Add safe variant of std::isspace

Author: trflynn89

fly/types/string/detail/string_classifier.hpp

@@ -143,6 +143,20 @@ class BasicStringClassifier
      */
     static constexpr bool is_x_digit(char_type ch);
 
+    /**
+     * Checks if the given character is a whitespace character as classified by the default C
+     * locale.
+     *
+     * The STL's std::isspace and std::iswspace require that the provided character fits into an
+     * unsigned char and unsigned wchar_t, respectively. Other values result in undefined behavior.
+     * This method has no such restriction.
+     *
+     * @param ch The character to classify.
+     *
+     * @return True if the character is a whitespace character.
+     */
+    static constexpr bool is_space(char_type ch);
+
 private:
     /**
      * Remove the 0x20 bit from the given character, effectively converting the a-z range of
@@ -161,6 +175,12 @@ class BasicStringClassifier
     static constexpr const auto s_upper_f = FLY_CHR(char_type, 'F');
     static constexpr const auto s_lower_a = FLY_CHR(char_type, 'a');
     static constexpr const auto s_lower_z = FLY_CHR(char_type, 'z');
+    static constexpr const auto s_space = FLY_CHR(char_type, ' ');
+    static constexpr const auto s_form_feed = FLY_CHR(char_type, '\f');
+    static constexpr const auto s_line_feed = FLY_CHR(char_type, '\n');
+    static constexpr const auto s_carriage_return = FLY_CHR(char_type, '\r');
+    static constexpr const auto s_horizontal_tab = FLY_CHR(char_type, '\t');
+    static constexpr const auto s_vertical_tab = FLY_CHR(char_type, '\v');
 
     static constexpr const auto s_case_bit = static_cast<int_type>(0x20);
     static constexpr const auto s_case_mask = static_cast<int_type>(~s_case_bit);
@@ -251,6 +271,14 @@ constexpr inline bool BasicStringClassifier<StringType>::is_x_digit(char_type ch
     return is_digit(ch) || ((alpha >= s_upper_a) && (alpha <= s_upper_f));
 }
 
+//==================================================================================================
+template <typename StringType>
+constexpr inline bool BasicStringClassifier<StringType>::is_space(char_type ch)
+{
+    return (ch == s_space) || (ch == s_form_feed) || (ch == s_line_feed) ||
+        (ch == s_carriage_return) || (ch == s_horizontal_tab) || (ch == s_vertical_tab);
+}
+
 //==================================================================================================
 template <typename StringType>
 constexpr inline auto BasicStringClassifier<StringType>::unify_az_characters(char_type ch)

fly/types/string/string.hpp

@@ -175,6 +175,20 @@ class BasicString
      */
     static constexpr bool is_x_digit(char_type ch);
 
+    /**
+     * Checks if the given character is a whitespace character as classified by the default C
+     * locale.
+     *
+     * The STL's std::isspace and std::iswspace require that the provided character fits into an
+     * unsigned char and unsigned wchar_t, respectively. Other values result in undefined behavior.
+     * This method has no such restriction.
+     *
+     * @param ch The character to classify.
+     *
+     * @return True if the character is a whitespace character.
+     */
+    static constexpr bool is_space(char_type ch);
+
     /**
      * Split a string into a vector of strings.
      *
@@ -547,6 +561,13 @@ constexpr inline bool BasicString<StringType>::is_x_digit(char_type ch)
     return classifier::is_x_digit(ch);
 }
 
+//==================================================================================================
+template <typename StringType>
+constexpr inline bool BasicString<StringType>::is_space(char_type ch)
+{
+    return classifier::is_space(ch);
+}
+
 //==================================================================================================
 template <typename StringType>
 std::vector<StringType> BasicString<StringType>::split(view_type input, char_type delimiter)
@@ -600,9 +621,9 @@ BasicString<StringType>::split(view_type input, char_type delimiter, size_type c
 template <typename StringType>
 void BasicString<StringType>::trim(StringType &target)
 {
-    auto is_non_space = [](int ch)
+    auto is_non_space = [](auto ch)
     {
-        return !std::isspace(ch);
+        return !is_space(ch);
     };
 
     // Remove leading whitespace.

test/types/string/string_classifier.cpp

@@ -197,4 +197,28 @@ CATCH_TEMPLATE_TEST_CASE(
             }
         }
     }
+
+    CATCH_SECTION("Check if a character is a whitespace character")
+    {
+        for (char_type ch = 0; (ch >= 0) && (ch < 0x80); ++ch)
+        {
+            CATCH_CHECK(
+                BasicString::is_space(ch) ==
+                static_cast<bool>(std::isspace(static_cast<unsigned char>(ch))));
+        }
+
+        if constexpr (sizeof(char_type) > 1)
+        {
+            // Spot check some values that incorrectly result in std::isspace returning true when
+            // cast to unsigned char (which is how the spec suggests to avoid undefined behavior).
+            CATCH_CHECK(std::isspace(static_cast<unsigned char>(0xaa20)));
+            CATCH_CHECK_FALSE(BasicString::is_space(0xaa20));
+
+            CATCH_CHECK(std::isspace(static_cast<unsigned char>(0xaa0a)));
+            CATCH_CHECK_FALSE(BasicString::is_space(0xaa0a));
+
+            CATCH_CHECK(std::isspace(static_cast<unsigned char>(0xaa09)));
+            CATCH_CHECK_FALSE(BasicString::is_space(0xaa09));
+        }
+    }
 }