Copilot commented on code in PR #50710:
URL: https://github.com/apache/arrow/pull/50710#discussion_r3678210221
##########
cpp/src/parquet/decoder.cc:
##########
@@ -1295,12 +1311,81 @@ class DictByteArrayDecoderImpl : public
DictDecoderImpl<ByteArrayType> {
int64_t valid_bits_offset,
typename EncodingTraits<ByteArrayType>::Accumulator*
out,
int* out_num_values) {
+ const auto* dict_values = dictionary_->data_as<ByteArray>();
+ const int values_to_decode = num_values - null_count;
+
+ switch (out->builder->type()->id()) {
+ case ::arrow::Type::BINARY:
+ case ::arrow::Type::STRING:
+ case ::arrow::Type::LARGE_BINARY:
+ case ::arrow::Type::LARGE_STRING: {
+ if (values_to_decode > 0) {
+ RETURN_NOT_OK(indices_scratch_space_->TypedResize<int32_t>(
+ values_to_decode, /*shrink_to_fit=*/false));
+ }
+ auto* decoded_indices =
indices_scratch_space_->mutable_data_as<int32_t>();
+ const int num_indices = idx_decoder_.GetBatch(decoded_indices,
values_to_decode);
+ if (ARROW_PREDICT_FALSE(num_indices != values_to_decode)) {
+ return Status::Invalid("Invalid or truncated dictionary index
stream: expected ",
+ values_to_decode, " indices but decoded ",
num_indices);
+ }
+
+ int64_t data_length = 0;
+ for (int i = 0; i < values_to_decode; ++i) {
+ const auto index = decoded_indices[i];
+ RETURN_NOT_OK(IndexInBounds(index));
+ if (ARROW_PREDICT_FALSE(AddWithOverflow(
+ data_length, static_cast<int64_t>(dict_values[index].len),
+ &data_length))) {
+ return Status::Invalid(
+ "excess expansion while decoding dictionary-encoded
BYTE_ARRAY");
+ }
+ }
+
+ auto append_predecoded = [&](auto* helper) {
+ int values_decoded = 0;
+ int pos_indices = 0;
+ int64_t remaining_data_length = data_length;
+
+ RETURN_NOT_OK(VisitBitRuns(
+ valid_bits, valid_bits_offset, num_values,
+ [&](int64_t position, int64_t length, bool valid) {
+ if (valid) {
+ for (int64_t i = 0; i < length; ++i) {
+ const auto& val =
dict_values[decoded_indices[pos_indices++]];
+ RETURN_NOT_OK(helper->AppendValue(
+ val.ptr, static_cast<int32_t>(val.len),
remaining_data_length));
+ remaining_data_length -= val.len;
+ }
+ values_decoded += static_cast<int>(length);
+ } else {
+ for (int64_t i = 0; i < length; ++i) {
+ helper->UnsafeAppendNull();
+ }
+ }
+ return Status::OK();
+ }));
+ DCHECK_EQ(pos_indices, values_to_decode);
+ DCHECK_EQ(remaining_data_length, 0);
+ *out_num_values = values_decoded;
+ return Status::OK();
Review Comment:
`values_to_decode` is derived from the `null_count` argument, but the append
loop advances `pos_indices` based on the validity bitmap. If `null_count` and
`valid_bits` are inconsistent (possible with corrupted inputs or upstream
bugs), this can read past `decoded_indices` in release builds (only guarded by
DCHECK). Add a runtime bounds check and return an Invalid status when the
bitmap’s non-null count doesn’t match `values_to_decode`.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
