Export of internal Abseil changes

--
509c39cb5aa70893a180e5625e06cd9f76061ecf by Shaindel Schwartz <shaindel@google.com>:

Release CivilTime parsing API.

PiperOrigin-RevId: 270262448
GitOrigin-RevId: 509c39cb5aa70893a180e5625e06cd9f76061ecf
Change-Id: I343eb3062cdf6a2c53e6fddcaa35eb25d7478b1a

4 files changed, 319 insertions, 0 deletions

diff --git a/absl/time/civil_time.cc b/absl/time/civil_time.cc
index 7527fc11..7b86fa8b 100644
--- a/absl/time/civil_time.cc
+++ b/absl/time/civil_time.cc
@@ -42,6 +42,58 @@ std::string FormatYearAnd(string_view fmt, CivilSecond cs) {
                 FormatTime(std::string(fmt), FromCivil(ncs, utc), utc));
 }
 
+template <typename CivilT>
+bool ParseYearAnd(string_view fmt, string_view s, CivilT* c) {
+  // Civil times support a larger year range than absl::Time, so we need to
+  // parse the year separately, normalize it, then use absl::ParseTime on the
+  // normalized std::string.
+  const std::string ss = std::string(s);  // TODO(absl-team): Avoid conversion.
+  const char* const np = ss.c_str();
+  char* endp;
+  errno = 0;
+  const civil_year_t y =
+      std::strtoll(np, &endp, 10);  // NOLINT(runtime/deprecated_fn)
+  if (endp == np || errno == ERANGE) return false;
+  const std::string norm = StrCat(NormalizeYear(y), endp);
+
+  const TimeZone utc = UTCTimeZone();
+  Time t;
+  if (ParseTime(StrCat("%Y", fmt), norm, utc, &t, nullptr)) {
+    const auto cs = ToCivilSecond(t, utc);
+    *c = CivilT(y, cs.month(), cs.day(), cs.hour(), cs.minute(), cs.second());
+    return true;
+  }
+
+  return false;
+}
+
+// Tries to parse the type as a CivilT1, but then assigns the result to the
+// argument of type CivilT2.
+template <typename CivilT1, typename CivilT2>
+bool ParseAs(string_view s, CivilT2* c) {
+  CivilT1 t1;
+  if (ParseCivilTime(s, &t1)) {
+    *c = CivilT2(t1);
+    return true;
+  }
+  return false;
+}
+
+template <typename CivilT>
+bool ParseLenient(string_view s, CivilT* c) {
+  // A fastpath for when the given std::string data parses exactly into the given
+  // type T (e.g., s="YYYY-MM-DD" and CivilT=CivilDay).
+  if (ParseCivilTime(s, c)) return true;
+  // Try parsing as each of the 6 types, trying the most common types first
+  // (based on csearch results).
+  if (ParseAs<CivilDay>(s, c)) return true;
+  if (ParseAs<CivilSecond>(s, c)) return true;
+  if (ParseAs<CivilHour>(s, c)) return true;
+  if (ParseAs<CivilMonth>(s, c)) return true;
+  if (ParseAs<CivilMinute>(s, c)) return true;
+  if (ParseAs<CivilYear>(s, c)) return true;
+  return false;
+}
 }  // namespace
 
 std::string FormatCivilTime(CivilSecond c) {
@@ -57,6 +109,44 @@ std::string FormatCivilTime(CivilDay c) { return FormatYearAnd("-%m-%d", c); }
 std::string FormatCivilTime(CivilMonth c) { return FormatYearAnd("-%m", c); }
 std::string FormatCivilTime(CivilYear c) { return FormatYearAnd("", c); }
 
+bool ParseCivilTime(string_view s, CivilSecond* c) {
+  return ParseYearAnd("-%m-%dT%H:%M:%S", s, c);
+}
+bool ParseCivilTime(string_view s, CivilMinute* c) {
+  return ParseYearAnd("-%m-%dT%H:%M", s, c);
+}
+bool ParseCivilTime(string_view s, CivilHour* c) {
+  return ParseYearAnd("-%m-%dT%H", s, c);
+}
+bool ParseCivilTime(string_view s, CivilDay* c) {
+  return ParseYearAnd("-%m-%d", s, c);
+}
+bool ParseCivilTime(string_view s, CivilMonth* c) {
+  return ParseYearAnd("-%m", s, c);
+}
+bool ParseCivilTime(string_view s, CivilYear* c) {
+  return ParseYearAnd("", s, c);
+}
+
+bool ParseLenientCivilTime(string_view s, CivilSecond* c) {
+  return ParseLenient(s, c);
+}
+bool ParseLenientCivilTime(string_view s, CivilMinute* c) {
+  return ParseLenient(s, c);
+}
+bool ParseLenientCivilTime(string_view s, CivilHour* c) {
+  return ParseLenient(s, c);
+}
+bool ParseLenientCivilTime(string_view s, CivilDay* c) {
+  return ParseLenient(s, c);
+}
+bool ParseLenientCivilTime(string_view s, CivilMonth* c) {
+  return ParseLenient(s, c);
+}
+bool ParseLenientCivilTime(string_view s, CivilYear* c) {
+  return ParseLenient(s, c);
+}
+
 namespace time_internal {
 
