The branch, eden has been updated
via 74af51d713ef381d62c39678eb62d07537abc88b (commit)
from 0c46f3a427d47852e9927347ff2e7c9dccc92bec (commit)
- Log -----------------------------------------------------------------
http://xbmc.git.sourceforge.net/git/gitweb.cgi?p=xbmc/scrapers;a=commit;h=74af51d713ef381d62c39678eb62d07537abc88b
commit 74af51d713ef381d62c39678eb62d07537abc88b
Author: olympia <[email protected]>
Date: Sun Feb 10 22:11:34 2013 +0100
sync with Frodo
diff --git a/metadata.common.imdb.com/addon.xml
b/metadata.common.imdb.com/addon.xml
index 5aff850..7a025d0 100644
--- a/metadata.common.imdb.com/addon.xml
+++ b/metadata.common.imdb.com/addon.xml
@@ -1,7 +1,7 @@
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<addon id="metadata.common.imdb.com"
- name="IMDB common scraper functions"
- version="2.7.0"
+ name="IMDB Scraper Library"
+ version="2.7.4"
provider-name="Team XBMC">
<requires>
<import addon="xbmc.metadata" version="1.0"/>
diff --git a/metadata.common.imdb.com/changelog.txt
b/metadata.common.imdb.com/changelog.txt
index c9ba9dc..9f42032 100644
--- a/metadata.common.imdb.com/changelog.txt
+++ b/metadata.common.imdb.com/changelog.txt
@@ -1,3 +1,15 @@
+[B]2.7.4[/B]
+- fixed: MetaCritic Score
+
+[B]2.7.3[/B]
+- fixed: orginal title won't scrape in some cases
+
+[B]2.7.2[/B]
+- changed: make sure to parse the intended page
+
+[B]2.7.1[/B]
+- fixed: IMDb genre after layout change
+
[B]2.7.0[/B]
- fixed: IMDb scraping after layout changes
diff --git a/metadata.common.imdb.com/imdb.xml
b/metadata.common.imdb.com/imdb.xml
index 885f575..3f1b64e 100644
--- a/metadata.common.imdb.com/imdb.xml
+++ b/metadata.common.imdb.com/imdb.xml
@@ -1,34 +1,34 @@
<scraperfunctions>
<GetIMDBGenresById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBGenres">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBGenres">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBGenresById>
<ParseIMDBGenres dest="5">
<RegExp input="$$2" output="<details>\1</details>"
dest="5">
<RegExp input="$$1"
output="<genre>\1</genre>" dest="2">
- <expression
repeat="yes">tt_str_gnr"\s>\s([^>]+)</expression>
+ <expression
repeat="yes">tt_stry_gnr"\s>\s([^<]+)</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</ParseIMDBGenres>
<GetIMDBRatingById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBRating">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBRating">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBRatingById>
<ParseIMDBRating dest="5">
<RegExp input="$$2" output="<details>\1</details>"
dest="5">
<RegExp input="$$1"
output="<rating>\1</rating><votes>\2</votes>" dest="2">
- <expression><span
itemprop="ratingValue">([0-9.]+).*?>.*?title="([0-9,]+)\s</expression>
+
<expression><span\sitemprop="ratingValue">([0-9.]+).*?>.*?title="([0-9,]+)\s</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</ParseIMDBRating>
<GetIMDBTOP250ById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBTOP250">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBTOP250">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBTOP250ById>
@@ -42,7 +42,7 @@
</ParseIMDBTOP250>
<GetIMDBStudioById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBStudio">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBStudio">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBStudioById>
@@ -56,7 +56,7 @@
</ParseIMDBStudio>
<GetIMDBCountryById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBCountry">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBCountry">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBCountryById>
@@ -70,35 +70,35 @@
</ParseIMDBCountry>
<GetMetaCriticRatingById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseMetaCriticRating">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseMetaCriticRating">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetMetaCriticRatingById>
<ParseMetaCriticRating dest="5">
<RegExp input="$$2" output="<details>\1</details>"
dest="5">
<RegExp input="$$1"
output="<rating>\1.\2</rating>" dest="2">
- <expression><a
href="criticreviews">(\d)(\d*)</expression>
+
<expression><a\shref="criticreviews[^"]*?"\stitle[^>]*?>\s(\d)(\d*)</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</ParseMetaCriticRating>
<GetIMDBPlotById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBPlot">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBPlot">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBPlotById>
<ParseIMDBPlot dest="5">
<RegExp input="$$2" output="<details>\1</details>"
dest="5">
<RegExp input="$$1"
output="<plot>\1</plot>" dest="2">
- <expression fixchars="1"
trim="1"><h2>Storyline</h2>\s+<div\sclass="inline
canwrap"\sitemprop="description">\s+<p>(.*?)<[^a/]</expression>
+ <expression fixchars="1"
trim="1"><h2>Storyline</h2>\s+<div\sclass="inline\scanwrap"\sitemprop="description">\s+<p>(.*?)<[^a/]</expression>
</RegExp>
<expression noclean="1" />
</RegExp>
</ParseIMDBPlot>
<GetIMDBTaglineById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBTagline">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBTagline">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBTaglineById>
