yeslogic_ucd_parse/extracted/
derived_joining_type.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 `extracted/DerivedJoiningType.txt` file.
12///
13/// This file gives the derived values of the Joining_Type property.
14#[derive(Clone, Debug, Default, Eq, PartialEq)]
15pub struct DerivedJoiningType {
16    /// The codepoint or codepoint range for this entry.
17    pub codepoints: Codepoints,
18    /// The derived Joining_Type of the codepoints in this entry.
19    pub joining_type: String,
20}
21
22impl UcdFile for DerivedJoiningType {
23    fn relative_file_path() -> &'static Path {
24        Path::new("extracted/DerivedJoiningType.txt")
25    }
26}
27
28impl UcdFileByCodepoint for DerivedJoiningType {
29    fn codepoints(&self) -> CodepointIter {
30        self.codepoints.into_iter()
31    }
32}
33
34impl std::str::FromStr for DerivedJoiningType {
35    type Err = Error;
36
37    fn from_str(line: &str) -> Result<DerivedJoiningType, Error> {
38        let (codepoints, joining_type) = parse_codepoint_association(line)?;
39        Ok(DerivedJoiningType {
40            codepoints,
41            joining_type: joining_type.to_string(),
42        })
43    }
44}
45
46#[cfg(test)]
47mod tests {
48    use super::DerivedJoiningType;
49
50    #[test]
51    fn parse_single() {
52        let line = "0628          ; D # Lo       ARABIC LETTER BEH\n";
53        let row: DerivedJoiningType = line.parse().unwrap();
54        assert_eq!(row.codepoints, 0x0628);
55        assert_eq!(row.joining_type, "D");
56    }
57
58    #[test]
59    fn parse_range() {
60        let line =  "1133B..1133C  ; T # Mn   [2] COMBINING BINDU BELOW..GRANTHA SIGN NUKTA\n";
61        let row: DerivedJoiningType = line.parse().unwrap();
62        assert_eq!(row.codepoints, (0x1133B, 0x1133C));
63        assert_eq!(row.joining_type, "T");
64    }
65}