Standardizes the RecordExtractor output type contract across every input format

[Jackie-Jiang]

Summary

Standardizes the RecordExtractor output type contract across every input format, including the date/time matrix. Every per-format extractor now produces the same Java type matrix, so downstream ingestion transforms see a uniform shape regardless of source.

The contract (now documented on RecordExtractor):

Logical value	Java output type
Boolean	Boolean (not stringified)
Small int (Byte / Short / int8 / int16 / int32)	Integer (small ints widen)
Long	Long
Float	Float
Double	Double
Decimal (BigDecimal, BigInteger)	BigDecimal (not stringified, not lossily converted to Double; BigInteger widens since Pinot has no BigInteger type)
String	String
Bytes (ByteBuffer, ByteString, Binary, ORC BytesColumnVector, …)	byte[]
Date	java.time.LocalDate (TZ-independent)
Time of day	java.time.LocalTime (nanosecond precision, TZ-independent)
Timestamp / instant	java.sql.Timestamp (sub-millisecond nanos preserved)
Multi-value (Collection, array except byte[])	Object[] (input shape preserved — null elements stay null, empty inputs round-trip as Object[0])
Map / nested record / struct	Map<Object, Object>
Null / unset	null
Anything else (Avro Utf8, ProtoBuf EnumValueDescriptor, Thrift TEnum, …)	value.toString()

Base contract additions (BaseRecordExtractor.convertSingleValue)

The base now recognizes the full modern java.time family so format-specific extractors don't each have to bridge them:

• LocalDate / LocalTime / Timestamp pass through (preserved).
• Instant / OffsetDateTime / ZonedDateTime bridge to Timestamp via Timestamp.from(instant) — lossless (sub-second nanos preserved).
• LocalDateTime bridges to Timestamp interpreting the wall-clock as UTC (matches Iceberg timestamp, Parquet isAdjustedToUTC=false, Snowflake TIMESTAMP_NTZ).
• The Temporal family is gated behind a single instanceof Temporal check so non-temporal values (the common case) skip 5 instanceof checks on the hot path.

BigInteger widens to BigDecimal (Pinot has no BigInteger data type; BigDecimal preserves arbitrary precision and downstream transforms handle it natively). This is reachable in practice from JSON parsers (Jackson produces BigInteger for integer literals overflowing Long).

PinotDataType additions

DATE (wraps LocalDate) and TIME (wraps LocalTime) are added alongside TIMESTAMP. Recognized via getSingleValueType(LocalDate.class) / getSingleValueType(LocalTime.class). Each handles convert(value, sourceType) for INTEGER / LONG / STRING / JSON / TIMESTAMP sources.

Production fixes to bring formats in line with the contract:

• ORC BOOLEAN → Boolean (was String).
• ORC DECIMAL → BigDecimal (was String).
• ORC DATE → LocalDate via LocalDate.ofEpochDay(days) (was Long days).
• ORC TIMESTAMP / TIMESTAMP_INSTANT → Timestamp with full sub-second nanos via setNanos (was Long epoch millis dropping sub-second precision).
• ORC LIST preserves input shape — null elements stay null, empty lists round-trip as Object[0] (was filtering nulls and surfacing empty / all-null lists as null, making "empty" and "null" indistinguishable downstream).
• Avro LocalDate / LocalTime pass through (handled by base).
• Parquet INT32 with DATE annotation → LocalDate (was Integer days).
• Parquet INT32 with TIME_MILLIS / INT64 with TIME_MICROS / TIME_NANOS → LocalTime (was raw int / long).
• Parquet INT96 and INT64 with TIMESTAMP_MILLIS / TIMESTAMP_MICROS / TIMESTAMP_NANOS → Timestamp (was epoch millis Long); sub-millisecond nanos preserved.
• Thrift / Avro / ProtoBuf Byte / Short → Integer (small ints unify behind a single Pinot type) — lifted into BaseRecordExtractor.convertSingleValue so all formats inherit the widening.
• ProtoBuf convertSingleValue simplified — only ByteString → byte[] is genuinely format-specific; everything else delegates to the base.
• BaseRecordExtractor.convertMap (and the matching ProtoBuf override) rechecks the converted key for null after convertSingleValue and drops the entry — format-specific overrides may translate format-native null sentinels into a null key, which HashMap.put would otherwise accept as a single null-keyed entry.