 std::ostream& operator<<(std::ostream& os, CivilYear y) {
diff --git a/absl/time/civil_time.h b/absl/time/civil_time.h
index beaf7d89..7c52586a 100644
--- a/absl/time/civil_time.h
+++ b/absl/time/civil_time.h
@@ -459,6 +459,57 @@ std::string FormatCivilTime(CivilDay c);
 std::string FormatCivilTime(CivilMonth c);
 std::string FormatCivilTime(CivilYear c);
 
+// absl::ParseCivilTime()
+//
+// Parses a civil-time value from the specified `absl::string_view` into the
+// passed output parameter. Returns `true` upon successful parsing.
+//
+// The expected form of the input string is as follows:
+//
+//  Type        | Format
+//  ---------------------------------
+//  CivilSecond | YYYY-MM-DDTHH:MM:SS
+//  CivilMinute | YYYY-MM-DDTHH:MM
+//  CivilHour   | YYYY-MM-DDTHH
+//  CivilDay    | YYYY-MM-DD
+//  CivilMonth  | YYYY-MM
+//  CivilYear   | YYYY
+//
+// Example:
+//
+//   absl::CivilDay d;
+//   bool ok = absl::ParseCivilTime("2018-01-02", &d); // OK
+//
+// Note that parsing will fail if the string's format does not match the
+// expected type exactly. `ParseLenientCivilTime()` below is more lenient.
+//
+bool ParseCivilTime(absl::string_view s, CivilSecond* c);
+bool ParseCivilTime(absl::string_view s, CivilMinute* c);
+bool ParseCivilTime(absl::string_view s, CivilHour* c);
+bool ParseCivilTime(absl::string_view s, CivilDay* c);
+bool ParseCivilTime(absl::string_view s, CivilMonth* c);
+bool ParseCivilTime(absl::string_view s, CivilYear* c);
+
+// ParseLenientCivilTime()
+//
+// Parses any of the formats accepted by `absl::ParseCivilTime()`, but is more
+// lenient if the format of the string does not exactly match the associated
+// type.
+//
+// Example:
+//
+//   absl::CivilDay d;
+//   bool ok = absl::ParseLenientCivilTime("1969-07-20", &d); // OK
+//   ok = absl::ParseLenientCivilTime("1969-07-20T10", &d);   // OK: T10 floored
+//   ok = absl::ParseLenientCivilTime("1969-07", &d);   // OK: day defaults to 1
+//
+bool ParseLenientCivilTime(absl::string_view s, CivilSecond* c);
+bool ParseLenientCivilTime(absl::string_view s, CivilMinute* c);
+bool ParseLenientCivilTime(absl::string_view s, CivilHour* c);
+bool ParseLenientCivilTime(absl::string_view s, CivilDay* c);
+bool ParseLenientCivilTime(absl::string_view s, CivilMonth* c);
+bool ParseLenientCivilTime(absl::string_view s, CivilYear* c);
+
 namespace time_internal {  // For functions found via ADL on civil-time tags.
 
 // Streaming Operators
diff --git a/absl/time/civil_time_benchmark.cc b/absl/time/civil_time_benchmark.cc
index 40869835..f04dbe20 100644
--- a/absl/time/civil_time_benchmark.cc
+++ b/absl/time/civil_time_benchmark.cc
@@ -66,6 +66,26 @@ void BM_Format(benchmark::State& state) {
 }
 BENCHMARK(BM_Format);
 
+void BM_Parse(benchmark::State& state) {
+  const std::string f = "2014-01-02T03:04:05";
+  absl::CivilSecond c;
+  while (state.KeepRunning()) {
+    const bool b = absl::ParseCivilTime(f, &c);
+    benchmark::DoNotOptimize(b);
+  }
+}
+BENCHMARK(BM_Parse);
+
+void BM_RoundTripFormatParse(benchmark::State& state) {
+  const absl::CivilSecond c(2014, 1, 2, 3, 4, 5);
+  absl::CivilSecond out;
+  while (state.KeepRunning()) {
+    const bool b = absl::ParseCivilTime(absl::FormatCivilTime(c), &out);
+    benchmark::DoNotOptimize(b);
+  }
+}
+BENCHMARK(BM_RoundTripFormatParse);
+
 template <typename T>
 void BM_CivilTimeAbslHash(benchmark::State& state) {
   const int kSize = 100000;
diff --git a/absl/time/civil_time_test.cc b/absl/time/civil_time_test.cc
index 03cd1f12..0ebd97ad 100644
--- a/absl/time/civil_time_test.cc
+++ b/absl/time/civil_time_test.cc
@@ -690,6 +690,69 @@ TEST(CivilTime, Format) {
   EXPECT_EQ("1970", absl::FormatCivilTime(y));
 }
 
