New character sets

1 files changed, 142 insertions, 16 deletions

diff --git a/src/csets.rs b/src/csets.rs
index 3ee5e53..f1b1227 100644
--- a/src/csets.rs
+++ b/src/csets.rs
@@ -134,6 +134,76 @@ impl CharacterSet for AnyCharacter {
 	}
 }
 
+/// Contains all Unicode alphabetic characters
+#[derive(Debug, Clone, Copy)]
+pub struct Alphabetic;
+
+impl CharacterSet for Alphabetic {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_alphabetic()
+	}
+}
+
+/// Contains all Unicode lowercase characters
+#[derive(Debug, Clone, Copy)]
+pub struct Lowercase;
+
+impl CharacterSet for Lowercase {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_lowercase()
+	}
+}
+
+/// Contains all Unicode uppercase characters
+#[derive(Debug, Clone, Copy)]
+pub struct Uppercase;
+
+impl CharacterSet for Uppercase {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_uppercase()
+	}
+}
+
+/// Contains all Unicode whitespace characters
+#[derive(Debug, Clone, Copy)]
+pub struct Whitespace;
+
+impl CharacterSet for Whitespace {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_whitespace()
+	}
+}
+
+/// Contains all Unicode alphabetic or numeric characters
+#[derive(Debug, Clone, Copy)]
+pub struct Alphanumeric;
+
+impl CharacterSet for Alphanumeric {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_alphanumeric()
+	}
+}
+
+/// Contains all Unicode control chartacters
+#[derive(Debug, Clone, Copy)]
+pub struct Control;
+
+impl CharacterSet for Control {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_control()
+	}
+}
+
+/// Contains all Unicode numeric characters
+#[derive(Debug, Clone, Copy)]
+pub struct Numeric;
+
+impl CharacterSet for Numeric {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_numeric()
+	}
+}
+
 /// Contains all ASCII characters
 #[derive(Debug, Clone, Copy)]
 pub struct Ascii;
@@ -144,23 +214,23 @@ impl CharacterSet for Ascii {
 	}
 }
 
-/// Contains the ASCII digits, 0-9
+/// Contains all ASCII letters: a-z, A-Z
 #[derive(Debug, Clone, Copy)]
-pub struct AsciiDigits;
+pub struct AsciiAlphabetic;
 
-impl CharacterSet for AsciiDigits {
+impl CharacterSet for AsciiAlphabetic {
 	fn contains(&self, ch: char) -> bool {
-		ch.is_ascii_digit()
+		ch.is_ascii_alphabetic()
 	}
 }
 
-/// Contains the ASCII whitespace characters
+/// Contains all uppercase ASCII letters, A-Z
 #[derive(Debug, Clone, Copy)]
-pub struct AsciiWhitespace;
+pub struct AsciiUppercase;
 
-impl CharacterSet for AsciiWhitespace {
+impl CharacterSet for AsciiUppercase {
 	fn contains(&self, ch: char) -> bool {
-		ch.is_ascii_whitespace()
+		ch.is_ascii_uppercase()
 	}
 }
 
@@ -174,23 +244,79 @@ impl CharacterSet for AsciiLowercase {
 	}
 }
 
-/// Contains all uppercase ASCII letters, A-Z
+/// Contains all ASCII characters a-z, A-Z, 0-9
 #[derive(Debug, Clone, Copy)]
-pub struct AsciiUppercase;
+pub struct AsciiAlphanumeric;
 
-impl CharacterSet for AsciiUppercase {
+impl CharacterSet for AsciiAlphanumeric {
 	fn contains(&self, ch: char) -> bool {
-		ch.is_ascii_uppercase()
+		ch.is_ascii_alphanumeric()
 	}
 }
 
-/// Containes all ASCII letters: a-z, A-Z
+/// Contains the ASCII digits, 0-9
 #[derive(Debug, Clone, Copy)]
-pub struct AsciiLetters;
+pub struct AsciiDigits;
 
-impl CharacterSet for AsciiLetters {
+impl CharacterSet for AsciiDigits {
 	fn contains(&self, ch: char) -> bool {
-		ch.is_ascii_alphabetic()
+		ch.is_ascii_digit()
+	}
+}
+
+/// Contains the ASCII hexadecimal digits: 0-9, a-f, A-F
+#[derive(Debug, Clone, Copy)]
+pub struct AsciiHexDigit;
+
+impl CharacterSet for AsciiHexDigit {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_ascii_hexdigit()
+	}
+}
+
+/// Contains ASCII punctuation characters:
+///
+/// - U+0021 ..= U+002F `! " # $ % & ' ( ) * + , - . /`
+/// - U+003A ..= U+0040 `: ; < = > ? @`
+/// - U+005B ..= U+0060 ``[ \ ] ^ _ ` ``
+/// - U+007B ..= U+007E `{ | } ~`
+#[derive(Debug, Clone, Copy)]
+pub struct AsciiPunctuation;
+
+impl CharacterSet for AsciiPunctuation {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_ascii_punctuation()
+	}
+}
+
+/// Contains the ASCII graphic characters: `U+0021 (!) ..= U+007E (~)`
+#[derive(Debug, Clone, Copy)]
+pub struct AsciiGraphic;
+
+impl CharacterSet for AsciiGraphic {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_ascii_graphic()
+	}
+}
+
+/// Contains the ASCII whitespace characters
+#[derive(Debug, Clone, Copy)]
+pub struct AsciiWhitespace;
+
+impl CharacterSet for AsciiWhitespace {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_ascii_whitespace()
+	}
+}
+
+/// Contains the ASCII control characters:
+/// `U+0000 (NUL) ..= U+001F (UNIT SEPARATOR)` or `U+007F (DELETE)`
+#[derive(Debug, Clone, Copy)]
+pub struct AsciiControl;
+
+impl CharacterSet for AsciiControl {
+	fn contains(&self, ch: char) -> bool {
+		ch.is_ascii_control()
 	}
 }