopping'}
contentType = ''
for key, value in contentTypeDict.iteritems():
@@ -918,7 +928,6 @@ def writeProgram(programdata):
if subprogramName :
print(' %s' % (subprogramName))
if addverbose=='y' :
- desc = re.sub(' +',' ', desc)
print(' %s' % (desc))
if actors or producers:
print(' ')
@@ -947,6 +956,14 @@ def printLog(*args):
def printError(*args):
print("Error : ", *args, file=sys.stderr)
+def replacement(match, txt):
+ if not(match is None):
+ programName = unescape(match.group(1)).replace('<','<').replace('>','>').strip()
+ programName = '| ' + programName + ' | '
+ return programName
+ else:
+ return '';
+
Settingfile = os.path.dirname(os.path.abspath(__file__)) + '/epg2xml.json'
ChannelInfos = []
try:
@@ -1061,7 +1078,7 @@ if default_fetch_limit :
sys.exit()
else :
period = int(default_fetch_limit)
- if period > 2 : period = 2
+ #if period > 2 : period = 2
else :
printError("epg2xml.json 파일의 default_fetch_limit항목이 없습니다.");
sys.exit()
| | |