+TEST(CivilTime, Parse) {
+  absl::CivilSecond ss;
+  absl::CivilMinute mm;
+  absl::CivilHour hh;
+  absl::CivilDay d;
+  absl::CivilMonth m;
+  absl::CivilYear y;
+
+  // CivilSecond OK; others fail
+  EXPECT_TRUE(absl::ParseCivilTime("2015-01-02T03:04:05", &ss));
+  EXPECT_EQ("2015-01-02T03:04:05", absl::FormatCivilTime(ss));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04:05", &mm));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04:05", &hh));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04:05", &d));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04:05", &m));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04:05", &y));
+
+  // CivilMinute OK; others fail
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04", &ss));
+  EXPECT_TRUE(absl::ParseCivilTime("2015-01-02T03:04", &mm));
+  EXPECT_EQ("2015-01-02T03:04", absl::FormatCivilTime(mm));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04", &hh));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04", &d));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04", &m));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03:04", &y));
+
+  // CivilHour OK; others fail
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03", &ss));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03", &mm));
+  EXPECT_TRUE(absl::ParseCivilTime("2015-01-02T03", &hh));
+  EXPECT_EQ("2015-01-02T03", absl::FormatCivilTime(hh));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03", &d));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03", &m));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02T03", &y));
+
+  // CivilDay OK; others fail
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02", &ss));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02", &mm));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02", &hh));
+  EXPECT_TRUE(absl::ParseCivilTime("2015-01-02", &d));
+  EXPECT_EQ("2015-01-02", absl::FormatCivilTime(d));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02", &m));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01-02", &y));
+
+  // CivilMonth OK; others fail
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01", &ss));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01", &mm));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01", &hh));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01", &d));
+  EXPECT_TRUE(absl::ParseCivilTime("2015-01", &m));
+  EXPECT_EQ("2015-01", absl::FormatCivilTime(m));
+  EXPECT_FALSE(absl::ParseCivilTime("2015-01", &y));
+
+  // CivilYear OK; others fail
+  EXPECT_FALSE(absl::ParseCivilTime("2015", &ss));
+  EXPECT_FALSE(absl::ParseCivilTime("2015", &mm));
+  EXPECT_FALSE(absl::ParseCivilTime("2015", &hh));
+  EXPECT_FALSE(absl::ParseCivilTime("2015", &d));
+  EXPECT_FALSE(absl::ParseCivilTime("2015", &m));
+  EXPECT_TRUE(absl::ParseCivilTime("2015", &y));
+  EXPECT_EQ("2015", absl::FormatCivilTime(y));
+}
+
 TEST(CivilTime, FormatAndParseLenient) {
   absl::CivilSecond ss;
   EXPECT_EQ("1970-01-01T00:00:00", absl::FormatCivilTime(ss));
@@ -708,6 +771,101 @@ TEST(CivilTime, FormatAndParseLenient) {
 
