svn commit: r1066576 - in /lenya/branches/BRANCH_2_1_X/src: modules/lucene/config/cocoon-xconf/analyzer_manager.xconf webapp/WEB-INF/analyzer/stopword/Arabic_ar.xml webapp/WEB-INF/analyzer/stopword/Russian_ru.xml

Author: rudolf
Date: Wed Feb  2 18:45:29 2011
New Revision: 1066576

URL: http://svn.apache.org/viewvc?rev=1066576&view=rev
Log:
add russian and arabic stopword files

Added:
    lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Arabic_ar.xml
Modified:
    lenya/branches/BRANCH_2_1_X/src/modules/lucene/config/cocoon-xconf/analyzer_manager.xconf
    lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Russian_ru.xml

Modified: lenya/branches/BRANCH_2_1_X/src/modules/lucene/config/cocoon-xconf/analyzer_manager.xconf
URL: http://svn.apache.org/viewvc/lenya/branches/BRANCH_2_1_X/src/modules/lucene/config/cocoon-xconf/analyzer_manager.xconf?rev=1066576&r1=1066575&r2=1066576&view=diff
==============================================================================
--- lenya/branches/BRANCH_2_1_X/src/modules/lucene/config/cocoon-xconf/analyzer_manager.xconf (original)
+++ lenya/branches/BRANCH_2_1_X/src/modules/lucene/config/cocoon-xconf/analyzer_manager.xconf Wed Feb  2 18:45:29 2011
@@ -21,6 +21,7 @@
 
 	<analyzer_manager logger="core.search.analyzer_manager">
 		<analyzer id="standard" class="org.apache.lucene.analysis.standard.StandardAnalyzer"/>
+	  <analyzer id="stopword_ar" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Arabic_ar.xml" checkupdate="true"/>
 	  <analyzer id="stopword_da" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Danish_da.xml" checkupdate="true"/>
 	  <analyzer id="stopword_nl" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Dutch_nl.xml" checkupdate="true"/>
 	  <analyzer id="stopword_en" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/English_en.xml" checkupdate="true"/>
@@ -31,9 +32,7 @@
 	  <analyzer id="stopword_no" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Norwegian_no.xml" checkupdate="true"/>
 	  <analyzer id="stopword_pl" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Polish_pl.xml" checkupdate="true"/>
 	  <analyzer id="stopword_pt" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Portuguese_pt.xml" checkupdate="true"/>
-	  <!--
 	  <analyzer id="stopword_ru" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Russian_ru.xml" checkupdate="true"/>
-	  -->
 	  <analyzer id="stopword_es" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Spanish_es.xml" checkupdate="true"/>
 	  <analyzer id="stopword_sv" class="org.apache.cocoon.components.search.analyzer.ConfigurableStopwordAnalyzer" configfile="context://WEB-INF/analyzer/stopword/Swedish_sv.xml" checkupdate="true"/>
 	  <analyzer id="analyzer_research" class="org.apache.cocoon.components.search.analyzer.ConfigurablePerFieldAnalyzer" configfile="context://WEB-INF/analyzer/perfieldconf.xml" checkupdate="false"/>

