Use statistics in Faker CTAS

plugin/trino-faker/src/main/java/io/trino/plugin/faker/ColumnInfo.java

@@ -13,9 +13,11 @@
  */
 package io.trino.plugin.faker;
 
+import com.google.common.collect.ImmutableMap;
 import io.trino.spi.connector.ColumnMetadata;
 import io.trino.spi.type.Type;
 
+import java.util.Map;
 import java.util.Optional;
 
 import static java.util.Objects.requireNonNull;
@@ -57,4 +59,16 @@ public ColumnInfo withComment(Optional<String> comment)
                 .setComment(comment)
                 .build());
     }
+
+    public ColumnInfo withHandle(FakerColumnHandle handle)
+    {
+        return new ColumnInfo(handle, metadata);
+    }
+
+    public ColumnInfo withProperties(Map<String, Object> properties)
+    {
+        return new ColumnInfo(handle, ColumnMetadata.builderFrom(metadata)
+                .setProperties(ImmutableMap.copyOf(properties))
+                .build());
+    }
 }

plugin/trino-faker/src/main/java/io/trino/plugin/faker/FakerColumnHandle.java

@@ -174,4 +174,19 @@ private static List<String> strings(Collection<?> values)
                 .map(String.class::cast)
                 .collect(toImmutableList());
     }
+
+    public FakerColumnHandle withNullProbability(double nullProbability)
+    {
+        return new FakerColumnHandle(columnIndex, name, type, nullProbability, generator, domain, step);
+    }
+
+    public FakerColumnHandle withDomain(Domain domain)
+    {
+        return new FakerColumnHandle(columnIndex, name, type, nullProbability, generator, domain, step);
+    }
+
+    public FakerColumnHandle withStep(ValueSet step)
+    {
+        return new FakerColumnHandle(columnIndex, name, type, nullProbability, generator, domain, step);
+    }
 }

plugin/trino-faker/src/main/java/io/trino/plugin/faker/FakerConfig.java

@@ -26,6 +26,8 @@ public class FakerConfig
     private double nullProbability = 0.5;
     private long defaultLimit = 1000L;
     private Locale locale = Locale.ENGLISH;
+    private double sequenceMinDistinctValuesRatio = 0.98;
+    private long maxDictionarySize = 1000L;
 
     @Max(1)
     @Min(0)
@@ -68,4 +70,39 @@ public FakerConfig setLocale(String value)
         this.locale = new Locale.Builder().setLanguageTag(value).build();
         return this;
     }
+
+    @Max(2)
+    @Min(0)
+    public double getSequenceMinDistinctValuesRatio()
+    {
+        return sequenceMinDistinctValuesRatio;
+    }
+
+    @Config("faker.sequence-min-distinct-values-ratio")
+    @ConfigDescription(
+            """
+            Minimum ratio of distinct values of a column to total number of rows in a table to treat the columns as a sequence
+            when creating a table using existing data. Set to a value greater than 1 to disable using sequences""")
+    public FakerConfig setSequenceMinDistinctValuesRatio(double value)
+    {
+        this.sequenceMinDistinctValuesRatio = value;
+        return this;
+    }
+
+    @Min(0)
+    public long getMaxDictionarySize()
+    {
+        return maxDictionarySize;
+    }
+
+    @Config("faker.max-dictionary-size")
+    @ConfigDescription(
+            """
+            Maximum size of randomly generated dictionaries to pick values from, used for columns with low number of approximate distinct values
+            observed during table creation using existing data. Set to zero to disable using dictionaries""")
+    public FakerConfig setMaxDictionarySize(long value)
+    {
+        this.maxDictionarySize = value;
+        return this;
+    }
 }

plugin/trino-faker/src/main/java/io/trino/plugin/faker/FakerConnector.java

@@ -125,6 +125,25 @@ public List<PropertyMetadata<?>> getSchemaProperties()
                         "Default limit of rows returned from any table in this schema, if not specified in the query",
                         null,
                         defaultLimit -> checkProperty(1 <= defaultLimit, INVALID_SCHEMA_PROPERTY, "default_limit value must be equal or greater than 1"),
+                        false),
+                doubleProperty(
+                        SchemaInfo.SEQUENCE_MIN_DISTINCT_VALUES_RATIO,
+                        """
+                                Minimum ratio of distinct values of a column to total number of rows in a table to treat the columns as a sequence
+                                when creating a table in this schema using existing data. Set to a value greater than 1 to disable using sequences""",
+                        null,
+                        sequenceMinDistinctValuesRatio -> checkProperty(
+                                0 <= sequenceMinDistinctValuesRatio && sequenceMinDistinctValuesRatio <= 2,
+                                INVALID_SCHEMA_PROPERTY,
+                                SchemaInfo.SEQUENCE_MIN_DISTINCT_VALUES_RATIO + " value must be between 0 and 2, inclusive"),
+                        false),
+                longProperty(
+                        SchemaInfo.MAX_DICTIONARY_SIZE,
+                        """
+                        Maximum size of randomly generated dictionaries to pick values from, used for columns with low number of approximate distinct values
+                        observed during table created in this schema using existing data. Set to zero to disable using dictionaries""",
+                        null,
+                        maxDictionarySize -> checkProperty(0 <= maxDictionarySize, INVALID_SCHEMA_PROPERTY, "max_dictionary_size value must be equal or greater than 0"),
                         false));
     }
 
@@ -143,6 +162,25 @@ public List<PropertyMetadata<?>> getTableProperties()
                         "Default limit of rows returned from this table if not specified in the query",
                         null,
                         defaultLimit -> checkProperty(1 <= defaultLimit, INVALID_TABLE_PROPERTY, "default_limit value must be equal or greater than 1"),
+                        false),
+                doubleProperty(
+                        TableInfo.SEQUENCE_MIN_DISTINCT_VALUES_RATIO,
+                        """
+                                Minimum ratio of distinct values of a column to total number of rows in a table to treat the columns as a sequence
+                                when creating a table using existing data. Set to a value greater than 1 to disable using sequences""",
+                        null,
+                        sequenceMinDistinctValuesRatio -> checkProperty(
+                                0 <= sequenceMinDistinctValuesRatio && sequenceMinDistinctValuesRatio <= 2,
+                                INVALID_TABLE_PROPERTY,
+                                TableInfo.SEQUENCE_MIN_DISTINCT_VALUES_RATIO + " value must be between 0 and 2, inclusive"),
+                        false),
+                longProperty(
+                        TableInfo.MAX_DICTIONARY_SIZE,
+                        """
+                        Maximum size of randomly generated dictionaries to pick values from, used for columns with low number of approximate distinct values
+                        observed during table creation using existing data. Set to zero to disable using dictionaries""",
+                        null,
+                        maxDictionarySize -> checkProperty(0 <= maxDictionarySize, INVALID_TABLE_PROPERTY, "max_dictionary_size value must be equal or greater than 0"),
                         false));
     }