Skip to content
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

Jump to bottom

Fix pyarrow segfaulting on fedora 39 #3213

Merged
merged 1 commit into from
Apr 5, 2024
Merged

Fix pyarrow segfaulting on fedora 39 #3213

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.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
10 changes: 5 additions & 5 deletions tools/python_api/src_cpp/include/pyarrow/pyarrow_scan.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,30 +22,30 @@ struct PyArrowTableScanSharedState final : public function::BaseScanSharedState
std::mutex lock;

PyArrowTableScanSharedState(
uint64_t numRows, std::vector<std::shared_ptr<ArrowArrayWrapper>>&& chunks)
uint64_t numRows, std::vector<std::shared_ptr<ArrowArrayWrapper>> chunks)
: BaseScanSharedState{numRows}, chunks{std::move(chunks)}, currentChunk{0} {}

ArrowArrayWrapper* getNextChunk();
};

struct PyArrowTableScanFunctionData final : public function::TableFuncBindData {
std::shared_ptr<ArrowSchemaWrapper> schema;
py::object table;
std::vector<std::shared_ptr<ArrowArrayWrapper>> arrowArrayBatches;
uint64_t numRows;

PyArrowTableScanFunctionData(std::vector<common::LogicalType> columnTypes,
std::shared_ptr<ArrowSchemaWrapper> schema, std::vector<std::string> columnNames,
py::handle table, uint64_t numRows)
std::vector<std::shared_ptr<ArrowArrayWrapper>> arrowArrayBatches, uint64_t numRows)
: TableFuncBindData{std::move(columnTypes), std::move(columnNames)},
schema{std::move(schema)}, table{py::reinterpret_borrow<py::object>(table)}, numRows{numRows} {}
schema{std::move(schema)}, arrowArrayBatches{arrowArrayBatches}, numRows{numRows} {}

~PyArrowTableScanFunctionData() override {}

std::unique_ptr<function::TableFuncBindData> copy() const override {
py::gil_scoped_acquire acquire;
// the schema is considered immutable so copying it by copying the shared_ptr is fine.
return std::make_unique<PyArrowTableScanFunctionData>(
columnTypes, schema, columnNames, *table, numRows);
columnTypes, schema, columnNames, arrowArrayBatches, numRows);
}
};

Expand Down
2 changes: 1 addition & 1 deletion tools/python_api/src_cpp/pandas/pandas_scan.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,9 +168,9 @@ static std::unique_ptr<ScanReplacementData> tryReplacePD(py::dict& dict, py::str
}

std::unique_ptr<ScanReplacementData> replacePD(const std::string& objectName) {
py::gil_scoped_acquire acquire;
auto pyTableName = py::str(objectName);
// Here we do an exhaustive search on the frame lineage.
py::gil_scoped_acquire acquire;
auto currentFrame = importCache->inspect.currentframe()();
while (hasattr(currentFrame, "f_locals")) {
auto localDict = py::reinterpret_borrow<py::dict>(currentFrame.attr("f_locals"));
Expand Down
19 changes: 10 additions & 9 deletions tools/python_api/src_cpp/pyarrow/pyarrow_scan.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,8 +28,16 @@ static std::unique_ptr<function::TableFuncBindData> bindFunc(
}
auto numRows = py::len(table);
auto schema = Pyarrow::bind(table, returnTypes, names);

py::list batches = table.attr("to_batches")(DEFAULT_VECTOR_CAPACITY);
std::vector<std::shared_ptr<ArrowArrayWrapper>> arrowArrayBatches;
for (auto& i: batches) {
arrowArrayBatches.push_back(std::make_shared<ArrowArrayWrapper>());
i.attr("_export_to_c")(reinterpret_cast<uint64_t>(arrowArrayBatches.back().get()));
}

return std::make_unique<PyArrowTableScanFunctionData>(
std::move(returnTypes), std::move(schema), std::move(names), std::move(table), numRows);
std::move(returnTypes), std::move(schema), std::move(names), arrowArrayBatches, numRows);
}

ArrowArrayWrapper* PyArrowTableScanSharedState::getNextChunk() {
Expand All @@ -46,16 +54,9 @@ static std::unique_ptr<function::TableFuncSharedState> initSharedState(
py::gil_scoped_acquire acquire;
PyArrowTableScanFunctionData* bindData =
dynamic_cast<PyArrowTableScanFunctionData*>(input.bindData);
py::list batches = bindData->table.attr("to_batches")(DEFAULT_VECTOR_CAPACITY);
std::vector<std::shared_ptr<ArrowArrayWrapper>> arrowArrayBatches;

for (auto& i : batches) {
arrowArrayBatches.push_back(std::make_shared<ArrowArrayWrapper>());
i.attr("_export_to_c")(reinterpret_cast<uint64_t>(arrowArrayBatches.back().get()));
}

return std::make_unique<PyArrowTableScanSharedState>(
bindData->numRows, std::move(arrowArrayBatches));
bindData->numRows, bindData->arrowArrayBatches);
}

static std::unique_ptr<function::TableFuncLocalState> initLocalState(
Expand Down