refactor(index): extract BasicTrainer trait from ScalarIndexPlugin

.claude/scheduled_tasks.lock

@@ -0,0 +1 @@
+{"sessionId":"23aa895c-8c08-462b-ae20-e4ad399e466f","pid":1,"procStart":"52028919","acquiredAt":1782134409076}

rust/lance-index/benches/geo.rs

@@ -15,7 +15,7 @@ use geoarrow_schema::Dimension;
 use lance_core::cache::LanceCache;
 use lance_core::{Error, ROW_ID};
 use lance_index::scalar::lance_format::LanceIndexStore;
-use lance_index::scalar::registry::ScalarIndexPlugin;
+use lance_index::scalar::registry::BasicTrainer;
 use lance_index::scalar::rtree::{BoundingBox, RTreeIndex, RTreeIndexPlugin, RTreeTrainingRequest};
 use lance_index::scalar::{GeoQuery, RelationQuery, ScalarIndex};
 use lance_io::object_store::ObjectStore;

rust/lance-index/src/scalar/bitmap.rs

@@ -48,7 +48,7 @@ use crate::{
         CreatedIndex, UpdateCriteria,
         expression::SargableQueryParser,
         registry::{
-            ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest,
+            BasicTrainer, ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest,
             VALUE_COLUMN_NAME,
         },
     },
@@ -1676,11 +1676,7 @@ pub async fn merge_bitmap_indices(
 }
 
 #[async_trait]
-impl ScalarIndexPlugin for BitmapIndexPlugin {
-    fn name(&self) -> &str {
-        "Bitmap"
-    }
-
+impl BasicTrainer for BitmapIndexPlugin {
     fn new_training_request(
         &self,
         params: &str,
@@ -1699,26 +1695,6 @@ impl ScalarIndexPlugin for BitmapIndexPlugin {
         Ok(Box::new(BitmapTrainingRequest::new(params)))
     }
 
-    fn provides_exact_answer(&self) -> bool {
-        true
-    }
-
-    fn version(&self) -> u32 {
-        BITMAP_INDEX_VERSION
-    }
-
-    fn new_query_parser(
-        &self,
-        index_name: String,
-        _index_details: &prost_types::Any,
-    ) -> Option<Box<dyn ScalarQueryParser>> {
-        Some(Box::new(SargableQueryParser::new(
-            index_name,
-            self.name().to_string(),
-            false,
-        )))
-    }
-
     async fn train_index(
         &self,
         data: SendableRecordBatchStream,
@@ -1760,6 +1736,37 @@ impl ScalarIndexPlugin for BitmapIndexPlugin {
             files: vec![file],
         })
     }
+}
+
+#[async_trait]
+impl ScalarIndexPlugin for BitmapIndexPlugin {
+    fn basic_trainer(&self) -> Option<&dyn BasicTrainer> {
+        Some(self)
+    }
+
+    fn name(&self) -> &str {
+        "Bitmap"
+    }
+
+    fn provides_exact_answer(&self) -> bool {
+        true
+    }
+
+    fn version(&self) -> u32 {
+        BITMAP_INDEX_VERSION
+    }
+
+    fn new_query_parser(
+        &self,
+        index_name: String,
+        _index_details: &prost_types::Any,
+    ) -> Option<Box<dyn ScalarQueryParser>> {
+        Some(Box::new(SargableQueryParser::new(
+            index_name,
+            self.name().to_string(),
+            false,
+        )))
+    }
 
     /// Load an index from storage
     async fn load_index(

rust/lance-index/src/scalar/bloomfilter.rs

@@ -9,7 +9,7 @@
 
 use crate::scalar::expression::{BloomFilterQueryParser, ScalarQueryParser};
 use crate::scalar::registry::{
-    ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest,
+    BasicTrainer, ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest,
 };
 use crate::scalar::{
     BloomFilterQuery, BuiltinIndexType, CreatedIndex, IndexFile, ScalarIndexParams, UpdateCriteria,
@@ -995,11 +995,7 @@ impl BloomFilterIndexPlugin {
 }
 
