MongoDB: Introduce a config flag to determine whether to skip using collation or not when querying

Signed-off-by: Sherif Ayad <[email protected]>

Author: sherifkayad

docs/src/main/sphinx/connector/mongodb.md

@@ -44,6 +44,7 @@ The following configuration properties are available:
 | `mongodb.connection-url`                 | The connection url that the driver uses to connect to a MongoDB deployment |
 | `mongodb.schema-collection`              | A collection which contains schema information                             |
 | `mongodb.case-insensitive-name-matching` | Match database and collection names case insensitively                     |
+| `mongodb.skip-collation`                 | Skip using `Collation` when querying the collections                       |
 | `mongodb.min-connections-per-host`       | The minimum size of the connection pool per host                           |
 | `mongodb.connections-per-host`           | The maximum size of the connection pool per host                           |
 | `mongodb.max-wait-time`                  | The maximum wait time                                                      |
@@ -96,6 +97,12 @@ Match database and collection names case insensitively.
 
 This property is optional; the default is `false`.
 
+### `mongodb.skip-collation`
+
+Skip using `Collation` when querying the collections. For more info check the [official documentation](https://www.mongodb.com/docs/manual/reference/collation/)
+
+This property is optional; the default is `false`.
+
 ### `mongodb.min-connections-per-host`
 
 The minimum number of connections per host for this MongoClient instance. Those connections are kept in a pool when idle, and the pool ensures over time that it contains at least this minimum number.

plugin/trino-mongodb/src/main/java/io/trino/plugin/mongodb/MongoClientConfig.java

@@ -31,6 +31,7 @@ public class MongoClientConfig
 {
     private String schemaCollection = "_schema";
     private boolean caseInsensitiveNameMatching;
+    private boolean skipCollation;
     private String connectionUrl;
 
     private int minConnectionsPerHost;
@@ -77,6 +78,18 @@ public MongoClientConfig setCaseInsensitiveNameMatching(boolean caseInsensitiveN
         return this;
     }
 
+    public boolean isSkipCollation()
+    {
+        return skipCollation;
+    }
+
+    @Config("mongodb.skip-collation")
+    public MongoClientConfig setSkipCollation(boolean skipCollation)
+    {
+        this.skipCollation = skipCollation;
+        return this;
+    }
+
     @NotNull
     public @Pattern(message = "Invalid connection URL. Expected mongodb:// or mongodb+srv://", regexp = "^mongodb(\\+srv)?://.*") String getConnectionUrl()
     {

plugin/trino-mongodb/src/main/java/io/trino/plugin/mongodb/MongoSession.java

@@ -176,6 +176,7 @@ public class MongoSession
 
     private final String schemaCollection;
     private final boolean caseInsensitiveNameMatching;
+    private final boolean skipCollation;
     private final int cursorBatchSize;
 
     private final Cache<SchemaTableName, MongoTable> tableCache;
@@ -187,6 +188,7 @@ public MongoSession(TypeManager typeManager, MongoClient client, MongoClientConf
         this.client = requireNonNull(client, "client is null");
         this.schemaCollection = requireNonNull(config.getSchemaCollection(), "config.getSchemaCollection() is null");
         this.caseInsensitiveNameMatching = config.isCaseInsensitiveNameMatching();
+        this.skipCollation = config.isSkipCollation();
         this.cursorBatchSize = config.getCursorBatchSize();
         this.implicitPrefix = requireNonNull(config.getImplicitRowFieldPrefix(), "config.getImplicitRowFieldPrefix() is null");
 
@@ -528,7 +530,12 @@ public MongoCursor<Document> execute(MongoTableHandle tableHandle, List<MongoCol
 
         MongoCollection<Document> collection = getCollection(tableHandle.remoteTableName());
         Document filter = buildFilter(tableHandle);
-        FindIterable<Document> iterable = collection.find(filter).projection(projection).collation(SIMPLE_COLLATION);
+        FindIterable<Document> iterable = collection.find(filter).projection(projection);
+
+        if (!skipCollation) {
+            iterable.collation(SIMPLE_COLLATION);
+        }
+
         tableHandle.limit().ifPresent(iterable::limit);
         log.debug("Find documents: collection: %s, filter: %s, projection: %s", tableHandle.schemaTableName(), filter, projection);
 

plugin/trino-mongodb/src/test/java/io/trino/plugin/mongodb/TestMongoClientConfig.java

@@ -34,6 +34,7 @@ public void testDefaults()
                 .setConnectionUrl(null)
                 .setSchemaCollection("_schema")
                 .setCaseInsensitiveNameMatching(false)
+                .setSkipCollation(false)
                 .setMinConnectionsPerHost(0)
                 .setConnectionsPerHost(100)
                 .setMaxWaitTime(120_000)
@@ -58,6 +59,7 @@ public void testExplicitPropertyMappings()
         Map<String, String> properties = ImmutableMap.<String, String>builder()
                 .put("mongodb.schema-collection", "_my_schema")
                 .put("mongodb.case-insensitive-name-matching", "true")
+                .put("mongodb.skip-collation", "true")
                 .put("mongodb.connection-url", "mongodb://router1.example.com:27017,router2.example2.com:27017,router3.example3.com:27017/")
                 .put("mongodb.min-connections-per-host", "1")
                 .put("mongodb.connections-per-host", "99")
@@ -79,6 +81,7 @@ public void testExplicitPropertyMappings()
         MongoClientConfig expected = new MongoClientConfig()
                 .setSchemaCollection("_my_schema")
                 .setCaseInsensitiveNameMatching(true)
+                .setSkipCollation(true)
                 .setConnectionUrl("mongodb://router1.example.com:27017,router2.example2.com:27017,router3.example3.com:27017/")
                 .setMinConnectionsPerHost(1)
                 .setConnectionsPerHost(99)