ucd_parse/
prop_list.rs

1use std::path::Path;
2
3use crate::{
4    common::{
5        parse_codepoint_association, CodepointIter, Codepoints, UcdFile,
6        UcdFileByCodepoint,
7    },
8    error::Error,
9};
10
11/// A single row in the `PropList.txt` file.
12///
13/// The `PropList.txt` file is the source of truth on several Unicode
14/// properties.
15#[derive(Clone, Debug, Default, Eq, PartialEq)]
16pub struct Property {
17    /// The codepoint or codepoint range for this entry.
18    pub codepoints: Codepoints,
19    /// The property name assigned to the codepoints in this entry.
20    pub property: String,
21}
22
23impl UcdFile for Property {
24    fn relative_file_path() -> &'static Path {
25        Path::new("PropList.txt")
26    }
27}
28
29impl UcdFileByCodepoint for Property {
30    fn codepoints(&self) -> CodepointIter {
31        self.codepoints.into_iter()
32    }
33}
34
35impl std::str::FromStr for Property {
36    type Err = Error;
37
38    fn from_str(line: &str) -> Result<Property, Error> {
39        let (codepoints, property) = parse_codepoint_association(line)?;
40        Ok(Property { codepoints, property: property.to_string() })
41    }
42}
43
44#[cfg(test)]
45mod tests {
46    use super::Property;
47
48    #[test]
49    fn parse_single() {
50        let line =
51            "061C          ; Bidi_Control # Cf       ARABIC LETTER MARK\n";
52        let row: Property = line.parse().unwrap();
53        assert_eq!(row.codepoints, 0x061C);
54        assert_eq!(row.property, "Bidi_Control");
55    }
56
57    #[test]
58    fn parse_range() {
59        let line = "0009..000D    ; White_Space # Cc   [5] <control-0009>..<control-000D>\n";
60        let row: Property = line.parse().unwrap();
61        assert_eq!(row.codepoints, (0x0009, 0x000D));
62        assert_eq!(row.property, "White_Space");
63    }
64}
ucd_parse/prop_list.rs

ucd_parse/
prop_list.rs