@@ -112,7 +112,7 @@
</ParseIMDBTagline>
<GetIMDBOutlineById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBOutline">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBOutline">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBOutlineById>
@@ -126,7 +126,7 @@
</ParseIMDBOutline>
<GetIMDBCastById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBCast">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBCast">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBCastById>
@@ -149,7 +149,7 @@
</ParseIMDBCast>
<GetIMDBDirectorsById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBDirectors">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBDirectors">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBDirectorsById>
@@ -166,7 +166,7 @@
</ParseIMDBDirectors>
<GetIMDBWritersById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBWriters">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBWriters">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBWritersById>
@@ -183,7 +183,7 @@
</ParseIMDBWriters>
<GetIMDBFullCastById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBFullCast">http://akas.imdb.com/title/$$1/combined</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBFullCast">http://akas.imdb.com/title/$$1/combined|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBFullCastById>
@@ -203,7 +203,7 @@
</ParseIMDBFullCast>
<GetIMDBFullDirectorsById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBFullDirectors">http://akas.imdb.com/title/$$1/combined</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBFullDirectors">http://akas.imdb.com/title/$$1/combined|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBFullDirectorsById>
@@ -220,14 +220,14 @@
</ParseIMDBFullDirectors>
<GetIMDBFullWritersById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBFullWriters">http://akas.imdb.com/title/$$1/combined</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBFullWriters">http://akas.imdb.com/title/$$1/combined|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBFullWritersById>
<ParseIMDBFullWriters dest="5">
<RegExp input="$$2" output="<details>\1</details>"
dest="5">
<RegExp input="$$1" output="\1" dest="6">
- <expression noclean="1">Writing
credits(.*?)name="</expression>
+ <expression
noclean="1">Writing\scredits(.*?)name="</expression>
</RegExp>
<RegExp input="$$6"
output="<credits>\1</credits>" dest="2">
<expression repeat="yes" fixchars="1"><a
href="/name/[^>]*>([^<]*)<</expression>
@@ -237,14 +237,14 @@
</ParseIMDBFullWriters>
<GetIMDBThumbsById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-posters.html"
function="ParseIMDBThumbs">http://akas.imdb.com/title/$$1/posters</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-posters.html"
function="ParseIMDBThumbs">http://akas.imdb.com/title/$$1/posters|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBThumbsById>
<ParseIMDBThumbs dest="5">
<RegExp input="$$6" output="<details>\1</details>"
dest="5">
<RegExp input="$$1" output="\1_SX1024_SY1024_\2"
dest="4">
- <expression noclean="1,2"><a
name="poster".*?src="(.*?)_S.*?(.jpg)".*?</a></expression>
+ <expression
noclean="1,2"><a\sname="poster".*?src="(.*?)_S.*?(.jpg)".*?</a></expression>
</RegExp>
<RegExp input="$$4" output="<thumb
aspect="poster">\1</thumb>" dest="6">
<expression
noclean="1">(.*?_SX[0-9]+_SY[0-9]+_.jpg)</expression>
@@ -254,18 +254,18 @@
</ParseIMDBThumbs>
<GetIMDBUSACert dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBUSACert">http://akas.imdb.com/title/$$1/</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-main.html"
function="ParseIMDBUSACert">http://akas.imdb.com/title/$$1/|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBUSACert>
<ParseIMDBUSACert dest="5">
<RegExp input="$$1"
output="<details><mpaa>$INFO[certprefix]\1</mpaa></details>"
dest="5">
- <expression>MPAA</a>\)</h4>\n?<span
itemprop="contentRating">Rated\s([^<]*)</expression>
+
<expression>MPAA</a>\)</h4>\n?<span\sitemprop="contentRating">Rated\s([^<]*)</expression>
</RegExp>
</ParseIMDBUSACert>
<GetIMDBCountryCert dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBCountryCert">http://akas.imdb.com/title/$$1/combined</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBCountryCert">http://akas.imdb.com/title/$$1/combined|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBCountryCert>
@@ -276,7 +276,7 @@
</ParseIMDBCountryCert>
<GetIMDBAKATitlesById dest="5">
- <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBAKATitles">http://akas.imdb.com/title/$$1/combined</url></details>";
dest="5">
+ <RegExp input="$$1" output="<details><url
cache="$$1-combined.html"
function="ParseIMDBAKATitles">http://akas.imdb.com/title/$$1/combined|accept-language=en-us</url></details>"
dest="5">
<expression noclean="1" />
</RegExp>
</GetIMDBAKATitlesById>
@@ -287,13 +287,13 @@
</RegExp>
<RegExp input="$$10" output="\1" dest="4">
<RegExp input="$$1" output="\2" dest="9">
- <expression fixchars="2"><meta
name="title" content="(IMDb -
)?(?:&#x22;)?([^"]*?)(?:&#x22;)? \([^\(]*?([0-9]{4})\)</expression>
+ <expression
fixchars="2"><meta\sname="title"\scontent="(IMDb\s-\s)?(?:&#x22;)?([^"]*?)(?:&#x22;)?
