97 lines
4.7 KiB
INI
Executable File
97 lines
4.7 KiB
INI
Executable File
**------------------------------------------------------------------------------------------------
|
|
* @header_start
|
|
* WebGrab+Plus ini for grabbing EPG data from TvGuide websites
|
|
* @Site: telerama.fr
|
|
* @MinSWversion: V2.1.1
|
|
* @Revision 11 - [23/08/2017] Blackbear199
|
|
* - site changes
|
|
* @Revision 10 - [09/01/2017] Blackbear199
|
|
* - channels.xml generation fix
|
|
* - other site changes
|
|
* @Revision 9 - [02/11/2015] Francis De Paemeleere
|
|
* - adjust .channels.xml generation
|
|
* uses cleanup(tags)
|
|
* @Revision 8 - [24/04/2013] Jan van Straaten
|
|
* removed the remaining tags in title
|
|
* @Revision 7 - [30/05/2012] Jan van Straaten
|
|
* title in html changed again
|
|
* @Revision 6 - []
|
|
* title and alternative title swapped
|
|
* @Revision 5 - []
|
|
* alternative title
|
|
* @Revision 4 - []
|
|
* improved index_showsplit
|
|
* @Revision 3 - []
|
|
* Full detail version
|
|
* @Remarks:
|
|
* possibly incomplete rating and video elements (see lookup 'list' in modify section)
|
|
* this site is very very slow in La Gomera
|
|
* @header_end
|
|
**------------------------------------------------------------------------------------------------
|
|
|
|
site {url=telerama.fr|timezone=Europe/Paris|maxdays=15|cultureinfo=fr-FR|charset=UTF-8|titlematchfactor=90}
|
|
site {episodesystem=onscreen|subtitlestype=deaf-signed|allowlastdayoverflow}
|
|
*
|
|
url_index{url|http://television.telerama.fr/tele/chaine-tv/|channel|.php?&date=|urldate|}
|
|
url_index.headers {customheader=Accept-Encoding=gzip,deflate}
|
|
*
|
|
urldate.format {datestring|yyyy-MM-dd}
|
|
*
|
|
index_urlchannellogo {url||<div class="tv--channel">|src="|"|</div>}
|
|
*
|
|
index_showsplit.scrub {multi|<div id="LISTETELE"|<div class="tv10-chaine-item">|<!--tv10-chaine-descri-->|<!--LISTETELE tv10-chaine-liste-->}
|
|
*
|
|
index_urlshow {url|http://television.telerama.fr|<h2 class="tv10-chaine-descri-tit">|href="|"|</h2>}
|
|
index_urlshow.headers {customheader=Accept-Encoding=gzip,deflate}
|
|
*
|
|
index_start.scrub {regex||<em class="placeholder">(\d{2}h\d{2})</em> à <em class="placeholder">\d{2}h\d{2}</em>||}
|
|
index_stop.scrub {regex||<em class="placeholder">\d{2}h\d{2}</em> à <em class="placeholder">(\d{2}h\d{2})</em>||}
|
|
index_title.scrub {single(separator=", " include=first)|<h2 class="tv10-chaine-descri-tit">||</h2>|</h2>}
|
|
index_category.scrub {single|<p class="tv10-chaine-descri-surt">||</p>|</p>}
|
|
*
|
|
index_start.modify {replace|h|:}
|
|
index_stop.modify {replace|h|:}
|
|
index_title.modify {cleanup(tags="<"">")}
|
|
*
|
|
title.scrub {regex||<p class="fiche--title"[^>]*>(.+?)</p>||}
|
|
subtitle.scrub {single| <p class="episode--id">||</p>|</p>}
|
|
temp_1.scrub {single| <p class="episode--title">||</p>|</p>}
|
|
description.scrub {single|Synopsis</h2>|">|</div>|</div>}
|
|
category.scrub {single|<p>Genre : ||</p>|</p>|
|
|
productiondate.scrub {single|<span itemprop="dateCreated"> (||)</span>|)</span>}
|
|
country.scrub {single|Nationalité : ||<|<}
|
|
showicon.scrub {single|<figure class="fiche--image">|src="|"|</figure>}
|
|
*
|
|
title.modify {cleanup(tags="<"">")}
|
|
subtitle.modify {set('temp_1' not "")|'temp_1'} *when temp_1 is not empty it contains the episode title
|
|
subtitle.modify {cleanup(tags="<"">")}
|
|
description.modify {cleanup(tags="<"">")}
|
|
country.modify {cleanup(style=name)}
|
|
country.modify {replace| |/}
|
|
showicon.modify {clear(~ "/trans.")}
|
|
showicon.modify {addstart(not~ "http")|http://television.telerama.fr/tele}
|
|
showicon.modify {clear("http://television.telerama.fr/tele")}
|
|
*
|
|
urlsubdetail.modify {addstart('index_urlshow' not "")|'index_urlshow'}
|
|
urlsubdetail.modify {replace(not "")|.php|,casting.php}
|
|
urlsubdetail.headers {customheader=Accept-Encoding=gzip,deflate}
|
|
*
|
|
subdetail_director.scrub {multi(max=3)|Réalisateurs :|class="obf">|</a>|</tbody>}
|
|
subdetail_actor.scrub {multi(max=7)|Acteurs :|class="obf">|</span>|</tbody>}
|
|
subdetail_presenter.scrub {multi(max=3)|Présentateurs :|class="obf">|</a>|</tbody>}
|
|
subdetail_writer.scrub {multi(max=3)|Scénaristes :|class="obf">|</a>|</tbody>}
|
|
*
|
|
subdetail_actor.modify {replace(type=regex)|</a>.+?<span class="lightgrey">[^\w]+|(role=}
|
|
subdetail_actor.modify {replace(~ "role=" type=regex)|\s*$|)}
|
|
subdetail_actor.modify {cleanup(tags="<"">")}
|
|
|
|
** _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
|
|
** ##### CHANNEL FILE CREATION (only to create the xxx-channel.xml file)
|
|
**
|
|
** @auto_xml_channel_start
|
|
*url_index{url|http://television.telerama.fr/tele/liste_chaines.php}
|
|
*index_site_id.scrub {regex||<span class="tv10-list-chn-chn">(?:.*?)(?:<a href="/tele/chaine-tv/([^\.]+).php">(?:.*?))*</a>||}
|
|
*index_site_channel.scrub {regex||<span class="tv10-list-chn-chn">(?:.*?)(?:<a href="/tele/chaine-tv/[^\.]+.php">([^<]+)</a>(?:.*?))*</span>||}
|
|
*index_site_id.modify {cleanup(removeduplicates=equal,100 link="index_site_channel")}
|
|
** @auto_xml_channel_end
|