Skip to content

GH-50627: [C++] Migrate from_string.cc to simdjson #50653

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

Merged
pitrou merged 10 commits into from
Jul 28, 2026
Merged

GH-50627: [C++] Migrate from_string.cc to simdjson #50653

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
c526a91
migrate `from_string.cc` to `simdjson`
taepper Jan 28, 2026
2f9559a
fix review comments
taepper Jul 27, 2026
6b76b3d
also add helper method for non-utf8 inputs to python/gdb.cc
taepper Jul 27, 2026
8627a6f
add helpers to fit simdjson into arrow results
taepper Jul 28, 2026
16ddada
use Status::operator& and linter error
taepper Jul 28, 2026
fe4158f
improve error context
taepper Jul 28, 2026
caeb78f
cleaner function names
taepper Jul 28, 2026
d318388
refactor ProcessJsonArrayElements
taepper Jul 28, 2026
cabf131
do not use static_assert(false,..) which is not portable before C++23
taepper Jul 28, 2026
2bbad73
remove superfluous comment
taepper Jul 28, 2026
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
23 changes: 3 additions & 20 deletions cpp/src/arrow/compute/kernels/scalar_cast_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
#include "arrow/chunked_array.h"
#include "arrow/extension_type.h"
#include "arrow/status.h"
#include "arrow/testing/builder.h"
#include "arrow/testing/extension_type.h"
#include "arrow/testing/gtest_util.h"
#include "arrow/testing/random.h"
Expand All @@ -56,33 +57,15 @@ using internal::checked_pointer_cast;
namespace compute {

static std::shared_ptr<Array> InvalidUtf8(std::shared_ptr<DataType> type) {
return ArrayFromJSON(type,
"["
R"(
"Hi",
"olá mundo",
"你好世界",
"",
)"
"\"\xa0\xa1\""
"]");
return BinaryArrayFromStrings(type, {"Hi", "olá mundo", "你好世界", "", "\xa0\xa1"});
}

static std::shared_ptr<Array> FixedSizeInvalidUtf8(std::shared_ptr<DataType> type) {
if (type->id() == Type::FIXED_SIZE_BINARY) {
// Assume a particular width for testing
EXPECT_EQ(3, checked_cast<const FixedSizeBinaryType&>(*type).byte_width());
}
return ArrayFromJSON(type,
"["
R"(
"Hi!",
"lá",
"你",
" ",
)"
"\"\xa0\xa1\xa2\""
"]");
return BinaryArrayFromStrings(type, {"Hi!", "lá", "你", " ", "\xa0\xa1\xa2"});
}

static std::vector<std::shared_ptr<DataType>> kNumericTypes = {
Expand Down
81 changes: 36 additions & 45 deletions cpp/src/arrow/compute/kernels/scalar_string_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1102,7 +1102,7 @@ TYPED_TEST(TestStringKernels, Utf8Reverse) {

// inputs with malformed utf8 chars would produce garbage output, but the end result
// would produce arrays with same lengths. Hence checking offset buffer equality
auto malformed_input = ArrayFromJSON(this->type(), "[\"ɑ\xFFɑa\", \"ɽ\xe1\xbdɽa\"]");
auto malformed_input = this->MakeArray({"ɑ\xFFɑa", "ɽ\xe1\xbdɽa"});
const Result<Datum>& res = CallFunction("utf8_reverse", {malformed_input});
ASSERT_TRUE(res->array()->buffers[1]->Equals(*malformed_input->data()->buffers[1]));
}
Expand Down Expand Up @@ -1225,7 +1225,7 @@ TYPED_TEST(TestStringKernels, Utf8Upper) {
this->CheckUnary("utf8_upper", "[\"ɑɑɑɑ\"]", this->type(), "[\"ⱭⱭⱭⱭ\"]");