\([^\(]*?([0-9]{4})(?:â\s)?\)</expression>
</RegExp>
<RegExp input="$$9" output="\1" dest="10">
<expression />
</RegExp>
<RegExp input="$$1" output="\1" dest="11">
- <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^<]+[<em>]?[^"]+Hong
Kong [<em>][^"]+English</expression>
+ <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^<]+[<em>]?[^"]+Hong
Kong\s[<em>][^"]+English</expression>
</RegExp>
<RegExp input="$$11" output="\1" dest="10">
<expression>(.+)</expression>
@@ -311,25 +311,25 @@
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="11">
- <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^"]+International\s(<em>)?\(English
title\)(</em>)?(,|<)( |b)</expression>
+ <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^"]+International\s(<em>)?\(English\stitle\)(</em>)?(,|<)(
|b)</expression>
</RegExp>
<RegExp input="$$11" output="\1" dest="10">
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="11">
- <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^"]+International\s(<em>)?\(English
title\)(</em>)? (<em>)?\(imdb</expression>
+ <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^"]+International\s(<em>)?\(English\stitle\)(</em>)?
(<em>)?\(imdb</expression>
</RegExp>
<RegExp input="$$11" output="\1" dest="10">
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="12">
- <expression fixchars="1"
clear="yes"><a
href="/country/[^>]+>(UK</a></div>)</expression>
+ <expression fixchars="1"
clear="yes"><a\shref="/country/[^>]+>(UK</a></div>)</expression>
</RegExp>
<RegExp input="$$12" output="$$9" dest="10">
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="12">
- <expression clear="yes"><a
href="/country/[^>]+>(USA</a></div>)</expression>
+ <expression
clear="yes"><a\shref="/country/[^>]+>(USA</a></div>)</expression>
</RegExp>
<RegExp input="$$12" output="$$9" dest="10">
<expression>(.+)</expression>
@@ -347,7 +347,7 @@
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="11">
- <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^<]+[<em>]?[^"]+((USA)\s(<em>)?\((new
title))</expression>
+ <expression fixchars="1"
clear="yes">>\s*?"([^<]+)"[^<]+[<em>]?[^"]+((USA)\s(<em>)?\((new\stitle))</expression>
</RegExp>
<RegExp input="$$11" output="\1" dest="10">
<expression>(.+)</expression>
@@ -364,13 +364,13 @@
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="4">
- <expression
fixchars="1">>\s*?"([^<]+)"[^<]+[<em>]?[^"]+$INFO[imdbakatitles]\s(<em>)?\((imdb
display)</expression>
+ <expression
fixchars="1">>\s*?"([^<]+)"[^<]+[<em>]?[^"]+$INFO[imdbakatitles]\s(<em>)?\((imdb\sdisplay)</expression>
</RegExp>
<RegExp input="$$4" output="\1" dest="2">
<expression>(.+)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="5">
- <expression><a
href="/country/[^>]+>($INFO[imdbakatitles])</expression>
+
<expression><a\shref="/country/[^>]+>($INFO[imdbakatitles])</expression>
</RegExp>
<RegExp input="$$5" output="$$9" dest="4">
<expression>($INFO[imdbakatitles])</expression>
diff --git a/metadata.common.port.hu/addon.xml
b/metadata.common.port.hu/addon.xml
index 611ef7e..5ef8605 100644
--- a/metadata.common.port.hu/addon.xml
+++ b/metadata.common.port.hu/addon.xml
@@ -1,7 +1,7 @@
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<addon id="metadata.common.port.hu"
- name="port.hu common scraper functions"
- version="1.0.0"
+ name="port.hu Scraper Library"
+ version="1.1.0"
provider-name="olympia">
<requires>
<import addon="xbmc.metadata" version="1.0"/>
diff --git a/metadata.common.port.hu/porthu.xml
b/metadata.common.port.hu/porthu.xml
index f518193..10cb766 100644
--- a/metadata.common.port.hu/porthu.xml
+++ b/metadata.common.port.hu/porthu.xml
@@ -2,22 +2,25 @@
<GetPortHUPlotByTitleAndYear clearbuffers="no" dest="4">
<RegExp input="$$5" output="<details>\1</details>"
dest="4">
- <RegExp input="$$1" output="\1" dest="8">
- <expression encode="1">(.+)::.+</expression>
+ <RegExp input="$$1" output="\1" dest="15">
+ <expression encode="1">(.+)::.+::.+</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="11">
- <expression>(.+)::.+</expression>
