1 | use std::path::Path; |
2 | |
3 | use 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/DerivedCombiningClass.txt` file. |
12 | /// |
13 | /// This file gives the derived values of the Decomposition_Type |
14 | /// property. |
15 | #[derive (Clone, Debug, Default, Eq, PartialEq)] |
16 | pub struct DerivedDecompositionType { |
17 | /// The codepoint or codepoint range for this entry. |
18 | pub codepoints: Codepoints, |
19 | /// The derived Decomposition_Type of the codepoints in this entry. |
20 | pub decomposition_type: String, |
21 | } |
22 | |
23 | impl UcdFile for DerivedDecompositionType { |
24 | fn relative_file_path() -> &'static Path { |
25 | Path::new("extracted/DerivedDecompositionType.txt" ) |
26 | } |
27 | } |
28 | |
29 | impl UcdFileByCodepoint for DerivedDecompositionType { |
30 | fn codepoints(&self) -> CodepointIter { |
31 | self.codepoints.into_iter() |
32 | } |
33 | } |
34 | |
35 | impl std::str::FromStr for DerivedDecompositionType { |
36 | type Err = Error; |
37 | |
38 | fn from_str(line: &str) -> Result<DerivedDecompositionType, Error> { |
39 | let (codepoints: Codepoints, decomposition_type: &str) = |
40 | parse_codepoint_association(line)?; |
41 | Ok(DerivedDecompositionType { |
42 | codepoints, |
43 | decomposition_type: decomposition_type.to_string(), |
44 | }) |
45 | } |
46 | } |
47 | |
48 | #[cfg (test)] |
49 | mod tests { |
50 | use super::DerivedDecompositionType; |
51 | |
52 | #[test ] |
53 | fn parse_single() { |
54 | let line = "00A0 ; Nobreak # Zs NO-BREAK SPACE \n" ; |
55 | let row: DerivedDecompositionType = line.parse().unwrap(); |
56 | assert_eq!(row.codepoints, 0x00A0); |
57 | assert_eq!(row.decomposition_type, "Nobreak" ); |
58 | } |
59 | |
60 | #[test ] |
61 | fn parse_range() { |
62 | let line = "3070..3071 ; Canonical # Lo [2] HIRAGANA LETTER BA..HIRAGANA LETTER PA \n" ; |
63 | let row: DerivedDecompositionType = line.parse().unwrap(); |
64 | assert_eq!(row.codepoints, (0x3070, 0x3071)); |
65 | assert_eq!(row.decomposition_type, "Canonical" ); |
66 | } |
67 | } |
68 | |