389 lines
24 KiB
XML
389 lines
24 KiB
XML
<?xml version="1.0" encoding="UTF-8"?>
|
||
<scraperfunctions>
|
||
<GetIMDBGenresById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBGenres">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBGenresById>
|
||
<ParseIMDBGenres dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression noclean="1">,"genre":\[([^\]]+)]</expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<genre>\1</genre>" dest="2">
|
||
<expression repeat="yes">"([^"]+)"</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression noclean="1">,\s*"genre":\s("[^"]+"),</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression noclean="1">"genre":\s\[(.*?)\]</expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<genre>\1</genre>" dest="2">
|
||
<expression repeat="yes">"([^"]+)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBGenres>
|
||
|
||
<GetIMDBRatingsById dest="5" clearbuffers="no">
|
||
<RegExp input="$$1" output="<details><url cache="\1-main.html" function="ParseIMDBRatings">https://www.imdb.com/title/\1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1">^(tt\d+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="default="true"" dest="3">
|
||
<expression clear="yes">\|default$</expression>
|
||
</RegExp>
|
||
</GetIMDBRatingsById>
|
||
<ParseIMDBRatings dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="<ratings><rating name="imdb" $$3><value>\2</value><votes>\1</votes></rating></ratings>" dest="2">
|
||
<expression>"ratingCount":([0-9,]+),*"bestRating":[^"]*"worstRating":[^"]*"ratingValue":([0-9.]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="<ratings><rating name="imdb" $$3><value>\2</value><votes>\1</votes></rating></ratings>" dest="2+">
|
||
<expression>"ratingCount":\s([0-9,]+),\s*"bestRating":\s"[^"]*",\s*"worstRating":\s"[^"]*",\s*"ratingValue":\s"([0-9.]+)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBRatings>
|
||
|
||
<GetIMDBTOP250ById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBTOP250">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBTOP250ById>
|
||
<ParseIMDBTOP250 dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="<top250>\1</top250>" dest="2">
|
||
<expression>Top\srated\smovie\s#([0-9]*)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="<top250>\1</top250>" dest="2+">
|
||
<expression>Top\sRated\sMovies\s#([0-9]*)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBTOP250>
|
||
|
||
<GetIMDBStudioById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBStudio">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBStudioById>
|
||
<ParseIMDBStudio dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="<studio>\1</studio>" dest="2">
|
||
<expression>"/company/[^"]+"\s*>([^<]+)</a></expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBStudio>
|
||
|
||
<GetIMDBCountryById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBCountry">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBCountryById>
|
||
<ParseIMDBCountry dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression clear="yes" noclean="1">Country:[^>]+>(.*?)</div></expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<country>\1</country>" dest="2">
|
||
<expression repeat="yes">href="/search/title/?\?country_of_origin[^>]+>(.*?)<</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="<country>\1</country>" dest="2">
|
||
<expression repeat="yes">{"id":"[^"]*","text":"([^"]*)","__typename":"CountryOfOrigin"}</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBCountry>
|
||
|
||
<GetMetaCriticRatingsById dest="5" clearbuffers="no">
|
||
<RegExp input="$$1" output="<details><url cache="\1-criticreviews.html" function="ParseMetaCriticRatings">https://www.imdb.com/title/\1/criticreviews/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1">(tt\d+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="default="true"" dest="3">
|
||
<expression clear="yes">\|default$</expression>
|
||
</RegExp>
|
||
</GetMetaCriticRatingsById>
|
||
<ParseMetaCriticRatings dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$4" output="<ratings><rating name="metacritic" max="100" $$3>\1</rating></ratings>" dest="2">
|
||
<RegExp input="$$1" output="<value>\1</value>" dest="4">
|
||
<expression clear="yes">"reviewCount":\d+,"score":(\d+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="<votes>\1</votes>" dest="4+">
|
||
<expression>"reviewCount":(\d+)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" clear="yes"/>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseMetaCriticRatings>
|
||
|
||
<GetIMDBPlotById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-reference.html" function="ParseIMDBPlot">https://www.imdb.com/title/$$1/reference|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBPlotById>
|
||
<ParseIMDBPlot dest="5">
|
||
<RegExp input="$$2" output="<details><plot>\1</plot></details>" dest="5">
|
||
