Skip to content

ARROW-11693: [C++] Add string length kernel #9786

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

Closed
edponce wants to merge 8 commits into from
Closed

ARROW-11693: [C++] Add string length kernel #9786

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
bbb32e7
Add utf8_length documentation
edponce Mar 23, 2021
8630d24
add a UTF8 case to binary_length, add an equivalent test for utf8_length
edponce Mar 23, 2021
05d6bde
add a test ensuring "\0" is invalid utf8
edponce Mar 23, 2021
6379ff4
add utf8_length kernels for STRING and LARGE_STRING
edponce Mar 23, 2021
0c81029
add optimized implementation for UTF8 strlen (does not depends on utf…
edponce Mar 24, 2021
16deddf
fix UTF8 tests, NUL is a valid character
edponce Mar 24, 2021
8e8f37b
add tests for utf8_length (emoji and mixed ASCII/UTF8)
edponce Mar 24, 2021
46a6b7a
fix format of C++ files
edponce Mar 24, 2021
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
4 changes: 4 additions & 0 deletions cpp/src/arrow/array/array_binary_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -301,14 +301,18 @@ class TestStringArray : public ::testing::Test {
auto st2 = ValidateFull(1, {0, 4}, "\xf4\x90\x80\x80");
// Single UTF8 character straddles two entries
auto st3 = ValidateFull(2, {0, 1, 2}, "\xc3\xa9");
// Null characters in the string
auto st4 = ValidateFull(1, {0, 4}, "\0\0\0\0");

@pitrou pitrou Mar 24, 2021

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.

Can you explain why this is invalid? Unicode character 0 is a valid unicode character.

@pitrou pitrou Mar 24, 2021

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.

Also, it works using PyArrow:

>>> arr = pa.array(["\0"])
>>> arr
<pyarrow.lib.StringArray object at 0x7f6befb354b0>
[
  ""
]
>>> arr.to_pylist()
['\x00']
>>> arr.validate(full=True)
>>>

@edponce edponce Mar 24, 2021 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

You are right, this case is valid. The UTF-8 standard (https://tools.ietf.org/html/rfc3629) states that characters consisting of 1-byte range from 0-7F (Section 3). But a multibyte UTF-8 character will not contain NUL as a continuation byte, the encoding does not allows it.
Valid: '\x00'
Invalid: '\xe10080'

@pitrou pitrou Mar 24, 2021

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.

Also, note that UTF8 validation is tested more thoroughly in https://github.com/apache/arrow/blob/master/cpp/src/arrow/util/utf8_util_test.cc

if (T::is_utf8) {
ASSERT_RAISES(Invalid, st1);
ASSERT_RAISES(Invalid, st2);
ASSERT_RAISES(Invalid, st3);
ASSERT_RAISES(Invalid, st4);
} else {
ASSERT_OK(st1);
ASSERT_OK(st2);
ASSERT_OK(st3);
ASSERT_OK(st4);
}
}

Expand Down
48 changes: 47 additions & 1 deletion cpp/src/arrow/compute/kernels/scalar_string.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,25 @@ struct BinaryLength {

#ifdef ARROW_WITH_UTF8PROC

struct Utf8Length {
template <typename OutValue, typename Arg0Value = util::string_view>
static OutValue Call(KernelContext*, Arg0Value val) {
auto str = reinterpret_cast<const utf8proc_uint8_t*>(val.data());
auto strlen = static_cast<utf8proc_ssize_t>(val.size());
utf8proc_int32_t codepoint;

OutValue length = 0;
while (strlen > 0) {
auto char_width = utf8proc_iterate(str, strlen, &codepoint);
Comment thread
pitrou marked this conversation as resolved.
Outdated
// XXX check for errmsg?
str += char_width;
strlen -= char_width;
++length;
}
return length;
}
};

// Direct lookup tables for unicode properties
constexpr uint32_t kMaxCodepointLookup =
0xffff; // up to this codepoint is in a lookup table
Expand Down Expand Up @@ -1569,9 +1588,16 @@ const FunctionDoc strptime_doc(

const FunctionDoc binary_length_doc(
"Compute string lengths",
("For each string in `strings`, emit its length. Null values emit null."),
("For each string in `strings`, emit the number of bytes. Null values emit null."),
{"strings"});

