Fix crash on parquet column type mismatch

[al13n321]

Changelog category (leave one):

• Bug Fix (user-visible misbehavior in an official stable release)

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Fixed crash in Parquet filtering when data types in the file substantially differ from requested types (e.g. ... FROM file('a.parquet', Parquet, 'x String'), but the file has x Int64). Without this fix, use input_format_parquet_filter_push_down = 0 as a workaround.

ParquetBlockInputFormat didn't cast Field-s to the requested types because I thought KeyCondition allows any types. I don't remember why I thought that. It was crashing in MergeTreeSetIndex::checkInRange() when FieldValue::update() tried to insert Field of wrong type into a column.

Fixes #68403

[robot-clickhouse]

This is an automated comment for commit 496a4c7 with description of existing statuses. It's updated for the latest CI running

✅ Click here to open a full report in a separate page

Successful checks

Check name	Description	Status
AST fuzzer	Runs randomly generated queries to catch program errors. The build type is optionally given in parenthesis. If it fails, ask a maintainer for help	✅ success
Bugfix validation	Checks that either a new test (functional or integration) or there some changed tests that fail with the binary built on master branch	✅ success
Builds	There's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS	✅ success
ClickBench	Runs [ClickBench](https://github.com/ClickHouse/ClickBench/) with instant-attach table	✅ success
Compatibility check	Checks that clickhouse binary runs on distributions with old libc versions. If it fails, ask a maintainer for help	✅ success
Docker keeper image	The check to build and optionally push the mentioned image to docker hub	✅ success
Docker server image	The check to build and optionally push the mentioned image to docker hub	✅ success
Fast test	Normally this is the first check that is ran for a PR. It builds ClickHouse and runs most of stateless functional tests, omitting some. If it fails, further checks are not started until it is fixed. Look at the report to see which tests fail, then reproduce the failure locally as described here	✅ success
Flaky tests	Checks if new added or modified tests are flaky by running them repeatedly, in parallel, with more randomization. Functional tests are run 100 times with address sanitizer, and additional randomization of thread scheduling. Integration tests are run up to 10 times. If at least once a new test has failed, or was too long, this check will be red. We don't allow flaky tests, read the doc	✅ success
Install packages	Checks that the built packages are installable in a clear environment	✅ success
Integration tests	The integration tests report. In parenthesis the package type is given, and in square brackets are the optional part/total tests	✅ success
Performance Comparison	Measure changes in query performance. The performance test report is described in detail here. In square brackets are the optional part/total tests	✅ success
Stateful tests	Runs stateful functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stateless tests	Runs stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stress test	Runs stateless functional tests concurrently from several clients to detect concurrency-related errors	✅ success
Style check	Runs a set of checks to keep the code style clean. If some of tests failed, see the related log from the report	✅ success
Unit tests	Runs the unit tests for different release types	✅ success
Upgrade check	Runs stress tests on server version from last release and then tries to upgrade it to the version from the PR. It checks if the new server can successfully startup without any errors, crashes or sanitizer asserts	✅ success

[jkartseva]

Looks good

[cpcloud]

Forgive the drive-by comment, but don't Parquet logical types provide the ability to make this distinction?

[al13n321]

They don't support integers wider than 64 bits. From parquet.thrift:

/**
 * Integer logical type annotation
 *
 * bitWidth must be 8, 16, 32, or 64.
 *
 * Allowed for physical types: INT32, INT64
 */
struct IntType {
  1: required i8 bitWidth
  2: required bool isSigned
}

[...]

/**
 * Decimal logical type annotation
 *
 * To maintain forward-compatibility in v1, implementations using this logical
 * type must also set scale and precision on the annotated SchemaElement.
 *
 * Allowed for physical types: INT32, INT64, FIXED, and BINARY
 */
struct DecimalType {
  1: required i32 scale
  2: required i32 precision
}

[...]

union LogicalType {
  1:  StringType STRING       // use ConvertedType UTF8
  2:  MapType MAP             // use ConvertedType MAP
  3:  ListType LIST           // use ConvertedType LIST
  4:  EnumType ENUM           // use ConvertedType ENUM
  5:  DecimalType DECIMAL     // use ConvertedType DECIMAL + SchemaElement.{scale, precision}
  6:  DateType DATE           // use ConvertedType DATE

  // use ConvertedType TIME_MICROS for TIME(isAdjustedToUTC = *, unit = MICROS)
  // use ConvertedType TIME_MILLIS for TIME(isAdjustedToUTC = *, unit = MILLIS)
  7:  TimeType TIME

  // use ConvertedType TIMESTAMP_MICROS for TIMESTAMP(isAdjustedToUTC = *, unit = MICROS)
  // use ConvertedType TIMESTAMP_MILLIS for TIMESTAMP(isAdjustedToUTC = *, unit = MILLIS)
  8:  TimestampType TIMESTAMP

  // 9: reserved for INTERVAL
  10: IntType INTEGER         // use ConvertedType INT_* or UINT_*
  11: NullType UNKNOWN        // no compatible ConvertedType
  12: JsonType JSON           // use ConvertedType JSON
  13: BsonType BSON           // use ConvertedType BSON
  14: UUIDType UUID           // no compatible ConvertedType
}

(Maybe we could write them as DecimalType with scale = 0, Decimal supports arbitrary length. The precision parameter would be awkward. E.g. UInt256 can hold some but not all 78-digit numbers. We could use precision = 77 but allow the value to be greater than 10^77, but then other parquet decoders may consider this an error. We could use precision = 78 and 33-byte-long value, but then our decoder can't know in advance whether all values will fit in UInt256. The latter seems ok though, maybe that would be the best way to do this. Also, signedness would probably be awkward in some way too.)