Added: lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Arabic_ar.xml
URL: http://svn.apache.org/viewvc/lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Arabic_ar.xml?rev=1066576&view=auto
==============================================================================
--- lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Arabic_ar.xml (added)
+++ lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Arabic_ar.xml Wed Feb  2 18:45:29 2011
@@ -0,0 +1,124 @@
+<?xml version="1.0" encoding="UTF-8"?>
+
+<stopwords>
+	<stopword>من</stopword>
+	<stopword>ومن</stopword>
+	<stopword>منها</stopword>
+	<stopword>منه</stopword>
+	<stopword>في</stopword>
+	<stopword>وفي</stopword>
+	<stopword>فيها</stopword>
+	<stopword>فيه</stopword>
+	<stopword>و</stopword>
+	<stopword>ف</stopword>
+	<stopword>ثم</stopword>
+	<stopword>او</stopword>
+	<stopword>أو</stopword>
+	<stopword>ب</stopword>
+	<stopword>بها</stopword>
+	<stopword>به</stopword>
+	<stopword>ا</stopword>
+	<stopword>Ø£</stopword>
+	<stopword>اى</stopword>
+	<stopword>اي</stopword>
+	<stopword>أي</stopword>
+	<stopword>أى</stopword>
+	<stopword>لا</stopword>
+	<stopword>ولا</stopword>
+	<stopword>الا</stopword>
+	<stopword>ألا</stopword>
+	<stopword>إلا</stopword>
+	<stopword>لكن</stopword>
+	<stopword>ما</stopword>
+	<stopword>وما</stopword>
+	<stopword>كما</stopword>
+	<stopword>فما</stopword>
+	<stopword>عن</stopword>
+	<stopword>مع</stopword>
+	<stopword>اذا</stopword>
+	<stopword>إذا</stopword>
+	<stopword>ان</stopword>
+	<stopword>أن</stopword>
+	<stopword>إن</stopword>
+	<stopword>انها</stopword>
+	<stopword>أنها</stopword>
+	<stopword>إنها</stopword>
+	<stopword>انه</stopword>
+	<stopword>أنه</stopword>
+	<stopword>إنه</stopword>
+	<stopword>بان</stopword>
+	<stopword>بأن</stopword>
+	<stopword>فان</stopword>
+	<stopword>فأن</stopword>
+	<stopword>وان</stopword>
+	<stopword>وأن</stopword>
+	<stopword>وإن</stopword>
+	<stopword>التى</stopword>
+	<stopword>التي</stopword>
+	<stopword>الذى</stopword>
+	<stopword>الذي</stopword>
+	<stopword>الذين</stopword>
+	<stopword>الى</stopword>
+	<stopword>الي</stopword>
+	<stopword>إلى</stopword>
+	<stopword>إلي</stopword>
+	<stopword>على</stopword>
+	<stopword>عليها</stopword>
+	<stopword>عليه</stopword>
+	<stopword>اما</stopword>
+	<stopword>أما</stopword>
+	<stopword>إما</stopword>
+	<stopword>ايضا</stopword>
+	<stopword>أيضا</stopword>
+	<stopword>كل</stopword>
+	<stopword>وكل</stopword>
+	<stopword>لم</stopword>
+	<stopword>ولم</stopword>
+	<stopword>لن</stopword>
+	<stopword>ولن</stopword>
+	<stopword>هى</stopword>
+	<stopword>هي</stopword>
+	<stopword>هو</stopword>
+	<stopword>وهى</stopword>
+	<stopword>وهي</stopword>
+	<stopword>وهو</stopword>
+	<stopword>فهى</stopword>
+	<stopword>فهي</stopword>
+	<stopword>فهو</stopword>
+	<stopword>انت</stopword>
+	<stopword>أنت</stopword>
+	<stopword>لك</stopword>
+	<stopword>لها</stopword>
+	<stopword>له</stopword>
+	<stopword>هذه</stopword>
+	<stopword>هذا</stopword>
+	<stopword>تلك</stopword>
+	<stopword>ذلك</stopword>
+	<stopword>هناك</stopword>
+	<stopword>كانت</stopword>
+	<stopword>كان</stopword>
+	<stopword>يكون</stopword>
+	<stopword>تكون</stopword>
+	<stopword>وكانت</stopword>
+	<stopword>وكان</stopword>
+	<stopword>غير</stopword>
+	<stopword>بعض</stopword>
+	<stopword>قد</stopword>
+	<stopword>نحو</stopword>
+	<stopword>بين</stopword>
+	<stopword>بينما</stopword>
+	<stopword>منذ</stopword>
+	<stopword>ضمن</stopword>
+	<stopword>حيث</stopword>
+	<stopword>الان</stopword>
+	<stopword>الآن</stopword>
+	<stopword>خلال</stopword>
+	<stopword>بعد</stopword>
+	<stopword>قبل</stopword>
+	<stopword>حتى</stopword>
+	<stopword>عند</stopword>
+	<stopword>عندما</stopword>
+	<stopword>لدى</stopword>
+	<stopword>جميع</stopword>
+</stopwords>
+

