jdbc-v2: ANTLR4 parser backends report a truncated VALUES list (and miscount value groups) for a value expression the grammar cannot parse

[polyglotAI-bot]

Describe the bug

With jdbc_sql_parser=ANTLR4 (or ANTLR4_PARAMS_PARSER), an INSERT ... VALUES (...) whose values list contains a value expression the bundled ANTLR grammar cannot parse gets a truncated values-list template, and PreparedStatement#executeBatch() then sends a syntactically broken statement to the server.

Two properties are read from the parse tree in SqlParserFacade and both come out wrong for such a statement:

• assignValuesListStartPosition / assignValuesListStopPosition (enterAssignmentValuesList, SqlParserFacade.java:322-325): the stop is reported at the closing parenthesis of a nested function call instead of the end of the value group;
• assignValuesGroups (enterDataClauseValues): a single value group is counted twice, and — worse — a two-group values list is counted as one.

PreparedStatementImpl slices the original SQL with those positions when assignValuesGroups == 1, so the batch template loses its closing parenthesis.

Steps to reproduce

hex(x'AB') is valid ClickHouse SQL (SELECT hex(x'AB') → AB) but the driver's ANTLR grammar has no hex-string literal, so the parse falls into error recovery:

Properties props = new Properties();
props.setProperty("jdbc_sql_parser", "ANTLR4");
try (Connection conn = DriverManager.getConnection(url, props)) {
    try (Statement s = conn.createStatement()) {
        s.execute("CREATE TABLE t (v1 Int32, v2 String) Engine MergeTree ORDER BY ()");
    }
    try (PreparedStatement ps = conn.prepareStatement("INSERT INTO t (v1, v2) VALUES (?, hex(x'AB'))")) {
        ps.setInt(1, 1);
        ps.addBatch();
        ps.setInt(1, 2);
        ps.addBatch();
        ps.executeBatch();
    }
}

Expected behaviour

Both rows are inserted (this is what the default JAVACC parser does).

Actual behaviour

Code: 62. DB::Exception: Cannot parse expression of type String here: hex(x'AB'),(2, hex(x'AB'):
While executing ValuesBlockInputFormat. (SYNTAX_ERROR)

The values-list template was cut one character short ((?, hex(x'AB') instead of (?, hex(x'AB'))).

Same shape with a JDBC escape sequence, i.e. the ANTLR-side counterpart of #3017:

INSERT INTO t (a, b) VALUES (?, toDate({d '2024-01-01'}))   -- stop reported at the inner ')'
INSERT INTO t (a, b) VALUES (?, hex(x'AB')), (?, hex(x'CD'))  -- reported as a single value group

Configuration

• Driver: jdbc-v2, main
• Property: jdbc_sql_parser=ANTLR4 / ANTLR4_PARAMS_PARSER (the default JAVACC is not affected by this one)
• Server: 26.5.1