[BUG] libcudf JSON reader crash with compressed data

[lithomas1]

Describe the bug A clear and concise description of what the bug is.

The libcudf JSON reader is "crashing" (not sure if its technically a crash, but I'm getting a CUDA error)

RuntimeError: CUDA error encountered at: /home/coder/cudf/cpp/src/io/json/read_json.cu:142: 1 cudaErrorInvalidValue invalid argument

Steps/Code to reproduce bug Follow this guide http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports to craft a minimal bug report. This helps us reproduce the issue you're having and resolve the issue more quickly.

import cudf
cudf.read_json("baddf.json.gz", orient="records", lines=True, engine="cudf") # Doesn't work :(
pd.read_json("baddf.json.gz", orient="records", lines=True) # OK

Expected behavior A clear and concise description of what you expected to happen.

Environment overview (please complete the following information)

• Environment location: [Bare-metal, Docker, Cloud(specify cloud provider)]
• Method of cuDF install: [conda, Docker, or from source]
  • If method of install is [Docker], provide docker pull & docker run commands used

Successful read, like with pandas.

Environment details Please run and paste the output of the cudf/print_env.sh script here, to gather any other relevant environment details

My cudf is the latest cudf (from main).

Additional context

I think the issue might be with the specific data values (they are all integers, even the string/floating columns). I'm pretty sure libcudf can write all the data types (even the nested struct/list ones).

baddf.json.gz

Also, if you uncompress the file by hand, you are able to read it with cudf

[wence-]

Compute-sanitizer:

========= COMPUTE-SANITIZER
========= Program hit cudaErrorInvalidValue (error 1) due to "invalid argument" on CUDA API call to cudaMemcpyAsync.
=========     Saved host backtrace up to driver entry point at error
=========     Host Frame: [0x445b06]
=========                in /usr/lib/x86_64-linux-gnu/libcuda.so.1
=========     Host Frame:cudaMemcpyAsync [0x6dabf]
=========                in /home/coder/.conda/envs/rapids/lib/libcudart.so.12
=========     Host Frame:cudf::io::json::detail::ingest_raw_input(cudf::device_span<char, 18446744073709551615ul>, cudf::host_span<std::unique_ptr<cudf::io::datasource, std::default_delete<cudf::io::datasource> >, 18446744073709551615ul>, cudf::io::compression_type, unsigned long, unsigned long, rmm::cuda_stream_view) [0x1decfdf]
=========                in /home/coder/cudf/cpp/build/conda/cuda-12.2/release/libcudf.so
=========     Host Frame:cudf::io::json::detail::get_record_range_raw_input(cudf::host_span<std::unique_ptr<cudf::io::datasource, std::default_delete<cudf::io::datasource> >, 18446744073709551615ul>, cudf::io::json_reader_options const&, rmm::cuda_stream_view) [0x1dee514]
=========                in /home/coder/cudf/cpp/build/conda/cuda-12.2/release/libcudf.so
=========     Host Frame:cudf::io::json::detail::read_batch(cudf::host_span<std::unique_ptr<cudf::io::datasource, std::default_delete<cudf::io::datasource> >, 18446744073709551615ul>, cudf::io::json_reader_options const&, rmm::cuda_stream_view, cuda::mr::__4::basic_resource_ref<(cuda::mr::__4::_AllocType)1, cuda::mr::__4::device_accessible>) [0x1deeb75]
=========                in /home/coder/cudf/cpp/build/conda/cuda-12.2/release/libcudf.so
=========     Host Frame:cudf::io::json::detail::read_json(cudf::host_span<std::unique_ptr<cudf::io::datasource, std::default_delete<cudf::io::datasource> >, 18446744073709551615ul>, cudf::io::json_reader_options const&, rmm::cuda_stream_view, cuda::mr::__4::basic_resource_ref<(cuda::mr::__4::_AllocType)1, cuda::mr::__4::device_accessible>) [0x1df030a]
=========                in /home/coder/cudf/cpp/build/conda/cuda-12.2/release/libcudf.so
=========     Host Frame:cudf::io::read_json(cudf::io::json_reader_options, rmm::cuda_stream_view, cuda::mr::__4::basic_resource_ref<(cuda::mr::__4::_AllocType)1, cuda::mr::__4::device_accessible>) [0x1d2de18]
=========                in /home/coder/cudf/cpp/build/conda/cuda-12.2/release/libcudf.so

[GregoryKimball]

Thank you @lithomas1 for sharing issue. We haven't done much testing with compressed JSON inputs. There could be a straightforward solution here, and we will take a closer look as soon as we can.