address review

Author: jimczi

plugins/analysis-nori/src/main/java/org/elasticsearch/index/analysis/NoriTokenizerFactory.java

@@ -30,10 +30,8 @@
 import java.io.IOException;
 import java.io.Reader;
 import java.io.StringReader;
-import java.util.HashSet;
 import java.util.List;
 import java.util.Locale;
-import java.util.Set;
 
 public class NoriTokenizerFactory extends AbstractTokenizerFactory {
     private static final String USER_DICT_PATH_OPTION = "user_dictionary";
@@ -49,18 +47,16 @@ public NoriTokenizerFactory(IndexSettings indexSettings, Environment env, String
     }
 
     public static UserDictionary getUserDictionary(Environment env, Settings settings) {
+        if (settings.get(USER_DICT_PATH_OPTION) != null && settings.get(USER_DICT_RULES_OPTION) != null) {
+            throw new IllegalArgumentException("It is not allowed to use [" + USER_DICT_PATH_OPTION + "] in conjunction" +
+                " with [" + USER_DICT_RULES_OPTION + "]");
+        }
         List<String> ruleList = Analysis.getWordList(env, settings, USER_DICT_PATH_OPTION, USER_DICT_RULES_OPTION);
         StringBuilder sb = new StringBuilder();
         if (ruleList == null || ruleList.isEmpty()) {
             return null;
         }
-        // check for duplicate terms
-        Set<String> terms = new HashSet<>();
         for (String line : ruleList) {
-            String[] split = line.split("\\s+");
-            if (terms.add(split[0]) == false) {
-                throw new IllegalArgumentException("Found duplicate term: [" + split[0] + "] in user dictionary. ");
-            }
             sb.append(line).append(System.lineSeparator());
         }
         try (Reader rulesReader = new StringReader(sb.toString())) {

plugins/analysis-nori/src/test/java/org/elasticsearch/index/analysis/NoriAnalysisTests.java

@@ -92,16 +92,6 @@ public void testNoriAnalyzerUserDict() throws Exception {
         }
     }
 
-    public void testNoriAnalyzerUserDictWithDuplicates() throws Exception {
-        Settings settings = Settings.builder()
-            .put("index.analysis.analyzer.my_analyzer.type", "nori")
-            .putList("index.analysis.analyzer.my_analyzer.user_dictionary_rules", "세종", "C샤프", "세종", "세종 세 종")
-            .build();
-        IllegalArgumentException exc = expectThrows(IllegalArgumentException.class, () -> createTestAnalysis(settings));
-        assertThat(exc.getMessage(), containsString("Found duplicate term: [세종]"));
-
-    }
-
     public void testNoriAnalyzerUserDictPath() throws Exception {
         Settings settings = Settings.builder()
             .put("index.analysis.analyzer.my_analyzer.type", "nori")
@@ -118,6 +108,17 @@ public void testNoriAnalyzerUserDictPath() throws Exception {
         }
     }
 
+    public void testNoriAnalyzerInvalidUserDictOption() throws Exception {
+        Settings settings = Settings.builder()
+            .put("index.analysis.analyzer.my_analyzer.type", "nori")
+            .put("index.analysis.analyzer.my_analyzer.user_dictionary", "user_dict.txt")
+            .putList("index.analysis.analyzer.my_analyzer.user_dictionary_rules", "c++", "C샤프", "세종", "세종시 세종 시")
+            .build();
+        IllegalArgumentException exc = expectThrows(IllegalArgumentException.class, () -> createTestAnalysis(settings));
+        assertThat(exc.getMessage(), containsString("It is not allowed to use [user_dictionary] in conjunction " +
+            "with [user_dictionary_rules]"));
+    }
+
     public void testNoriTokenizer() throws Exception {
         Settings settings = Settings.builder()
             .put("index.analysis.tokenizer.my_tokenizer.type", "nori_tokenizer")