Project

General

Profile

Bug #258 » imdb.xml

patched system/scrapers/video/common/imdb.xml (v2) - foresto, 22/10/2013 06:37 AM

 
<scraperfunctions>
<GetIMDBThumbs dest="5">
<RegExp input="$$6" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="&lt;thumb&gt;\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2&lt;/thumb&gt;" dest="6">
<expression noclean="1,2">&lt;img.*?class=&quot;poster&quot;.*?src=&quot;([^&quot;]*?)_S[^&quot;]*?(.jpg)&quot;</expression>
</RegExp>
<RegExp input="$$1" output="&lt;thumb&gt;\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2&lt;/thumb&gt;" dest="6+">
<expression repeat="yes" noclean="1">&lt;a href=.*?itemprop=.thumbnailUrl.*?src=&quot;([^&quot;]*?)_S[^&quot;]*?(.jpg)&quot;</expression>
</RegExp>
<expression noclean="1"/>
</RegExp>
</GetIMDBThumbs>
<GetIMDBTrailer dest="5">
<RegExp input="$$2" output="&lt;details&gt;&lt;trailer urlencoded=&quot;yes&quot;&gt;\1&lt;/trailer&gt;&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="2">
<expression noclean="1">&quot;file&quot;[^&quot;]*&quot;([^&quot;]*)</expression>
</RegExp>
<RegExp input="$$1" output="$$2/\1" dest="2">
<expression noclean="1">&quot;id&quot;[^&quot;]*&quot;([^&quot;]*)</expression>
</RegExp>
<expression noclean="1"/>
</RegExp>
</GetIMDBTrailer>

<GetIMDBCast dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="6">
<expression noclean="1">&lt;table\sclass=&quot;cast_list&quot;&gt;(.*?)&lt;/table&gt;</expression>
</RegExp>
<RegExp input="$$6" output="&lt;actor&gt;&lt;thumb&gt;\2\3SX1024_SY1024_.jpg&lt;/thumb&gt;&lt;name&gt;\1&lt;/name&gt;&lt;role&gt;\5&lt;/role&gt;&lt;/actor&gt;" dest="7">
<expression repeat="yes" clear="yes" trim="3,4" noclean="1,2">&lt;img[^a]*alt=&quot;([^&quot;]*)&quot;[^&quot;]*&quot;[^&quot;]*&quot;[^s]*src=&quot;(?:[^&quot;]*\.[^&quot;]*&quot;class=&quot;loadlate\shidden\s&quot;\sloadlate=&quot;([^&quot;]*)(_V._))[^&gt;]*.*?ter&quot;&gt;[^&gt;]*&gt;\n\s*(&lt;[^&gt;]*&gt;)?([^&lt;\(]*)?</expression>
</RegExp>
<RegExp input="$$6" output="&lt;actor&gt;&lt;thumb&gt;&lt;/thumb&gt;&lt;name&gt;\1&lt;/name&gt;&lt;role&gt;\3&lt;/role&gt;&lt;/actor&gt;" dest="7+">
<expression repeat="yes" trim="1,2,3" fixchars="3" noclean="1,2">&lt;img[^a]*alt=&quot;([^&quot;]*)&quot;[^&quot;]*&quot;[^&quot;]*&quot;[^s]*src=&quot;(?:[^&quot;]*\.[^&quot;]*\.png)&quot;\sclass=&quot;.*?&quot;\s/&gt;*.*?ter&quot;&gt;[^&gt;]*&gt;\n\s*(&lt;[^&gt;]*&gt;)?([^&lt;\(]*)?</expression>
</RegExp>
<RegExp input="$$7" output="&lt;actor&gt;&lt;thumb&gt;\1&lt;/thumb&gt;\2&lt;/actor&gt;" dest="2+">
<expression repeat="yes" clear="yes" noclean="1,2,3">&lt;actor&gt;&lt;thumb&gt;(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)?&lt;/thumb&gt;(.*?)&lt;/actor&gt;</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</GetIMDBCast>