 #[async_trait]
-impl ScalarIndexPlugin for BloomFilterIndexPlugin {
-    fn name(&self) -> &str {
-        "BloomFilter"
-    }
-
+impl BasicTrainer for BloomFilterIndexPlugin {
     fn new_training_request(
         &self,
         params: &str,
@@ -1082,6 +1078,13 @@ impl ScalarIndexPlugin for BloomFilterIndexPlugin {
             files: vec![file],
         })
     }
+}
+
+#[async_trait]
+impl ScalarIndexPlugin for BloomFilterIndexPlugin {
+    fn basic_trainer(&self) -> Option<&dyn BasicTrainer> {
+        Some(self)
+    }
 
     fn provides_exact_answer(&self) -> bool {
         false
@@ -1091,6 +1094,10 @@ impl ScalarIndexPlugin for BloomFilterIndexPlugin {
         BLOOMFILTER_INDEX_VERSION
     }
 
+    fn name(&self) -> &str {
+        "BloomFilter"
+    }
+
     fn new_query_parser(
         &self,
         index_name: String,

rust/lance-index/src/scalar/btree.rs

@@ -23,7 +23,9 @@ use crate::{
     scalar::{
         CreatedIndex, UpdateCriteria,
         expression::{SargableQueryParser, ScalarQueryParser},
-        registry::{ScalarIndexPlugin, TrainingOrdering, TrainingRequest, VALUE_COLUMN_NAME},
+        registry::{
+            BasicTrainer, ScalarIndexPlugin, TrainingOrdering, TrainingRequest, VALUE_COLUMN_NAME,
+        },
     },
 };
 use crate::{metrics::NoOpMetricsCollector, scalar::registry::TrainingCriteria};
@@ -3202,11 +3204,7 @@ impl TrainingRequest for BTreeTrainingRequest {
 pub struct BTreeIndexPlugin;
 
 #[async_trait]
-impl ScalarIndexPlugin for BTreeIndexPlugin {
-    fn name(&self) -> &str {
-        "BTree"
-    }
-
+impl BasicTrainer for BTreeIndexPlugin {
     fn new_training_request(
         &self,
         params: &str,
@@ -3222,26 +3220,6 @@ impl ScalarIndexPlugin for BTreeIndexPlugin {
         Ok(Box::new(BTreeTrainingRequest::new(params)))
     }
 
-    fn provides_exact_answer(&self) -> bool {
-        true
-    }
-
-    fn version(&self) -> u32 {
-        BTREE_INDEX_VERSION
-    }
-
-    fn new_query_parser(
-        &self,
-        index_name: String,
-        _index_details: &prost_types::Any,
-    ) -> Option<Box<dyn ScalarQueryParser>> {
-        Some(Box::new(SargableQueryParser::new(
-            index_name,
-            self.name().to_string(),
-            false,
-        )))
-    }
-
     async fn train_index(
         &self,
         data: SendableRecordBatchStream,
@@ -3286,6 +3264,37 @@ impl ScalarIndexPlugin for BTreeIndexPlugin {
             files,
         })
     }
+}
+
+#[async_trait]
+impl ScalarIndexPlugin for BTreeIndexPlugin {
+    fn basic_trainer(&self) -> Option<&dyn BasicTrainer> {
+        Some(self)
+    }
+
+    fn name(&self) -> &str {
+        "BTree"
+    }
+
+    fn provides_exact_answer(&self) -> bool {
+        true
+    }
+
+    fn version(&self) -> u32 {
+        BTREE_INDEX_VERSION
+    }
+
+    fn new_query_parser(
+        &self,
+        index_name: String,
+        _index_details: &prost_types::Any,
+    ) -> Option<Box<dyn ScalarQueryParser>> {
+        Some(Box::new(SargableQueryParser::new(
+            index_name,
+            self.name().to_string(),
+            false,
+        )))
+    }
 
     async fn load_index(
         &self,

rust/lance-index/src/scalar/fmindex.rs

@@ -39,8 +39,8 @@ use crate::metrics::MetricsCollector;
 use crate::pb;
 use crate::scalar::expression::{ScalarQueryParser, TextQueryParser};
 use crate::scalar::registry::{
-    DefaultTrainingRequest, ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest,
-    VALUE_COLUMN_NAME,
+    BasicTrainer, DefaultTrainingRequest, ScalarIndexPlugin, TrainingCriteria, TrainingOrdering,
+    TrainingRequest, VALUE_COLUMN_NAME,
 };
 use crate::scalar::{
     AnyQuery, BuiltinIndexType, CreatedIndex, IndexFile, IndexStore, OldIndexDataFilter,
@@ -1676,10 +1676,7 @@ async fn write_empty_fmindex_partition(store: &dyn IndexStore) -> Result<IndexFi
 pub struct FMIndexPlugin;
 