+ <expression>(.+)::.+::.+</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="17">
- <expression encode="1">.+::(.+)</expression>
+ <expression encode="1">.+::(.+)::.+</expression>
+ </RegExp>
+ <RegExp input="$$1" output="\1" dest="8">
+ <expression encode="1">.+::.+::(.+)</expression>
</RegExp>
- <RegExp input="$$8" output="<url
function="ParsePortHUSearchResults"
cache="porthusearch-\1.html">http://www.port.hu/pls/ci/cinema.film_creator?i_text=\1&amp;i_film_creator=1</url>";
dest="5">
+ <RegExp input="$$15" output="<url
function="ParsePortHUSearchResults"
cache="porthusearch-\1.html">http://www.port.hu/pls/ci/cinema.film_creator?i_text=\1&amp;i_film_creator=1</url>";
dest="5">
<expression />
</RegExp>
<expression noclean="1" />
</RegExp>
</GetPortHUPlotByTitleAndYear>
- <ParsePortHUSearchResults dest="5">
+ <ParsePortHUSearchResults clearbuffers="no" dest="5">
<RegExp input="$$2" output="<details>\1</details>"
dest="5">
<RegExp input="$$1" output="\1" dest="9">
<expression
noclean="1">i_film_id=([\d]*)&amp;i_city_id=-1&amp;i_county_id=-1&amp;i_topic_id=2"
target="_top">[^<]*[^>]*>\s\($$11\)(((?!br).)*($$17)\)).*?</span></expression>
@@ -25,15 +28,21 @@
<RegExp input="$$1" output="\1" dest="9">
<expression
noclean="1">i_film_id=([\d]*)&amp;i_city_id=-1&amp;i_county_id=-1&amp;i_topic_id=2"
target="_top">$$11<(((?!br).)*($$17)\)).*?</span></expression>
</RegExp>
- <RegExp input="$$9" output="<url
function="ParsePortHUPlot"
cache="porthu-\1.html">http://www.port.hu/pls/fi/films.film_page?i_where=2&i_film_id=\1</url>";
dest="2">
- <expression>(.+)</expression>
- </RegExp>
- <RegExp input="$$1"
output="<plot>\1</plot>" dest="2">
+ <RegExp input="$$1" output="\1" dest="18">
<expression>class="txt">\($$11\).*?div
class="separator"></div>\n<span
class="txt">(.*?)</span></expression>
</RegExp>
- <RegExp input="$$1"
output="<plot>\1</plot>" dest="2">
+ <RegExp input="$$1" output="\1" dest="18">
<expression>class="blackbigtitle">$$11.*?div
class="separator"></div>\n<span
class="txt">(.*?)</span></expression>
</RegExp>
+ <RegExp input="$$18"
output="<plot>\1</plot>" dest="2">
+ <expression>(.+)</expression>
+ </RegExp>
+ <RegExp input="$$18" output="<url
function="ParseTMDBPlot"
cache="tmdb-hu-$$8.json">http://api.themoviedb.org/3/movie/$$8?api_key=57983e31fb435df4df77afb854740ea9&amp;language=hu</url>";
dest="2">
+ <expression>^$</expression>
+ </RegExp>
+ <RegExp input="$$9" output="<url
function="ParsePortHUPlot"
cache="porthu-\1.html">http://www.port.hu/pls/fi/films.film_page?i_where=2&i_film_id=\1</url>";
dest="2">
+ <expression>(.+)</expression>
+ </RegExp>
<expression noclean="1" />
</RegExp>
</ParsePortHUSearchResults>
@@ -51,13 +60,5 @@
<expression noclean="1" />
</RegExp>
</ParsePortHUPlot>
- <ParseFallbackTMDBPlot dest="5">
- <RegExp input="$$2" output="<details>\1</details>"
dest="5">
- <RegExp input="$$1"
output="<plot>\1</plot>" dest="2">
- <expression clear="yes"
fixchars="1">"overview":"(.*?)","</expression>
- </RegExp>
- <expression noclean="1" />
- </RegExp>
- </ParseFallbackTMDBPlot>
</scraperfunctions>
diff --git a/metadata.common.rt.com/addon.xml b/metadata.common.rt.com/addon.xml
index 521d859..bbca637 100644
--- a/metadata.common.rt.com/addon.xml
+++ b/metadata.common.rt.com/addon.xml
@@ -1,7 +1,7 @@
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<addon id="metadata.common.rt.com"
name="Rotten Tomatoes common scraper functions"
- version="1.3.0"
+ version="1.4.0"
provider-name="mortstar">
<requires>
<import addon="xbmc.metadata" version="1.0"/>
diff --git a/metadata.common.rt.com/changelog.txt
b/metadata.common.rt.com/changelog.txt
index 7ce357e..9c1fccb 100644
--- a/metadata.common.rt.com/changelog.txt
+++ b/metadata.common.rt.com/changelog.txt
@@ -1,3 +1,6 @@
+[B]1.4.0[/B]
+- Added: Common function to grab critics' consensus to 'plot' of movie
+
[B]1.3.0[/B]
- Added: Common function to grab plot
diff --git a/metadata.common.rt.com/rt.xml b/metadata.common.rt.com/rt.xml
index b328f7d..b67c3ec 100644
--- a/metadata.common.rt.com/rt.xml
+++ b/metadata.common.rt.com/rt.xml
@@ -18,20 +18,6 @@
