**------------------------------------------------------------------------------------------------
* @header_start
* WebGrab+Plus ini for grabbing EPG data from TvGuide websites
* @Site: plus.es
* @MinSWversion: V1.1.1/5405
* @Revision 1 - [23/07/2014] Jan van Straaten
*   - fixed start time errors and added episode, rating and productiondate
* @Revision 0 - [16/07/2014] Francis De Paemeleere/ Jan van Straaten
*   - creation
* @Remarks:
* @header_end
**------------------------------------------------------------------------------------------------
*
site {url=plus.es|timezone=UTC+01:00|maxdays=28|cultureinfo=es-ES|charset=iso-8859-15,UTF-8|titlematchfactor=50|subtitlestype=onscreen|ratingsystem=MPAA|episodesystem=xmltv_ns|}
*
url_index{url()|http://www1.plus.es/modulo/index.html?params=fecha%3D|urldate|%26hora%3D|subpage|%26cadenas%3D|channel|%26vista%3Dampliada%26modulo%3DGUIParrilla}
*http://www.plus.es/modulo/index.html?params=fecha%3D17072014%26hora%3D2000%26cadenas%3DTVE%26vista%3Dampliada%26modulo%3DGUIParrilla

urldate.format {datestring|ddMMyyyy}
subpage.format {list|0100|0400|0700|1000|1300|1600|1900|2200}

*
scope.range {(splitindex)|end}

index_showsplit.scrub {multi||||} * copy index pages
index_showsplit.modify {remove|\n}
index_showsplit.modify {remove|  }
* insert a dummy at each subpage start
index_showsplit.modify {replace(type=regex)|"(<div id=\"contenedorParrilla\">.+?<div class=\"canal_guide\">)"|<div class="time time_XXgen_cn"> subpage start , remove </div>}
* split into index_showsplit
index_showsplit.modify {substring(type=regex)|"<div class=\"time (.+?</div>)"}
*
*index_showsplit.modify {addstart|} * for inspection only
* the following loop replaces the showdata of the first show after a subpage if it is the same show as the last show of the previous subpage,
* but the time_ component is kept because it is needed in the next loop as duration
index_temp_1.modify {calculate(format=F0 type=element)|'index_showsplit' #} * the loop counter
loop {('index_temp_1' > "1" max=200)|end}
index_temp_2.modify {clear} * will contain the showid of the show after the subpage
index_temp_3.modify {clear} * will contain the data of the show before the subpage
index_temp_1.modify {calculate(format=F0)|1 -} * next index
index_temp_5.modify {calculate(format=F0)|0 *} *reset the condition 0 not on a subpage +1 on a subpage 2 the show before and after the subpage contain the same show
index_temp_6.modify {substring(type=element)|'index_showsplit' 'index_temp_1' 1} * the index_showsplit
index_temp_5.modify {calculate('index_temp_6' ~ "subpage" format=F0)|1 +} * we are on a subpage, examine the show before and after
index_temp_1.modify {calculate('index_temp_5' = "1" format=F0)|1 +} * the index of the show after the subpage
index_temp_6.modify {substring('index_temp_5' = "1" type=element)|'index_showsplit' 'index_temp_1' 1} * the show after the subpage
index_temp_6.modify {addend('index_temp_5' = "1")|<first show of page>} * to identify it for the next loop
index_showsplit.modify {replace('index_temp_5' = "1" type=element)|'index_showsplit' 'index_temp_1' 1|'index_temp_6'}
index_temp_2.modify {substring('index_temp_5' = "1" type=regex)|'index_temp_6' "<a href=\"/programacion/.+?/(R\d{8})/"} *the showid of the show after the subpage
index_temp_1.modify {calculate('index_temp_5' = "1" format=F0)|2 -} *the index of the show before the subpage
index_temp_3.modify {substring('index_temp_5' = "1" type=element)|'index_showsplit' 'index_temp_1' 1} * the show before the subpage