   absl::CivilYear y;
   EXPECT_EQ("1970", absl::FormatCivilTime(y));
+
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-01-02T03:04:05", &ss));
+  EXPECT_EQ("2015-01-02T03:04:05", absl::FormatCivilTime(ss));
+
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-01-02T03:04:05", &mm));
+  EXPECT_EQ("2015-01-02T03:04", absl::FormatCivilTime(mm));
+
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-01-02T03:04:05", &hh));
+  EXPECT_EQ("2015-01-02T03", absl::FormatCivilTime(hh));
+
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-01-02T03:04:05", &d));
+  EXPECT_EQ("2015-01-02", absl::FormatCivilTime(d));
+
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-01-02T03:04:05", &m));
+  EXPECT_EQ("2015-01", absl::FormatCivilTime(m));
+
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-01-02T03:04:05", &y));
+  EXPECT_EQ("2015", absl::FormatCivilTime(y));
+}
+
+TEST(CivilTime, ParseEdgeCases) {
+  absl::CivilSecond ss;
+  EXPECT_TRUE(
+      absl::ParseLenientCivilTime("9223372036854775807-12-31T23:59:59", &ss));
+  EXPECT_EQ("9223372036854775807-12-31T23:59:59", absl::FormatCivilTime(ss));
+  EXPECT_TRUE(
+      absl::ParseLenientCivilTime("-9223372036854775808-01-01T00:00:00", &ss));
+  EXPECT_EQ("-9223372036854775808-01-01T00:00:00", absl::FormatCivilTime(ss));
+
+  absl::CivilMinute mm;
+  EXPECT_TRUE(
+      absl::ParseLenientCivilTime("9223372036854775807-12-31T23:59", &mm));
+  EXPECT_EQ("9223372036854775807-12-31T23:59", absl::FormatCivilTime(mm));
+  EXPECT_TRUE(
+      absl::ParseLenientCivilTime("-9223372036854775808-01-01T00:00", &mm));
+  EXPECT_EQ("-9223372036854775808-01-01T00:00", absl::FormatCivilTime(mm));
+
+  absl::CivilHour hh;
+  EXPECT_TRUE(
+      absl::ParseLenientCivilTime("9223372036854775807-12-31T23", &hh));
+  EXPECT_EQ("9223372036854775807-12-31T23", absl::FormatCivilTime(hh));
+  EXPECT_TRUE(
+      absl::ParseLenientCivilTime("-9223372036854775808-01-01T00", &hh));
+  EXPECT_EQ("-9223372036854775808-01-01T00", absl::FormatCivilTime(hh));
+
+  absl::CivilDay d;
+  EXPECT_TRUE(absl::ParseLenientCivilTime("9223372036854775807-12-31", &d));
+  EXPECT_EQ("9223372036854775807-12-31", absl::FormatCivilTime(d));
+  EXPECT_TRUE(absl::ParseLenientCivilTime("-9223372036854775808-01-01", &d));
+  EXPECT_EQ("-9223372036854775808-01-01", absl::FormatCivilTime(d));
+
+  absl::CivilMonth m;
+  EXPECT_TRUE(absl::ParseLenientCivilTime("9223372036854775807-12", &m));
+  EXPECT_EQ("9223372036854775807-12", absl::FormatCivilTime(m));
+  EXPECT_TRUE(absl::ParseLenientCivilTime("-9223372036854775808-01", &m));
+  EXPECT_EQ("-9223372036854775808-01", absl::FormatCivilTime(m));
+
+  absl::CivilYear y;
+  EXPECT_TRUE(absl::ParseLenientCivilTime("9223372036854775807", &y));
+  EXPECT_EQ("9223372036854775807", absl::FormatCivilTime(y));
+  EXPECT_TRUE(absl::ParseLenientCivilTime("-9223372036854775808", &y));
+  EXPECT_EQ("-9223372036854775808", absl::FormatCivilTime(y));
+
+  // Tests some valid, but interesting, cases
+  EXPECT_TRUE(absl::ParseLenientCivilTime("0", &ss)) << ss;
+  EXPECT_EQ(absl::CivilYear(0), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime("0-1", &ss)) << ss;
+  EXPECT_EQ(absl::CivilMonth(0, 1), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime(" 2015 ", &ss)) << ss;
+  EXPECT_EQ(absl::CivilYear(2015), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime(" 2015-6 ", &ss)) << ss;
+  EXPECT_EQ(absl::CivilMonth(2015, 6), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-6-7", &ss)) << ss;
+  EXPECT_EQ(absl::CivilDay(2015, 6, 7), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime(" 2015-6-7 ", &ss)) << ss;
+  EXPECT_EQ(absl::CivilDay(2015, 6, 7), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime("2015-06-07T10:11:12 ", &ss)) << ss;
+  EXPECT_EQ(absl::CivilSecond(2015, 6, 7, 10, 11, 12), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime(" 2015-06-07T10:11:12 ", &ss)) << ss;
+  EXPECT_EQ(absl::CivilSecond(2015, 6, 7, 10, 11, 12), ss);
+  EXPECT_TRUE(absl::ParseLenientCivilTime("-01-01", &ss)) << ss;
+  EXPECT_EQ(absl::CivilMonth(-1, 1), ss);
+
+  // Tests some invalid cases
+  EXPECT_FALSE(absl::ParseLenientCivilTime("01-01-2015", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015-", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("0xff-01", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015-02-30T04:05:06", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015-02-03T04:05:96", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("X2015-02-03T04:05:06", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015-02-03T04:05:003", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015 -02-03T04:05:06", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015-02-03-04:05:06", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("2015:02:03T04-05-06", &ss)) << ss;
+  EXPECT_FALSE(absl::ParseLenientCivilTime("9223372036854775808", &y)) << y;
 }
 
 TEST(CivilTime, OutputStream) {