</RegExp>
</GetRTRatingById>
- <!-- Get Rotten Tomatoes URL from IMDb ID for outline -->
- <GetRTOutlineById dest="5">
- <RegExp input="$$1" output="<details><url
function="ParseRTOutline"
cache="\1-rt.html">http://www.rottentomatoes.com/alias?type=imdbid&s=\1</url></details>";
dest="5">
- <expression>([0-9]{7})</expression>
- </RegExp>
- </GetRTOutlineById>
-
- <!-- Get Rotten Tomatoes URL from IMDb ID for plot -->
- <GetRTPlotById dest="5">
- <RegExp input="$$1" output="<details><url
function="ParseRTPlot"
cache="\1-rt.html">http://www.rottentomatoes.com/alias?type=imdbid&s=\1</url></details>";
dest="5">
- <expression>([0-9]{7})</expression>
- </RegExp>
- </GetRTPlotById>
-
<!-- TomatoMeter from All Critics -->
<ParseRTTomAll dest="5">
<RegExp input="$$1" output="<details><rating
max="100">\1</rating><votes>\2</votes></details>"
dest="5">
@@ -69,6 +55,11 @@
</ParseRTAveTop>
<!-- Critics' Consensus -->
+ <GetRTOutlineById dest="5">
+ <RegExp input="$$1" output="<details><url
function="ParseRTOutline"
cache="\1-rt.html">http://www.rottentomatoes.com/alias?type=imdbid&s=\1</url></details>";
dest="5">
+ <expression>([0-9]{7})</expression>
+ </RegExp>
+ </GetRTOutlineById>
<ParseRTOutline dest="5">
<RegExp input="$$1"
output="<details><outline>\1</outline></details>"
dest="5">
<expression
noclean="1">critic_consensus">([^<]+)</p></expression>
@@ -78,7 +69,27 @@
</RegExp>
</ParseRTOutline>
+ <!-- Critics' Consensus -->
+ <GetRTCConsensusToPlotById dest="5">
+ <RegExp input="$$1" output="<details><url
function="ParseRTCConsensusToPlot"
cache="\1-rt.html">http://www.rottentomatoes.com/alias?type=imdbid&s=\1</url></details>";
dest="5">
+ <expression>([0-9]{7})</expression>
+ </RegExp>
+ </GetRTCConsensusToPlotById>
+ <ParseRTCConsensusToPlot dest="5">
+ <RegExp input="$$1"
output="<details><plot>\1</plot></details>" dest="5">
+ <expression
noclean="1">critic_consensus">([^<]+)</p></expression>
+ </RegExp>
+ <RegExp input="$$5" output="" dest="5">
+ <expression><details><plot>No consensus
yet.</plot></details></expression>
+ </RegExp>
+ </ParseRTCConsensusToPlot>
+
<!-- Plot -->
+ <GetRTPlotById dest="5">
+ <RegExp input="$$1" output="<details><url
function="ParseRTPlot"
cache="\1-rt.html">http://www.rottentomatoes.com/alias?type=imdbid&s=\1</url></details>";
dest="5">
+ <expression>([0-9]{7})</expression>
+ </RegExp>
+ </GetRTPlotById>
<ParseRTPlot dest="5">
<RegExp input="$$1"
output="<details><plot>\1\2</plot></details>" dest="5">
<expression trim="1">class="movie_synopsis"
itemprop="description">[^\w]*([^\n]*)[^>]*>\s([^~<]*)</expression>
diff --git a/metadata.common.themoviedb.org/addon.xml
b/metadata.common.themoviedb.org/addon.xml
index 1c00e60..b58e150 100644
--- a/metadata.common.themoviedb.org/addon.xml
+++ b/metadata.common.themoviedb.org/addon.xml
@@ -1,7 +1,7 @@
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<addon id="metadata.common.themoviedb.org"
- name="The MovieDB common scraper functions"
- version="2.9.1"
+ name="The Movie Database Scraper Library"
+ version="2.9.2"
provider-name="Team XBMC">
<requires>
<import addon="xbmc.metadata" version="1.0"/>
diff --git a/metadata.common.themoviedb.org/tmdb.xml
b/metadata.common.themoviedb.org/tmdb.xml
index ab8b3df..6b6243f 100644
--- a/metadata.common.themoviedb.org/tmdb.xml
+++ b/metadata.common.themoviedb.org/tmdb.xml
@@ -132,17 +132,23 @@
</RegExp>
</ParseFallbackTMDBTagline>
- <GetTMDBSetByIdChain dest="4">
+ <GetTMDBSetByIdChain clearbuffers="no" dest="4">
<RegExp input="$$5" output="<details>\1</details>"
dest="4">
- <RegExp input="$$1" output="<url
function="ParseTMDBSet"
cache="tmdb-$INFO[language]-\1.json">http://api.themoviedb.org/3/movie/\1?api_key=57983e31fb435df4df77afb854740ea9&amp;language=$INFO[language]</url>";
dest="5">
+ <RegExp input="$$1" output="\1" dest="8">
+ <expression clear="yes" noclean="1" />
+ </RegExp>
+ <RegExp input="$$8" output="<url
function="ParseTMDBSet"
cache="tmdb-$INFO[language]-\1.json">http://api.themoviedb.org/3/movie/\1?api_key=57983e31fb435df4df77afb854740ea9&amp;language=$INFO[language]</url>";
