Add method to parse date from null-terminated string

dremio · Jun 11, 2020 · e722790 · e722790
1 parent a262cf1
commit e722790
Show file tree

Hide file tree

Showing 4 changed files with 72 additions and 2 deletions.
diff --git a/cpp/src/arrow/util/value_parsing.h b/cpp/src/arrow/util/value_parsing.h
@@ -561,6 +561,39 @@ static inline bool ParseTimestampStrptime(const char* buf, size_t length,
  return true;
 }
 
+/// \brief Returns time since the UNIX epoch in the requested unit. Takes null terminated
+/// buffer as argument
+static inline bool ParseTimestampStrptimeFromCString(const char* buf, size_t length,
+ const char* format,
+ bool ignore_time_in_day,
+ bool allow_trailing_chars,
+ TimeUnit::type unit, int64_t* out) {
+ // NOTE: strptime() is more than 10x faster than arrow_vendored::date::parse().
+ struct tm result;
+ memset(&result, 0, sizeof(struct tm));
+#ifdef _WIN32
+ char* ret = arrow_strptime(buf, format, &result);
+#else
+ char* ret = strptime(buf, format, &result);
+#endif
+ if (ret == NULLPTR) {
+ return false;
+ }
+ if (!allow_trailing_chars && static_cast<size_t>(ret - buf) != length) {
+ return false;
+ }
+ // ignore the time part
+ arrow_vendored::date::sys_seconds secs =
+ arrow_vendored::date::sys_days(arrow_vendored::date::year(result.tm_year + 1900) /
+ (result.tm_mon + 1) / result.tm_mday);
+ if (!ignore_time_in_day) {
+ secs += (std::chrono::hours(result.tm_hour) + std::chrono::minutes(result.tm_min) +
+ std::chrono::seconds(result.tm_sec));
+ }
+ *out = detail::ConvertTimePoint(secs, unit);
+ return true;
+}
+
 /// \brief Parsing options for timestamps
 struct ParseTimestampContext {
  TimeUnit::type unit;

diff --git a/cpp/src/gandiva/tests/generate_data.h b/cpp/src/gandiva/tests/generate_data.h
@@ -129,4 +129,16 @@ class FastUtf8DataGenerator : public DataGenerator<std::string> {
  char cur_char_;
 };
 
+class Utf8DateDataGenerator : public DataGenerator<std::string> {
+ public:
+ Utf8DateDataGenerator() {}
+
+ std::string GenerateData() {
+ return "1990-0" + std::to_string(random_.next() / 9 + 1) + "-1" +
+ std::to_string(random_.next() / 9 + 1);
+ }
+
+ private:
+ Random random_;
+};
 } // namespace gandiva
diff --git a/cpp/src/gandiva/tests/micro_benchmarks.cc b/cpp/src/gandiva/tests/micro_benchmarks.cc
@@ -277,6 +277,31 @@ static void TimedTestInExpr(benchmark::State& state) {
  ASSERT_OK(status);
 }
 
+static void TimedTestToDate(benchmark::State& state) {
+ auto field_a = field("a", utf8());
+ auto schema = arrow::schema({field_a});
+ auto pool = arrow::default_memory_pool();
+
+ auto field_result = field("res", arrow::date64());
+
+ auto node_a = TreeExprBuilder::MakeField(field_a);
+ auto date_pattern = TreeExprBuilder::MakeStringLiteral("YYYY-MM-DD");
+ auto suppress_literal = TreeExprBuilder::MakeLiteral(1);
+ auto fn = TreeExprBuilder::MakeFunction(
+ "to_date", {node_a, date_pattern, suppress_literal}, arrow::date64());
+ auto expr = TreeExprBuilder::MakeExpression(fn, field_result);
+
+ std::shared_ptr<Projector> projector;
+ ASSERT_OK(Projector::Make(schema, {expr}, TestConfiguration(), &projector));
+
+ Utf8DateDataGenerator data_generator;
+ ProjectEvaluator evaluator(projector);
+
+ Status status = TimedEvaluate<arrow::StringType, std::string>(
+ schema, evaluator, data_generator, pool, 1 * MILLION, 16 * THOUSAND, state);
+ ASSERT_TRUE(status.ok());
+}
+
 static void DoDecimalAdd3(benchmark::State& state, int32_t precision, int32_t scale,
  bool large = false) {
  // schema for input fields
@@ -398,6 +423,7 @@ BENCHMARK(TimedTestFilterLike)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(TimedTestAllocs)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(TimedTestMultiOr)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(TimedTestInExpr)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
+BENCHMARK(TimedTestToDate)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(DecimalAdd2Fast)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(DecimalAdd2LeadingZeroes)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(DecimalAdd2LeadingZeroesWithDiv)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
@@ -406,5 +432,4 @@ BENCHMARK(DecimalAdd3Fast)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(DecimalAdd3LeadingZeroes)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(DecimalAdd3LeadingZeroesWithDiv)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
 BENCHMARK(DecimalAdd3Large)->MinTime(1.0)->Unit(benchmark::kMicrosecond);
-
 } // namespace gandiva
diff --git a/cpp/src/gandiva/to_date_holder.cc b/cpp/src/gandiva/to_date_holder.cc
@@ -83,7 +83,7 @@ int64_t ToDateHolder::operator()(ExecutionContext* context, const std::string& d
  // 1. processes date that do not match the format.
  // 2. does not process time in format +08:00 (or) id.
  int64_t seconds_since_epoch = 0;
- if (!::arrow::internal::ParseTimestampStrptime(
+ if (!::arrow::internal::ParseTimestampStrptimeFromCString(
  data.c_str(), data.length(), pattern_.c_str(),
  /*ignore_time_in_day=*/true, /*allow_trailing_chars=*/true,
  ::arrow::TimeUnit::SECOND, &seconds_since_epoch)) {