time: Add is_datetime() for RFC 3339 Internet Date/Time Format

src/core/time/CMakeLists.txt

@@ -1,4 +1,5 @@
-sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME time SOURCES gmt.cc)
+sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME time
+  SOURCES gmt.cc datetime.cc)
 
 if(SOURCEMETA_CORE_INSTALL)
   sourcemeta_library_install(NAMESPACE sourcemeta PROJECT core NAME time)

src/core/time/datetime.cc

@@ -0,0 +1,208 @@
+#include <sourcemeta/core/time.h>
+
+#include <array> // std::array
+
+namespace sourcemeta::core {
+
+static constexpr auto is_digit(const char character) -> bool {
+  return character >= '0' && character <= '9';
+}
+
+static constexpr auto is_leap_year(const unsigned int year) -> bool {
+  return (year % 4 == 0) && (year % 100 != 0 || year % 400 == 0);
+}
+
+static constexpr auto max_day_in_month(const unsigned int month,
+                                       const unsigned int year)
+    -> unsigned int {
+  constexpr std::array<unsigned int, 13> days{
+      {0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31}};
+  if (month == 2 && is_leap_year(year)) {
+    return 29;
+  }
+  return days[month];
+}
+
+auto is_datetime(const std::string_view value) -> bool {
+  const auto size{value.size()};
+
+  // Minimum valid date-time: "YYYY-MM-DDTHH:MM:SSZ" = 20 characters
+  if (size < 20) {
+    return false;
+  }
+
+  std::string_view::size_type position{0};
+
+  // --- full-date: date-fullyear "-" date-month "-" date-mday ---
+
+  // date-fullyear = 4DIGIT
+  if (!is_digit(value[0]) || !is_digit(value[1]) || !is_digit(value[2]) ||
+      !is_digit(value[3])) {
+    return false;
+  }
+  const auto year{static_cast<unsigned int>(value[0] - '0') * 1000 +
+                  static_cast<unsigned int>(value[1] - '0') * 100 +
+                  static_cast<unsigned int>(value[2] - '0') * 10 +
+                  static_cast<unsigned int>(value[3] - '0')};
+  position = 4;
+
+  // "-"
+  if (value[position] != '-') {
+    return false;
+  }
+  position += 1;
+
+  // date-month = 2DIGIT ; 01-12
+  if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+    return false;
+  }
+  const auto month{static_cast<unsigned int>(value[position] - '0') * 10 +
+                   static_cast<unsigned int>(value[position + 1] - '0')};
+  if (month < 1 || month > 12) {
+    return false;
+  }
+  position += 2;
+
+  // "-"
+  if (value[position] != '-') {
+    return false;
+  }
+  position += 1;
+
+  // date-mday = 2DIGIT ; 01-28/29/30/31 based on month/year
+  if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+    return false;
+  }
+  const auto day{static_cast<unsigned int>(value[position] - '0') * 10 +
+                 static_cast<unsigned int>(value[position + 1] - '0')};
+  position += 2;
+
+  // --- "T" or "t" separator ---
+  if (value[position] != 'T' && value[position] != 't') {
+    return false;
+  }
+  position += 1;
+
+  // --- partial-time: time-hour ":" time-minute ":" time-second ---
+
+  // time-hour = 2DIGIT ; 00-23
+  if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+    return false;
+  }
+  const auto hour{static_cast<unsigned int>(value[position] - '0') * 10 +
+                  static_cast<unsigned int>(value[position + 1] - '0')};
+  if (hour > 23) {
+    return false;
+  }
+  position += 2;
+
+  // ":"
+  if (value[position] != ':') {
+    return false;
+  }
+  position += 1;
+
+  // time-minute = 2DIGIT ; 00-59
+  if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+    return false;
+  }
+  const auto minute{static_cast<unsigned int>(value[position] - '0') * 10 +
+                    static_cast<unsigned int>(value[position + 1] - '0')};
+  if (minute > 59) {
+    return false;
+  }
+  position += 2;
+
+  // ":"
+  if (value[position] != ':') {
+    return false;
+  }
+  position += 1;
+
+  // time-second = 2DIGIT ; 00-60 (60 = leap second per §5.7)
+  if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+    return false;
+  }
+  const auto second{static_cast<unsigned int>(value[position] - '0') * 10 +
+                    static_cast<unsigned int>(value[position + 1] - '0')};
+  if (second > 60) {
+    return false;
+  }
+  position += 2;
+
+  // --- [time-secfrac] = "." 1*DIGIT ---
+  if (position < size && value[position] == '.') {
+    position += 1;
+    if (position >= size || !is_digit(value[position])) {
+      // "." must be followed by at least 1 digit
+      return false;
+    }
+    while (position < size && is_digit(value[position])) {
+      position += 1;
+    }
+  }
+
+  // --- time-offset = "Z" / time-numoffset ---
+  if (position >= size) {
+    // No time offset present — invalid
+    return false;
+  }
+
+  if (value[position] == 'Z' || value[position] == 'z') {
+    position += 1;
+  } else if (value[position] == '+' || value[position] == '-') {
+    position += 1;
+
+    // time-numoffset = ("+" / "-") time-hour ":" time-minute
+    if (position + 5 > size) {
+      return false;
+    }
+
+    // Offset time-hour = 2DIGIT ; 00-23
+    if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+      return false;
+    }
+    const auto offset_hour{
+        static_cast<unsigned int>(value[position] - '0') * 10 +
+        static_cast<unsigned int>(value[position + 1] - '0')};
+    if (offset_hour > 23) {
+      return false;
+    }
+    position += 2;
+
+    // ":" — REQUIRED (colonless offsets like +0530 are invalid per §5.6)
+    if (value[position] != ':') {
+      return false;
+    }
+    position += 1;
+
+    // Offset time-minute = 2DIGIT ; 00-59
+    if (!is_digit(value[position]) || !is_digit(value[position + 1])) {
+      return false;
+    }
+    const auto offset_minute{
+        static_cast<unsigned int>(value[position] - '0') * 10 +
+        static_cast<unsigned int>(value[position + 1] - '0')};
+    if (offset_minute > 59) {
+      return false;
+    }
+    position += 2;
+  } else {
+    // Not Z, not +/-, invalid character for time-offset
+    return false;
+  }
+
+  // String must be fully consumed — no trailing characters
+  if (position != size) {
+    return false;
+  }
+
+  // --- Validate date-mday against month/year (§5.7) ---
+  if (day < 1 || day > max_day_in_month(month, year)) {
+    return false;
+  }
+
+  return true;
+}
+
+} // namespace sourcemeta::core