dest="5">
<expression />
</RegExp>
<expression noclean="1" />
</RegExp>
</GetTMDBSetByIdChain>
- <GetTMDBLangSetByIdChain dest="4">
+ <GetTMDBLangSetByIdChain clearbuffers="no" dest="4">
<RegExp input="$$5" output="<details>\1</details>"
dest="4">
- <RegExp input="$$1" output="<url
function="ParseTMDBSet"
cache="tmdb-$INFO[tmdbsetlanguage]-\1.json">http://api.themoviedb.org/3/movie/\1?api_key=57983e31fb435df4df77afb854740ea9&amp;language=$INFO[tmdbsetlanguage]</url>";
dest="5">
+ <RegExp input="$$1" output="\1" dest="8">
+ <expression clear="yes" noclean="1" />
+ </RegExp>
+ <RegExp input="$$8" output="<url
function="ParseTMDBSet"
cache="tmdb-$INFO[tmdbtaglinelanguage]-\1.json">http://api.themoviedb.org/3/movie/\1?api_key=57983e31fb435df4df77afb854740ea9&amp;language=$INFO[tmdbtaglinelanguage]</url>";
dest="5">
<expression />
</RegExp>
<expression noclean="1" />
diff --git a/metadata.universal/addon.xml b/metadata.universal/addon.xml
index d97c63f..219c64a 100644
--- a/metadata.universal/addon.xml
+++ b/metadata.universal/addon.xml
@@ -1,7 +1,7 @@
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<addon id="metadata.universal"
- name="Universal Scraper"
- version="2.1.8"
+ name="Universal Movie Scraper"
+ version="2.3.0"
provider-name="Olympia, Team XBMC">
<requires>
<import addon="xbmc.metadata" version="1.0"/>
@@ -11,7 +11,7 @@
<import addon="metadata.common.impa.com" version="1.0.3"/>
<import addon="metadata.common.hdtrailers.net" version="1.0.7"/>
<import addon="metadata.common.youtubetrailers" version="1.0.4"/>
- <import addon="metadata.common.rt.com" version="1.3.0"/>
+ <import addon="metadata.common.rt.com" version="1.4.0"/>
<import addon="metadata.common.ofdb.de" version="1.0.0"/>
<import addon="metadata.common.trakt.tv" version="1.0.0"/>
<import addon="metadata.common.port.hu" version="1.0.0"/>
@@ -56,7 +56,7 @@
<description lang="da">Denne scraper indsamler information fra følgende
understøttede sider: IMDb, TMDb, Rotten Tomatoes, Trakt.tv, OFDb.de. Første
søgning sker altid i TMDb, men efterfølgende kan der felt for felt vælges,
hvilken side den specifikke infomation ønskes fra.</description>
<description lang="de">Dieser Sammler sammelt Informationen von den
folgenden, unterstützen Seiten: IMDb, TMDb, Rotten Tomatoes, Trakt.tv,
OFDb.de. Es fürt jedes mal die Anfangssuche auf TMDb aus, aber danach kann
Feld für Feld ausgewählt werden, woher die Information geholt werden
sollen.</description>
<description lang="el">ÎÏ
ÏÏ Ïο scraper ÏÏ
λλÎγει
ÏληÏοÏοÏÎ¯ÎµÏ Î±ÏÏ ÏÎ¹Ï Î±ÎºÏλοÏ
Î¸ÎµÏ Î¹ÏÏοÏελίδεÏ:
IMDb, TMDb, Rotten Tomatoes, Trakt.tv, OFDb.de. ΠαÏÏική
αναζήÏηÏη γίνεÏαι ÏάνÏα ÏÏο TMDb, αλλά ÎÏειÏα
μÏοÏεί να ÏÏ
θμιÏÏεί Ïεδίο ανά Ïεδίο αÏÏ
Ïοια Ïελίδα να λαμβάνονÏαι
ÏληÏοÏοÏίεÏ.</description>
- <description lang="en">This scraper collecting information from the
following supported sites: IMDb, TMDb, Rotten Tomatoes, Trakt.tv, OFDb.de. It
always does the initial search on TMDb, but following that it can be set field
by field that from which site you want that specific information.</description>
+ <description lang="en">Universal Scraper is currently the most
customizable scraper by collecting information from the following supported
sites: IMDb, TMDb, Rotten Tomatoes, Trakt.tv, OFDb.de, port.hu. This scraper is
currently the flagship of the Team-XBMC scrapers. The initial search can be
done either on TMDb or IMDb (according to the settings), but following that it
can be set field by field that from which site you want that specific
information.</description>
<description lang="es">Este buscador obtiene información de los
siguientes sitios: IMDb, TMDb, Rotten Tomatoes, Trakt.tv, OFDb.de. La primera
búsqueda siempre se realiza en TMDb, pero se puede elegir, campo por campo, de
qué sitio se tiene que obtener la información.</description>
<description lang="es_AR">Este buscador obtiene información de los
siguientes sitios: IMDb, TMDb, Rotten tomatoes, Trakt.tv, OFDb.de. Realiza la
búsqueda inicial en TMDb, pero a partir de allà se puede especificar desde
que sitio se desea obtener cada campo.</description>
