This commit is contained in:
2026-08-05 18:33:30 -04:00
parent 7ce5ed118d
commit cd47d868c1
20 changed files with 307 additions and 181 deletions
-3
View File
@@ -23,7 +23,4 @@ pub enum SearchError {
#[error(transparent)]
SparqlSyntax(#[from] oxigraph::sparql::SparqlSyntaxError),
#[error("The provided DocType is not valid for this operation")]
InvalidDocType,
}
+29 -3
View File
@@ -12,6 +12,7 @@ pub use tantivy::schema::document::{Document, Value};
pub use tantivy::indexer::IndexWriter;
pub use error::{Result, SearchError};
use gl_graph::vocab;
pub use index::{SearchIndex, SearchIndexBuilder};
pub use schema::Schema;
@@ -20,12 +21,37 @@ pub enum DocType {
RdfsClass = 1,
SkosConcept = 2,
Person = 3,
CorporateBody = 4,
Work = 5,
Expression = 6,
Manifestation = 7,
}
impl DocType {
pub fn from_named_node(node: NamedNodeRef) -> Option<Self> {
match node.as_str() {
"http://rdaregistry.info/Elements/c/C10004" => Some(DocType::Person),
pub fn to_named_node(&self) -> NamedNodeRef<'_> {
match self {
DocType::RdfProperty => vocab::rdf::PROPERTY,
DocType::RdfsClass => vocab::rdfs::CLASS,
DocType::SkosConcept => vocab::skos::CONCEPT,
DocType::Person => vocab::rdac::PERSON,
DocType::CorporateBody => vocab::rdac::CORPORATE_BODY,
DocType::Work => vocab::rdac::WORK,
DocType::Expression => vocab::rdac::EXPRESSION,
DocType::Manifestation => vocab::rdac::MANIFESTATION,
}
}
pub fn from_named_node<'a>(node: impl Into<NamedNodeRef<'a>>) -> Option<Self> {
let node = node.into();
match node {
vocab::rdf::PROPERTY => Some(DocType::RdfProperty),
vocab::rdfs::CLASS => Some(DocType::RdfsClass),
vocab::skos::CONCEPT => Some(DocType::SkosConcept),
vocab::rdac::PERSON => Some(DocType::Person),
vocab::rdac::CORPORATE_BODY => Some(DocType::CorporateBody),
vocab::rdac::WORK => Some(DocType::Work),
vocab::rdac::EXPRESSION => Some(DocType::Expression),
vocab::rdac::MANIFESTATION => Some(DocType::Manifestation),
_ => None,
}
}
+68 -33
View File
@@ -5,10 +5,11 @@ use oxigraph::sparql::{QueryResults, SparqlEvaluator};
use oxigraph::store::Store;
use tantivy::IndexWriter;
use tantivy::schema::{Field, OwnedValue};
use tracing::debug;
use gl_graph::{vocab, CurieHelper};
use crate::DocType;
use crate::language::LanguageCondition;
use crate::schema::Schema;
use crate::SearchError::InvalidDocType;
fn term_to_named_node(term: &Term) -> Option<&NamedNode> {
if let Term::NamedNode(node) = term {
@@ -52,16 +53,15 @@ fn term_ref_as_str(term: TermRef<'_>) -> Option<&str> {
}
}
const GIVEN_NAME: NamedNodeRef = NamedNodeRef::new_unchecked("http://rdaregistry.info/Elements/a/datatype/P50292");
const SURNAME: NamedNodeRef = NamedNodeRef::new_unchecked("http://rdaregistry.info/Elements/a/datatype/P50291");
pub fn index_entity(dataset: &Dataset, language: &LanguageCondition, writer: &IndexWriter<HashMap<Field, OwnedValue>>) -> crate::Result<usize> {
let curie_helper = CurieHelper::new((&*gl_graph::PREFIXES).clone());
let mut subject_type_map = HashMap::new();
for quad in dataset.quads_for_pattern(None, Some(rdf::TYPE), None, None) {
if let NamedOrBlankNodeRef::NamedNode(subject) = quad.subject {
subject_type_map.entry(subject)
.and_modify(|types: &mut Vec<_>| types.push(quad.object))
.or_insert_with(Vec::new);
.or_insert_with(|| vec![quad.object]);
}
}
@@ -69,16 +69,24 @@ pub fn index_entity(dataset: &Dataset, language: &LanguageCondition, writer: &In
for (subject, types) in subject_type_map {
for type_ in types {
if let Some(doc_type) = term_ref_as_named_node(type_).and_then(DocType::from_named_node) {
match doc_type {
DocType::Person => {
let mut doc = index_person(dataset, subject, language);
doc.insert(Schema::discriminant_field(), OwnedValue::U64(doc_type as u64));
doc.insert(Schema::iri_field(), OwnedValue::Str(subject.as_str().to_string()));
writer.add_document(doc)?;
counter += 1;
}
_ => {}
}
let mut doc = match doc_type {
DocType::Person => index_person(dataset, subject, language),
DocType::CorporateBody => index_corporate_body(dataset, subject, language),
_ => continue
};
let subject_str = subject.as_str();
doc.insert(Schema::discriminant_field(), OwnedValue::U64(doc_type as u64));
doc.insert(Schema::iri_field(), OwnedValue::Str(subject_str.to_string()));
let curie = curie_helper.abbreviate(None, subject_str)
.map(OwnedValue::Str)
.unwrap_or(OwnedValue::Null);