Architectural changes

• New ORCRecordExtractor extracted from ORCRecordReader (a thin Record wrapper around VectorizedRowBatch + TypeDescription + rowId); ORC was the only format without a separate extractor class.
• convert and convertSingleValue annotated @Nullable: a non-null Java input may still represent a logical null (format-native sentinel), and format-specific overrides are free to translate to null. The base impl never returns null.
• RecordExtractor.init's recordExtractorConfig parameter annotated @Nullable (formats that don't need config pass null).

Tests

Every per-format *RecordExtractorTest is rewritten to a self-contained, type-driven pattern:

• One @Test per format-native input type, ordered to match the format's native type list (documented in each extractor's class Javadoc).
• Self-contained: no inheritance from AbstractRecordExtractorTest (deleted — no remaining users).
• Markdown Javadoc throughout (///, [X] references, backtick code).
• Each per-format test focuses on what is format-specific. Cross-format common behavior (init filtering, primitive preservation, Temporal family handling, multi-value shape preservation, map null-key dropping) is covered once in BaseRecordExtractorTest.
• testListPreservesNullElements and testEmptyListExtractedAsEmptyArray added at the format level for inputs that can carry the shape natively (ORC, JSON, Avro, Parquet, Thrift).
• New ParquetNativeRecordExtractorTest and ParquetAvroRecordExtractorTest with in-memory Group / GenericRecord inputs (no temp files).
• Class-level Javadoc of every extractor now documents the source type → Java input → Java output matrix, replacing the previous one-liner.

Test correctness fixes

• ThriftRecordReaderTest.testPrimitiveTypePreservation was asserting Short.class for i16 elements ("stay Short — not promoted to Integer") — directly contradicting the new contract. Rewritten to expect Integer per the widening.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 55.38%. Comparing base (6d85277) to head (4e21a72).

❗ There is a different number of reports uploaded between BASE (6d85277) and HEAD (4e21a72). Click for more details.

HEAD has 4 uploads less than BASE

Flag	BASE (6d85277)	HEAD (4e21a72)
java-21	5	4
unittests	2	1
temurin	5	4
unittests2	1	0

Additional details and impacted files

@@             Coverage Diff              @@
##             master   #18378      +/-   ##
============================================
- Coverage     63.45%   55.38%   -8.07%     
+ Complexity     1683      843     -840     
============================================
  Files          3253     2546     -707     
  Lines        198884   147413   -51471     
  Branches      30798    23738    -7060     
============================================
- Hits         126196    81650   -44546     
+ Misses        62620    58758    -3862     
+ Partials      10068     7005    -3063     

Flag	Coverage Δ
custom-integration1	100.00% <ø> (ø)
integration	100.00% <ø> (ø)
integration1	100.00% <ø> (ø)
integration2	0.00% <ø> (ø)
java-21	55.38% <100.00%> (-8.07%)	⬇️
temurin	55.38% <100.00%> (-8.07%)	⬇️
unittests	55.38% <100.00%> (-8.07%)	⬇️
unittests1	55.38% <100.00%> (+<0.01%)	⬆️
unittests2	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

Pull request overview

Standardizes the RecordExtractor output type contract across Pinot input formats so ingestion transforms and downstream type coercion see a uniform Java type matrix independent of source format.

Changes:

• Documents a unified RecordExtractor type contract and updates BaseRecordExtractor conversion behavior (notably widening Byte/Short → Integer and clarifying nullability semantics).
• Refactors ORC ingestion to use a dedicated ORCRecordExtractor (moving per-row extraction logic out of ORCRecordReader).
• Rewrites/introduces per-format extractor tests (Avro/JSON/CSV/ProtoBuf/Thrift/ORC/Parquet/CLP-Log) to validate contract adherence; removes old shared test base classes.

