Skip to content

[SPARK-59352][SQL] Validate prefix lengths in the vectorized DELTA_BYTE_ARRAY reader #58639

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
holdenk wants to merge 1 commit into
base: master
Choose a base branch
from
+71 −1
Open

[SPARK-59352][SQL] Validate prefix lengths in the vectorized DELTA_BYTE_ARRAY reader #58639

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 22 additions & 0 deletions ...va/org/apache/spark/sql/execution/datasources/parquet/VectorizedDeltaByteArrayReader.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@
import org.apache.parquet.bytes.ByteBufferInputStream;
import org.apache.parquet.column.values.RequiresPreviousReader;
import org.apache.parquet.column.values.ValuesReader;
import org.apache.parquet.io.ParquetDecodingException;
import org.apache.parquet.io.api.Binary;
import org.apache.spark.sql.execution.vectorized.OnHeapColumnVector;
import org.apache.spark.sql.execution.vectorized.WritableColumnVector;
Expand Down Expand Up @@ -71,6 +72,24 @@ public Binary readBinary(int len) {
return Binary.fromConstantByteArray(binaryValVector.getBinary(0));
}

/**
* The prefix length is read from the file, so validate it against the previous value before
* using it to copy bytes out of that value. A corrupt page must fail the read instead of
* producing a malformed value.
*/
private void checkPrefixLength(int prefixLength) {
if (prefixLength < 0) {
throw new ParquetDecodingException(
"Corrupted DELTA_BYTE_ARRAY page: negative prefix length: " + prefixLength);
}
int previousLength = previous == null ? 0 : previous.remaining();
if (prefixLength > previousLength) {
throw new ParquetDecodingException(
"Corrupted DELTA_BYTE_ARRAY page: prefix length " + prefixLength
+ " is larger than the previous value's length " + previousLength);
}
}

private void readValues(int total, WritableColumnVector c, int rowId) {
for (int i = 0; i < total; i++) {
// NOTE: due to PARQUET-246, it is important that we
Expand All @@ -79,6 +98,7 @@ private void readValues(int total, WritableColumnVector c, int rowId) {
// value of the page should have an empty prefix, it may not
// because of PARQUET-246.
int prefixLength = prefixLengthVector.getInt(currentRow);
checkPrefixLength(prefixLength);
ByteBuffer suffix = suffixReader.getBytes(currentRow);
byte[] suffixArray = suffix.array();
int suffixLength = suffix.limit() - suffix.position();
Expand Down Expand Up @@ -120,6 +140,7 @@ private void readGeoData(int total, WritableColumnVector c, int rowId, int srid,
WKBConverterStrategy converter) {
for (int i = 0; i < total; i++) {
int prefixLength = prefixLengthVector.getInt(currentRow);
checkPrefixLength(prefixLength);

@HyukjinKwon HyukjinKwon Sep 9, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The new tests cover the readBinary and skipBinary paths but not this geo path (readGeometry/readGeography), which copies the prefix with previous.get(wkb, 0, prefixLength) rather than the appendBytes(...) the other two use. A corrupt-prefix geometry/geography case would exercise that distinct copy behind the same guard. Consider a geo craft-page test, or parametrizing craftPage/the existing tests over the read method. Non-blocking.

ByteBuffer suffix = suffixReader.getBytes(currentRow);
int suffixLength = suffix.limit() - suffix.position();
int length = prefixLength + suffixLength;
Expand Down Expand Up @@ -165,6 +186,7 @@ public void skipBinary(int total) {

for (int i = 0; i < total; i++) {
int prefixLength = prefixLengthVector.getInt(currentRow);
checkPrefixLength(prefixLength);
ByteBuffer suffix = suffixReader.getBytes(currentRow);
byte[] suffixArray = suffix.array();
int suffixLength = suffix.limit() - suffix.position();
Expand Down
50 changes: 49 additions & 1 deletion ...g/apache/spark/sql/execution/datasources/parquet/ParquetDeltaByteArrayEncodingSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,9 +16,13 @@
*/
package org.apache.spark.sql.execution.datasources.parquet

import org.apache.parquet.bytes.DirectByteBufferAllocator
import org.apache.parquet.bytes.{ByteBufferInputStream, BytesInput, DirectByteBufferAllocator}
import org.apache.parquet.column.values.Utils
import org.apache.parquet.column.values.delta.DeltaBinaryPackingValuesWriterForInteger
import org.apache.parquet.column.values.deltalengthbytearray.DeltaLengthByteArrayValuesWriter
import org.apache.parquet.column.values.deltastrings.DeltaByteArrayWriter
import org.apache.parquet.io.ParquetDecodingException
import org.apache.parquet.io.api.Binary

import org.apache.spark.sql.catalyst.util.STUtils
import org.apache.spark.sql.execution.vectorized.{OnHeapColumnVector, WritableColumnVector}
Expand Down Expand Up @@ -59,6 +63,37 @@ class ParquetDeltaByteArrayEncodingSuite extends ParquetCompatibilityTest with S
assertReadWriteWithSkipN(writer, reader, randvalues)
}

test("prefix length larger than the previous value is rejected") {
// Craft a page by hand: a benign 2-byte first value, then a value whose prefix length
// claims 65536 bytes of a 2-byte previous value.
val is = craftPage(prefixLengths = Array(0, 65536), suffixes = Array("ab", ""))
reader.initFromPage(2, is)
writableColumnVector = new OnHeapColumnVector(2, StringType)
val e = intercept[ParquetDecodingException] {
reader.readBinary(2, writableColumnVector, 0)
}
assert(e.getMessage.contains("prefix length 65536"))
}

test("negative prefix length is rejected") {
val is = craftPage(prefixLengths = Array(0, -1), suffixes = Array("ab", "cd"))
reader.initFromPage(2, is)
writableColumnVector = new OnHeapColumnVector(2, StringType)
val e = intercept[ParquetDecodingException] {
reader.readBinary(2, writableColumnVector, 0)
}
assert(e.getMessage.contains("negative prefix length"))
}

test("prefix length larger than the previous value is rejected when skipping") {
val is = craftPage(prefixLengths = Array(0, 65536), suffixes = Array("ab", ""))
reader.initFromPage(2, is)
val e = intercept[ParquetDecodingException] {
reader.skipBinary(2)
}
assert(e.getMessage.contains("prefix length 65536"))
}

test("test lengths") {
var reader = new VectorizedDeltaBinaryPackedReader
Utils.writeData(writer, values)
Expand Down Expand Up @@ -100,6 +135,19 @@ class ParquetDeltaByteArrayEncodingSuite extends ParquetCompatibilityTest with S
geoType)
}

/** Builds a raw DELTA_BYTE_ARRAY page from explicit prefix lengths and suffixes. */
private def craftPage(
prefixLengths: Array[Int],
suffixes: Array[String]): ByteBufferInputStream = {
val allocator = new DirectByteBufferAllocator
val prefixWriter =
new DeltaBinaryPackingValuesWriterForInteger(128, 4, 64 * 1024, 64 * 1024, allocator)
val suffixWriter = new DeltaLengthByteArrayValuesWriter(64 * 1024, 64 * 1024, allocator)
prefixLengths.foreach(prefixWriter.writeInteger)
suffixes.foreach(s => suffixWriter.writeBytes(Binary.fromString(s)))
BytesInput.concat(prefixWriter.getBytes, suffixWriter.getBytes).toInputStream
}

private def assertGeoReadWrite(
writer: DeltaByteArrayWriter,
reader: VectorizedDeltaByteArrayReader,
Expand Down