Model experimental Schema (#1016)

Adds an experimental model for PartiQL Schema; as part of working on this PR the following spec issues created as pre-requisite for finalizing the SchemaType:

- Define PartiQL Type semantics: https://github.com/partiql/partiql-spec/issues/49
- Create PartiQLSchema specification: https://github.com/partiql/partiql-docs/issues/37

With this change we can model the `SQL` and `NoSQL` schemas on collections with collection constraints. In addition, with this model, we're able to represent other schemas like nested, array, scalar, and possibly Graph schemas as collection of Graph types:

```
// Example of nested schema
<<{'a': INT, 'b': [{'x': DECIMAL, 'y': DECIMAL}]>>

// Example of Scalar schema
<<INT>>

// Example of Array schema
<<[INT, DECIMAL]>>
```

As StaticType currently does not model the Orderedness of Bag and List as
constraints, leaving the Orderedness to the BAG and LIST types themselves.

Why the experimental model is a collection?

We model the Schema (E.g. an SQL Table`) as a collection to solve the orthogonality problem that SQL has with Table (as entities that define the shape of data) which implicitly defines them as a collection.

CHANGELOG.md

@@ -33,6 +33,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   but brings it in conformance with the specification.
 - Added `partiql-plan` package which contains experimental PartiQL Plan data structures.
 - Initializes SPI Framework under `partiql-spi`.
+- Models experimental `Schema` with constraints.
+  With this change, we're introducing `Tuple` and `Collection` constraints to be able to model the shape of data as 
+  constraints.
 - Introduces the PartiQLSchemaInferencer and PlannerSession
   - The PlannerSession describes the current session and is used by the PartiQLSchemaInferencer.
   - The PartiQLSchemaInferencer provides a function, `infer`, to aid in inferring the output `StaticType` of a

partiql-lang/src/main/kotlin/org/partiql/lang/planner/transforms/plan/PlanTyper.kt

@@ -66,6 +66,7 @@ import org.partiql.types.StaticType
 import org.partiql.types.StringType
 import org.partiql.types.StructType
 import org.partiql.types.SymbolType
+import org.partiql.types.TupleConstraint
 
 /**
  * Types a given logical plan.
@@ -438,7 +439,8 @@ internal object PlanTyper : PlanRewriter<PlanTyper.Context>() {
                             fields = input.getTypeEnv().associate { attribute ->
                                 attribute.name to attribute.type
                             },
-                            contentClosed = true
+                            contentClosed = true,
+                            constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                         )
                     )
                 )
@@ -648,7 +650,14 @@ internal object PlanTyper : PlanRewriter<PlanTyper.Context>() {
             TODO("Duplicate keys in struct is not yet handled")
         }
 
-        return node.copy(type = StructType(structFields.toMap(), contentClosed = closedContent), fields = fields)
+        return node.copy(
+            type = StructType(
+                structFields.toMap(),
+                contentClosed = closedContent,
+                constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
+            ),
+            fields = fields
+        )
     }
 
     override fun visitArgValue(node: Arg.Value, ctx: Context): PlanNode {

partiql-lang/src/test/kotlin/org/partiql/lang/planner/transforms/PartiQLSchemaInferencerTests.kt

@@ -24,6 +24,7 @@ import org.partiql.types.ListType
 import org.partiql.types.StaticType
 import org.partiql.types.StaticType.Companion.unionOf
 import org.partiql.types.StructType
+import org.partiql.types.TupleConstraint
 import java.net.URL
 import java.time.Instant
 import java.util.stream.Stream
@@ -58,23 +59,30 @@ class PartiQLSchemaInferencerTests {
                     "id" to TYPE_AWS_DDB_PETS_ID,
                     "breed" to TYPE_AWS_DDB_PETS_BREED
                 ),
-                contentClosed = true
+                contentClosed = true,
+                constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
             )
         )
         val TABLE_AWS_DDB_B = BagType(
             StructType(
                 fields = mapOf("identifier" to StaticType.STRING),
-                contentClosed = true
+                contentClosed = true,
+                constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
             )
         )
         val TABLE_AWS_B_B = BagType(
             StructType(
                 fields = mapOf("identifier" to StaticType.INT),
-                contentClosed = true
+                contentClosed = true,
+                constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
             )
         )
         val TYPE_B_B_B_B_B = StaticType.INT
-        private val TYPE_B_B_B_B = StructType(mapOf("b" to TYPE_B_B_B_B_B), contentClosed = true)
+        private val TYPE_B_B_B_B = StructType(
+            mapOf("b" to TYPE_B_B_B_B_B),
+            contentClosed = true,
+            constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
+        )
         val TYPE_B_B_B_C = StaticType.INT
         val TYPE_B_B_C = StaticType.INT
         val TYPE_B_B_B =
@@ -83,7 +91,8 @@ class PartiQLSchemaInferencerTests {
                     "b" to TYPE_B_B_B_B,
                     "c" to TYPE_B_B_B_C
                 ),
-                contentClosed = true
+                contentClosed = true,
+                constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
             )
     }
 
