phone-search analyzer: don't emit sip/tel prefix

in line with the previous two commits, this is something else the search analyzer shouldn't emit since otherwise searching for any number with such a prefix will match _any_ document with the same prefix. Signed-off-by: Ralph Ursprung <[email protected]>
rursprung · Jan 10, 2025 · ed0014f · ed0014f
1 parent ff3c8da
commit ed0014f
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 4 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -92,8 +92,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Always use `constant_score` query for `match_only_text` field ([#16964](https://github.com/opensearch-project/OpenSearch/pull/16964))
 - Fix Shallow copy snapshot failures on closed index ([#16868](https://github.com/opensearch-project/OpenSearch/pull/16868))
 - Fix multi-value sort for unsigned long ([#16732](https://github.com/opensearch-project/OpenSearch/pull/16732))
-- The `phone-search` analyzer no longer emits the international calling code as a token ([#16993](https://github.com/opensearch-project/OpenSearch/pull/16993))
-- The `phone-search` analyzer no longer emits extension numbers and unformatted input as a token ([#16993](https://github.com/opensearch-project/OpenSearch/pull/16993))
+- The `phone-search` analyzer no longer emits the tel/sip prefix, international calling code, extension numbers and unformatted input as a token ([#16993](https://github.com/opensearch-project/OpenSearch/pull/16993))
 
 ### Security
 

diff --git a/...sis-phonenumber/src/main/java/org/opensearch/analysis/phone/PhoneNumberTermTokenizer.java b/...sis-phonenumber/src/main/java/org/opensearch/analysis/phone/PhoneNumberTermTokenizer.java
@@ -98,7 +98,9 @@ private Set<String> getTokens() throws IOException {
 
         // Rip off the "tel:" or "sip:" prefix
         if (input.indexOf("tel:") == 0 || input.indexOf("sip:") == 0) {
-            tokens.add(input.substring(0, 4));
+            if (addNgrams) {
+                tokens.add(input.substring(0, 4));
+            }
             input = input.substring(4);
         }
 

diff --git a/...sis-phonenumber/src/test/java/org/opensearch/analysis/phone/PhoneNumberAnalyzerTests.java b/...sis-phonenumber/src/test/java/org/opensearch/analysis/phone/PhoneNumberAnalyzerTests.java
@@ -87,7 +87,7 @@ public void testEuropeDetailled() throws IOException {
      * Test for all tokens which are emitted by the "phone" analyzer.
      */
     public void testEuropeDetailledSearch() throws IOException {
-        assertTokensAreInAnyOrder(phoneSearchAnalyzer, "tel:+441344840400", Arrays.asList("tel:+441344840400", "tel:", "441344840400"));
+        assertTokensAreInAnyOrder(phoneSearchAnalyzer, "tel:+441344840400", Arrays.asList("tel:+441344840400", "441344840400"));
     }
 
     public void testEurope() throws IOException {
@@ -162,6 +162,10 @@ public void testTelPrefix() throws IOException {
         assertTokensInclude("tel:+1228", Arrays.asList("1228", "122", "228"));
     }
 
+    public void testTelPrefixSearch() throws IOException {
+        assertTokensInclude("tel:+1228", Arrays.asList("1228"));
+    }
+
     public void testNumberPrefix() throws IOException {
         assertTokensInclude("+1228", Arrays.asList("1228", "122", "228"));
     }

diff --git a/...is-phonenumber/src/yamlRestTest/resources/rest-api-spec/test/analysis-phone/20_search.yml b/...is-phonenumber/src/yamlRestTest/resources/rest-api-spec/test/analysis-phone/20_search.yml
@@ -54,6 +54,11 @@
         index:  test
         id:     5
         body:   { "phone": "+1-888-280-4331", "phone-ch": "+1-888-280-4331" }
+  - do:
+      index:
+        index:  test
+        id:     6
+        body:   { "phone": "tel:+441344840400", "phone-ch": "tel:+441344840400" }
   - do:
       indices.refresh: {}
 
@@ -155,3 +160,36 @@
             match:
               "phone": "888 280 4331"
   - match: { hits.total: 0 }
+
+  # document & search have a tel: prefix
+  - do:
+      search:
+        rest_total_hits_as_int: true
+        index: test
+        body:
+          query:
+            match:
+              "phone": "tel:+441344840400"
+  - match: { hits.total: 1 }
+
+  # only document has a tel: prefix
+  - do:
+      search:
+        rest_total_hits_as_int: true
+        index: test
+        body:
+          query:
+            match:
+              "phone": "+441344840400"
+  - match: { hits.total: 1 }
+
+  # only search has a tel: prefix
+  - do:
+      search:
+        rest_total_hits_as_int: true
+        index: test
+        body:
+          query:
+            match:
+              "phone": "tel:+1 888 280 4331"
+  - match: { hits.total: 1 }