doc.insert(Schema::curie_field(), curie);
writer.add_document(doc)?;
counter += 1;
}
}
}
@@ -87,7 +95,7 @@ pub fn index_entity(dataset: &Dataset, language: &LanguageCondition, writer: &In
}
fn index_person(dataset: &Dataset, subject: NamedNodeRef<'_>, language: &LanguageCondition) -> HashMap<Field, OwnedValue> {
let given_name = dataset.quads_for_pattern(Some(subject.into()), Some(GIVEN_NAME), None, None)
let given_name = dataset.quads_for_pattern(Some(subject.into()), Some(vocab::rdaad::GIVEN_NAME), None, None)
.map(|q| q.object)
.filter(|term| language.primary_matches_term(*term))
.filter_map(term_ref_as_str)
@@ -96,7 +104,7 @@ fn index_person(dataset: &Dataset, subject: NamedNodeRef<'_>, language: &Languag
.next()
.unwrap_or_else(|| OwnedValue::Null);
let surname = dataset.quads_for_pattern(Some(subject.into()), Some(SURNAME), None, None)
let surname = dataset.quads_for_pattern(Some(subject.into()), Some(vocab::rdaad::SURNAME), None, None)
.map(|q| q.object)
.filter(|term| language.primary_matches_term(*term))
.filter_map(term_ref_as_str)
@@ -111,18 +119,30 @@ fn index_person(dataset: &Dataset, subject: NamedNodeRef<'_>, language: &Languag
])
}
pub fn index_triples(doc_type: DocType, store: Store, language: &LanguageCondition, writer: &IndexWriter<HashMap<Field, OwnedValue>>) -> crate::Result<usize> {
fn index_corporate_body(dataset: &Dataset, subject: NamedNodeRef<'_>, language: &LanguageCondition) -> HashMap<Field, OwnedValue> {
let name = dataset.quads_for_pattern(Some(subject.into()), Some(vocab::rdaad::NAME_OF_CORPORATE_BODY), None, None)
.map(|q| q.object)
.filter(|term| language.primary_matches_term(*term))
.filter_map(term_ref_as_str)
.map(String::from)
.map(OwnedValue::Str)
.next()
.unwrap_or_else(|| OwnedValue::Null);
HashMap::from_iter([
(Schema::field("corporate_name", language.primary_language()), name),
])
}
pub fn index_schema(store: Store, language: &LanguageCondition, writer: &IndexWriter<HashMap<Field, OwnedValue>>) -> crate::Result<usize> {
let curie_helper = CurieHelper::new((&*gl_graph::PREFIXES).clone());
let label_filter = language.to_filter_expression("label");
let description_filter = language.to_filter_expression("description");
let rdf_type = match doc_type {
DocType::RdfsClass => Ok("rdfs:Class"),
DocType::RdfProperty => Ok("rdf:Property"),
DocType::SkosConcept => Ok("skos:Concept"),
_ => Err(InvalidDocType),
}?;
let query = format!(r#"SELECT ?subject ?label ?description WHERE {{
?subject a {rdf_type} ;
let query = format!(r#"SELECT ?class ?subject ?label ?description WHERE {{
VALUES ?class {{ rdf:Property rdfs:Class skos:Concept }}
?subject a ?class ;
rdfs:label ?label .
{label_filter}
@@ -141,17 +161,32 @@ pub fn index_triples(doc_type: DocType, store: Store, language: &LanguageConditi
let mut counter = 0usize;
let query_results = sparql.on_store(&store).execute()?;
if let QueryResults::Solutions(solutions) = query_results {
let discriminant = OwnedValue::U64(doc_type as u64);
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);
document.insert(Schema::discriminant_field(), discriminant.clone());
let subject = solution.get("subject")
let discriminant = solution.get("class")
.and_then(term_to_named_node)
.map(NamedNode::as_str)
.map(OwnedValue::from)
.and_then(DocType::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);
@@ -159,13 +194,13 @@ pub fn index_triples(doc_type: DocType, store: Store, language: &LanguageConditi
.and_then(term_as_str)
.map(OwnedValue::from)
.unwrap_or(OwnedValue::Null);
document.insert(Schema::field("label", primary_language), label);
document.insert(label_field, label);
let definition = solution.get("description")
.and_then(term_as_str)
.map(OwnedValue::from)
.unwrap_or(OwnedValue::Null);
document.insert(Schema::field("definition", primary_language), definition);
document.insert(definition_field, definition);
writer.add_document(document)?;
counter += 1;
+1
View File
@@ -33,6 +33,7 @@ impl Schema {
schema_builder.add_text_field("surname:en", stored_ngram32.clone());
schema_builder.add_text_field("given_name:en", stored_ngram32.clone());
schema_builder.add_text_field("corporate_name:en", stored_ngram32.clone());
schema_builder.add_u64_field("subtitle_start", schema::STORED);
schema_builder.add_u64_field("subtitle_end", schema::STORED);