MSEARCH-466 Add char_filter for indexes (#318)

* Add char_filter for instance/authority (cherry picked from commit be43e28)
folio-org · Dec 6, 2022 · d18b3b2 · d18b3b2
1 parent 0da948c
commit d18b3b2
Show file tree

Hide file tree

Showing 5 changed files with 40 additions and 11 deletions.
diff --git a/src/main/java/org/folio/search/cql/builders/AllTermQueryBuilder.java b/src/main/java/org/folio/search/cql/builders/AllTermQueryBuilder.java
@@ -14,20 +14,19 @@
 @Component
 public class AllTermQueryBuilder extends FulltextQueryBuilder {
 
+  private static final String WHITE_SPACE = "\\s+";
+
   @Override
   public QueryBuilder getQuery(Object term, String resource, String... fields) {
     if (term instanceof String) {
       var stringTerm = (String) term;
-      var terms = stringTerm.split("\\s+");
+      var terms = stringTerm.split(WHITE_SPACE);
+
       if (terms.length == 1) {
         return getMultiMatchQuery(terms[0], fields);
+      } else {
+        return getBoolQuery(terms, fields);
       }
-
-      var boolQuery = boolQuery();
-      for (var singleTerm : terms) {
-        boolQuery.must(getMultiMatchQuery(singleTerm, fields));
-      }
-      return boolQuery;
     }
 
     return getMultiMatchQuery(term, fields);
@@ -51,4 +50,12 @@ public Set<String> getSupportedComparators() {
   private QueryBuilder getMultiMatchQuery(Object term, String... fieldNames) {
     return multiMatchQuery(term, fieldNames).operator(AND).type(CROSS_FIELDS);
   }
+
+  private QueryBuilder getBoolQuery(String[] terms, String... fieldNames) {
+    var boolQuery = boolQuery();
+    for (var singleTerm : terms) {
+      boolQuery.must(getMultiMatchQuery(singleTerm, fieldNames));
+    }
+    return boolQuery;
+  }
 }
diff --git a/src/main/resources/elasticsearch/index/authority.json b/src/main/resources/elasticsearch/index/authority.json
@@ -23,9 +23,19 @@
       "source_analyzer": {
         "tokenizer": "icu_tokenizer",
         "filter": [ "folio_word_delimiter_graph", "icu_folding" ],
+        "char_filter": [ "authority_char_filter" ],
         "type": "custom"
       }
     },
-    "tokenizers": { }
+    "tokenizers": { },
+    "char_filter": {
+      "authority_char_filter": {
+        "type": "mapping",
+        "mappings": [
+          "& => and",
+          ": => \\s+"
+        ]
+      }
+    }
   }
 }
diff --git a/src/main/resources/elasticsearch/index/instance.json b/src/main/resources/elasticsearch/index/instance.json
@@ -31,9 +31,19 @@
       "source_analyzer": {
         "tokenizer": "icu_tokenizer",
         "filter": [ "folio_word_delimiter_graph", "icu_folding" ],
+        "char_filter": [ "instance_char_filter" ],
         "type": "custom"
       }
     },
-    "tokenizers": { }
+    "tokenizers": { },
+    "char_filter": {
+      "instance_char_filter": {
+        "type": "mapping",
+        "mappings": [
+          "& => and",
+          ": => \\s+"
+        ]
+      }
+    }
   }
 }
diff --git a/src/test/java/org/folio/search/controller/SearchInstanceIT.java b/src/test/java/org/folio/search/controller/SearchInstanceIT.java
@@ -286,7 +286,9 @@ private static Stream<Arguments> testDataProvider() {
       arguments("holdingsIdentifiers all {value}", "e3ff6133-b9a2-4d4c-a1c9-dc1867d4df19"),
 
       //search by multiple different parameters
-      arguments("(keyword all {value})", "wolves matthew 9781609383657")
+      arguments("(keyword all {value})", "wolves matthew 9781609383657"),
+      arguments("(keyword all {value})", "A semantic web primer & wolves"),
+      arguments("(title all {value})", "A semantic web primer : 0747-0850")
     );
   }
 }
diff --git a/src/test/resources/samples/semantic-web-primer/instance.json b/src/test/resources/samples/semantic-web-primer/instance.json
@@ -2,7 +2,7 @@
   "id": "5bf370e0-8cca-4d9c-82e4-5170ab2a0a39",
   "hrid": "inst000000000022",
   "source": "FOLIO",
-  "title": "A semantic web primer 0747-0850 wolves",
+  "title": "A semantic web primer :0747-0850 & wolves",
   "indexTitle": "Semantic web primer",
   "alternativeTitles": [
     {