Make sequence threshold configurable

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

@@ -26,6 +26,7 @@ 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)
@@ -70,6 +71,24 @@ public FakerConfig setLocale(String value)
         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()
     {

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

@@ -126,6 +126,17 @@ public List<PropertyMetadata<?>> getSchemaProperties()
                         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,
                         """
@@ -152,6 +163,17 @@ public List<PropertyMetadata<?>> getTableProperties()
                         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,
                         """

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

@@ -115,6 +115,7 @@ public class FakerMetadata
     private final List<SchemaInfo> schemas = new ArrayList<>();
     private final double nullProbability;
     private final long defaultLimit;
+    private final double sequenceMinDistinctValuesRatio;
     private final long maxDictionarySize;
     private final FakerFunctionProvider functionsProvider;
 
@@ -132,6 +133,7 @@ public FakerMetadata(FakerConfig config, FakerFunctionProvider functionProvider)
         this.schemas.add(new SchemaInfo(SCHEMA_NAME, Map.of()));
         this.nullProbability = config.getNullProbability();
         this.defaultLimit = config.getDefaultLimit();
+        this.sequenceMinDistinctValuesRatio = config.getSequenceMinDistinctValuesRatio();
         this.maxDictionarySize = config.getMaxDictionarySize();
         this.functionsProvider = requireNonNull(functionProvider, "functionProvider is null");
         this.random = new Random(1);
@@ -494,6 +496,10 @@ else if (metadata.getStatisticType().equals(ColumnStatisticType.NUMBER_OF_DISTIN
         }
 
         Map<String, List<Object>> columnValues = getColumnValues(tableName, info, distinctValues, minimums, maximums);
+        SchemaInfo schema = getSchema(tableName.getSchemaName());
+        double schemaMinSequenceRatio = (double) schema.properties().getOrDefault(SchemaInfo.SEQUENCE_MIN_DISTINCT_VALUES_RATIO, sequenceMinDistinctValuesRatio);
+        double tableMinSequenceRatio = (double) info.properties().getOrDefault(TableInfo.SEQUENCE_MIN_DISTINCT_VALUES_RATIO, schemaMinSequenceRatio);
+
         return info.withColumns(columns.stream().map(column -> {
             if (isCharacterType(column.type()) || !minimums.containsKey(column.name())) {
                 return column;
@@ -517,7 +523,7 @@ else if (metadata.getStatisticType().equals(ColumnStatisticType.NUMBER_OF_DISTIN
             // Only include types that support generating sequences in FakerPageSource,
             // but don't include types with configurable precision, dates, or intervals.
             // The number of distinct values is an approximation, so compare it with a margin.
-            if (isIntegerType(column.type()) && (double) distinctValues.get(column.name()) / rowCount.get() >= 0.98) {
+            if (isIntegerType(column.type()) && (double) distinctValues.get(column.name()) / rowCount.get() >= tableMinSequenceRatio) {
                 handle = handle.withStep(ValueSet.of(column.type(), 1L));
                 properties.put(STEP_PROPERTY, "1");
             }

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

@@ -23,6 +23,7 @@ public record SchemaInfo(String name, Map<String, Object> properties)
 {
     public static final String NULL_PROBABILITY_PROPERTY = "null_probability";
     public static final String DEFAULT_LIMIT_PROPERTY = "default_limit";
+    public static final String SEQUENCE_MIN_DISTINCT_VALUES_RATIO = "sequence_min_distinct_values_ratio";
     public static final String MAX_DICTIONARY_SIZE = "max_dictionary_size";
 
     public SchemaInfo

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

@@ -27,6 +27,7 @@ public record TableInfo(List<ColumnInfo> columns, Map<String, Object> properties
 {
     public static final String NULL_PROBABILITY_PROPERTY = "null_probability";
     public static final String DEFAULT_LIMIT_PROPERTY = "default_limit";
+    public static final String SEQUENCE_MIN_DISTINCT_VALUES_RATIO = "sequence_min_distinct_values_ratio";
     public static final String MAX_DICTIONARY_SIZE = "max_dictionary_size";
 
     public TableInfo

plugin/trino-faker/src/test/java/io/trino/plugin/faker/TestFakerConfig.java

@@ -31,6 +31,7 @@ void testDefaults()
                 .setNullProbability(0.5)
                 .setDefaultLimit(1000L)
                 .setLocale("en")
+                .setSequenceMinDistinctValuesRatio(0.98)
                 .setMaxDictionarySize(1000L));
     }
 
@@ -41,13 +42,15 @@ void testExplicitPropertyMappings()
                 .put("faker.null-probability", "1.0")
                 .put("faker.default-limit", "10")
                 .put("faker.locale", "pl-PL")
+                .put("faker.sequence-min-distinct-values-ratio", "2")
                 .put("faker.max-dictionary-size", "0")
                 .buildOrThrow();
 
         FakerConfig expected = new FakerConfig()
                 .setNullProbability(1.0)
                 .setDefaultLimit(10L)
                 .setLocale("pl-PL")
+                .setSequenceMinDistinctValuesRatio(2.0)
                 .setMaxDictionarySize(0L);
 
         assertFullMapping(properties, expected);

plugin/trino-faker/src/test/java/io/trino/plugin/faker/TestFakerQueries.java

@@ -495,14 +495,14 @@ void testCreateTableAsSelect()
 
         for (TestDataType testCase : testCases) {
             try (TestTable sourceTable = new TestTable(getQueryRunner()::execute, "ctas_src_" + testCase.name(), "(%s) WITH (null_probability = 0, default_limit = 1000)".formatted(testCase.columnSchema()));
-                    TestTable table = new TestTable(getQueryRunner()::execute, "ctas_" + testCase.name(), "WITH (null_probability = 0, default_limit = 1000, max_dictionary_size = 0) AS SELECT %s FROM %s".formatted(testCase.name(), sourceTable.getName()))) {
+                    TestTable table = new TestTable(getQueryRunner()::execute, "ctas_" + testCase.name(), "WITH (null_probability = 0, default_limit = 1000, max_dictionary_size = 0, sequence_min_distinct_values_ratio = 2) AS SELECT %s FROM %s".formatted(testCase.name(), sourceTable.getName()))) {
                 assertQuery("SELECT %s FROM %s".formatted(testCase.queryExpression(), table.getName()), "VALUES (%s)".formatted(testCase.expectedValue()));
             }
         }
 
         for (TestDataType testCase : testCases) {
             try (TestTable sourceTable = new TestTable(getQueryRunner()::execute, "ctas_src_" + testCase.name(), "(%s) WITH (null_probability = 0, default_limit = 2)".formatted(testCase.name() + " " + testCase.type()));
-                    TestTable table = new TestTable(getQueryRunner()::execute, "ctas_" + testCase.name(), "WITH (null_probability = 0, default_limit = 1000, max_dictionary_size = 2) AS SELECT %s FROM %s".formatted(testCase.name(), sourceTable.getName()))) {
+                    TestTable table = new TestTable(getQueryRunner()::execute, "ctas_" + testCase.name(), "WITH (null_probability = 0, default_limit = 1000, max_dictionary_size = 2, sequence_min_distinct_values_ratio = 2) AS SELECT %s FROM %s".formatted(testCase.name(), sourceTable.getName()))) {
                 assertQuery("SELECT %s FROM %s".formatted(testCase.queryExpression(), table.getName()), "VALUES (%s)".formatted(testCase.expectedValue()));
             }
         }