update vendored/datetime

apache · Jul 11, 2024 · bfe0792 · bfe0792
1 parent edfa343
commit bfe0792
Show file tree

Hide file tree

Showing 7 changed files with 113 additions and 55 deletions.
diff --git a/cpp/src/arrow/vendored/datetime/README.md b/cpp/src/arrow/vendored/datetime/README.md
@@ -17,7 +17,7 @@ copies or substantial portions of the Software.
 Sources for datetime are adapted from Howard Hinnant's date library
 (https://github.com/HowardHinnant/date).
 
-Sources are taken from changeset cc4685a21e4a4fdae707ad1233c61bbaff241f93
+Sources are taken from changeset 1ead6715dec030d340a316c927c877a3c4e5a00c
 of the above project.
 
 The following changes are made:

diff --git a/cpp/src/arrow/vendored/datetime/date.h b/cpp/src/arrow/vendored/datetime/date.h
@@ -84,8 +84,6 @@
 #   pragma warning(disable : 4127)
 #endif
 
-namespace arrow_vendored
-{
 namespace date
 {
 
@@ -4230,7 +4228,7 @@ inline
 std::basic_ostream<CharT, Traits>&
 operator<<(std::basic_ostream<CharT, Traits>& os, const local_time<Duration>& ut)
 {
-    return (os << sys_time<Duration>{ut.time_since_epoch()});
+    return (date::operator<<(os, sys_time<Duration>{ut.time_since_epoch()}));
 }
 
 namespace detail
@@ -6353,7 +6351,10 @@ read_signed(std::basic_istream<CharT, Traits>& is, unsigned m = 1, unsigned M =
         if (('0' <= c && c <= '9') || c == '-' || c == '+')
         {
             if (c == '-' || c == '+')
+            {
                 (void)is.get();
+                --M;
+            }
             auto x = static_cast<int>(read_unsigned(is, std::max(m, 1u), M));
             if (!is.fail())
             {
@@ -6526,7 +6527,14 @@ read(std::basic_istream<CharT, Traits>& is, int a0, Args&& ...args)
             *e++ = static_cast<CharT>(CharT(u % 10) + CharT{'0'});
             u /= 10;
         } while (u > 0);
+#if defined(__GNUC__) && __GNUC__ >= 11
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wstringop-overflow"
+#endif
         std::reverse(buf, e);
+#if defined(__GNUC__) && __GNUC__ >= 11
+#pragma GCC diagnostic pop
+#endif
         for (auto p = buf; p != e && is.rdstate() == std::ios::goodbit; ++p)
             read(is, *p);
     }
@@ -6592,7 +6600,7 @@ from_stream(std::basic_istream<CharT, Traits>& is, const CharT* fmt,
 
         CONSTDATA int not_a_year = numeric_limits<short>::min();
         CONSTDATA int not_a_2digit_year = 100;
-        CONSTDATA int not_a_century = not_a_year / 100;
+        CONSTDATA int not_a_century = numeric_limits<int>::min();
         CONSTDATA int not_a_month = 0;
         CONSTDATA int not_a_day = 0;
         CONSTDATA int not_a_hour = numeric_limits<int>::min();
@@ -7519,7 +7527,12 @@ from_stream(std::basic_istream<CharT, Traits>& is, const CharT* fmt,
                     {
                         auto c = static_cast<char>(Traits::to_char_type(ic));
                         if (c == '-')
+                        {
                             neg = true;
+                            (void)is.get();
+                        }
+                        else if (c == '+')
+                            (void)is.get();
                     }
                     if (modified == CharT{})
                     {
@@ -7735,9 +7748,7 @@ from_stream(std::basic_istream<CharT, Traits>& is, const CharT* fmt,
                 year_month_day ymd_trial = sys_days(year{Y}/January/Sunday[1]) +
                                            weeks{U-1} +
                                            (weekday{static_cast<unsigned>(wd)} - Sunday);
-                if (Y == not_a_year)
-                    Y = static_cast<int>(ymd_trial.year());
-                else if (year{Y} != ymd_trial.year())
+                if (year{Y} != ymd_trial.year())
                     goto broken;
                 if (m == not_a_month)
                     m = static_cast<int>(static_cast<unsigned>(ymd_trial.month()));
@@ -7754,9 +7765,7 @@ from_stream(std::basic_istream<CharT, Traits>& is, const CharT* fmt,
                 year_month_day ymd_trial = sys_days(year{Y}/January/Monday[1]) +
                                            weeks{W-1} +
                                            (weekday{static_cast<unsigned>(wd)} - Monday);
-                if (Y == not_a_year)
-                    Y = static_cast<int>(ymd_trial.year());
-                else if (year{Y} != ymd_trial.year())
+                if (year{Y} != ymd_trial.year())
                     goto broken;
                 if (m == not_a_month)
                     m = static_cast<int>(static_cast<unsigned>(ymd_trial.month()));
@@ -8224,7 +8233,6 @@ operator<<(std::basic_ostream<CharT, Traits>& os,
 }
 