<description lang="fr">Ce scraper collecte les informations depuis les
sites supportés suivants : IMDb, TMDb, Rotten Tomatoes, Trakt.tv, OFDb.de. La
recherche initiale est toujours effectuée depuis TMDb, mais vous pouvez
indiquer les sites sources des champs un par un pour avoir une information plus
spécifique.</description>
diff --git a/metadata.universal/changelog.txt b/metadata.universal/changelog.txt
index fdf8d2d..2a13096 100644
--- a/metadata.universal/changelog.txt
+++ b/metadata.universal/changelog.txt
@@ -1,5 +1,19 @@
+[B]2.3.0[/B]
+- fixed: various original title fixes
+- fixed: year won't scrape in certain cases
+- added: option to insert IMDb Outline as plot
+
+[B]2.2.0[/B]
+- fixed: original title won't scrape
+- fixed: year in case of TV Movies
+- added: option to insert RT critics concensus as plot
+- added: option to not scrape tagline
+
+[B]2.1.9[/B]
+- fixed: issue with original title
+
[B]2.1.8[/B]
-- fixed: brokage due to wrong versioning; this is the last update for Eden
+- cosmetic: synced version number with Eden
[B]2.1.7[/B]
- changed: added 'Short' to search results
diff --git a/metadata.universal/resources/settings.xml
b/metadata.universal/resources/settings.xml
index 59341d3..5633787 100644
--- a/metadata.universal/resources/settings.xml
+++ b/metadata.universal/resources/settings.xml
@@ -17,10 +17,10 @@
<category label="30400">
<setting type="lsep" label="30401"/>
<setting type="sep"/>
- <setting label="30014" type="labelenum"
values="IMDb|themoviedb.org|Rotten Tomatoes|trakt.tv|OFDb.de|port.hu"
id="plotsource" default="IMDb"/>
+ <setting label="30014" type="labelenum"
values="IMDb|themoviedb.org|Rotten Tomatoes|trakt.tv|OFDb.de|port.hu|Rotten
Tomatoes / Critics' Consensus|IMDb Outline" id="plotsource" default="IMDb"/>
<setting label="30020" type="labelenum"
values="bg|cs|da|de|el|en|es|fi|fr|he|hr|hu|it|ja|ko|nl|no|pl|pt|ru|sl|sv|tr|zh"
id="tmdbplotlanguage" default="en" visible="eq(-1,1)"/>
<setting type="sep"/>
- <setting label="30042" type="labelenum" values="IMDb|themoviedb.org"
id="taglinesource" default="IMDb"/>
+ <setting label="30042" type="labelenum"
values="IMDb|themoviedb.org|None" id="taglinesource" default="IMDb"/>
<setting label="30043" type="labelenum"
values="bg|cs|da|de|el|en|es|fi|fr|he|hr|hu|it|ja|ko|nl|no|pl|pt|ru|sl|sv|tr|zh"
id="tmdbtaglinelanguage" default="en" visible="eq(-1,1)"/>
<setting type="sep"/>
<setting label="30045" type="labelenum" values="IMDb|Rotten Tomatoes /
Critics' Consensus|None" id="outlinesource" default="IMDb"/>
diff --git a/metadata.universal/universal.xml b/metadata.universal/universal.xml
index eaba0d4..c449857 100644
--- a/metadata.universal/universal.xml
+++ b/metadata.universal/universal.xml
@@ -1,5 +1,5 @@
<?xml version="1.0" encoding="UTF-8"?>
-<scraper framework="1.1" date="2012-04-28">
+<scraper framework="1.1" date="2013-02-03">
<NfoUrl dest="3">
<RegExp input="$$1"
output="<url>http://akas.imdb.com/title/tt\1/</url><id>tt\1</id>";
dest="3">
<expression clear="yes"
noclean="1">imdb....?/Title\?([0-9]*)</expression>
@@ -43,10 +43,10 @@
<expression
clear="yes">/title/([t0-9]*)/(combined|faq|releaseinfo|vote)</expression>
</RegExp>
<RegExp input="$$1"
output="<entity><title>\1</title><year>\2</year><url
cache="$$7-main.html">http://akas.imdb.com/title/$$7/</url><id>$$7</id></entity>";
dest="5">
- <expression clear="yes" noclean="1"><meta
name="title"
content="(?:&#x22;)?([^"]*?)(?:&#x22;)?
\([^\(]*?([0-9]{4})\)</expression>
+ <expression clear="yes" noclean="1"><meta
name="title"
content="(?:&#x22;)?([^"]*?)(?:&#x22;)?
\([^\(]*?([0-9]{4})(?:â\s)?\)</expression>
</RegExp>
<RegExp input="$$1"
output="<entity><title>\2</title><year>\1</year><url
cache="$$7-main.html">http://akas.imdb.com/title/$$7/</url><id>$$7</id></entity>";
dest="5+">
- <expression fixchars="2" noclean="1"><meta
name="title" content="(?:&#x22;)?[^"]*?(?:&#x22;)?
\([^\(]*?([0-9]{4})\).*?Also Known As:</h4>([^\n]*)</expression>
+ <expression fixchars="2" noclean="1"><meta
name="title" content="(?:&#x22;)?[^"]*?(?:&#x22;)?