// Test invalid data
auto invalid_input = ArrayFromJSON(this->type(), "[\"ɑa\xFFɑ\", \"ɽ\xe1\xbdɽaa\"]");
auto invalid_input = this->MakeArray({"ɑa\xFFɑ", "ɽ\xe1\xbdɽaa"});
EXPECT_RAISES_WITH_MESSAGE_THAT(Invalid, testing::HasSubstr("Invalid UTF8 sequence"),
CallFunction("utf8_upper", {invalid_input}));
}
Expand All @@ -1247,7 +1247,7 @@ TYPED_TEST(TestStringKernels, Utf8Lower) {
this->CheckUnary("utf8_lower", "[\"ȺȺȺȺ\"]", this->type(), "[\"ⱥⱥⱥⱥ\"]");

// Test invalid data
auto invalid_input = ArrayFromJSON(this->type(), "[\"Ⱥa\xFFⱭ\", \"Ɽ\xe1\xbdⱤaA\"]");
auto invalid_input = this->MakeArray({"Ⱥa\xFFⱭ", "Ɽ\xe1\xbdⱤaA"});
EXPECT_RAISES_WITH_MESSAGE_THAT(Invalid, testing::HasSubstr("Invalid UTF8 sequence"),
CallFunction("utf8_lower", {invalid_input}));
}
Expand All @@ -1267,7 +1267,7 @@ TYPED_TEST(TestStringKernels, Utf8SwapCase) {
"[\"HeLLo, wOrLD!\", \"$. a35?\"]");

// Test invalid data
auto invalid_input = ArrayFromJSON(this->type(), "[\"Ⱥa\xFFⱭ\", \"Ɽ\xe1\xbdⱤaA\"]");
auto invalid_input = this->MakeArray({"Ⱥa\xFFⱭ", "Ɽ\xe1\xbdⱤaA"});
EXPECT_RAISES_WITH_MESSAGE_THAT(Invalid, testing::HasSubstr("Invalid UTF8 sequence"),
CallFunction("utf8_swapcase", {invalid_input}));
}
Expand Down Expand Up @@ -2516,101 +2516,92 @@ TYPED_TEST(TestBinaryKernels, SliceBytesPosPos) {
SliceOptions options{2, 4};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"a\xc2\xa2\", \"ab\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"\", \"\xa2\", \"\xc2\xa2\", \"\xc2\xff\"]", &options);
this->MakeArray({"", "a", "ab", "a\xc2\xa2", "ab\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "", "\xa2", "\xc2\xa2", "\xc2\xff"}), &options);
SliceOptions options_step{1, 5, 2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"a\xc2\xa2\", \"ab\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"b\", \"\xc2\", \"b\xa2\", \"b\xff\"]", &options_step);
this->MakeArray({"", "a", "ab", "a\xc2\xa2", "ab\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "b", "\xc2", "b\xa2", "b\xff"}), &options_step);
SliceOptions options_step_neg{5, 1, -2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"a\xc2\xa2\", \"ab\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"\", \"\xa2\", \"\xa2\", \"Z\xc2\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "a\xc2\xa2", "ab\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "", "\xa2", "\xa2", "Z\xc2"}), &options_step_neg);
options_step_neg.stop = 0;
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"a\xc2\xa2\", \"aZ\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"b\", \"\xa2\", \"\xa2Z\", \"Z\xc2\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "a\xc2\xa2", "aZ\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "b", "\xa2", "\xa2Z", "Z\xc2"}), &options_step_neg);
}

TYPED_TEST(TestBinaryKernels, SliceBytesPosNeg) {
SliceOptions options{2, -1};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"a\xc2\xa2\", \"aZ\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"\", \"\", \"\xc2\", \"\xc2\xff\"]", &options);
this->MakeArray({"", "a", "ab", "a\xc2\xa2", "aZ\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "", "", "\xc2", "\xc2\xff"}), &options);
SliceOptions options_step{1, -1, 2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"a\xc2\xa2\", \"aZ\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"\", \"\xc2\", \"Z\", \"b\xff\"]", &options_step);
this->MakeArray({"", "a", "ab", "a\xc2\xa2", "aZ\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "", "\xc2", "Z", "b\xff"}), &options_step);
SliceOptions options_step_neg{3, -4, -2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"a\", \"b\", \"\xa2Z\", \"\xa2Z\", \"\xff\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "a", "b", "\xa2Z", "\xa2Z", "\xff"}), &options_step_neg);
options_step_neg.stop = -5;
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"a\", \"b\", \"\xa2Z\", \"\xa2Z\", \"\xffP\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "a", "b", "\xa2Z", "\xa2Z", "\xffP"}), &options_step_neg);
}

