tools/graph/src/indexer.rs

use rayon::iter::ParallelIterator;
use std::collections::{HashMap, HashSet};
use oxigraph::model::{Graph, NamedNode, TermRef, NamedOrBlankNodeRef, NamedNodeRef};
use rayon::iter::IntoParallelRefIterator;
use gl_search::{Field, OwnedValue, Schema};
use crate::class::Class;
use crate::{helpers, vocab, CurieHelper};
use crate::language::LanguageCondition;
use crate::ontology::ResourceDescription;

pub struct Indexer<'a> {
    language: LanguageCondition,
    curie_helper: &'a CurieHelper,
}

impl<'a> Indexer<'a> {
    pub fn new(language: LanguageCondition, curie_helper: &'a CurieHelper) -> Self {
        Self {
            language,
            curie_helper,
        }
    }

    fn extract_string(&self, graph: &Graph, subject: NamedNodeRef, predicate: NamedNodeRef) -> OwnedValue {
        graph.object_for_subject_predicate(subject, predicate)
            .filter(|term| self.language.primary_matches_term(*term))
            .and_then(helpers::term_ref_as_str)
            .map(String::from)
            .map(OwnedValue::Str)
            .unwrap_or_else(|| OwnedValue::Null)
    }

    fn person(&self, graph: &Graph, subject: NamedNodeRef) -> HashMap<Field, OwnedValue> {
        HashMap::from_iter([
            (
                Schema::field("given_name", self.language.primary_language()),
                self.extract_string(graph, subject, vocab::rdaad::GIVEN_NAME),
            ),
            (
                Schema::field("surname", self.language.primary_language()),
                self.extract_string(graph, subject, vocab::rdaad::SURNAME),
            ),
        ])
    }

    fn corporate_body(&self, graph: &Graph, subject: NamedNodeRef) -> HashMap<Field, OwnedValue> {
        HashMap::from_iter([
            (
                Schema::field("corporate_name", self.language.primary_language()),
                self.extract_string(graph, subject, vocab::rdaad::NAME_OF_CORPORATE_BODY),
            ),
        ])
    }

    fn work(&self, graph: &Graph, subject: NamedNodeRef) -> HashMap<Field, OwnedValue> {
        HashMap::from_iter([
            (
                Schema::field("title", self.language.primary_language()),
                self.extract_string(graph, subject, vocab::rdawd::TITLE_OF_WORK),
            ),
        ])
    }

    fn expression(&self, graph: &Graph, subject: NamedNodeRef) -> HashMap<Field, OwnedValue> {
        HashMap::from_iter([
            (
                Schema::field("title", self.language.primary_language()),
                self.extract_string(graph, subject, vocab::rdaed::TITLE_OF_EXPRESSION),
            ),
        ])
    }

    fn manifestation(&self, graph: &Graph, subject: NamedNodeRef) -> HashMap<Field, OwnedValue> {
        HashMap::from_iter([
            (
                Schema::field("title", self.language.primary_language()),
                self.extract_string(graph, subject, vocab::rdamd::TITLE_OF_MANIFESTATION),
            ),
        ])
    }

    pub fn graph(&self, graph: &Graph) -> Vec<HashMap<Field, OwnedValue>> {
        let mut entities_and_types = HashSet::new();
        let triples = graph.triples_for_predicate(vocab::rdf::TYPE);
        for triple in triples {
            if let NamedOrBlankNodeRef::NamedNode(subject) = triple.subject &&
                let TermRef::NamedNode(class) = triple.object {
                entities_and_types.insert((subject, class));
            }
        }

        entities_and_types.par_iter()
            .filter_map(|(subject, class)| {
                Class::try_from_named_node(*class).and_then(|class| {
                    match class {
                        Class::RdfProperty => None,
                        Class::RdfsClass => None,
                        Class::SkosConcept => None,
                        Class::Work => Some(self.work(graph, *subject)),
                        Class::Person => Some(self.person(graph, *subject)),
                        Class::CorporateBody => Some(self.corporate_body(graph, *subject)),
                        Class::Expression => Some(self.expression(graph, *subject)),
                        Class::Manifestation => Some(self.manifestation(graph, *subject)),
                    }.map(|mut document| {
                        document.insert(Schema::discriminant_field(), OwnedValue::from(class as u64));
                        document.insert(Schema::iri_field(), OwnedValue::Str(subject.as_str().to_string()));

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

                        document
                    })
                })
            }).collect()
    }

    pub fn ontology(&self, entities: HashMap<NamedNode, ResourceDescription>) -> Vec<HashMap<Field, OwnedValue>> {
        let label_field = Schema::field("label", self.language.primary_language());
        let definition_field = Schema::field("definition", self.language.primary_language());

        entities.iter()
            .map(|(subject, description)| {
                let mut document = HashMap::with_capacity(4);

                let discriminant = description.class
                    .as_ref()
                    .and_then(|node| Class::try_from_named_node(node.as_ref()))
                    .map(|doc_type| doc_type as u64)
                    .map(OwnedValue::U64)
                    .unwrap_or(OwnedValue::Null);
                document.insert(Schema::discriminant_field(), discriminant);

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

                let subject = OwnedValue::from(subject.as_str());
                document.insert(Schema::iri_field(), subject);

                let label = description.label.clone()
                    .map(OwnedValue::from)
                    .unwrap_or(OwnedValue::Null);
                document.insert(label_field, label);

                let definition = description.description.clone()
                    .map(OwnedValue::from)
                    .unwrap_or(OwnedValue::Null);
                document.insert(definition_field, definition);

                document
            }).collect()
    }
}