From a73292bbf74397e392413ba1a3ac154315d31beb Mon Sep 17 00:00:00 2001
From: Enrico Seiler <enrico.seiler@hotmail.de>
Date: Tue, 14 Nov 2023 16:08:26 +0100
Subject: [PATCH] [REVIEW] char_predicate alternative

---
 .../char_operations/predicate_detail.hpp      | 20 ++++++++++---
 .../char_operations/char_predicate.cpp        | 28 +++++++++----------
 2 files changed, 30 insertions(+), 18 deletions(-)

diff --git a/include/seqan3/utility/char_operations/predicate_detail.hpp b/include/seqan3/utility/char_operations/predicate_detail.hpp
index 02b0140539..8cefdc735b 100644
--- a/include/seqan3/utility/char_operations/predicate_detail.hpp
+++ b/include/seqan3/utility/char_operations/predicate_detail.hpp
@@ -19,6 +19,7 @@
 #include <stdexcept>
 #include <string>
 
+#include <seqan3/utility/concept.hpp>
 #include <seqan3/utility/detail/type_name_as_string.hpp>
 #include <seqan3/utility/type_traits/basic.hpp>
 
@@ -194,10 +195,21 @@ struct char_predicate_base
     constexpr bool operator()(value_t const val) const noexcept
         requires (sizeof(value_t) != 1)
     {
-        // std::char_traits is only defined for char types. libc++ deprecates other specialisations in llvm-17, and
-        // removes them in llvm-18. The int specialisation is needed for seqan3::is_eof(EOF).
-        using char_traits_value_t = std::conditional_t<std::same_as<value_t, int>, char, value_t>;
-        using char_trait = std::char_traits<char_traits_value_t>;
+        // std::char_traits is only guaranteed to be defined for character types.
+        // libc++ deprecates other specialisations in llvm-17, and removes them in llvm-18.
+        // We map the non-character types to corresponding chracter types.
+        // For example, `seqan3::is_eof(EOF)` will call this function with `value_t == int`.
+        // clang-format off
+        using char_value_t = std::conditional_t<seqan3::builtin_character<value_t>, value_t,
+                             std::conditional_t<std::same_as<value_t, std::char_traits<char>::int_type>, char,
+                             std::conditional_t<std::same_as<value_t, std::char_traits<wchar_t>::int_type>, wchar_t,
+                             std::conditional_t<std::same_as<value_t, std::char_traits<char8_t>::int_type>, char8_t,
+                             std::conditional_t<std::same_as<value_t, std::char_traits<char16_t>::int_type>, char16_t,
+                             std::conditional_t<std::same_as<value_t, std::char_traits<char32_t>::int_type>, char32_t,
+                             void>>>>>>;
+        // clang-format on
+        static_assert(!std::same_as<char_value_t, void>, "There is no valid character representation.");
+        using char_trait = std::char_traits<char_value_t>;
         return (static_cast<std::make_unsigned_t<value_t>>(val) < 256) ? operator()(static_cast<uint8_t>(val))
              : (char_trait::eq_int_type(val, char_trait::eof()))       ? derived_t::data[256]
                                                                        : false;
diff --git a/test/snippet/utility/char_operations/char_predicate.cpp b/test/snippet/utility/char_operations/char_predicate.cpp
index 14a6c743a8..92f8c633a2 100644
--- a/test/snippet/utility/char_operations/char_predicate.cpp
+++ b/test/snippet/utility/char_operations/char_predicate.cpp
@@ -4,55 +4,55 @@
 int main()
 {
     //! [is_eof]
-    static_assert(seqan3::is_eof(EOF));  // returns true
-    static_assert(!seqan3::is_eof('C')); // returns false
+    static_assert(seqan3::is_eof(EOF));
+    static_assert(!seqan3::is_eof('C'));
     //! [is_eof]
 
     //! [is_cntrl]
-    static_assert(seqan3::is_cntrl('\0')); // returns true.
+    static_assert(seqan3::is_cntrl('\0'));
     //! [is_cntrl]
 
     //! [is_print]
-    static_assert(seqan3::is_print(' ')); // returns true.
+    static_assert(seqan3::is_print(' '));
     //! [is_print]
 
     //! [is_space]
-    static_assert(seqan3::is_space('\n')); // returns true.
+    static_assert(seqan3::is_space('\n'));
     //! [is_space]
 
     //! [is_blank]
-    static_assert(seqan3::is_blank('\t')); // returns true.
+    static_assert(seqan3::is_blank('\t'));
     //! [is_blank]
 
     //! [is_graph]
-    static_assert(seqan3::is_graph('%')); // returns true.
+    static_assert(seqan3::is_graph('%'));
     //! [is_graph]
 
     //! [is_punct]
-    static_assert(seqan3::is_punct(':')); // returns true.
+    static_assert(seqan3::is_punct(':'));
     //! [is_punct]
 
     //! [is_alnum]
-    static_assert(seqan3::is_alnum('9')); // returns true.
+    static_assert(seqan3::is_alnum('9'));
     //! [is_alnum]
 
     //! [is_alpha]
-    static_assert(seqan3::is_alpha('z')); // returns true.
+    static_assert(seqan3::is_alpha('z'));
     //! [is_alpha]
 
     //! [is_upper]
-    static_assert(seqan3::is_upper('K')); // returns true.
+    static_assert(seqan3::is_upper('K'));
     //! [is_upper]
 
     //! [is_lower]
-    static_assert(seqan3::is_lower('a')); // returns true.
+    static_assert(seqan3::is_lower('a'));
     //! [is_lower]
 
     //! [is_digit]
-    static_assert(seqan3::is_digit('1')); // returns true.
+    static_assert(seqan3::is_digit('1'));
     //! [is_digit]
 
     //! [is_xdigit]
-    static_assert(seqan3::is_xdigit('e')); // returns true.
+    static_assert(seqan3::is_xdigit('e'));
     //! [is_xdigit]
 }