Avatical Removal Part1: remove Avatica dependency from metadata conversion flow

[carsheng]

This change accomplishes Part 1 of removing Avatica from the datacloud-jdbc driver - specifically focusing on marshalling of column metadata. It removes Avatica dependencies from metadata conversion and introduces a custom SimpleResultSet implementation for database metadata queries which includes much of the JDBc boilerplate that was originally handled by Avatica.
QueryMetadataUtil now uses SimpleMetadataResultSet.of() instead of Avatica-based constructors, and type mappings were updated to use standard JDBCType instead of Avatica's SqlType enum.

[carsheng]

@mkaufmann confirm whether FLOAT --> double conversion is correct?

[mkaufmann]

It's pretty unclear where the names here are coming from as they don't seem to be SQL types. The only use case of this map seems to be for private static List<Object> constructColumnData(ResultSet resultSet) throws SQLException {

That method doesn't really make sense though as it seems to be only used for convering a ResultSet in the end to a result set. So I think this + constructColumnData can just be completely removed. Instead of all that complication instead this function

    public static ResultSet createColumnResultSet(
            String schemaPattern, String tableNamePattern, String columnNamePattern, Connection connection)
            throws SQLException {

        final List<Object> data;

        try (val statement = connection.createStatement()) {
            val getColumnsQuery = getColumnsQueryInner(schemaPattern, tableNamePattern, columnNamePattern);
            val resultSet = statement.executeQuery(getColumnsQuery);
            data = constructColumnData(resultSet);
        }

        return getMetadataResultSet(QueryDBMetadata.GET_COLUMNS, data);
    }

Should be adjusted so that it directly return the resultSet from statement.executeQuery(getColumnsQuery); and to enable that the query needs to be adjusted to directly return the columns with the expected names and types for GET_COLUMNS. That can be achieved by using CAST( ... AS <TYPE>) and as <COLUMN_NAME> in the top level select

[carsheng]

@mkaufmann i think we can remove some amount of the code through refactoring, but the base problem still exists (i.e. we still need to determine the correct type mapping).

The names which we're seeing in the above map (such as int2, float4...etc) are base hyper types (we see the same in Postgres) that are fetched by querying the internal db table pg_catalog.pg_attribute for column metadata. Agreed, they're not SQL types, but they are the way column types are stored in this internal db table, and so we need to have a mapping between such base db types and SQL types.

If we do away with the mapping, when running metadata.getColumns() the customer will see non-SQL types such as int4 which would break the JDBC contract

[codecov]

Codecov Report

❌ Patch coverage is 69.88235% with 128 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.43%. Comparing base (b8c5eb9) to head (c28d02a).
⚠️ Report is 5 commits behind head on main.

Files with missing lines	Patch %	Lines
...datacloud/jdbc/core/resultset/SimpleResultSet.java	63.46%	20 Missing and 18 partials ⚠️
...e/datacloud/jdbc/core/SimpleMetadataResultSet.java	57.83%	20 Missing and 15 partials ⚠️
...core/resultset/ResultSetWithPositionalGetters.java	35.29%	22 Missing ⚠️
...force/datacloud/jdbc/core/metadata/ColumnType.java	24.00%	18 Missing and 1 partial ⚠️
...ud/jdbc/core/metadata/SimpleResultSetMetaData.java	81.57%	3 Missing and 4 partials ⚠️
...loud/jdbc/core/resultset/ForwardOnlyResultSet.java	75.00%	3 Missing and 1 partial ⚠️
...esforce/datacloud/jdbc/core/QueryMetadataUtil.java	90.47%	2 Missing ⚠️
...tacloud/jdbc/core/resultset/ReadOnlyResultSet.java	98.93%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##               main     #136      +/-   ##
============================================
- Coverage     83.47%   82.43%   -1.04%     
- Complexity     1271     1509     +238     
============================================
  Files           104      113       +9     
  Lines          3765     4214     +449     
  Branches        375      420      +45     
============================================
+ Hits           3143     3474     +331     
- Misses          462      535      +73     
- Partials        160      205      +45     

Components	Coverage Δ
JDBC Core	83.47% <69.88%> (-2.09%)	⬇️
JDBC Main	40.69% <ø> (ø)
JDBC HTTP	91.09% <ø> (ø)
JDBC Utilities	66.07% <ø> (+5.10%)	⬆️
Spark Datasource	∅ <ø> (∅)

Files with missing lines	Coverage Δ
.../datacloud/jdbc/core/resultset/ColumnAccessor.java	100.00% <100.00%> (ø)
...e/datacloud/jdbc/util/ArrowToColumnTypeMapper.java	62.26% <ø> (ø)
...com/salesforce/datacloud/jdbc/util/ArrowUtils.java	84.12% <ø> (ø)
...tacloud/jdbc/core/resultset/ReadOnlyResultSet.java	98.93% <98.93%> (ø)
...esforce/datacloud/jdbc/core/QueryMetadataUtil.java	91.48% <90.47%> (-0.15%)	⬇️
...loud/jdbc/core/resultset/ForwardOnlyResultSet.java	75.00% <75.00%> (ø)
...ud/jdbc/core/metadata/SimpleResultSetMetaData.java	81.57% <81.57%> (ø)
...force/datacloud/jdbc/core/metadata/ColumnType.java	68.96% <24.00%> (ø)
...core/resultset/ResultSetWithPositionalGetters.java	35.29% <35.29%> (ø)
...e/datacloud/jdbc/core/SimpleMetadataResultSet.java	57.83% <57.83%> (ø)
... and 1 more

... and 21 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.