Reviewed changes

Copilot reviewed 18 out of 18 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
pinot-spi/src/main/java/org/apache/pinot/spi/data/readers/RecordExtractor.java	Adds/expands the documented cross-format output-type contract and clarifies conversion expectations.
pinot-spi/src/main/java/org/apache/pinot/spi/data/readers/BaseRecordExtractor.java	Updates base conversion semantics and documents default dispatch behavior.
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/BaseRecordExtractorTest.java	Reworks tests to assert the updated base contract (multi-value, map recursion, nested-record default behavior, etc.).
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/AbstractRecordReaderTest.java	Minor cleanup (uses imported Schema type instead of FQCN).
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/AbstractRecordExtractorTest.java	Deletes unused shared extractor test base.
pinot-plugins/pinot-input-format/pinot-orc/src/main/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordReader.java	Delegates row extraction to ORCRecordExtractor; keeps file IO + include filtering.
pinot-plugins/pinot-input-format/pinot-orc/src/main/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordExtractor.java	New ORC per-row extractor implementing the standardized contract (primitives, maps, structs, timestamps, decimals).
pinot-plugins/pinot-input-format/pinot-orc/src/test/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordExtractorTest.java	New in-memory vectorized tests for ORC extraction behavior (no file IO).
pinot-plugins/pinot-input-format/pinot-thrift/src/test/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordExtractorTest.java	Rewrites Thrift tests to be type-driven and contract-focused (including nested structs, lists/sets/maps, optionals).
pinot-plugins/pinot-input-format/pinot-protobuf/src/test/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractorTest.java	Rewrites ProtoBuf tests using dynamic descriptors/messages for isolated, type-driven verification.
pinot-plugins/pinot-input-format/pinot-protobuf/src/test/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractorLowLevelTest.java	Removes older low-level test suite in favor of the new contract-based tests.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetNativeRecordExtractorTest.java	New Parquet-Group-based tests for primitive/logical/complex mappings (LIST/MAP/struct/INT96).
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetAvroRecordExtractorTest.java	New focused test for Parquet-Avro INT96 override behavior.
pinot-plugins/pinot-input-format/pinot-json/src/test/java/org/apache/pinot/plugin/inputformat/json/JSONRecordExtractorTest.java	Rewrites JSON tests to validate the realistic Jackson type matrix and recursive conversions.
pinot-plugins/pinot-input-format/pinot-csv/src/test/java/org/apache/pinot/plugin/inputformat/csv/CSVRecordExtractorTest.java	Rewrites CSV tests around CSV’s untyped nature + MV delimiter behavior (string/empty/null semantics).
pinot-plugins/pinot-input-format/pinot-clp-log/src/test/java/org/apache/pinot/plugin/inputformat/clplog/CLPLogRecordExtractorTest.java	Updates CLP-Log tests for encoding config parsing + topic-name preservation with cleaner parameterization.
pinot-plugins/pinot-input-format/pinot-avro-base/src/test/java/org/apache/pinot/plugin/inputformat/avro/AvroRecordExtractorTest.java	Rewrites Avro tests to cover primitives, bytes/fixed, enums, arrays/maps/nested records, nullable unions, and logical types.
pinot-plugins/pinot-input-format/pinot-avro-base/src/test/java/org/apache/pinot/plugin/inputformat/avro/AvroRecordExtractorComplexTypesTest.java	Removes separate complex-type IO-based test class (coverage moved into the new type-driven tests).

[Copilot]

Pull request overview

This PR standardizes the RecordExtractor output-type contract across Pinot input formats so ingestion transforms see a consistent Java type matrix regardless of source format (including modern date/time handling).

Changes:

• Documents and enforces a unified RecordExtractor output contract, with shared conversions implemented in BaseRecordExtractor (numeric widening, bytes materialization, java.time bridging).
• Adds DATE (LocalDate) and TIME (LocalTime) to PinotDataType, plus corresponding tests and type inference updates.
• Refactors and aligns format-specific extractors/readers (notably ORC) and rewrites per-format extractor tests to be type-driven and self-contained.