<GetIMDBDirectors dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="6">
<expression clear="yes" noclean="1">Director(?:s)?:.*?&lt;/h4&gt;(.*?)&lt;div class</expression>
</RegExp>
<RegExp input="$$6" output="&lt;director&gt;\1&lt;/director&gt;" dest="2">
<expression repeat="yes" fixchars="1" clear="yes">href=&quot;/name[^&gt;]*&gt;[^&lt;]*&lt;[^&gt;]*&gt;([^&lt;]*)&lt;</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</GetIMDBDirectors>

<GetIMDBWriters dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="6">
<expression clear="yes" noclean="1">Writer(?:s)?:.*?&lt;/h4&gt;(.*?)&lt;div class</expression>
</RegExp>
<RegExp input="$$6" output="&lt;credits&gt;\1&lt;/credits&gt;" dest="2">
<expression repeat="yes" fixchars="1" clear="yes">href=&quot;/name[^&gt;]*&gt;[^&lt;]*&lt;[^&gt;]*&gt;([^&lt;]*)&lt;</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</GetIMDBWriters>

<GetIMDBFullCast dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="6">
<expression noclean="1">&lt;table class=&quot;cast&quot;&gt;(.*?)&lt;/table&gt;</expression>
</RegExp>
<RegExp input="$$6" output="&lt;actor&gt;&lt;thumb&gt;\1_SX1024_SY1024_\2&lt;/thumb&gt;&lt;name&gt;\3&lt;/name&gt;&lt;role&gt;\5&lt;/role&gt;&lt;/actor&gt;" dest="7">
<expression repeat="yes" clear="yes" fixchars="3,5" trim="3,5" noclean="1,2">&lt;img src="(?:([^&quot;]*\.)[^&quot;]*(\.jpg))?[^&gt;]*[^&quot;]*&quot;nm&quot;&gt;&lt;a href=&quot;[^&quot;]*[^&gt;]*&gt;([^&lt;]*)&lt;[^&quot;]*&quot;ddd&quot;&gt;([^&lt;]&lt;)?[^&quot;]*&quot;char&quot;&gt;(.*?)&lt;/td&gt;</expression>
</RegExp>
<RegExp input="$$7" output="&lt;actor&gt;&lt;thumb&gt;\1&lt;/thumb&gt;\2&lt;/actor&gt;" dest="2+">
<expression repeat="yes" clear="yes" noclean="1,2,3">&lt;actor&gt;&lt;thumb&gt;(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)&lt;/thumb&gt;(.*?)&lt;/actor&gt;</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</GetIMDBFullCast>

<GetIMDBFullDirectors dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="6">
<expression clear="yes" noclean="1">Directed by.*?&lt;/h5&gt;(.*?)&lt;/table</expression>
</RegExp>
<RegExp input="$$6" output="&lt;director&gt;\1&lt;/director&gt;" dest="2">
<expression repeat="yes" clear="yes">href=&quot;/name[^&gt;]*&gt;([^&lt;]*)&lt;</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</GetIMDBFullDirectors>

<GetIMDBFullWriters dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="\1" dest="6">
<expression noclean="1">Writing\scredits(.*?)name="</expression>
</RegExp>
<RegExp input="$$6" output="&lt;credits&gt;\1&lt;/credits&gt;" dest="2">
<expression repeat="yes" fixchars="1">&lt;a href=&quot;/name/[^&gt;]*&gt;([^&lt;]*)&lt;</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</GetIMDBFullWriters>

<GetIMDBPlot dest="5">
<RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
<RegExp input="$$1" output="&lt;plot&gt;\1&lt;/plot&gt;" dest="2">
<expression clear="yes">&lt;div class="desc"&gt;.*?&lt;p&gt;\n([^\n]*)</expression>
</RegExp>
<expression noclean="1"/>
</RegExp>
</GetIMDBPlot>
</scraperfunctions>
(5-5/7)