@@ -127,7 +136,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("pets" to StaticType.ANY),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 ),
                 problemHandler = assertProblemExists {
@@ -144,7 +154,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("pets" to StaticType.ANY),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 ),
                 problemHandler = assertProblemExists {
@@ -195,7 +206,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("pets" to StaticType.ANY),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 ),
                 problemHandler = assertProblemExists {
@@ -539,7 +551,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("unknown_col" to AnyType()),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 ),
                 problemHandler = assertProblemExists {
@@ -618,7 +631,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("cast_breed" to unionOf(StaticType.INT, StaticType.MISSING)),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -630,7 +644,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("upper_breed" to StaticType.STRING),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -640,7 +655,8 @@ class PartiQLSchemaInferencerTests {
                 expected = BagType(
                     StructType(
                         fields = mapOf("a" to ListType(unionOf(StaticType.INT, StaticType.DECIMAL))),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -672,7 +688,8 @@ class PartiQLSchemaInferencerTests {
                             "a" to StaticType.INT,
                             "b" to StaticType.DECIMAL,
                         ),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -685,7 +702,8 @@ class PartiQLSchemaInferencerTests {
                             "a" to StaticType.INT,
                             "b" to StaticType.DECIMAL,
                         ),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -698,7 +716,8 @@ class PartiQLSchemaInferencerTests {
                             "b" to StaticType.DECIMAL,
                             "a" to StaticType.INT,
                         ),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -713,7 +732,8 @@ class PartiQLSchemaInferencerTests {
                             "s" to StaticType.INT,
                             "m" to StaticType.INT,
                         ),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),
@@ -728,7 +748,8 @@ class PartiQLSchemaInferencerTests {
                             "s" to StaticType.DECIMAL,
                             "m" to StaticType.DECIMAL,
                         ),
-                        contentClosed = true
+                        contentClosed = true,
+                        constraints = setOf(TupleConstraint.Open(false), TupleConstraint.UniqueAttrs(true))
                     )
                 )
             ),

partiql-types/src/main/kotlin/org/partiql/types/StaticType.kt

@@ -81,7 +81,7 @@ sealed class StaticType {
             LIST,
             SEXP,
             STRUCT,
-            BAG
+            BAG,
         )
     }
 
@@ -209,8 +209,20 @@ class UnsupportedTypeCheckException(message: String) : RuntimeException(message)
  */
 sealed class CollectionType : SingleType() {
     abstract val elementType: StaticType
+    abstract val constraints: Set<CollectionConstraint>
+
+    internal fun validateCollectionConstraints() {
+        if (elementType !is StructType && constraints.any { it is TupleCollectionConstraint }) {
+            throw UnsupportedTypeConstraint("Only collection of tuples can have tuple constraints")
+        }
+    }
 }
 
+/**
+ * Exception thrown when a [StaticType] is initialized with an unsupported type constraint.
+ */
+class UnsupportedTypeConstraint(message: String) : Exception(message)
+
 // Single types from ExprValueType.
 