 #[async_trait]
-impl ScalarIndexPlugin for FMIndexPlugin {
-    fn name(&self) -> &str {
-        "Fm"
-    }
+impl BasicTrainer for FMIndexPlugin {
     fn new_training_request(
         &self,
         _params: &str,
@@ -1713,6 +1710,17 @@ impl ScalarIndexPlugin for FMIndexPlugin {
             files,
         })
     }
+}
+
+#[async_trait]
+impl ScalarIndexPlugin for FMIndexPlugin {
+    fn basic_trainer(&self) -> Option<&dyn BasicTrainer> {
+        Some(self)
+    }
+
+    fn name(&self) -> &str {
+        "Fm"
+    }
     fn provides_exact_answer(&self) -> bool {
         true
     }
@@ -1766,6 +1774,7 @@ mod tests {
     use std::sync::Arc;
 
     use crate::scalar::lance_format::LanceIndexStore;
+    use crate::scalar::registry::BasicTrainer;
 
     #[test]
     fn test_fmindex_build_and_search() {

rust/lance-index/src/scalar/inverted.rs

@@ -119,7 +119,9 @@ use crate::{
     scalar::{
         CreatedIndex, ScalarIndex,
         expression::{FtsQueryParser, ScalarQueryParser},
-        registry::{ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest},
+        registry::{
+            BasicTrainer, ScalarIndexPlugin, TrainingCriteria, TrainingOrdering, TrainingRequest,
+        },
     },
 };
 
@@ -193,11 +195,7 @@ impl TrainingRequest for InvertedIndexTrainingRequest {
 }
 
 #[async_trait]
-impl ScalarIndexPlugin for InvertedIndexPlugin {
-    fn name(&self) -> &str {
-        "Inverted"
-    }
-
+impl BasicTrainer for InvertedIndexPlugin {
     fn new_training_request(
         &self,
         params: &str,
@@ -219,33 +217,6 @@ impl ScalarIndexPlugin for InvertedIndexPlugin {
         Ok(Box::new(InvertedIndexTrainingRequest::new(params)))
     }
 
-    fn provides_exact_answer(&self) -> bool {
-        false
-    }
-
-    fn version(&self) -> u32 {
-        max_supported_fts_format_version().index_version()
-    }
-
-    fn new_query_parser(
-        &self,
-        index_name: String,
-        _index_details: &prost_types::Any,
-    ) -> Option<Box<dyn ScalarQueryParser>> {
-        let Ok(index_details) = _index_details.to_msg::<pbold::InvertedIndexDetails>() else {
-            return None;
-        };
-
-        if Self::can_accelerate_queries(&index_details) {
-            Some(Box::new(FtsQueryParser::new(
-                index_name,
-                self.name().to_string(),
-            )))
-        } else {
-            None
-        }
-    }
-
     /// Train a new index
     ///
     /// The provided data must fulfill all the criteria returned by `training_criteria`.
@@ -280,6 +251,44 @@ impl ScalarIndexPlugin for InvertedIndexPlugin {
         )
         .await
     }
+}
+
+#[async_trait]
+impl ScalarIndexPlugin for InvertedIndexPlugin {
+    fn basic_trainer(&self) -> Option<&dyn BasicTrainer> {
+        Some(self)
+    }
+
+    fn name(&self) -> &str {
+        "Inverted"
+    }
+
+    fn provides_exact_answer(&self) -> bool {
+        false
+    }
+
+    fn version(&self) -> u32 {
+        max_supported_fts_format_version().index_version()
+    }
+
+    fn new_query_parser(
+        &self,
+        index_name: String,
+        _index_details: &prost_types::Any,
+    ) -> Option<Box<dyn ScalarQueryParser>> {
+        let Ok(index_details) = _index_details.to_msg::<pbold::InvertedIndexDetails>() else {
+            return None;
+        };
+
+        if Self::can_accelerate_queries(&index_details) {
+            Some(Box::new(FtsQueryParser::new(
+                index_name,
+                self.name().to_string(),
+            )))
+        } else {
+            None
+        }
+    }
 
     /// Load an index from storage
     ///