Reviewed changes

Copilot reviewed 29 out of 29 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
pinot-spi/src/main/java/org/apache/pinot/spi/data/readers/RecordExtractor.java	Defines/documentes the cross-format output type matrix and makes init config nullable.
pinot-spi/src/main/java/org/apache/pinot/spi/data/readers/BaseRecordExtractor.java	Implements shared conversions (small-int widening, BigInteger→BigDecimal, ByteBuffer→byte[], Temporal→Timestamp, map key dropping rules).
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/BaseRecordExtractorTest.java	Adds contract-level tests for single/multi/map/temporal conversion behavior.
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/AbstractRecordReaderTest.java	Simplifies schema references (uses imported Schema).
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/AbstractRecordExtractorTest.java	Removes obsolete shared extractor test base.
pinot-common/src/main/java/org/apache/pinot/common/utils/PinotDataType.java	Adds DATE/TIME types and updates class-to-type inference.
pinot-common/src/test/java/org/apache/pinot/common/utils/PinotDataTypeTest.java	Adds unit tests for new DATE/TIME conversions and inference.
pinot-plugins/pinot-input-format/pinot-parquet/src/main/java/org/apache/pinot/plugin/inputformat/parquet/ParquetNativeRecordExtractor.java	Aligns Parquet logical types to the contract (LocalDate/LocalTime/Timestamp, INT96→Timestamp).
pinot-plugins/pinot-input-format/pinot-parquet/src/main/java/org/apache/pinot/plugin/inputformat/parquet/ParquetAvroRecordExtractor.java	Updates deprecated INT96 handling to return Timestamp via the native extractor helper.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetNativeRecordExtractorTest.java	New in-memory tests for Parquet primitive/logical/list/map/struct extraction.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetAvroRecordExtractorTest.java	New focused tests for parquet-avro INT96 override behavior.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetCollectionAndEquivalenceTest.java	Updates fixture expectations for date/timestamp outputs to match the new contract.
pinot-plugins/pinot-input-format/pinot-orc/src/main/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordReader.java	Delegates row extraction to a dedicated ORCRecordExtractor and updates supported type checks.
pinot-plugins/pinot-input-format/pinot-orc/src/main/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordExtractor.java	New ORC-specific extractor enforcing the contract (Boolean/BigDecimal/LocalDate/Timestamp, complex types).
pinot-plugins/pinot-input-format/pinot-orc/src/test/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordExtractorTest.java	Replaces IO-based tests with direct vectorized batch extractor tests.
pinot-plugins/pinot-input-format/pinot-protobuf/src/main/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractor.java	Simplifies convertSingleValue to only special-case ByteString, delegates other conversions to base; aligns map key dropping semantics.
pinot-plugins/pinot-input-format/pinot-protobuf/src/test/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractorTest.java	Rewrites tests to be descriptor-driven and type-focused without temp files.
pinot-plugins/pinot-input-format/pinot-protobuf/src/test/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractorLowLevelTest.java	Removes redundant low-level protobuf extractor tests.
pinot-plugins/pinot-input-format/pinot-thrift/src/main/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordExtractor.java	Updates extractor documentation to match the contract (incl. small-int widening).
pinot-plugins/pinot-input-format/pinot-thrift/src/test/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordExtractorTest.java	Rewrites tests to validate the contract directly via extractor output.
pinot-plugins/pinot-input-format/pinot-thrift/src/test/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordReaderTest.java	Updates reader tests for widened small ints and contract-aligned equality.
pinot-plugins/pinot-input-format/pinot-json/src/main/java/org/apache/pinot/plugin/inputformat/json/JSONRecordExtractor.java	Documents JSON-to-Java type behavior and relies on base conversions for widening.
pinot-plugins/pinot-input-format/pinot-json/src/test/java/org/apache/pinot/plugin/inputformat/json/JSONRecordExtractorTest.java	Rewrites tests to be contract-driven without JSON file IO.
pinot-plugins/pinot-input-format/pinot-csv/src/main/java/org/apache/pinot/plugin/inputformat/csv/CSVRecordExtractor.java	Documents CSV extractor semantics (untyped strings, empty→null, optional MV splitting).
pinot-plugins/pinot-input-format/pinot-csv/src/test/java/org/apache/pinot/plugin/inputformat/csv/CSVRecordExtractorTest.java	Rewrites tests to focus on extractor behavior using in-memory parsing.
pinot-plugins/pinot-input-format/pinot-clp-log/src/test/java/org/apache/pinot/plugin/inputformat/clplog/CLPLogRecordExtractorTest.java	Updates CLP log extractor tests to align with contract/expectations and reduces setup.
pinot-plugins/pinot-input-format/pinot-avro-base/src/main/java/org/apache/pinot/plugin/inputformat/avro/AvroRecordExtractor.java	Updates documentation and removes now-redundant Instant→Timestamp handling (delegated to base).
pinot-plugins/pinot-input-format/pinot-avro-base/src/test/java/org/apache/pinot/plugin/inputformat/avro/AvroRecordExtractorComplexTypesTest.java	Removes legacy complex-type tests superseded by the new test approach.

