83 lines
4.5 KiB
INI
Executable File
83 lines
4.5 KiB
INI
Executable File
**------------------------------------------------------------------------------------------------
|
|
* @header_start
|
|
* WebGrab+Plus ini for grabbing EPG data from TvGuide websites
|
|
* @Site: sol.no
|
|
* @MinSWversion: V1.1.1/53.6
|
|
* uses span in remove duplicates
|
|
* @Revision 2 - [29/12/2013] Jan van Straaten
|
|
* remove of 'subshow' within a 'show' (with same start) in the index that causes dayjumps
|
|
* @Revision 1 - [27/02/2012] Willy de Wilde
|
|
* Showsplit.scrub and stop.scrub
|
|
* @Revision 0 - [27/01/2011] Alberto Miguel
|
|
* none
|
|
* @Remarks:
|
|
* none
|
|
* @header_end
|
|
**------------------------------------------------------------------------------------------------
|
|
|
|
site {url=sol.no|timezone=UTC+01:00|maxdays=6|cultureinfo=NO|charset=utf-8|titlematchfactor=90|ratingsystem=NO}
|
|
url_index{url|http://www.sol.no/underholdning/tv/guiden/index.cgi?StartTime=0500&EndTime=2900&Date=|urldate|&Categories=*&Channels=|channel}
|
|
urldate.format {datestring|yyyy-MM-dd}
|
|
*
|
|
index_urlshow {url|http://www.sol.no/|<td><strong>|<a href="http://www.sol.no//|" target="_blank"|</a>} *Sometimes the // is problematic.
|
|
index_showsplit.scrub {multi()|<table id="tableau"|<td class="border" bgcolor="#d9d7ca">|</tr|</table>}
|
|
*
|
|
* it happens that some shows are listed that occur within another show
|
|
* the following removes the extra show
|
|
scope.range {(splitindex)|end}
|
|
index_showsplit.modify {|} * dummy required to initiate the index_showsplit operations
|
|
index_temp_2.modify {substring(type=regex)|'index_showsplit' "(.+<span color=.+</td><td><strong>).+"} * the part that is equal
|
|
index_temp_2.modify {remove(type=regex)|"<span color=\"#4d4d40\">(\d{2}:\d{2})</span>"} * remove the stop time (not equal)
|
|
index_temp_2.modify {cleanup(removeduplicates link=index_showsplit)} * span=2 , removes only when next or next-next are equal
|
|
end_scope
|
|
*
|
|
index_start.scrub {single|<td>|<strong>|</strong><br>|<span color=}
|
|
index_stop.scrub {single()|<td>|<span color="#4d4d40">|</span><br/>|</td}
|
|
index_title.scrub {multi(exclude="<a href")|<td>|<strong>|<br>|</td>}
|
|
index_temp_1.scrub {single|<a href="http://www.sol.no//underholdning/|return false;">|<br>|}
|
|
index_category.scrub {single|</span><br/>|</td><td>|</td>|}
|
|
index_description.scrub {multi(exclude="<span")|<td>|<br>|</td>|<td align="right"}
|
|
*
|
|
* enable the following two lines to create a channel list file
|
|
*index_site_channel.scrub {multi(exclude="ALLE KANALER" "Norske kanaler" "-------")|<select class="dropdown" name="Channels">|">|</option>|</select>}
|
|
*index_site_id.scrub {multi(exclude="*" "_norske" "nrktv1" "separator")|<select class="dropdown" name="Channels">|<option value="|"|</select>}
|
|
*
|
|
category.scrub {single|<td>|<font color=#606060> |</td>|<td>}
|
|
title.scrub {multi(exclude="<font")|<td>|<font face="verdana,arial,helvetica" size="1"><b>|<!--(0)-->|</td>}
|
|
description.scrub {single(separator="Medvirkende:" include=first)|<td colspan=4>|<font face="verdana,arial,helvetica" size="1">|</td>|</tr>}
|
|
titleoriginal.scrub {single(lang=xx)|<td colspan=4>|(|)|</td>}
|
|
productiondate.scrub {single|<td colspan=4>|fra|.|</td>}
|
|
presenter.scrub {single(separator=", ")|<td colspan=4>|Programleder|.|</td>}
|
|
actor.scrub {single(separator=", ")|<td colspan=4>|Medvirkende:|.|</td>}
|
|
director.scrub {single(separator=", ")|<td colspan=4>|Regi:|.|</td>}
|
|
rating.scrub {single(exclude="</td>")|<td colspan=4>|. (|. år)|</td>} *Sometimes on series get something like this (2:4)
|
|
*
|
|
index_title.modify {remove(null)|</strong>}
|
|
index_title.modify {remove(null)|'index_start'}
|
|
*
|
|
index_temp_1.modify {remove(null)|</a></strong>} *I have to use this because cleanup delete sometimes the last ) on titles with final (R)
|
|
index_title.modify {addstart(null)|'index_temp_1'}
|
|
*
|
|
title.modify {remove(null)|</b>} *I have to use this because cleanup delete sometimes the last ) on titles with final (R)
|
|
*
|
|
description.modify {remove(null)|'index_description'} *There is duplicated description when exists index_description
|
|
description.modify {remove(null)|('titleoriginal')} *There is duplicated description when exists index_description
|
|
description.modify {remove(null)|Programleder'presenter'.}
|
|
*
|
|
titleoriginal.modify{replace()|``|\'}
|
|
titleoriginal.modify{addend(notnull)|.!-}
|
|
titleoriginal.modify{remove(null)|..!-} *Sometimes there is a final . on the original title
|
|
titleoriginal.modify{remove(null)|.!-}
|
|
*
|
|
presenter.modify {remove(null)|:}
|
|
presenter.modify {remove(null)|er }
|
|
presenter.modify {remove(null)|.}
|
|
*
|
|
description.modify{replace()|``|\'}
|
|
*
|
|
director.modify{replace()|``|\'}
|
|
*
|
|
actor.modify{replace()|``|\'}
|
|
*
|
|
rating.modify {addend(notnull)|. år}
|