diff options
Diffstat (limited to 'vendor/ucd-parse/src/script_extensions.rs')
-rw-r--r-- | vendor/ucd-parse/src/script_extensions.rs | 68 |
1 files changed, 68 insertions, 0 deletions
diff --git a/vendor/ucd-parse/src/script_extensions.rs b/vendor/ucd-parse/src/script_extensions.rs new file mode 100644 index 000000000..050e1f039 --- /dev/null +++ b/vendor/ucd-parse/src/script_extensions.rs @@ -0,0 +1,68 @@ +use std::path::Path; +use std::str::FromStr; + +use crate::common::{ + parse_codepoint_association, CodepointIter, Codepoints, UcdFile, + UcdFileByCodepoint, +}; +use crate::error::Error; + +/// A single row in the `ScriptExtensions.txt` file. +#[derive(Clone, Debug, Default, Eq, PartialEq)] +pub struct ScriptExtension { + /// The codepoint or codepoint range for this entry. + pub codepoints: Codepoints, + /// The script extension names assigned to the codepoints in this entry. + pub scripts: Vec<String>, +} + +impl UcdFile for ScriptExtension { + fn relative_file_path() -> &'static Path { + Path::new("ScriptExtensions.txt") + } +} + +impl UcdFileByCodepoint for ScriptExtension { + fn codepoints(&self) -> CodepointIter { + self.codepoints.into_iter() + } +} + +impl FromStr for ScriptExtension { + type Err = Error; + + fn from_str(line: &str) -> Result<ScriptExtension, Error> { + let (codepoints, scripts) = parse_codepoint_association(line)?; + Ok(ScriptExtension { + codepoints, + scripts: scripts.split_whitespace().map(str::to_string).collect(), + }) + } +} + +#[cfg(test)] +mod tests { + use super::ScriptExtension; + + #[test] + fn parse_single() { + let line = "060C ; Arab Syrc Thaa # Po ARABIC COMMA\n"; + let row: ScriptExtension = line.parse().unwrap(); + assert_eq!(row.codepoints, 0x060C); + assert_eq!(row.scripts, vec!["Arab", "Syrc", "Thaa"]); + } + + #[test] + fn parse_range() { + let line = "A836..A837 ; Deva Gujr Guru Kthi Mahj Modi Sind Takr Tirh # So [2] NORTH INDIC QUARTER MARK..NORTH INDIC PLACEHOLDER MARK\n"; + let row: ScriptExtension = line.parse().unwrap(); + assert_eq!(row.codepoints, (0xA836, 0xA837)); + assert_eq!( + row.scripts, + vec![ + "Deva", "Gujr", "Guru", "Kthi", "Mahj", "Modi", "Sind", + "Takr", "Tirh", + ] + ); + } +} |