#ifdef ARROW_WITH_UTF8PROC
const FunctionDoc utf8_length_doc("Compute utf8 string lengths",
("For each string in `strings`, emit the number of "
"utf8 characters. Null values emit null."),
{"strings"});
#endif // ARROW_WITH_UTF8PROC

void AddStrptime(FunctionRegistry* registry) {
auto func = std::make_shared<ScalarFunction>("strptime", Arity::Unary(), &strptime_doc);
DCHECK_OK(func->AddKernel({utf8()}, OutputType(StrptimeResolve),
Expand All @@ -1597,6 +1623,23 @@ void AddBinaryLength(FunctionRegistry* registry) {
DCHECK_OK(registry->AddFunction(std::move(func)));
}

#ifdef ARROW_WITH_UTF8PROC
void AddUtf8Length(FunctionRegistry* registry) {
auto func =
std::make_shared<ScalarFunction>("utf8_length", Arity::Unary(), &utf8_length_doc);

ArrayKernelExec exec_offset_32 =
applicator::ScalarUnaryNotNull<Int32Type, StringType, Utf8Length>::Exec;
DCHECK_OK(func->AddKernel({utf8()}, int32(), std::move(exec_offset_32)));

ArrayKernelExec exec_offset_64 =
applicator::ScalarUnaryNotNull<Int64Type, LargeStringType, Utf8Length>::Exec;
DCHECK_OK(func->AddKernel({large_utf8()}, int64(), std::move(exec_offset_64)));

DCHECK_OK(registry->AddFunction(std::move(func)));
}
#endif // ARROW_WITH_UTF8PROC

template <template <typename> class ExecFunctor>
void MakeUnaryStringBatchKernel(
std::string name, FunctionRegistry* registry, const FunctionDoc* doc,
Expand Down Expand Up @@ -1866,6 +1909,9 @@ void RegisterScalarStringAscii(FunctionRegistry* registry) {

AddSplit(registry);
AddBinaryLength(registry);
#ifdef ARROW_WITH_UTF8PROC
AddUtf8Length(registry);
#endif
AddMatchSubstring(registry);
AddStrptime(registry);
}
Expand Down
9 changes: 7 additions & 2 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 @@ -61,8 +61,8 @@ class TestBinaryKernels : public BaseTestStringKernels<TestType> {};
TYPED_TEST_SUITE(TestBinaryKernels, BinaryTypes);

TYPED_TEST(TestBinaryKernels, BinaryLength) {
this->CheckUnary("binary_length", R"(["aaa", null, "", "b"])", this->offset_type(),
"[3, null, 0, 1]");
this->CheckUnary("binary_length", R"(["aaa", null, "áéíóú", "", "b"])",
this->offset_type(), "[3, null, 10, 0, 1]");
}

template <typename TestType>
Expand Down Expand Up @@ -103,6 +103,11 @@ TEST(TestStringKernels, LARGE_MEMORY_TEST(Utf8Upper32bitGrowth)) {

#ifdef ARROW_WITH_UTF8PROC

TYPED_TEST(TestStringKernels, Utf8Length) {
this->CheckUnary("utf8_length", R"(["aaa", null, "áéíóú", "", "b"])",
this->offset_type(), "[3, null, 5, 0, 1]");
Comment thread
pitrou marked this conversation as resolved.
Outdated
}

TYPED_TEST(TestStringKernels, Utf8Upper) {
this->CheckUnary("utf8_upper", "[\"aAazZæÆ&\", null, \"\", \"b\"]", this->type(),
"[\"AAAZZÆÆ&\", null, \"\", \"B\"]");
Expand Down
11 changes: 8 additions & 3 deletions docs/source/cpp/compute.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -435,9 +435,11 @@ String transforms
+--------------------------+------------+-------------------------+---------------------+---------+
| binary_length | Unary | Binary- or String-like | Int32 or Int64 | \(2) |
+--------------------------+------------+-------------------------+---------------------+---------+
| utf8_lower | Unary | String-like | String-like | \(3) |
| utf8_length | Unary | String-like | Int32 or Int64 | \(3) |
+--------------------------+------------+-------------------------+---------------------+---------+
| utf8_upper | Unary | String-like | String-like | \(3) |
| utf8_lower | Unary | String-like | String-like | \(4) |
+--------------------------+------------+-------------------------+---------------------+---------+
| utf8_upper | Unary | String-like | String-like | \(4) |
+--------------------------+------------+-------------------------+---------------------+---------+


Expand All @@ -447,7 +449,10 @@ String transforms
* \(2) Output is the physical length in bytes of each input element. Output
type is Int32 for Binary / String, Int64 for LargeBinary / LargeString.

* \(3) Each UTF8-encoded character in the input is converted to lowercase or
* \(3) Output is the number of characters (not bytes) of each input element.
Output type is Int32 for String, Int64 for LargeString.

* \(4) Each UTF8-encoded character in the input is converted to lowercase or
uppercase.


Expand Down