Initial commit

2026-06-08 19:33:49 -04:00
commit 6906fb973a
24 changed files with 9382 additions and 0 deletions
@@ -0,0 +1,77 @@
+use std::sync::OnceLock;
+use tantivy::schema;
+use tantivy::schema::{
+    Field, IndexRecordOption, Schema as TantivySchema, TextFieldIndexing, TextOptions,
+};
+
+static SCHEMA: OnceLock<TantivySchema> = OnceLock::new();
+
+pub struct Schema;
+
+impl Schema {
+    pub fn schema() -> &'static TantivySchema {
+        SCHEMA.get_or_init(|| {
+            let ngram_32 = TextOptions::default().set_indexing_options(
+                TextFieldIndexing::default()
+                    .set_index_option(IndexRecordOption::WithFreqsAndPositions)
+                    .set_tokenizer("ngram_32"),
+            );
+
+            let en_stem = TextOptions::default().set_indexing_options(
+                TextFieldIndexing::default()
+                    .set_index_option(IndexRecordOption::WithFreqsAndPositions)
+                    .set_tokenizer("en_stem"),
+            );
+
+            let mut schema_builder = TantivySchema::builder();
+            schema_builder.add_u64_field("type", schema::FAST | schema::INDEXED);
+            schema_builder.add_text_field("iri", schema::STORED | schema::STRING);
+            schema_builder.add_text_field("label", ngram_32.clone());
+            schema_builder.add_text_field("comment", en_stem.clone());
+
+            schema_builder.add_text_field("given name", ngram_32.clone());
+            schema_builder.add_text_field("surname", ngram_32);
+
+            schema_builder.add_text_field("title", en_stem.clone());
+            schema_builder.add_text_field("description", en_stem.clone());
+            schema_builder.add_text_field("content", en_stem);
+
+            schema_builder.add_u64_field("page", schema::STORED);
+            schema_builder.add_u64_field("book", schema::STORED);
+            schema_builder.add_u64_field("chapter", schema::STORED);
+            schema_builder.add_u64_field("verse", schema::STORED);
+
+            schema_builder.build()
+        })
+    }
+
+    pub fn type_field() -> Field {
+        Self::schema().get_field("type").unwrap()
+    }
+
+    pub fn iri_field() -> Field {
+        Self::schema().get_field("iri").unwrap()
+    }
+
+    pub fn label_field() -> Field {
+        Self::schema().get_field("label").unwrap()
+    }
+
+    pub fn comment_field() -> Field {
+        Self::schema().get_field("comment").unwrap()
+    }
+
+    pub fn ontology_fields() -> Vec<Field> {
+        vec![Self::label_field(), Self::comment_field()]
+    }
+
+    pub fn default_fields() -> Vec<Field> {
+        vec![
+            Self::schema().get_field("given name").unwrap(),
+            Self::schema().get_field("surname").unwrap(),
+            Self::schema().get_field("title").unwrap(),
+            Self::schema().get_field("description").unwrap(),
+            Self::schema().get_field("content").unwrap(),
+        ]
+    }
+}