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/DerivedGeneralCategory.txt` file.
12///
13/// This file gives the derived values of the General_Category property.
14#[derive(Clone, Debug, Default, Eq, PartialEq)]
15pub struct DerivedGeneralCategory {
16 /// The codepoint or codepoint range for this entry.
17 pub codepoints: Codepoints,
18 /// The derived General_Category of the codepoints in this entry.
19 pub general_category: String,
20}
21
22impl UcdFile for DerivedGeneralCategory {
23 fn relative_file_path() -> &'static Path {
24 Path::new("extracted/DerivedGeneralCategory.txt")
25 }
26}
27
28impl UcdFileByCodepoint for DerivedGeneralCategory {
29 fn codepoints(&self) -> CodepointIter {
30 self.codepoints.into_iter()
31 }
32}
33
34impl std::str::FromStr for DerivedGeneralCategory {
35 type Err = Error;
36
37 fn from_str(line: &str) -> Result<DerivedGeneralCategory, Error> {
38 let (codepoints: Codepoints, general_category: &str) =
39 parse_codepoint_association(line)?;
40 Ok(DerivedGeneralCategory {
41 codepoints,
42 general_category: general_category.to_string(),
43 })
44 }
45}
46
47#[cfg(test)]
48mod tests {
49 use super::DerivedGeneralCategory;
50
51 #[test]
52 fn parse_single() {
53 let line = "04D9 ; Ll # CYRILLIC SMALL LETTER SCHWA\n";
54 let row: DerivedGeneralCategory = line.parse().unwrap();
55 assert_eq!(row.codepoints, 0x04D9);
56 assert_eq!(row.general_category, "Ll");
57 }
58
59 #[test]
60 fn parse_range() {
61 let line = "0660..0669 ; Nd # [10] ARABIC-INDIC DIGIT ZERO..ARABIC-INDIC DIGIT NINE";
62 let row: DerivedGeneralCategory = line.parse().unwrap();
63 assert_eq!(row.codepoints, (0x0660, 0x0669));
64 assert_eq!(row.general_category, "Nd");
65 }
66}
67