pest_tmp/unicode/
mod.rs

1//! Character inclusion in binary or General_Category value Unicode sets.
2//!
3//! We rely on dead code elimination to remove the tables that aren't needed.
4
5#![allow(bad_style)]
6#![allow(clippy::all)]
7
8macro_rules! char_property_functions {
9    {$(
10        mod $module:ident;
11        [$(
12            $prop:ident,
13        )*];
14    )*} => {$(
15        #[allow(unused)]
16        mod $module;
17        $(pub fn $prop(c: char) -> bool {
18            self::$module::$prop.contains_char(c)
19        })*
20    )*};
21}
22
23char_property_functions! {
24    mod binary;
25    [
26        // ASCII_HEX_DIGIT, // let this one be stripped out -- the full trie is wasteful for ASCII
27        ALPHABETIC, BIDI_CONTROL, CASE_IGNORABLE, CASED, CHANGES_WHEN_CASEFOLDED,
28        CHANGES_WHEN_CASEMAPPED, CHANGES_WHEN_LOWERCASED, CHANGES_WHEN_TITLECASED,
29        CHANGES_WHEN_UPPERCASED, DASH, DEFAULT_IGNORABLE_CODE_POINT, DEPRECATED, DIACRITIC,
30        EXTENDER, GRAPHEME_BASE, GRAPHEME_EXTEND, GRAPHEME_LINK, HEX_DIGIT, HYPHEN,
31        IDS_BINARY_OPERATOR, IDS_TRINARY_OPERATOR, ID_CONTINUE, ID_START, IDEOGRAPHIC, JOIN_CONTROL,
32        LOGICAL_ORDER_EXCEPTION, LOWERCASE, MATH, NONCHARACTER_CODE_POINT, OTHER_ALPHABETIC,
33        OTHER_DEFAULT_IGNORABLE_CODE_POINT, OTHER_GRAPHEME_EXTEND, OTHER_ID_CONTINUE,
34        OTHER_ID_START, OTHER_LOWERCASE, OTHER_MATH, OTHER_UPPERCASE, PATTERN_SYNTAX,
35        PATTERN_WHITE_SPACE, PREPENDED_CONCATENATION_MARK, QUOTATION_MARK, RADICAL,
36        REGIONAL_INDICATOR, SENTENCE_TERMINAL, SOFT_DOTTED, TERMINAL_PUNCTUATION, UNIFIED_IDEOGRAPH,
37        UPPERCASE, VARIATION_SELECTOR, WHITE_SPACE, XID_CONTINUE, XID_START,
38    ];
39
40    mod category;
41    [
42        CASED_LETTER, CLOSE_PUNCTUATION, CONNECTOR_PUNCTUATION, CONTROL, CURRENCY_SYMBOL,
43        DASH_PUNCTUATION, DECIMAL_NUMBER, ENCLOSING_MARK, FINAL_PUNCTUATION, FORMAT,
44        INITIAL_PUNCTUATION, LETTER, LETTER_NUMBER, LINE_SEPARATOR, LOWERCASE_LETTER, MARK,
45        MATH_SYMBOL, MODIFIER_LETTER, MODIFIER_SYMBOL, NONSPACING_MARK, NUMBER, OPEN_PUNCTUATION,
46        OTHER, OTHER_LETTER, OTHER_NUMBER, OTHER_PUNCTUATION, OTHER_SYMBOL, PARAGRAPH_SEPARATOR,
47        PRIVATE_USE, PUNCTUATION, SEPARATOR, SPACE_SEPARATOR, SPACING_MARK, SURROGATE, SYMBOL,
48        TITLECASE_LETTER, UNASSIGNED, UPPERCASE_LETTER,
49    ];
50}
51
52pub fn by_name(name: &str) -> Option<Box<Fn(char) -> bool>> {
53    for property in binary::BY_NAME {
54        if name == property.0.to_uppercase() {
55            return Some(Box::new(move |c| property.1.contains_char(c)));
56        }
57    }
58
59    for property in category::BY_NAME {
60        if name == property.0.to_uppercase() {
61            return Some(Box::new(move |c| property.1.contains_char(c)));
62        }
63    }
64
65    None
66}