 /**
@@ -396,8 +408,13 @@ data class ClobType(override val metas: Map<String, Any> = mapOf()) : SingleType
  */
 data class ListType(
     override val elementType: StaticType = ANY,
-    override val metas: Map<String, Any> = mapOf()
+    override val metas: Map<String, Any> = mapOf(),
+    override val constraints: Set<CollectionConstraint> = setOf()
 ) : CollectionType() {
+
+    init {
+        validateCollectionConstraints()
+    }
     override fun flatten(): StaticType = this
 
     override val allTypes: List<StaticType>
@@ -411,8 +428,12 @@ data class ListType(
  */
 data class SexpType(
     override val elementType: StaticType = ANY,
-    override val metas: Map<String, Any> = mapOf()
+    override val metas: Map<String, Any> = mapOf(),
+    override val constraints: Set<CollectionConstraint> = setOf(),
 ) : CollectionType() {
+    init {
+        validateCollectionConstraints()
+    }
     override fun flatten(): StaticType = this
 
     override val allTypes: List<StaticType>
@@ -426,8 +447,12 @@ data class SexpType(
  */
 data class BagType(
     override val elementType: StaticType = ANY,
-    override val metas: Map<String, Any> = mapOf()
+    override val metas: Map<String, Any> = mapOf(),
+    override val constraints: Set<CollectionConstraint> = setOf(),
 ) : CollectionType() {
+    init {
+        this.validateCollectionConstraints()
+    }
     override fun flatten(): StaticType = this
 
     override val allTypes: List<StaticType>
@@ -438,9 +463,18 @@ data class BagType(
 
 data class StructType(
     val fields: Map<String, StaticType> = mapOf(),
+    // `TupleConstraint` already has `Open` constraint which overlaps with `contentClosed`.
+    // In addition, `primaryKeyFields` must not exist on the `StructType` as `PrimaryKey`
+    // is a property of collection of tuples. As we have plans to define PartiQL types in
+    // more details it's foreseeable to have an refactor of our types in future and have a
+    // new definition of this type as `Tuple`. See the following issue for more details:
+    // https://github.com/partiql/partiql-spec/issues/49
+    // TODO remove `contentClosed` and `primaryKeyFields` if after finalizing our type specification we're
+    // still going with `StructType`.
     val contentClosed: Boolean = false,
     val primaryKeyFields: List<String> = listOf(),
-    override val metas: Map<String, Any> = mapOf()
+    val constraints: Set<TupleConstraint> = setOf(),
+    override val metas: Map<String, Any> = mapOf(),
 ) : SingleType() {
     override fun flatten(): StaticType = this
 
@@ -451,8 +485,8 @@ data class StructType(
         val entries = fields.entries
         val firstSeveral = entries.toList().take(3).joinToString { "${it.key}: ${it.value}" }
         return when {
-            entries.size <= 3 -> "struct($firstSeveral)"
-            else -> "struct($firstSeveral, ... and ${entries.size - 3} other field(s))"
+            entries.size <= 3 -> "struct($firstSeveral, $constraints)"
+            else -> "struct($firstSeveral, ... and ${entries.size - 3} other field(s), $constraints)"
         }
     }
 }
@@ -518,6 +552,33 @@ sealed class NumberConstraint {
     }
 }
 
+/**
+ * Represents Tuple constraints; this is still experimental.
+ * and subject to change upon finalization of the following:
+ * - https://github.com/partiql/partiql-spec/issues/49
+ * - https://github.com/partiql/partiql-docs/issues/37
+ */
+sealed class TupleConstraint {
+    data class UniqueAttrs(val value: Boolean) : TupleConstraint()
+    data class Open(val value: Boolean) : TupleConstraint()
+}
+
+/**
+ * An Interface for constraints that are only applicable to collection of tuples, e.g. `PrimaryKey`.
+ */
+interface TupleCollectionConstraint
+
+/**
+ * Represents Collection constraints; this is still experimental.
+ * and subject to change upon finalization of the following:
+ * - https://github.com/partiql/partiql-spec/issues/49
+ * - https://github.com/partiql/partiql-docs/issues/37
+ */
+sealed class CollectionConstraint {
+    data class PrimaryKey(val keys: Set<String>) : TupleCollectionConstraint, CollectionConstraint()
+    data class PartitionKey(val keys: Set<String>) : TupleCollectionConstraint, CollectionConstraint()
+}
+
 internal fun StaticType.isNullOrMissing(): Boolean = (this is NullType || this is MissingType)
 internal fun StaticType.isNumeric(): Boolean = (this is IntType || this is FloatType || this is DecimalType)
 internal fun StaticType.isText(): Boolean = (this is SymbolType || this is StringType)