Changeset 31859 for gs3-extensions/solr/trunk

Timestamp:

2017-08-08T02:00:54+12:00 (7 years ago)

Author:

Georgiy Litvinov

Message:

Replaced russian morphology analyzer by much more efficient version.

Location:

gs3-extensions/solr/trunk/src

Files:

• build.xml (modified) (1 diff)
• conf/lang/pymorphy2_dicts (added)
• conf/lang/pymorphy2_dicts/grammemes.json (added)
• conf/lang/pymorphy2_dicts/gramtab-opencorpora-ext.json (added)
• conf/lang/pymorphy2_dicts/gramtab-opencorpora-int.json (added)
• conf/lang/pymorphy2_dicts/meta.json (added)
• conf/lang/pymorphy2_dicts/p_t_given_w.intdawg (added)
• conf/lang/pymorphy2_dicts/paradigms.array (added)
• conf/lang/pymorphy2_dicts/prediction-suffixes-0.dawg (added)
• conf/lang/pymorphy2_dicts/prediction-suffixes-1.dawg (added)
• conf/lang/pymorphy2_dicts/prediction-suffixes-2.dawg (added)
• conf/lang/pymorphy2_dicts/suffixes.json (added)
• conf/lang/pymorphy2_dicts/words.dawg (added)
• conf/schema.xml.in (modified) (1 diff)
• lib/russianmorphology/commons-codec-1.10.jar (added)
• lib/russianmorphology/dawg-0.2.0.jar (added)
• lib/russianmorphology/guava-18.0.jar (added)
• lib/russianmorphology/jmorphy2-core-0.2.0.jar (added)
• lib/russianmorphology/jmorphy2-dicts-ru-0.2.0.jar (added)
• lib/russianmorphology/jmorphy2-dicts-uk-0.2.0.jar (added)
• lib/russianmorphology/jmorphy2-lucene-0.2.0.jar (added)
• lib/russianmorphology/jmorphy2-nlp-0.2.0.jar (added)
• lib/russianmorphology/morph-4.7.2.jar (deleted)
• lib/russianmorphology/noggit-0.7.jar (added)
• lib/russianmorphology/russian-4.7.2.jar (deleted)

gs3-extensions/solr/trunk/src/build.xml

@@ -88 +88 @@
            serializer.jar"/>
   <property name="russian-morph-jars"
-            value="russian-4.7.2.jar
-                   morph-4.7.2.jar"/>
+        value="jmorphy2-lucene-0.2.0.jar
+                   jmorphy2-nlp-0.2.0.jar
+                   jmorphy2-core-0.2.0.jar
+                   dawg-0.2.0.jar
+                   jmorphy2-dicts-uk-0.2.0.jar
+                   guava-18.0.jar
+                   commons-codec-1.10.jar
+                   noggit-0.7.jar
+                   jmorphy2-dicts-ru-0.2.0.jar"/>
+
 
   <property name="java-service-files" 

gs3-extensions/solr/trunk/src/conf/schema.xml.in

@@ -1164 +1164 @@
         <filter class="solr.LowerCaseFilterFactory"/>
         <filter class="solr.StopFilterFactory" ignoreCase="true" words="lang/stopwords_ru.txt" format="snowball" />
-        <filter class="org.apache.lucene.morphology.russian.RussianFilterFactory"/>
+        <filter class="company.evo.jmorphy2.lucene.Jmorphy2StemFilterFactory" dict="lang/pymorphy2_dicts"/>
       </analyzer>
     </fieldType>