index_temp_3.modify {substring('index_temp_5' = "1" type=regex)|'index_temp_3' "\">(<a href=.+?\</div>)"} * the data excl the time_ component of the show before the subpage
index_temp_5.modify {calculate('index_temp_3' ~ 'index_temp_2' format=F0)|1 +} * condition "2" ,same show before and after the subpage
index_showsplit.modify {replace('index_temp_5' = "2" type=element)|'index_showsplit' 'index_temp_1' 1|removed} * indicate the show before the subpage as removeduplicates
index_temp_6.modify {replace('index_temp_5' = "2" type=regex)|"\">(<a href=.+?\</div>)"|'index_temp_3'} * replace the data of the show after the subpage with that of before that, keep the time_ value
index_temp_1.modify {calculate('index_temp_5' = "2" format=F0)|2 +} * back to the show after the subpage
index_showsplit.modify {replace('index_temp_5' = "2" type=element)|'index_showsplit' 'index_temp_1' 1|'index_temp_6'}
*index_temp_1.modify {calculate('index_temp_5' = "2" format=F0)|3 -} * jump over the show before the subpage
index_temp_1.modify {calculate('index_temp_5' = "2" format=F0)|2 -} * jump over the show before the subpage
end_loop
*
*remove the first index_show (that's a subpage)
index_showsplit.modify {remove(type=element)|0 1}
*
* the next loop calculates the start and stop times from the 'duration' which is the value of time_nn at the beginning of each indexshow
* the first 'duration value of each subpage is counted from the subpage starttime :
* The first start time is always(?) 38 minutes before the first subpage time , here 01:00 - 00:38 = 00:22
index_variable_element.modify {clear}
index_variable_element.modify {addstart|00:22} * the 'base' startime of the first subpage
index_temp_1.modify {clear}
index_temp_1.modify {addstart|00:22} * the first start time
*
index_temp_3.modify {calculate(format=F0)|0 *} *subpage counter 0, 3, 6 etc
index_temp_5.modify {clear} * will contain the new buildup of the index_showsplit
*
loop {(each 'index_temp_6' in 'index_showsplit')|end}
index_temp_2.modify {calculate(format=F0)|0 *} * will contain the duration of the show
* calculate the 'base' starttime at each subpage
index_temp_3.modify {calculate('index_temp_6' ~ "subpage" format=F0)|3 +} * increase subpage counter if on subpage
*index_temp_1.modify {addstart|}
index_variable_element.modify {calculate('index_temp_6' ~ "subpage" format=time,HH:mm)|'index_temp_3' 00:22 +} * the new 'base' startime
* add the start time of this show
* calculate the starttime
index_temp_4.modify {clear} * will contain the new show info
index_temp_4.modify {addend('index_temp_6' not ~ "remove")|'index_temp_6'}
index_temp_4.modify {addend|<strong class="WG_hour_start">}
index_temp_4.modify {addend|'index_temp_1'}
index_temp_4.modify {addend|</strong>}
* get the duration of this show
index_temp_2.modify {substring('index_temp_6' not ~ "remove" type=regex)|'index_temp_6' "\Atime_(\d+)"}
* calculate the stop time of this show
index_temp_2.modify {calculate(format=time)|60 /} * converted into hh:mm
index_temp_1.modify {calculate('index_temp_6' not ~ "<first show of page>" format=time)|'index_temp_1' 'index_temp_2' +}
index_temp_1.modify {calculate('index_temp_6' ~ "<first show of page>" format=time)|'index_variable_element' 'index_temp_2' +}
*
index_temp_4.modify{addend|<strong class="WG_hour_stop">}
index_temp_4.modify{addend|'index_temp_1'}
index_temp_4.modify{addend|</strong>}
*
index_temp_5.modify{addend('index_temp_4' ~ "<a href=\"/programacion")|'index_temp_4'##_split_##}
*index_temp_5.modify{addend()|'index_temp_4'##_split_##}
end_loop
index_showsplit.modify {clear}
index_showsplit.modify{addend|'index_temp_5'}
