search in common module, now fixing layer mixing issue

common/src/lib.rs

@@ -1,4 +1,7 @@
 // common/src/lib.rs
+
+pub mod search;
+
 pub mod proto {
     pub mod multieko2 {
         pub mod adresar {

common/src/search.rs

@@ -0,0 +1,78 @@
+// common/src/search.rs
+
+use tantivy::schema::*;
+use tantivy::tokenizer::*;
+use tantivy::Index;
+
+/// Creates a hybrid Slovak search schema with optimized prefix fields.
+pub fn create_search_schema() -> Schema {
+    let mut schema_builder = Schema::builder();
+
+    schema_builder.add_u64_field("pg_id", INDEXED | STORED);
+
+    // FIELD 1: For prefixes (1-15 chars).
+    let short_prefix_indexing = TextFieldIndexing::default()
+        .set_tokenizer("slovak_prefix_edge")
+        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
+    let short_prefix_options = TextOptions::default()
+        .set_indexing_options(short_prefix_indexing)
+        .set_stored();
+    schema_builder.add_text_field("prefix_edge", short_prefix_options);
+
+    // FIELD 2: For the full word.
+    let full_word_indexing = TextFieldIndexing::default()
+        .set_tokenizer("slovak_prefix_full")
+        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
+    let full_word_options = TextOptions::default()
+        .set_indexing_options(full_word_indexing)
+        .set_stored();
+    schema_builder.add_text_field("prefix_full", full_word_options);
+
+    // NGRAM FIELD: For substring matching.
+    let ngram_field_indexing = TextFieldIndexing::default()
+        .set_tokenizer("slovak_ngram")
+        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
+    let ngram_options = TextOptions::default()
+        .set_indexing_options(ngram_field_indexing)
+        .set_stored();
+    schema_builder.add_text_field("text_ngram", ngram_options);
+
+    schema_builder.build()
+}
+
+/// Registers all necessary Slovak tokenizers with the index.
+///
+/// This must be called by ANY process that opens the index
+/// to ensure the tokenizers are loaded into memory.
+pub fn register_slovak_tokenizers(index: &Index) -> tantivy::Result<()> {
+    let tokenizer_manager = index.tokenizers();
+
+    // TOKENIZER for `prefix_edge`: Edge N-gram (1-15 chars)
+    let edge_tokenizer =
+        TextAnalyzer::builder(NgramTokenizer::new(1, 15, true)?)
+            .filter(RemoveLongFilter::limit(40))
+            .filter(LowerCaser)
+            .filter(AsciiFoldingFilter)
+            .build();
+    tokenizer_manager.register("slovak_prefix_edge", edge_tokenizer);
+
+    // TOKENIZER for `prefix_full`: Simple word tokenizer
+    let full_tokenizer =
+        TextAnalyzer::builder(SimpleTokenizer::default())
+            .filter(RemoveLongFilter::limit(40))
+            .filter(LowerCaser)
+            .filter(AsciiFoldingFilter)
+            .build();
+    tokenizer_manager.register("slovak_prefix_full", full_tokenizer);
+
+    // NGRAM TOKENIZER: For substring matching.
+    let ngram_tokenizer =
+        TextAnalyzer::builder(NgramTokenizer::new(3, 3, false)?)
+            .filter(RemoveLongFilter::limit(40))
+            .filter(LowerCaser)
+            .filter(AsciiFoldingFilter)
+            .build();
+    tokenizer_manager.register("slovak_ngram", ngram_tokenizer);
+
+    Ok(())
+}