\([^\(]*?([0-9]{4})(?:â\s)?\).*?Also Known As:</h4>([^\n]*)</expression>
</RegExp>
<RegExp input="$$1" output="\1" dest="4">
<expression noclean="1"><table
class="findList"(.*?)</div</expression>
@@ -76,11 +76,14 @@
</GetAKASearchResults>
<GetDetails dest="3">
<RegExp input="$$5" output="<details>\1</details>"
dest="3">
- <RegExp conditional="imdbsearch" input="$$1"
output="\2" dest="19">
- <expression fixchars="2"><meta
name="title" content="(IMDb -
)?(?:&#x22;)?([^"]*?)(?:&#x22;)? \([^\(]*?([0-9]{4})\)</expression>
+ <RegExp input="$$1" output="\2" dest="19">
+ <expression fixchars="2"><meta
name="title" content="(IMDb -
)?(?:&#x22;)?([^"]*?)(?:&#x22;)?
\([^\(]*?([0-9]{4})(?:â\s)?\)</expression>
</RegExp>
- <RegExp conditional="imdbsearch" input="$$1"
output="\2" dest="20">
- <expression><meta name="title"
content="([^"]*?) \([^\(]*?([0-9]{4})\)</expression>
+ <RegExp input="$$1" output="\1" dest="19">
+ <expression fixchars="1"
trim="1"><br/><span\sclass="title-extra">([^<]*)<i>\(original
title\)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="\2" dest="20">
+ <expression><meta
name="title"\scontent="([^"]*?)\s\([^\(]*?([0-9]{4})(?:â\s)?\)</expression>
</RegExp>
<RegExp conditional="tmdbsearch" input="$$1"
output="\1" dest="2">
<expression
noclean="1">"id":([0-9]*),"imdb_id</expression>
@@ -88,10 +91,10 @@
<RegExp conditional="tmdbsearch" input="$$1"
output="\1" dest="2">
<expression
noclean="1">"id":[0-9]*,"imdb_id":"([^"]*)</expression>
</RegExp>
- <RegExp conditional="tmdbsearch" input="$$1"
output="\1" dest="19">
+ <RegExp input="$$1" output="\1" dest="19">
<expression
fixchars="1">"original_title":"([^"]*)</expression>
</RegExp>
- <RegExp conditional="tmdbsearch" input="$$1"
output="\1" dest="20">
+ <RegExp input="$$1" output="\1" dest="20">
<expression
noclean="1">"release_date":"([0-9]+)-</expression>
</RegExp>
<RegExp input="$$2" output="<id>\1</id>"
dest="5">
@@ -172,6 +175,12 @@
<RegExp input="$INFO[plotsource]" output="<chain
function="GetTMDBLangPlotByIdChain">$$2</chain>" dest="5+">
<expression>themoviedb.org</expression>
</RegExp>
+ <RegExp input="$INFO[plotsource]" output="<chain
function="GetRTCConsensusToPlotById">$$2</chain>" dest="5+">
+ <expression>Rotten Tomatoes / Critics'
Consensus</expression>
+ </RegExp>
+ <RegExp input="$INFO[plotsource]" output="<chain
function="GetIMDBOutlineById">$$2</chain>" dest="5+">
+ <expression>IMDb Outline</expression>
+ </RegExp>
<RegExp input="$INFO[taglinesource]" output="<chain
function="GetIMDBTaglineById">$$2</chain>" dest="5+">
<expression>IMDb</expression>
</RegExp>
@@ -187,7 +196,7 @@
<RegExp input="$INFO[plotsource]" output="<chain
function="GetTRAKTTVPlotByIdChain">$$2</chain>" dest="5+">
<expression>trakt.tv</expression>
</RegExp>
- <RegExp input="$INFO[plotsource]" output="<chain
function="GetPortHUPlotByTitleAndYear">$$19::$$20</chain>"
dest="5+">
+ <RegExp input="$INFO[plotsource]" output="<chain
function="GetPortHUPlotByTitleAndYear">$$19::$$20::$$2</chain>"
dest="5+">
<expression>port.hu</expression>
</RegExp>
<RegExp input="$INFO[creditssource]" output="<chain
function="GetIMDBCastById">$$2</chain>" dest="5+">
-----------------------------------------------------------------------
Summary of changes:
metadata.common.imdb.com/addon.xml | 4 +-
metadata.common.imdb.com/changelog.txt | 12 +++++
metadata.common.imdb.com/imdb.xml | 70 ++++++++++++++--------------
metadata.common.port.hu/addon.xml | 4 +-
metadata.common.port.hu/porthu.xml | 39 ++++++++--------
metadata.common.rt.com/addon.xml | 2 +-
metadata.common.rt.com/changelog.txt | 3 +
metadata.common.rt.com/rt.xml | 39 ++++++++++------
metadata.common.themoviedb.org/addon.xml | 4 +-
metadata.common.themoviedb.org/tmdb.xml | 14 ++++--
metadata.universal/addon.xml | 8 ++--
metadata.universal/changelog.txt | 16 ++++++-
metadata.universal/resources/settings.xml | 4 +-
metadata.universal/universal.xml | 29 ++++++++----
14 files changed, 152 insertions(+), 96 deletions(-)
hooks/post-receive
--
Scrapers
------------------------------------------------------------------------------
Free Next-Gen Firewall Hardware Offer
Buy your Sophos next-gen firewall before the end March 2013
and get the hardware for free! Learn more.
http://p.sf.net/sfu/sophos-d2d-feb
_______________________________________________
Xbmc-addons mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xbmc-addons