<RegExp input="$$1" output="\1" dest="2">
|
||
<expression fixchars="1" trim="1" clear="yes">"summaries":.*?"plaidHtml":"([^"]*)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBPlot>
|
||
|
||
<GetIMDBTaglineById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-reference.html" function="ParseIMDBTagline">https://www.imdb.com/title/$$1/reference|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBTaglineById>
|
||
<ParseIMDBTagline dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="<tagline>\1</tagline>" dest="2">
|
||
<expression fixchars="1" trim="1">"taglines":\{"edges":\[\{"node":\{"text":"([^"]+)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBTagline>
|
||
|
||
<GetIMDBOutlineById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBOutline">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBOutlineById>
|
||
<ParseIMDBOutline dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="<outline>\1</outline>" dest="2">
|
||
<expression fixchars="1" trim="1">data-testid="plot-xl"[^>]*>([^<]*)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBOutline>
|
||
|
||
<GetIMDBOutlineToPlotById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBOutlineToPlot">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBOutlineToPlotById>
|
||
<ParseIMDBOutlineToPlot 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">data-testid="plot-xl"[^>]*>([^<]*)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBOutlineToPlot>
|
||
|
||
<GetIMDBCastById dest="5"> <!-- Only Topcast -->
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBCast">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBCastById>
|
||
<ParseIMDBCast dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="<actor><thumb>\1\2</thumb><name>\3</name><role>\4</role></actor>" dest="2">
|
||
<expression repeat="yes" fixchars="1" noclean="1">data-testid="title-cast-item".*?(?:src="([^"]*)".*?data-testid="title-cast-item__actor"|data-testid="title-cast-item__actor".*?src="([^"]*)") href="[^"]*"[^>]*>([^<]+)</a>.*?/characters/[^"]*".*?><span[^>]*>([^<]+)</span></expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBCast>
|
||
|
||
<GetIMDBDirectorsById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBDirectors">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBDirectorsById>
|
||
<ParseIMDBDirectors dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression clear="yes" noclean="1">"grouping":\{"text":"Director".*?"credits":\[(.*?)\]</expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<director>\1</director>" dest="2">
|
||
<expression repeat="yes" fixchars="yes" clear="yes">"nameText":\{"text":"([^"]+)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBDirectors>
|
||
|
||
<GetIMDBWritersById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBWriters">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBWritersById>
|
||
<ParseIMDBWriters dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression clear="yes" noclean="1">"grouping":\{"text":"Writers".*?"credits":\[(.*?)\]</expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<credits>\1</credits>" dest="2">
|
||
<expression repeat="yes" fixchars="yes" clear="yes">"nameText":\{"text":"([^"]+)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBWriters>
|
||
|
||
<GetIMDBFullCastById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBCast">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBFullCastById>
|
||
|
||
<GetIMDBFullDirectorsById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-fullcredits.html" function="ParseIMDBFullDirectors">https://www.imdb.com/title/$$1/fullcredits/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=text/html</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBFullDirectorsById>
|
||
<ParseIMDBFullDirectors dest="5">
|
||
<RegExp input="$$2" output="<details>\1</details>" dest="5">
|
||
<RegExp input="$$1" output="\1" dest="6">
|
||
<expression clear="yes" noclean="1">"directorsPageTitle":\[\{"credits":\[(.*?)\]\}</expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<director>\1</director>" dest="2">
|
||
<expression repeat="yes" fixchars="1" clear="yes">"nameText":\{"text":"([^"]+)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBFullDirectors>
|
||
|
||
<GetIMDBFullWritersById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-fullcredits.html" function="ParseIMDBFullWriters">https://www.imdb.com/title/$$1/fullcredits/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=text/html</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 clear="yes" noclean="1">"Writers"(.+?)"Cast"</expression>
|
||
</RegExp>
|
||
<RegExp input="$$6" output="<credits>\1</credits>" dest="2">
|
||
<expression repeat="yes" fixchars="yes" clear="yes">"nameText":\{"text":"([^"]+)"</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBFullWriters>
|
||
|
||
<GetIMDBThumbsById dest="5"> <!-- Broken -->
|
||
<RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBThumbs">https://www.imdb.com/title/$$1/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</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="<thumb aspect="poster">\1\2.jpg</thumb>" dest="6">
|
||
<expression noclean="1">Poster"\s*src="(.*?)(_V._)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBThumbs>
|
||
|
||
<GetIMDBUSACert dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-reference.html" function="ParseIMDBUSACert">https://www.imdb.com/title/$$1/reference/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBUSACert>
|
||
<ParseIMDBUSACert dest="5">
|
||
<RegExp input="$$1" output="<details><mpaa>$INFO[certprefix]\1</mpaa></details>" dest="5">
|
||
<expression>"rating":"([^"]+)",[^}]*?"ratingsBody":\{"id":"MPAA"[^\}]*\},"country":\{"id":"US"</expression>
|
||
</RegExp>
|
||
</ParseIMDBUSACert>
|
||
|
||
<GetIMDBCountryCert dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-reference.html" function="ParseIMDBCountryCert">https://www.imdb.com/title/$$1/reference/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
<RegExp input="$INFO[imdbcertcountry]" output="$$2" dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-reference.html" function="ParseIMDBUSACert">https://www.imdb.com/title/$$1/reference/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="2">
|
||
<expression noclean="1"/>
|
||
</RegExp>
|
||
<expression>United States</expression>
|
||
</RegExp>
|
||
</GetIMDBCountryCert>
|
||
<ParseIMDBCountryCert dest="5">
|
||
<RegExp input="$$1" output="<details><mpaa>$INFO[certprefix]\1</mpaa></details>" dest="5">
|
||
<expression>"rating":"([^"]+)".*?"country":\{"id":"[A-Z]+","text":"$INFO[imdbcertcountry]"</expression>
|
||
</RegExp>
|
||
</ParseIMDBCountryCert>
|
||
|
||
<GetIMDBAKATitlesById dest="5">
|
||
<RegExp input="$$1" output="<details><url cache="$$1-releaseinfo.html" function="ParseIMDBAKATitles">https://www.imdb.com/title/$$1/releaseinfo/|accept-language=en-us&User-Agent=Mozilla/5.0%20(Windows%20NT%2010.0;%20Win64;%20x64)%20AppleWebKit/537.36%20(KHTML,%20like%20Gecko)%20Chrome/87.0.4280.88%20Safari/537.36&Accept=application/json</url></details>" dest="5">
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</GetIMDBAKATitlesById>
|
||
<ParseIMDBAKATitles dest="5">
|
||
<RegExp input="$$2" output="<details><title>\1</title></details>" dest="5">
|
||
<RegExp input="$$1" output="\2" dest="2">
|
||
<expression>'og:title'\scontent="(IMDb\s-\s)?(?:&#x22;)?([^"]*?)(?:&#x22;)? \([^\(]*?([0-9]{4})(?:–\s)?\)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$10" output="\1" dest="4">
|
||
<RegExp input="$$1" output="\1" dest="9">
|
||
<expression>name">\s\(original\stitle\)[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="9">
|
||
<expression>class="originalTitle">([^<]*)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$9" output="\1" dest="10">
|
||
<expression />
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="11">
|
||
<expression>'og:title'\scontent="(IMDb\s-\s)?(?:&#x22;)?([^"]*?)(?:&#x22;)? \([^\(]*?([0-9]{4})(?:–\s)?\)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$11" output="\1" dest="10">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="11">
|
||
<expression clear="yes">name">Hong Kong\s\(English[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$11" output="\1" dest="10">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="11">
|
||
<expression clear="yes">name">Canada\s\(English[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$11" output="\1" dest="10">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="11">
|
||
<expression clear="yes">name">UK<[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$11" output="\1" dest="10">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="11">
|
||
<expression clear="yes">name">World-wide\s\(English\stitle\)<[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$11" output="\1" dest="10">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="11">
|
||
<expression clear="yes">name">USA<[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$11" output="\1" dest="10">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
<RegExp input="$$4" output="\1" dest="2">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="4">
|
||
<expression>name">$INFO[imdbakatitles][^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$1" output="\1" dest="4">
|
||
<expression>name">$INFO[imdbakatitles]<[^>]+>[^>]+>([^<]+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$$4" output="\1" dest="2">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<RegExp input="$INFO[imdbakatitles]" output="$$9" dest="4">
|
||
<expression>Keep Original</expression>
|
||
</RegExp>
|
||
<RegExp input="$$4" output="\1" dest="2">
|
||
<expression>(.+)</expression>
|
||
</RegExp>
|
||
<expression noclean="1" />
|
||
</RegExp>
|
||
</ParseIMDBAKATitles>
|
||
|
||
</scraperfunctions> |