[Copilot]

Pull request overview

This PR standardizes the RecordExtractor output type contract across Pinot input formats by centralizing more conversion behavior in BaseRecordExtractor, aligning format-specific extractors (notably ORC/Parquet/ProtoBuf/Thrift/JSON/CSV) to the same Java type matrix, and updating tests accordingly.

Changes:

• Expanded BaseRecordExtractor.convertSingleValue() to unify small-int widening, BigInteger → BigDecimal, and modern java.time handling (LocalDate/LocalTime passthrough; Instant/OffsetDateTime/ZonedDateTime/LocalDateTime → Timestamp).
• Added PinotDataType.DATE and PinotDataType.TIME plus tests, and updated RecordExtractor documentation to formally define the contract.
• Refactored ORC ingestion to use a new ORCRecordExtractor (separating extraction from file IO), and rewrote per-format extractor tests to be type-driven/self-contained (removing legacy shared test bases).

Reviewed changes

Copilot reviewed 29 out of 29 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/BaseRecordExtractorTest.java	Rewritten to cover the updated base conversion/type contract (including temporal + shape-preservation cases).
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/AbstractRecordReaderTest.java	Cleans up schema type usage (imports Schema directly).
pinot-spi/src/test/java/org/apache/pinot/spi/data/readers/AbstractRecordExtractorTest.java	Deleted legacy shared extractor test base (no remaining users).
pinot-spi/src/main/java/org/apache/pinot/spi/data/readers/RecordExtractor.java	Documents the standardized output-type contract; makes init config nullable; convert return nullable.
pinot-spi/src/main/java/org/apache/pinot/spi/data/readers/BaseRecordExtractor.java	Centralizes widened numeric + modern java.time conversions; improves map key null handling; better default record error.
pinot-plugins/pinot-input-format/pinot-thrift/src/test/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordReaderTest.java	Updates expectations for small-int widening and multi-value extraction shape.
pinot-plugins/pinot-input-format/pinot-thrift/src/test/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordExtractorTest.java	Rewritten as direct extractor tests with explicit type-matrix coverage (including nested/map/list/set).
pinot-plugins/pinot-input-format/pinot-thrift/src/main/java/org/apache/pinot/plugin/inputformat/thrift/ThriftRecordExtractor.java	Updates class-level contract documentation to match standardized outputs.
pinot-plugins/pinot-input-format/pinot-protobuf/src/test/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractorTest.java	Rewritten to use dynamic descriptors/messages; covers presence semantics + map/repeated/nested cases.
pinot-plugins/pinot-input-format/pinot-protobuf/src/test/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractorLowLevelTest.java	Deleted older low-level tests in favor of the new direct/type-driven suite.
pinot-plugins/pinot-input-format/pinot-protobuf/src/main/java/org/apache/pinot/plugin/inputformat/protobuf/ProtoBufRecordExtractor.java	Aligns init signature; map-key null handling; delegates almost all single-value conversion to base.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetNativeRecordExtractorTest.java	New in-memory Group tests covering Parquet primitives, logical types, list/map/struct, and timestamp rounding.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetCollectionAndEquivalenceTest.java	Updates fixture assertions for LocalDate/Timestamp outputs and canonicalization docs.
