tools/graph/src/language.rs

use oxigraph::model::TermRef;
use oxilangtag::LanguageTag;
use std::sync::LazyLock;

pub const ENGLISH_PRIMARY: &str = "en";

pub static ENGLISH_OR_UNTAGGED: LazyLock<LanguageCondition> = LazyLock::new(|| {
    LanguageCondition::ExactMatchOrUntagged(
        LanguageTag::parse(ENGLISH_PRIMARY.to_string()).unwrap(),
    )
});

#[derive(Clone)]
pub enum LanguageCondition {
    ExactMatchOnly(LanguageTag<String>),
    ExactMatchOrUntagged(LanguageTag<String>),
    UntaggedOnly,
    AnyOrNone,
}

impl LanguageCondition {
    pub fn primary_matches_term<'a>(&self, term: impl Into<TermRef<'a>>) -> bool {
        if let TermRef::Literal(literal) = term.into() {
            let tag = literal
                .language()
                .map(LanguageTag::parse_and_normalize)
                .and_then(Result::ok);

            match (tag, self) {
                (Some(language), LanguageCondition::ExactMatchOnly(expectation))
                | (Some(language), LanguageCondition::ExactMatchOrUntagged(expectation)) => {
                    language.primary_language() == expectation.primary_language()
                }
                (None, LanguageCondition::ExactMatchOrUntagged(_)) => true,
                (None, LanguageCondition::UntaggedOnly) => true,
                (_, LanguageCondition::AnyOrNone) => true,
                _ => false,
            }
        } else {
            false
        }
    }

    pub fn primary_language(&self) -> Option<&str> {
        match self {
            LanguageCondition::ExactMatchOnly(tag) => Some(tag.primary_language()),
            LanguageCondition::ExactMatchOrUntagged(tag) => Some(tag.primary_language()),
            _ => None,
        }
    }

    pub fn to_filter_expression(&self, var: &str) -> String {
        match self {
            LanguageCondition::ExactMatchOnly(language) => {
                format!(r#"FILTER(langMATCHES(LANG(?{var}), "{language}"))"#)
            }
            LanguageCondition::ExactMatchOrUntagged(language) => {
                format!(r#"FILTER(langMATCHES(LANG(?{var}), "{language}") || !hasLANG(?{var}))"#)
            }
            LanguageCondition::UntaggedOnly => format!("FILTER(!hasLANG(?{var}))"),
            LanguageCondition::AnyOrNone => "".to_string(),
        }
    }
}