index_showsplit.modify{replace()|##_split_##|\|}
end_scope
index_showsplit.modify {addstart|} * just for inspection of the loop result
index_start.scrub {single|<strong class="WG_hour_start">||</strong>}
*index_start.modify {addstart|'previous_stop'}
index_stop.scrub {single|<strong class="WG_hour_stop">||</strong>}
*index_subtitle.scrub {regex||<strong class=\"hour_start\">(.*?)</span>||} * optional starttime check
*index_subtitle.modify {cleanup}
index_title.scrub {single()|<strong class="title">||</strong>|</strong>}
index_category.scrub {single|<strong class="category">||</strong>|</strong>}
*
index_urlshow {url|http://www.plus.es|<a href="||">|">}
index_urlshow.headers {customheader=Accept-Encoding=gzip,deflate}     * to speedup the downloading of the detail pages
*
title.scrub {single|<div class="title-image">|>|<|}
subtitle.scrub {single(separator=": " include=last exclude="Episodio")|<h1 itemprop="name" class="|h-epsilon"> |</h1>|</div>}
description.scrub {multi|<div itemprop="description"|<p>|</p>|</div>} 
episode.scrub {single|<div class="title-image">| TEMPORADA |</p>|} 
temp_1.scrub {single|<h1 itemprop="name" class="h-epsilon">|Episodio |</h1>|</div>} 
titleoriginal.scrub {single(lang=en)|<p class="h-gamma">||</p>|</div>} 
*temp_2.scrub {single|<dt>Título Original Serie:</dt>|<dd>|</dd>|} 
productiondate.scrub {single|<li><img src="/assets/images/icons/svg|(|)</p>|<div class="moral">} 
composer.scrub {multi|<h3 class="heading">COMPOSITOR</h3>|<span itemprop="name">|</span>|</p>} 
actor.scrub {multi|<p itemprop="actor" itemscope itemtype="http://schema.org/Person">|<span itemprop="name">|</span>|</p>} 
director.scrub {multi|<p itemprop="director" itemscope itemtype="http://schema.org/Person">|<span itemprop="name">|</span>|</p>} 
writer.scrub {multi|<p itemprop="author" itemscope itemtype="http://schema.org/Person">|<span itemprop="name">|</span>|</p>} 
*presenter.scrub {single(separator=", ")|<dt>PRESENTADOR:</dt>|<dd>|</dd>|} 
producer.scrub {multi|<h3 class="heading">PRODUCTOR</h3>|<span itemprop="name">|</span>|</p>} 
rating.scrub {single|<h3 class="heading">Calificación</h3>|alt="|">|</div>}
subtitles.scrub {single|<span>Subtítulos:|(|)|</span></p>} 
*premiere.scrub {single|<dt>Tipo de pase:</dt>|<dd|>Estr|</dd>}
*previousshown.scrub {single|<dt>Tipo de pase:</dt>|<dd|>Multidifusió|</dd>}
showicon.scrub {url||<div class="cover">|<img src="|" alt="|</div>}
*
temp_1.modify {remove|: 'subtitle'}
title.modify {remove|. TEMPORADA 'episode'}
titleoriginal.modify {remove|. TEMPORADA 'episode'}
description.modify {addstart('temp_1' not "")|Episodio 'temp_1' - }
description.modify {addstart('episode' not "")|Temporada 'episode' }
*titleoriginal.modify {replace('temp_2' not "")|'titleoriginal'|'temp_2'}
temp_1.modify {calculate('temp_1' not "" format=F0)|1 -}
episode.modify {calculate('episode' not "" format=F0)|1 -}
episode.modify {addend('episode' not "")|.'temp_1'.0/1}
*rating.modify {remove(notnull)|</p>">}
subtitles.modify {replace(not "")|'subtitles'|true}
*premiere.modify {replace(not "")|'premiere'|true}
*previousshown.modify {replace(not "")|'previousshown'|true}
rating.modify {replace|No recomendada para menores de doce años|PG}
rating.modify {replace|No recomendada para menores de siete años|TV-Y7}
rating.modify {replace|No recomendada para menores de trece años|TV-14}
rating.modify {replace|No recomendada para menores de dieciocho años|TV-MA}
rating.modify {replace|Apta para todos los públicos|TV-G}
rating.modify {replace|No recomendada para menores de dieciséis años|TV-14}