TYPED_TEST(TestBinaryKernels, SliceBytesNegNeg) {
SliceOptions options{-2, -1};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"ab\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"a\", \"\xc2\", \"\xc2\", \"\xff\"]", &options);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "ab\xc2\xffZ"}),
this->MakeArray({"", "", "a", "\xc2", "\xc2", "\xff"}), &options);
SliceOptions options_step{-4, -1, 2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"a\", \"Z\", \"a\xc2\", \"P\xff\"]", &options_step);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "", "a", "Z", "a\xc2", "P\xff"}), &options_step);
SliceOptions options_step_neg{-1, -3, -2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"a\", \"b\", \"\xa2\", \"\xa2\", \"Z\"]", &options_step_neg);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "a", "b", "\xa2", "\xa2", "Z"}), &options_step_neg);
options_step_neg.stop = -4;
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"a\", \"b\", \"\xa2Z\", \"\xa2Z\", \"Z\xc2\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "a", "b", "\xa2Z", "\xa2Z", "Z\xc2"}), &options_step_neg);
}

TYPED_TEST(TestBinaryKernels, SliceBytesNegPos) {
SliceOptions options{-2, 4};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"a\", \"ab\", \"\xc2\xa2\", \"\xc2\xa2\", \"\xff\"]",
&options);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "a", "ab", "\xc2\xa2", "\xc2\xa2", "\xff"}), &options);
SliceOptions options_step{-4, 4, 2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"a\", \"a\", \"Z\xa2\", \"a\xc2\", \"P\xff\"]",
&options_step);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "a", "a", "Z\xa2", "a\xc2", "P\xff"}), &options_step);
SliceOptions options_step_neg{-1, 1, -2};
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"\", \"\xa2\", \"\xa2\", \"Z\xc2\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "", "", "\xa2", "\xa2", "Z\xc2"}), &options_step_neg);
options_step_neg.stop = 0;
this->CheckUnary(
"binary_slice",
"[\"\", \"a\", \"ab\", \"Z\xc2\xa2\", \"aZ\xc2\xa2\", \"aP\xc2\xffZ\"]",
this->type(), "[\"\", \"\", \"b\", \"\xa2\", \"\xa2Z\", \"Z\xc2\"]",
&options_step_neg);
this->MakeArray({"", "a", "ab", "Z\xc2\xa2", "aZ\xc2\xa2", "aP\xc2\xffZ"}),
this->MakeArray({"", "", "b", "\xa2", "\xa2Z", "Z\xc2"}), &options_step_neg);
}

TYPED_TEST(TestStringKernels, PadAscii) {
Expand Down
4 changes: 3 additions & 1 deletion cpp/src/arrow/extension/json_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,7 @@
#include "arrow/array/validate.h"
#include "arrow/ipc/test_common.h"
#include "arrow/record_batch.h"
#include "arrow/testing/builder.h"
#include "arrow/testing/gtest_util.h"
#include "parquet/exception.h"

Expand Down Expand Up @@ -64,7 +65,8 @@ TEST_F(TestJsonExtensionType, JsonRoundtrip) {
TEST_F(TestJsonExtensionType, InvalidUTF8) {
for (const auto& storage_type : {utf8(), large_utf8(), utf8_view()}) {
auto json_type = json(storage_type);
auto invalid_input = ArrayFromJSON(storage_type, "[\"Ⱥa\xFFⱭ\", \"Ɽ\xe1\xbdⱤaA\"]");
auto invalid_input =
BinaryArrayFromStrings(storage_type, {"Ⱥa\xFFⱭ", "Ɽ\xe1\xbdⱤaA"});
auto ext_arr = ExtensionType::WrapArray(json_type, invalid_input);

ASSERT_RAISES_WITH_MESSAGE(Invalid,
Expand Down
Loading
Loading