Modified: lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Russian_ru.xml
URL: http://svn.apache.org/viewvc/lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Russian_ru.xml?rev=1066576&r1=1066575&r2=1066576&view=diff
==============================================================================
--- lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Russian_ru.xml (original)
+++ lenya/branches/BRANCH_2_1_X/src/webapp/WEB-INF/analyzer/stopword/Russian_ru.xml Wed Feb  2 18:45:29 2011
@@ -1,101 +1,164 @@
 <?xml version="1.0" encoding="UTF-8"?>
+
 <stopwords>
-<stopword>а</stopword>
-<stopword>без</stopword>
-<stopword>более</stopword>
-<stopword>бы</stopword>
-<stopword>был</stopword>
-<stopword>была</stopword>
-<stopword>были</stopword>
-<stopword>было</stopword>
-<stopword>быть</stopword>
-<stopword>в</stopword>
-<stopword>вам</stopword>
-<stopword>ваÑ?</stopword>
-<stopword>веÑ?ь</stopword>
-<stopword>во</stopword>
-<stopword>вот</stopword>
-<stopword>вÑ?е</stopword>
-<stopword>вÑ?его</stopword>
-<stopword>вÑ?ех</stopword>
-<stopword>вы</stopword>
-<stopword>где</stopword>
-<stopword>да</stopword>
-<stopword>даже</stopword>
-<stopword>длÑ?</stopword>
-<stopword>до</stopword>
-<stopword>его</stopword>
-<stopword>ее</stopword>
-<stopword>еÑ?ли</stopword>
-<stopword>еÑ?ть</stopword>
-<stopword>еще</stopword>
-<stopword>же</stopword>
-<stopword>за</stopword>
-<stopword>здеÑ?ь</stopword>
-<stopword>и</stopword>
-<stopword>из</stopword>
-<stopword>или</stopword>
-<stopword>им</stopword>
-<stopword>их</stopword>
-<stopword>к</stopword>
-<stopword>как</stopword>
-<stopword>ко</stopword>
-<stopword>когда</stopword>
-<stopword>кто</stopword>
-<stopword>ли</stopword>
-<stopword>либо</stopword>
-<stopword>мне</stopword>
-<stopword>может</stopword>
-<stopword>мы</stopword>
-<stopword>на</stopword>
-<stopword>надо</stopword>
-<stopword>наш</stopword>
-<stopword>не</stopword>
-<stopword>него</stopword>
-<stopword>нее</stopword>
-<stopword>нет</stopword>
-<stopword>ни</stopword>
-<stopword>них</stopword>
-<stopword>но</stopword>
-<stopword>ну</stopword>
-<stopword>о</stopword>
-<stopword>об</stopword>
-<stopword>однако</stopword>
-<stopword>он</stopword>
-<stopword>она</stopword>
-<stopword>они</stopword>
-<stopword>оно</stopword>
-<stopword>от</stopword>
-<stopword>очень</stopword>
-<stopword>по</stopword>
-<stopword>под</stopword>
-<stopword>при</stopword>
-<stopword>Ñ?</stopword>
-<stopword>Ñ?о</stopword>
-<stopword>так</stopword>
-<stopword>также</stopword>
-<stopword>такой</stopword>
-<stopword>там</stopword>
-<stopword>те</stopword>
-<stopword>тем</stopword>
-<stopword>то</stopword>
-<stopword>того</stopword>
-<stopword>тоже</stopword>
-<stopword>той</stopword>
-<stopword>только</stopword>
-<stopword>том</stopword>
-<stopword>ты</stopword>
-<stopword>у</stopword>
-<stopword>уже</stopword>
-<stopword>хотÑ?</stopword>
-<stopword>чего</stopword>
-<stopword>чей</stopword>
-<stopword>чем</stopword>
-<stopword>что</stopword>
-<stopword>чтобы</stopword>
-<stopword>чье</stopword>
-<stopword>чьÑ?</stopword>
-<stopword>Ñ?та</stopword>
-<stopword>Ñ?ти</stopword>
-<stopword>Ñ?то</stopword>
-<stopword>Ñ?</stopword></stopwords>
+	<stopword>и</stopword>
+	<stopword>в</stopword>
+	<stopword>во</stopword>
+	<stopword>не</stopword>
+	<stopword>что</stopword>
+	<stopword>он</stopword>
+	<stopword>на</stopword>
+	<stopword>я</stopword>
+	<stopword>с</stopword>
+	<stopword>со</stopword>
+	<stopword>как</stopword>
+	<stopword>а</stopword>
+	<stopword>то</stopword>
+	<stopword>все</stopword>
+	<stopword>она</stopword>
+	<stopword>так</stopword>
+	<stopword>его</stopword>
+	<stopword>но</stopword>
+	<stopword>да</stopword>
+	<stopword>ты</stopword>
+	<stopword>к</stopword>
+	<stopword>у</stopword>
+	<stopword>же</stopword>
+	<stopword>вы</stopword>