src/core/time/include/sourcemeta/core/time.h

@@ -5,12 +5,13 @@
 #include <sourcemeta/core/time_export.h>
 #endif
 
-#include <chrono> // std::chrono::system_clock::time_point
-#include <string> // std::string
+#include <chrono>      // std::chrono::system_clock::time_point
+#include <string>      // std::string
+#include <string_view> // std::string_view
 
 /// @defgroup time Time
-/// @brief A growing implementation of time-related utilities for standard such
-/// as RFC 7231 (GMT).
+/// @brief A growing implementation of time-related utilities for standards
+/// such as RFC 7231 (GMT) and RFC 3339 (Internet Date/Time Format).
 ///
 /// This functionality is included as follows:
 ///
@@ -79,6 +80,31 @@ auto to_gmt(const std::chrono::system_clock::time_point time) -> std::string;
 SOURCEMETA_CORE_TIME_EXPORT
 auto from_gmt(const std::string &time) -> std::chrono::system_clock::time_point;
 
+/// @ingroup time
+/// Check whether the given string is a valid date-time value per RFC 3339
+/// Section 5.6 (Internet Date/Time Format). This implements the full
+/// `date-time` production rule:
+///
+/// ```
+/// date-time = full-date "T" full-time
+/// ```
+///
+/// where "T" may also be lowercase "t" (per RFC 3339 §5.6 NOTE). For example:
+///
+/// ```cpp
+/// #include <sourcemeta/core/time.h>
+///
+/// #include <cassert>
+///
+/// assert(sourcemeta::core::is_datetime("1985-04-12T23:20:50.52Z"));
+/// assert(sourcemeta::core::is_datetime("1996-12-19T16:39:57-08:00"));
+/// assert(sourcemeta::core::is_datetime("1990-12-31T23:59:60Z"));
+/// assert(!sourcemeta::core::is_datetime("2024-01-15T14:30:00"));
+/// assert(!sourcemeta::core::is_datetime("2024-01-15 14:30:00Z"));
+/// ```
+SOURCEMETA_CORE_TIME_EXPORT
+auto is_datetime(std::string_view value) -> bool;
+
 } // namespace sourcemeta::core
 
 #endif

test/time/CMakeLists.txt

@@ -1,5 +1,5 @@
 sourcemeta_googletest(NAMESPACE sourcemeta PROJECT core NAME time
-  SOURCES gmt_test.cc)
+  SOURCES gmt_test.cc datetime_test.cc)
 
 target_link_libraries(sourcemeta_core_time_unit
   PRIVATE sourcemeta::core::time)