pinot-plugins/pinot-input-format/pinot-parquet/src/test/java/org/apache/pinot/plugin/inputformat/parquet/ParquetAvroRecordExtractorTest.java	New tests for parquet-avro INT96 override producing Timestamp.
pinot-plugins/pinot-input-format/pinot-parquet/src/main/java/org/apache/pinot/plugin/inputformat/parquet/ParquetNativeRecordExtractor.java	Implements Parquet logical-type conversions to LocalDate/LocalTime/Timestamp; preserves sub-ms nanos.
pinot-plugins/pinot-input-format/pinot-parquet/src/main/java/org/apache/pinot/plugin/inputformat/parquet/ParquetAvroRecordExtractor.java	Narrows scope to INT96 handling and aligns INT96 conversion output to Timestamp.
pinot-plugins/pinot-input-format/pinot-orc/src/test/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordExtractorTest.java	Rewritten to be in-memory VectorizedRowBatch tests directly against the new ORC extractor.
pinot-plugins/pinot-input-format/pinot-orc/src/main/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordReader.java	Refactored to delegate per-row extraction to ORCRecordExtractor (reader now focuses on IO/filtering).
pinot-plugins/pinot-input-format/pinot-orc/src/main/java/org/apache/pinot/plugin/inputformat/orc/ORCRecordExtractor.java	New extractor implementing ORC category → standardized Java output types (incl. timestamps with nanos).
pinot-plugins/pinot-input-format/pinot-json/src/test/java/org/apache/pinot/plugin/inputformat/json/JSONRecordExtractorTest.java	Rewritten to direct extractor tests emphasizing standardized type outputs and shape preservation.
pinot-plugins/pinot-input-format/pinot-json/src/main/java/org/apache/pinot/plugin/inputformat/json/JSONRecordExtractor.java	Adds contract documentation for Jackson’s parsed JSON shapes and numeric behaviors.
pinot-plugins/pinot-input-format/pinot-csv/src/test/java/org/apache/pinot/plugin/inputformat/csv/CSVRecordExtractorTest.java	Rewritten to extractor-only tests (no file IO), focusing on cell handling + delimiter semantics.
pinot-plugins/pinot-input-format/pinot-csv/src/main/java/org/apache/pinot/plugin/inputformat/csv/CSVRecordExtractor.java	Updates contract documentation and init signature for nullable fields; behavior unchanged.
pinot-plugins/pinot-input-format/pinot-clp-log/src/test/java/org/apache/pinot/plugin/inputformat/clplog/CLPLogRecordExtractorTest.java	Simplifies tests and updates assertions consistent with the standardized conversion behavior.
pinot-plugins/pinot-input-format/pinot-avro-base/src/test/java/org/apache/pinot/plugin/inputformat/avro/AvroRecordExtractorComplexTypesTest.java	Deleted legacy complex-type test suite (replaced by newer format-specific coverage elsewhere).
pinot-plugins/pinot-input-format/pinot-avro-base/src/main/java/org/apache/pinot/plugin/inputformat/avro/AvroRecordExtractor.java	Updates Avro contract docs; relies on base for temporal conversion; keeps Avro fixed → byte[].
pinot-common/src/test/java/org/apache/pinot/common/utils/PinotDataTypeTest.java	Adds DATE/TIME conversion tests and updates getSingleValueType coverage.
pinot-common/src/main/java/org/apache/pinot/common/utils/PinotDataType.java	Adds DATE and TIME PinotDataTypes and updates class-to-type detection.