 }  // namespace date
-}  // namespace arrow_vendored
 
 #ifdef _MSC_VER
 #   pragma warning(pop)

diff --git a/cpp/src/arrow/vendored/datetime/ios.h b/cpp/src/arrow/vendored/datetime/ios.h
@@ -32,8 +32,6 @@
 # if TARGET_OS_IPHONE
 #   include <string>
 
-    namespace arrow_vendored
-    {
     namespace date
     {
     namespace iOSUtils
@@ -44,7 +42,6 @@
 
     }  // namespace iOSUtils
     }  // namespace date
-    }  // namespace arrow_vendored
 
 # endif  // TARGET_OS_IPHONE
 #else   // !__APPLE__

diff --git a/cpp/src/arrow/vendored/datetime/ios.mm b/cpp/src/arrow/vendored/datetime/ios.mm
@@ -22,7 +22,7 @@
 // SOFTWARE.
 //
 
-#include "ios.h"
+#include "date/ios.h"
 
 #if TARGET_OS_IPHONE
 
@@ -47,8 +47,6 @@
 #define TAR_SIZE_POSITION               124
 #define TAR_SIZE_SIZE                   12
 
-namespace arrow_vendored
-{
 namespace date
 {
     namespace iOSUtils
@@ -335,6 +333,5 @@ bool writeFile(const std::string &tzdataPath, const std::string &fileName,
 
     }  // namespace iOSUtils
 }  // namespace date
-}  // namespace arrow_vendored
 
 #endif  // TARGET_OS_IPHONE
diff --git a/cpp/src/arrow/vendored/datetime/tz.cpp b/cpp/src/arrow/vendored/datetime/tz.cpp
@@ -30,10 +30,6 @@
 // been invented (that would involve another several millennia of evolution).
 // We did not mean to shout.
 
-// NOTE(ARROW): This is required so that symbols are properly exported from the DLL
-#include "visibility.h"
-
-
 #ifdef _WIN32
    // windows.h will be included directly and indirectly (e.g. by curl).
    // We need to define these macros to prevent windows.h bringing in
@@ -87,15 +83,19 @@
 #  include <windows.h>
 #endif  // _WIN32
 
-#include "tz_private.h"
+#include "date/tz_private.h"
 
 #ifdef __APPLE__
-#  include "ios.h"
+#  include "date/ios.h"
 #else
 #  define TARGET_OS_IPHONE 0
 #  define TARGET_OS_SIMULATOR 0
 #endif
 
+#if defined(ANDROID) || defined(__ANDROID__)
+#include <sys/system_properties.h>
+#endif
+
 #if USE_OS_TZDB
 #  include <dirent.h>
 #endif
@@ -118,7 +118,7 @@
 #include <vector>
 #include <sys/stat.h>
 
-// unistd.h is used on some platforms as part of the means to get
+// unistd.h is used on some platforms as part of the the means to get
 // the current time zone. On Win32 windows.h provides a means to do it.
 // gcc/mingw supports unistd.h on Win32 but MSVC does not.
 
@@ -384,8 +384,7 @@ class file_streambuf
 };
 
 #endif  // !USE_OS_TZDB
-namespace arrow_vendored
-{
+
 namespace date
 {
 // +---------------------+
@@ -2709,7 +2708,8 @@ operator<<(std::ostream& os, const time_zone& z)
         os.width(8);
         os << s.format_ << "   ";
         os << s.until_year_ << ' ' << s.until_date_;
-        os << "   " << s.until_utc_ << " UTC";
+        os << "   ";
+        date::operator<<(os, s.until_utc_) << " UTC";
         os << "   " << s.until_std_ << " STD";
         os << "   " << s.until_loc_;
         os << "   " << make_time(s.initial_save_);
@@ -2734,8 +2734,7 @@ operator<<(std::ostream& os, const time_zone& z)
 std::ostream&
 operator<<(std::ostream& os, const leap_second& x)
 {
-    using namespace date;
-    return os << x.date_ << "  +";
+    return date::operator<<(os, x.date_) << "  +";
 }
 
 #if USE_OS_TZDB
@@ -3716,13 +3715,78 @@ get_tzdb()
     return get_tzdb_list().front();
 }
 
