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 `ScriptExtensions.txt` file.
12#[derive(Clone, Debug, Default, Eq, PartialEq)]
13pub struct ScriptExtension {
14 /// The codepoint or codepoint range for this entry.
15 pub codepoints: Codepoints,
16 /// The script extension names assigned to the codepoints in this entry.
17 pub scripts: Vec<String>,
18}
19
20impl UcdFile for ScriptExtension {
21 fn relative_file_path() -> &'static Path {
22 Path::new("ScriptExtensions.txt")
23 }
24}
25
26impl UcdFileByCodepoint for ScriptExtension {
27 fn codepoints(&self) -> CodepointIter {
28 self.codepoints.into_iter()
29 }
30}
31
32impl std::str::FromStr for ScriptExtension {
33 type Err = Error;
34
35 fn from_str(line: &str) -> Result<ScriptExtension, Error> {
36 let (codepoints: Codepoints, scripts: &str) = parse_codepoint_association(line)?;
37 Ok(ScriptExtension {
38 codepoints,
39 scripts: scripts.split_whitespace().map(str::to_string).collect(),
40 })
41 }
42}
43
44#[cfg(test)]
45mod tests {
46 use super::ScriptExtension;
47
48 #[test]
49 fn parse_single() {
50 let line = "060C ; Arab Syrc Thaa # Po ARABIC COMMA\n";
51 let row: ScriptExtension = line.parse().unwrap();
52 assert_eq!(row.codepoints, 0x060C);
53 assert_eq!(row.scripts, vec!["Arab", "Syrc", "Thaa"]);
54 }
55
56 #[test]
57 fn parse_range() {
58 let line = "A836..A837 ; Deva Gujr Guru Kthi Mahj Modi Sind Takr Tirh # So [2] NORTH INDIC QUARTER MARK..NORTH INDIC PLACEHOLDER MARK\n";
59 let row: ScriptExtension = line.parse().unwrap();
60 assert_eq!(row.codepoints, (0xA836, 0xA837));
61 assert_eq!(
62 row.scripts,
63 vec![
64 "Deva", "Gujr", "Guru", "Kthi", "Mahj", "Modi", "Sind",
65 "Takr", "Tirh",
66 ]
67 );
68 }
69}
70