+	<stopword>за</stopword>
+	<stopword>бы</stopword>
+	<stopword>по</stopword>
+	<stopword>только</stopword>
+	<stopword>ее</stopword>
+	<stopword>мне</stopword>
+	<stopword>было</stopword>
+	<stopword>вот</stopword>
+	<stopword>от</stopword>
+	<stopword>меня</stopword>
+	<stopword>еще</stopword>
+	<stopword>нет</stopword>
+	<stopword>о</stopword>
+	<stopword>из</stopword>
+	<stopword>ему</stopword>
+	<stopword>теперь</stopword>
+	<stopword>когда</stopword>
+	<stopword>даже</stopword>
+	<stopword>ну</stopword>
+	<stopword>вдруг</stopword>
+	<stopword>ли</stopword>
+	<stopword>если</stopword>
+	<stopword>уже</stopword>
+	<stopword>или</stopword>
+	<stopword>ни</stopword>
+	<stopword>быть</stopword>
+	<stopword>был</stopword>
+	<stopword>него</stopword>
+	<stopword>до</stopword>
+	<stopword>вас</stopword>
+	<stopword>нибудь</stopword>
+	<stopword>опять</stopword>
+	<stopword>уж</stopword>
+	<stopword>вам</stopword>
+	<stopword>сказал</stopword>
+	<stopword>ведь</stopword>
+	<stopword>там</stopword>
+	<stopword>потом</stopword>
+	<stopword>себя</stopword>
+	<stopword>ничего</stopword>
+	<stopword>ей</stopword>
+	<stopword>может</stopword>
+	<stopword>они</stopword>
+	<stopword>тут</stopword>
+	<stopword>где</stopword>
+	<stopword>есть</stopword>
+	<stopword>надо</stopword>
+	<stopword>ней</stopword>
+	<stopword>для</stopword>
+	<stopword>мы</stopword>
+	<stopword>тебя</stopword>
+	<stopword>их</stopword>
+	<stopword>чем</stopword>
+	<stopword>была</stopword>
+	<stopword>сам</stopword>
+	<stopword>чтоб</stopword>
+	<stopword>без</stopword>
+	<stopword>будто</stopword>
+	<stopword>человек</stopword>
+	<stopword>чего</stopword>
+	<stopword>раз</stopword>
+	<stopword>тоже</stopword>
+	<stopword>себе</stopword>
+	<stopword>под</stopword>
+	<stopword>жизнь</stopword>
+	<stopword>будет</stopword>
+	<stopword>ж</stopword>
+  <stopword>тогда</stopword>
+	<stopword>кто</stopword>
+	<stopword>этот</stopword>
+	<stopword>говорил</stopword>
+	<stopword>того</stopword>
+	<stopword>потому</stopword>
+	<stopword>этого</stopword>
+	<stopword>какой</stopword>
+	<stopword>совсем</stopword>
+	<stopword>ним</stopword>
+	<stopword>здесь</stopword>
+	<stopword>этом</stopword>
+	<stopword>один</stopword>
+	<stopword>почти</stopword>
+	<stopword>мой</stopword>
+	<stopword>тем</stopword>
+	<stopword>чтобы</stopword>
+	<stopword>нее</stopword>
+	<stopword>кажется</stopword>
+	<stopword>сейчас</stopword>
+	<stopword>были</stopword>
+	<stopword>куда</stopword>
+	<stopword>зачем</stopword>
+	<stopword>сказать</stopword>
+	<stopword>всех</stopword>
+	<stopword>никогда</stopword>
+	<stopword>сегодня</stopword>
+	<stopword>можно</stopword>
+	<stopword>при</stopword>
+	<stopword>наконец</stopword>
+	<stopword>два</stopword>
+	<stopword>об</stopword>
+	<stopword>другой</stopword>
+	<stopword>хоть</stopword>
+	<stopword>после</stopword>
+	<stopword>над</stopword>
+	<stopword>больше</stopword>
+	<stopword>тот</stopword>
+	<stopword>через</stopword>
+	<stopword>эти</stopword>
+	<stopword>нас</stopword>
+	<stopword>про</stopword>
+	<stopword>всего</stopword>
+	<stopword>них</stopword>
+	<stopword>какая</stopword>
+	<stopword>много</stopword>
+	<stopword>разве</stopword>
+	<stopword>сказала</stopword>
+	<stopword>три</stopword>
+	<stopword>эту</stopword>
+  <stopword>моя</stopword>
+	<stopword>впрочем</stopword>
+	<stopword>хорошо</stopword>
+	<stopword>свою</stopword>
+	<stopword>этой</stopword>
+	<stopword>перед</stopword>
+	<stopword>иногда</stopword>
+	<stopword>лучше</stopword>
+	<stopword>чуть</stopword>
+	<stopword>том</stopword>
+	<stopword>нельзя</stopword>
+	<stopword>такой</stopword>
+	<stopword>им</stopword>
+	<stopword>более</stopword>
+	<stopword>всегда</stopword>
+	<stopword>конечно</stopword>
+	<stopword>всю</stopword>
+	<stopword>между</stopword>
+</stopwords>
+