.

search/src/index.rs

@@ -1,13 +1,13 @@
-use crate::error;
+use crate::{error, SearchDocument};
 use crate::error::SearchError;
 use crate::schema::Schema;
 use std::path::PathBuf;
 use tantivy::collector::TopDocs;
 use tantivy::directory::{ManagedDirectory, MmapDirectory};
 use tantivy::query::{BooleanQuery, Occur, QueryParser, TermQuery};
-use tantivy::schema::{Field, IndexRecordOption, Value};
-use tantivy::tokenizer::{NgramTokenizer, TokenizerManager};
-use tantivy::{Index, IndexReader, IndexWriter, ReloadPolicy, TantivyDocument, Term};
+use tantivy::schema::{Field, IndexRecordOption, NamedFieldDocument, Value};
+use tantivy::tokenizer::{LowerCaser, NgramTokenizer, TextAnalyzer, TokenizerManager};
+use tantivy::{Document, Index, IndexReader, IndexWriter, ReloadPolicy, Score, TantivyDocument, Term};
 
 #[derive(Default)]
 pub struct SearchIndexBuilder {
@@ -23,8 +23,12 @@ impl SearchIndexBuilder {
     pub fn build(self) -> error::Result<SearchIndex> {
         if let Some(path) = self.path {
             let ngram_32 = NgramTokenizer::new(1, 32, false)?;
+            let ngram_32_lowercase = TextAnalyzer::builder(ngram_32)
+                .filter(LowerCaser)
+                .build();
+
             let tokenizer_manager = TokenizerManager::default();
-            tokenizer_manager.register("ngram_32", ngram_32);
+            tokenizer_manager.register("ngram_32", ngram_32_lowercase);
 
             let mmap_directory = MmapDirectory::open(path)?;
             let managed_directory = ManagedDirectory::wrap(Box::new(mmap_directory))?;
@@ -82,7 +86,7 @@ impl SearchIndex {
         type_: Option<u64>,
         user_query: &str,
         default_fields: Vec<Field>,
-    ) -> error::Result<Vec<String>> {
+    ) -> error::Result<Vec<NamedFieldDocument>> {
         let parser = QueryParser::for_index(&self.index, default_fields);
         let (user_query, _) = parser.parse_query_lenient(user_query);
 
@@ -98,16 +102,12 @@ impl SearchIndex {
 
         let query = BooleanQuery::new(subqueries);
         let searcher = self.reader.searcher();
-        let results = searcher.search(&query, &TopDocs::with_limit(10000).order_by_score())?;
-        let mut iris = vec![];
-        for (_score, address) in results.iter() {
-            let doc: TantivyDocument = searcher.doc(*address)?;
-            if let Some(doc_iri) = doc.get_first(Schema::iri_field()) {
-                let doc_iri_string = doc_iri.as_str().unwrap_or("???").to_string();
-                iris.push(doc_iri_string);
-            }
-        }
-
-        Ok(iris)
+        let results = searcher.search(&query, &TopDocs::with_limit(10000).order_by_score())?
+            .iter()
+            .map(|(_, address)| searcher.doc(*address))
+            .filter_map(Result::ok)
+            .map(|doc: SearchDocument| doc.to_named_doc(Schema::schema()))
+            .collect();
+        Ok(results)
     }
 }

search/src/lib.rs

@@ -4,6 +4,7 @@ mod schema;
 
 pub use tantivy::TantivyDocument as SearchDocument;
 pub use tantivy::doc;
+pub use tantivy::schema::{NamedFieldDocument, OwnedValue};
 
 pub use error::{Result, SearchError};
 pub use index::{SearchIndex, SearchIndexBuilder};

search/src/schema.rs

@@ -11,36 +11,31 @@ pub struct Schema;
 impl Schema {
     pub fn schema() -> &'static TantivySchema {
         SCHEMA.get_or_init(|| {
-            let ngram_32 = TextOptions::default().set_indexing_options(
+            let stored_ngram32 = TextOptions::default().set_indexing_options(
                 TextFieldIndexing::default()
                     .set_index_option(IndexRecordOption::WithFreqsAndPositions)
                     .set_tokenizer("ngram_32"),
-            );
+            ).set_stored();
 
-            let en_stem = TextOptions::default().set_indexing_options(
+            let stored_en_stem = TextOptions::default().set_indexing_options(
                 TextFieldIndexing::default()
                     .set_index_option(IndexRecordOption::WithFreqsAndPositions)
                     .set_tokenizer("en_stem"),
-            );
+            ).set_stored();
 
             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("label", stored_ngram32.clone());
+            schema_builder.add_text_field("comment", stored_en_stem.clone());
 
-            schema_builder.add_text_field("title", en_stem.clone());
+            schema_builder.add_text_field("givenName", stored_ngram32.clone());
+            schema_builder.add_text_field("surname", stored_ngram32);
+
+            /*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.add_text_field("content", en_stem);*/
             schema_builder.build()
         })
     }