Extracted vector-type from sqrl project

[velo]

No description provided.

[Copilot]

Pull Request Overview

This PR extracts the vector type from the sqrl project and extends PostgreSQL connector functionality. Key changes include the introduction of FlinkVectorType with its serializer, new PostgreSQL type serializers and converters, and updates to the JDBC dynamic table factory for Postgres integration.

Reviewed Changes

Copilot reviewed 13 out of 21 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
types/vector-type/src/main/java/com/datasqrl/types/vector/FlinkVectorTypeSerializer.java	Adds a custom serializer for the vector type with methods for serialization/deserialization.
types/vector-type/src/main/java/com/datasqrl/types/vector/FlinkVectorType.java	Introduces the Flink vector type class with a public double array.
connectors/postgresql-connector/src/test/java/com/datasqrl/connector/postgresql/jdbc/FlinkJdbcTest.java	Provides an integration test for writing and reading data via PostgreSQL.
connectors/postgresql-connector/src/main/java/com/datasqrl/connector/postgresql/type/*	Implements several PostgreSQL-specific type serializers and converters.
connectors/postgresql-connector/src/main/java/com/datasqrl/connector/postgresql/jdbc/*	Updates Postgres dialect, dynamic table factory, and base row converter logic to support the new types.

Files not reviewed (8)

• connectors/pom.xml: Language not supported
• connectors/postgresql-connector/pom.xml: Language not supported
• connectors/postgresql-connector/src/main/resources/META-INF/services/com.datasqrl.connector.postgresql.type.JdbcTypeSerializer: Language not supported
• connectors/postgresql-connector/src/main/resources/META-INF/services/org.apache.flink.table.factories.Factory: Language not supported
• flink-sql-runner/pom.xml: Language not supported
• pom.xml: Language not supported
• types/pom.xml: Language not supported
• types/vector-type/pom.xml: Language not supported

[Copilot]

The implementation of the copy(DataInputView, DataOutputView) method does not follow the serialization format used in serialize(), which writes an int and a series of doubles. Use corresponding read/write operations (i.e. read an int then the double values) to ensure the vector data is copied correctly.

Suggested change

	target.writeUTF(source.readUTF());
	int length = source.readInt(); // Read the length of the array
	target.writeInt(length); // Write the length of the array
	for (int i = 0; i < length; i++) {
	target.writeDouble(source.readDouble()); // Read and write each double value
	}

[Copilot]

The serializer declares its dialect type name as 'jsonb' but sets the PGobject type to 'json'; ensure consistency by setting the type to 'jsonb' if that is the intended behavior.

Suggested change

	pgObject.setType("json");
	pgObject.setType("jsonb");

[Copilot]

The deserializer converter currently returns null for any input value, which may lead to data loss if not intentional. Implement a proper conversion so that the input value is correctly deserialized.

Suggested change

	return (val) -> null;
	};
	return (val) -> {
	if (val == null) {
	return null;
	}
	try {
	ObjectMapper mapper = new ObjectMapper();
	JsonNode jsonNode = mapper.readTree(val.toString());
	// Convert JsonNode to Row or other appropriate structure
	// Assuming a simple conversion to Row for demonstration
	Row row = new Row(jsonNode.size());
	int index = 0;
	for (JsonNode field : jsonNode) {
	row.setField(index++, field.asText());
	}
	return row;
	} catch (Exception e) {
	throw new RuntimeException("Failed to deserialize value: " + val, e);
	}
	};