+namespace {
+
+class recursion_limiter
+{
+    unsigned depth_ = 0;
+    unsigned limit_;
+
+    class restore_recursion_depth;
+
+public:
+    recursion_limiter(recursion_limiter const&) = delete;
+    recursion_limiter& operator=(recursion_limiter const&) = delete;
+
+    explicit constexpr recursion_limiter(unsigned limit) noexcept;
+
+    restore_recursion_depth count();
+};
+
+class recursion_limiter::restore_recursion_depth
+{
+    recursion_limiter* rc_;
+
+public:
+    ~restore_recursion_depth();
+    restore_recursion_depth(restore_recursion_depth&&) = default;
+
+    explicit restore_recursion_depth(recursion_limiter* rc) noexcept;
+};
+
+inline
+recursion_limiter::restore_recursion_depth::~restore_recursion_depth()
+{
+    --(rc_->depth_);
+}
+
+inline
+recursion_limiter::restore_recursion_depth::restore_recursion_depth(recursion_limiter* rc)
+                                                                                  noexcept
+    : rc_{rc}
+{}
+
+inline
+constexpr
+recursion_limiter::recursion_limiter(unsigned limit) noexcept
+    : limit_{limit}
+{
+}
+
+inline
+recursion_limiter::restore_recursion_depth
+recursion_limiter::count()
+{
+    ++depth_;
+    if (depth_ > limit_)
+        throw std::runtime_error("recursion limit of " +
+                                  std::to_string(limit_) + " exceeded");
+    return restore_recursion_depth{this};
+}
+
+}  // unnamed namespace
+
 const time_zone*
 #if HAS_STRING_VIEW
 tzdb::locate_zone(std::string_view tz_name) const
 #else
 tzdb::locate_zone(const std::string& tz_name) const
 #endif
 {
+    // If a link-to-link chain exceeds this limit, give up
+    thread_local recursion_limiter rc{10};
+    auto restore_count = rc.count();
+
     auto zi = std::lower_bound(zones.begin(), zones.end(), tz_name,
 #if HAS_STRING_VIEW
         [](const time_zone& z, const std::string_view& nm)
@@ -3746,13 +3810,7 @@ tzdb::locate_zone(const std::string& tz_name) const
         });
         if (li != links.end() && li->name() == tz_name)
         {
-            zi = std::lower_bound(zones.begin(), zones.end(), li->target(),
-                [](const time_zone& z, const std::string& nm)
-                {
-                    return z.name() < nm;
-                });
-            if (zi != zones.end() && zi->name() == li->target())
-                return &*zi;
+            return locate_zone(li->target());
         }
 #endif  // !USE_OS_TZDB
         throw std::runtime_error(std::string(tz_name) + " not found in timezone database");
@@ -4038,6 +4096,18 @@ tzdb::current_zone() const
         if (!result.empty())
             return locate_zone(result);
 #endif
+    // Fall through to try other means.
+    }
+    {
+    // On Android, it is not possible to use file based approach either,
+    // we have to ask the value of `persist.sys.timezone` system property
+#if defined(ANDROID) || defined(__ANDROID__)
+    char sys_timezone[PROP_VALUE_MAX];
+    if (__system_property_get("persist.sys.timezone", sys_timezone) > 0)
+    {
+        return locate_zone(sys_timezone);
+    }
+#endif // defined(ANDROID) || defined(__ANDROID__)
     // Fall through to try other means.
     }
     {
@@ -4072,7 +4142,6 @@ current_zone()
 }
 
 }  // namespace date
-}  // namespace arrow_vendored
 
 #if defined(__GNUC__) && __GNUC__ < 5
 # pragma GCC diagnostic pop

diff --git a/cpp/src/arrow/vendored/datetime/tz.h b/cpp/src/arrow/vendored/datetime/tz.h
@@ -43,12 +43,6 @@
 // required. On Windows, the names are never "Standard" so mapping is always required.
 // Technically any OS may use the mapping process but currently only Windows does use it.
 
-// NOTE(ARROW): If this is not set, then the library will attempt to
-// use libcurl to obtain a timezone database, and we probably do not want this.
-#ifndef _WIN32
-#define USE_OS_TZDB 1
-#endif
-
 #ifndef USE_OS_TZDB
 #  define USE_OS_TZDB 0
 #endif
@@ -140,8 +134,6 @@ static_assert(HAS_REMOTE_API == 0 ? AUTO_DOWNLOAD == 0 : true,
 #  endif
 #endif
 
-namespace arrow_vendored
-{
 namespace date
 {
 
@@ -239,8 +231,8 @@ nonexistent_local_time::make_msg(local_time<Duration> tp, const local_info& i)
        << i.first.abbrev << " and\n"
        << local_seconds{i.second.begin.time_since_epoch()} + i.second.offset << ' '
        << i.second.abbrev
-       << " which are both equivalent to\n"
-       << i.first.end << " UTC";
+       << " which are both equivalent to\n";
+    date::operator<<(os, i.first.end) << " UTC";
     return os.str();
 }
 
@@ -2796,6 +2788,5 @@ to_gps_time(const tai_time<Duration>& t)
 }
 
 }  // namespace date
-}  // namespace arrow_vendored
 
 #endif  // TZ_H