tools/graph/src/language.rs

use std::env::var;
use oxigraph::model::{Literal, TermRef};
use oxilangtag::LanguageTag;
use std::sync::LazyLock;
use spargebra::algebra::{Expression, Function, GraphPattern};
use spargebra::term::{NamedNode, NamedNodePattern, TermPattern, TriplePattern, Variable};
use tracing::debug;

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, Debug)]
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 filter(conditions: impl IntoIterator<Item = (String, Self)>) -> String {
        let exact_expression = |variable: String, language| Expression::FunctionCall(
            Function::LangMatches, vec![
                Expression::FunctionCall(Function::Lang, vec![
                    Expression::Variable(Variable::new_unchecked(variable))
                ]),
                Expression::Literal(Literal::new_simple_literal(language))
            ],
        );

        let untagged_expression = |variable| Expression::Not(
            Box::new(Expression::FunctionCall(Function::HasLang, vec![
                Expression::Variable(Variable::new_unchecked(variable))
            ]))
        );

        let condition_to_expression = |variable, condition| match condition {
            LanguageCondition::ExactMatchOnly(language) =>
                Some(exact_expression(variable, language.to_string())),
            LanguageCondition::ExactMatchOrUntagged(language) => {
                Some(Expression::Or(
                    Box::new(exact_expression(variable.clone(), language.to_string())),
                    Box::new(untagged_expression(variable)),
                ))
            }
            LanguageCondition::UntaggedOnly => Some(untagged_expression(variable)),
            LanguageCondition::AnyOrNone => None,
        };

        let mut iter = conditions.into_iter();
        let first_expression = iter.next()
            .and_then(|(variable, condition)| condition_to_expression(variable, condition));
        if let Some(expression) = first_expression {
            let expr = iter.filter_map(|(variable, condition)| condition_to_expression(variable, condition))
                .fold(expression, |acc, next_exp| Expression::Or(Box::new(acc), Box::new(next_exp)));
            GraphPattern::Filter {
                expr,
                inner: Box::new(GraphPattern::Bgp { patterns: vec![] }),
            }.to_string()
        } else {
            String::default()
        }
    }
}