**------------------------------------------------------------------------------------------------
* @header_start
* WebGrab+Plus ini for grabbing EPG data from TvGuide websites
* @Site: mediaset.it
* @MinSWversion: V1.1.1/52
*   none
* @Revision 2 - [21/04/2015] Francis De Paemeleere
*    adding channel icon
* @Revision 1 - [29/09/2014] Francis De Paemeleere
*	fix pre-6am shwo grabbing
*   fix actor/dirctor grabbing + remove info from description
* @Revision 0 - [09/02/2014] Steve Wildmore
*   creation
* @Remarks:
*   none
* @header_end
**------------------------------------------------------------------------------------------------

site {url=mediaset.it|timezone=Europe/Rome|maxdays=7|cultureinfo=it-IT|charset=UTF-8|titlematchfactor=90|episodesystem=onscreen|ratingsystem=MPAA}
index_variable_element.modify {calculate(scope=urlindex format=date,yyyyMM)|'urldate'}
url_index{url|http://www.mediaset.it/guidatv/inc/canali/######/|urldate|_|channel|.sjson}
url_index.modify {replace|######|'index_variable_element'}
urldate.format {datestring|yyyyMMdd}

*
index_showsplit.scrub {multi(separator="},")|{"events":|[|]|}
index_showsplit.modify {replace|}|}

index_urlchannellogo.modify {addstart|http://www.mediaset.it/include/img/loghi/'config_site_id'.png}

index_start.scrub 	{single|"startTime":"||"|"}
index_stop.scrub 	{single|"endTime":"||"|"}

index_title.scrub 	 {regex||"displayTitle"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
index_category.scrub {regex||"genere"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
index_category.scrub {regex||"sottogenere"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
index_temp_1.scrub	 {regex||"channel"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}

* http://servizi.mediaset.it/Contenuti/guidaTv/programDetails.jsp?programId=201402072112_C5&jsoncallback=?
index_urlshow {url|servizi.mediaset.it/Contenuti/guidaTv/programDetails.jsp?programId=########|"startTime":"||"|"}

index_temp_2.modify {calculate(format=date,yyyyMMdd)|'showdate'}
index_temp_2.modify {calculate('index_start' < "06:00" format=date,yyyyMMdd)|'showdate' 01:00 -}		* shows before 6am, are called by the previous day
index_urlshow.modify {replace|########|'index_temp_2'}
index_urlshow.modify {addend|_'index_temp_1'&jsoncallback=?}
index_urlshow.modify {replace|:|}
index_urlshow.modify {addstart|http://}

index_title.modify {remove(type=regex)|"\s*-\s*PRIMA TV\s*"}
index_title.modify {remove(type=regex)|"\s*\(\s*DI\s+[^)]*\)"}

title.scrub {regex||"displayTitle"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
title.modify {remove(type=regex)|"\s*\(\s*DI\s+[^)]*\)"}
description.scrub {regex||"trama"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
temp_1.scrub {regex||"description"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
description.modify {addend| 'temp_1'}
description.modify {cleanup(style=jsondecode)}

productiondate.scrub {regex||"anno"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
premiere.scrub {regex||"prima"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}
premiere.modify {replace('title' ~ "PRIMA TV")|'premiere'|true}
title.modify {remove(type=regex)|"\s*-\s*PRIMA TV\s*"}

director.scrub {regex||Regia:\s*(.*?)\s*-||}                                * director info form description
director.scrub {regex||"regia"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)"||}			* director info form json data
director.modify {replace|,|;}
director.modify {replace|;|\|}
director.modify {replace|.|. }
director.modify {replace|  | }
director.modify {cleanup(style=name)}
director.modify {cleanup(removeduplicates=name)}

actor.scrub {regex||Cast:(?:\s(.*?),)*\s*(.*?(?:\w{3,}))(?:\s+-\s+\|\.)||}	* actor info form description
actor.scrub {regex||cast"\s*:\s*"([^"\\]*(?:\\.[^"\\]*)*)||}				* actor info form json data
actor.modify {replace|,|;}
actor.modify {replace|;|\|}
actor.modify {replace|.|. }
actor.modify {replace|  | }
actor.modify {cleanup(style=name)}
actor.modify {cleanup(removeduplicates=name)}

description.modify {remove(type=regex)|Regia:\s*.*? - }
description.modify {remove(type=regex)|(Cast:(?:\s(?:.*?),)*\s*.*?(?:\w{3,})(?:\s+-\s+\|\.))}

* rating
rating.modify {addstart('description' ~ "VISIONE ADATTA A TUTTI")|G}
rating.modify {addstart('description' ~ "VISIONE CONSIGLIATA CON LA PRESENZA DI UN ADULTO")|PG}
rating.modify {addstart('description' ~ "VISIONE CONSIGLIATA AL SOLO PUBBLICO ADULTO")|R}
rating.modify {addstart('description' ~ "VIETATO AI MINORI DI 14 ANNI")|PG-13}
rating.modify {addstart('description' ~ "VIETATO AI MINORI DI 18 ANNI")|PG-17}

description.modify {remove(type=regex)|\s*VISIONE ADATTA A TUTTI\s*\.\s*}
description.modify {remove(type=regex)|\s*VISIONE CONSIGLIATA CON LA PRESENZA DI UN ADULTO\s*\.\s*}
description.modify {remove(type=regex)|\s*VISIONE CONSIGLIATA AL SOLO PUBBLICO ADULTO\s*\.\s*}
description.modify {remove(type=regex)|\s*VIETATO AI MINORI DI 14 ANNI\s*\.\s*}
description.modify {remove(type=regex)|\s*VIETATO AI MINORI DI 18 ANNI\s*\.\s*}

*
**  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _  _
**      #####  CHANNEL FILE CREATION (only to create the xxx-channel.xml file)
**
** @auto_xml_channel_start
*url_index{url|http://www.mediaset.it/guidatv/}
*index_site_channel.scrub {multi(exclude="tutte")|<select id="canale">|">|</option>|</select>}
*index_site_id.scrub {multi|<select id="canale">|<option value="|">|</select>}
*scope.range {(channellist)|end}
*index_site_id.modify {cleanup(removeduplicates=equal,100 link="index_site_channel")}
*end_scope
** @auto_xml_channel_end