tools/graph/src/index.rs

use std::collections::HashMap;
use oxigraph::model::NamedNode;
use oxigraph::sparql::results::{QueryResultsFormat, QueryResultsParser, SliceQueryResultsParserOutput};
use gl_inference::tonic::transport::Channel;
use gl_inference::proto::ontology_client::OntologyClient;
use gl_inference::proto::OntologyQueryRequest;
use gl_search::{Field, OwnedValue, Schema};
use crate::class::Class;
use crate::{curie, CurieHelper};
use crate::helpers::{term_as_str, term_to_named_node};
use crate::language::LanguageCondition;

pub async fn index_ontology(
    client: &mut OntologyClient<Channel>,
    language: &LanguageCondition,
) -> crate::Result<Vec<HashMap<Field, OwnedValue>>> {
    let curie_helper = CurieHelper::new((&*curie::PREFIXES).clone());

    let label_filter = language.to_filter_expression("label");
    let description_filter = language.to_filter_expression("description");

    let mut request = OntologyQueryRequest::default();
    request.sparql_query = format!(r#"PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT ?class ?subject ?label ?description WHERE {{
    VALUES ?class {{ rdf:Property rdfs:Class skos:Concept }}
    ?subject a ?class ;
    	rdfs:label ?label .
    {label_filter}

    OPTIONAL {{ ?subject rdfs:comment ?comment }}
    OPTIONAL {{ ?subject skos:definition ?definition }}
    BIND(COALESCE(?definition, ?comment) AS ?description)
    {description_filter}
}}"#);

    let response = client.query(request).await.unwrap();
    let parser_output = QueryResultsParser::from_format(QueryResultsFormat::Json)
        .for_slice(&response.get_ref().results)
        .unwrap();

    let mut results = Vec::new();
    if let SliceQueryResultsParserOutput::Solutions(solutions) = parser_output {
        let primary_language = language.primary_language();
        let label_field = Schema::field("label", primary_language);
        let definition_field = Schema::field("definition", primary_language);
        for solution in solutions.filter_map(Result::ok) {
            let mut document = HashMap::with_capacity(4);

            let discriminant = solution
                .get("class")
                .and_then(term_to_named_node)
                .and_then(Class::try_from_named_node)
                .map(|doc_type| doc_type as u64)
                .map(OwnedValue::U64)
                .unwrap_or(OwnedValue::Null);
            document.insert(Schema::discriminant_field(), discriminant);

            let subject_str = solution
                .get("subject")
                .and_then(term_to_named_node)
                .map(NamedNode::as_str);

            let curie = subject_str
                .and_then(|subject| curie_helper.abbreviate(None, subject))
                .map(OwnedValue::Str)
                .unwrap_or(OwnedValue::Null);
            document.insert(Schema::curie_field(), curie);

            let subject = subject_str
                .map(OwnedValue::from)
                .unwrap_or(OwnedValue::Null);
            document.insert(Schema::iri_field(), subject);

            let label = solution
                .get("label")
                .and_then(term_as_str)
                .map(OwnedValue::from)
                .unwrap_or(OwnedValue::Null);
            document.insert(label_field, label);

            let definition = solution
                .get("description")
                .and_then(term_as_str)
                .map(OwnedValue::from)
                .unwrap_or(OwnedValue::Null);
            document.insert(definition_field, definition);

            results.push(document);
        }
    }
    Ok(results)
}