Issue #2259 - Add missing IsAscii* helper functions in mozilla/TextUtils.h

mfbt/TextUtils.h

@@ -35,6 +35,86 @@ public:
 
 } // namespace detail
 
+// The overloads below are not templated in order to make
+// implicit conversions to span work as expected for the Span
+// overloads.
+
+/** Returns true iff |aChar| is ASCII, i.e. in the range [0, 0x80). */
+inline constexpr bool IsAscii(unsigned char aChar) { return aChar < 0x80; }
+
+/** Returns true iff |aChar| is ASCII, i.e. in the range [0, 0x80). */
+inline constexpr bool IsAscii(signed char aChar) {
+  return IsAscii(static_cast<unsigned char>(aChar));
+}
+
+/** Returns true iff |aChar| is ASCII, i.e. in the range [0, 0x80). */
+inline constexpr bool IsAscii(char aChar) {
+  return IsAscii(static_cast<unsigned char>(aChar));
+}
+
+/** Returns true iff |aChar| is ASCII, i.e. in the range [0, 0x80). */
+inline constexpr bool IsAscii(char16_t aChar) { return aChar < 0x80; }
+
+/** Returns true iff |aChar| is ASCII, i.e. in the range [0, 0x80). */
+inline constexpr bool IsAscii(char32_t aChar) { return aChar < 0x80; }
+
+/**
+ * Returns true iff every character in the null-terminated string pointed to by
+ * |aChar| is ASCII, i.e. in the range [0, 0x80).
+ */
+template <typename Char>
+constexpr bool IsAsciiNullTerminated(const Char* aChar) {
+  while (Char c = *aChar++) {
+    if (!IsAscii(c)) {
+      return false;
+    }
+  }
+  return true;
+}
+
+/**
+ * Returns true iff |aChar| matches Ascii Whitespace.
+ *
+ * This function is intended to match the Infra standard
+ * (https://infra.spec.whatwg.org/#ascii-whitespace)
+ */
+template <typename Char>
+constexpr bool IsAsciiWhitespace(Char aChar) {
+  using UnsignedChar = typename detail::MakeUnsignedChar<Char>::Type;
+  auto uc = static_cast<UnsignedChar>(aChar);
+  return uc == 0x9 || uc == 0xA || uc == 0xC || uc == 0xD || uc == 0x20;
+}
+
+/**
+ * Returns true iff |aChar| matches [a-z].
+ *
+ * This function is basically what you thought islower was, except its behavior
+ * doesn't depend on the user's current locale.
+ */
+template<typename Char>
+constexpr bool
+IsAsciiLowercaseAlpha(Char aChar)
+{
+  using UnsignedChar = typename detail::MakeUnsignedChar<Char>::Type;
+  auto uc = static_cast<UnsignedChar>(aChar);
+  return 'a' <= uc && uc <= 'z';
+}
+
+/**
+ * Returns true iff |aChar| matches [A-Z].
+ *
+ * This function is basically what you thought isupper was, except its behavior
+ * doesn't depend on the user's current locale.
+ */
+template<typename Char>
+constexpr bool
+IsAsciiUppercaseAlpha(Char aChar)
+{
+  using UnsignedChar = typename detail::MakeUnsignedChar<Char>::Type;
+  auto uc = static_cast<UnsignedChar>(aChar);
+  return 'A' <= uc && uc <= 'Z';
+}
+
 /**
  * Returns true iff |aChar| matches [a-zA-Z].
  *
@@ -44,12 +124,73 @@ public:
 template<typename Char>
 constexpr bool
 IsAsciiAlpha(Char aChar)
+{
+  return IsAsciiLowercaseAlpha(aChar) || IsAsciiUppercaseAlpha(aChar);
+}
+
+/**
+ * Returns true iff |aChar| matches [0-9].
+ *
+ * This function is basically what you thought isdigit was, except its behavior
+ * doesn't depend on the user's current locale.
+ */
+template<typename Char>
+constexpr bool
+IsAsciiDigit(Char aChar)
 {
   using UnsignedChar = typename detail::MakeUnsignedChar<Char>::Type;
-  return ('a' <= static_cast<UnsignedChar>(aChar) &&
+  auto uc = static_cast<UnsignedChar>(aChar);
-          static_cast<UnsignedChar>(aChar) <= 'z') ||
+  return '0' <= uc && uc <= '9';
-         ('A' <= static_cast<UnsignedChar>(aChar) &&
+}
-          static_cast<UnsignedChar>(aChar) <= 'Z');
+
+/**
+ * Returns true iff |aChar| matches [0-9a-fA-F].
+ *
+ * This function is basically isxdigit, but guaranteed to be only for ASCII.
+ */
+template <typename Char>
+constexpr bool IsAsciiHexDigit(Char aChar) {
+  using UnsignedChar = typename detail::MakeUnsignedChar<Char>::Type;
+  auto uc = static_cast<UnsignedChar>(aChar);
+  return ('0' <= uc && uc <= '9') || ('a' <= uc && uc <= 'f') ||
+         ('A' <= uc && uc <= 'F');
+}
+
+/**
+ * Returns true iff |aChar| matches [a-zA-Z0-9].
+ *
+ * This function is basically what you thought isalnum was, except its behavior
+ * doesn't depend on the user's current locale.
+ */
+template <typename Char>
+constexpr bool IsAsciiAlphanumeric(Char aChar) {
+  return IsAsciiDigit(aChar) || IsAsciiAlpha(aChar);
+}
+
+/**
+ * Converts an ASCII alphanumeric digit [0-9a-zA-Z] to number as if in base-36.
+ * (This function therefore works for decimal, hexadecimal, etc.).
+ */
+template <typename Char>
+uint8_t AsciiAlphanumericToNumber(Char aChar) {
+  using UnsignedChar = typename detail::MakeUnsignedChar<Char>::Type;
+  auto uc = static_cast<UnsignedChar>(aChar);
+
+  if ('0' <= uc && uc <= '9') {
+    return uc - '0';
+  }
+
+  if ('A' <= uc && uc <= 'Z') {
+    return uc - 'A' + 10;
+  }
+
+  // Ideally this function would be constexpr, but unfortunately gcc at least as
+  // of 6.4 forbids non-constexpr function calls in unevaluated constexpr
+  // function calls.  See bug 1453456.  So for now, just assert and leave the
+  // entire function non-constexpr.
+  MOZ_ASSERT('a' <= uc && uc <= 'z',
+             "non-ASCII alphanumeric character can't be converted to number");
+  return uc - 'a' + 10;
 }
 
 } // namespace mozilla