|
| 1 | +use std::fmt; |
| 2 | + |
| 3 | +use isolang::Language; |
| 4 | +use oxilangtag::LanguageTag; |
| 5 | + |
| 6 | +#[derive(Debug, Clone, PartialEq, Eq)] |
| 7 | +pub enum LanguageCodeError { |
| 8 | + InvalidBcp47Tag { tag: String, message: String }, |
| 9 | + UnsupportedLanguage { language: String }, |
| 10 | +} |
| 11 | + |
| 12 | +impl fmt::Display for LanguageCodeError { |
| 13 | + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 14 | + match self { |
| 15 | + LanguageCodeError::InvalidBcp47Tag { tag, message } => { |
| 16 | + write!(f, "Invalid BCP 47 tag '{tag}': {message}") |
| 17 | + } |
| 18 | + LanguageCodeError::UnsupportedLanguage { language } => { |
| 19 | + write!(f, "Unsupported language subtag '{language}'") |
| 20 | + } |
| 21 | + } |
| 22 | + } |
| 23 | +} |
| 24 | + |
| 25 | +impl std::error::Error for LanguageCodeError {} |
| 26 | + |
| 27 | +/// Converts a BCP 47 language tag into its ISO 639-3 language code. |
| 28 | +/// |
| 29 | +/// The conversion uses the primary language subtag only and ignores script, region, variant, |
| 30 | +/// and extension subtags. |
| 31 | +pub fn bcp47_to_iso639_3(tag: &str) -> Result<&'static str, LanguageCodeError> { |
| 32 | + let parsed = LanguageTag::parse(tag).map_err(|error| LanguageCodeError::InvalidBcp47Tag { |
| 33 | + tag: tag.to_string(), |
| 34 | + message: error.to_string(), |
| 35 | + })?; |
| 36 | + |
| 37 | + let primary_language = parsed.primary_language(); |
| 38 | + let language = match primary_language.len() { |
| 39 | + 2 => Language::from_639_1(primary_language), |
| 40 | + 3 => Language::from_639_3(primary_language), |
| 41 | + _ => None, |
| 42 | + }; |
| 43 | + |
| 44 | + language |
| 45 | + .map(|x| x.to_639_3()) |
| 46 | + .ok_or_else(|| LanguageCodeError::UnsupportedLanguage { |
| 47 | + language: primary_language.to_string(), |
| 48 | + }) |
| 49 | +} |
| 50 | + |
| 51 | +#[cfg(test)] |
| 52 | +mod tests { |
| 53 | + use super::*; |
| 54 | + |
| 55 | + #[test] |
| 56 | + fn bcp47_to_iso639_3_for_primary_language_tags() { |
| 57 | + assert_eq!(bcp47_to_iso639_3("en").unwrap(), "eng"); |
| 58 | + assert_eq!(bcp47_to_iso639_3("de").unwrap(), "deu"); |
| 59 | + assert_eq!(bcp47_to_iso639_3("fr").unwrap(), "fra"); |
| 60 | + } |
| 61 | + |
| 62 | + #[test] |
| 63 | + fn bcp47_to_iso639_3_ignores_non_primary_subtags() { |
| 64 | + assert_eq!(bcp47_to_iso639_3("en-US").unwrap(), "eng"); |
| 65 | + assert_eq!(bcp47_to_iso639_3("zh-Hant-TW").unwrap(), "zho"); |
| 66 | + } |
| 67 | + |
| 68 | + #[test] |
| 69 | + fn bcp47_to_iso639_3_rejects_malformed_tags() { |
| 70 | + let err = bcp47_to_iso639_3("en--US").unwrap_err(); |
| 71 | + assert!(matches!(err, LanguageCodeError::InvalidBcp47Tag { .. })); |
| 72 | + } |
| 73 | + |
| 74 | + #[test] |
| 75 | + fn bcp47_to_iso639_3_rejects_unsupported_primary_language() { |
| 76 | + let err = bcp47_to_iso639_3("qaa").unwrap_err(); |
| 77 | + assert_eq!( |
| 78 | + err, |
| 79 | + LanguageCodeError::UnsupportedLanguage { |
| 80 | + language: "qaa".to_string(), |
| 81 | + } |
| 82 | + ); |
| 83 | + } |
| 84 | +} |
0 commit comments