chore: commit should be in doc.attribute

Signed-off-by: Wei Zhang <[email protected]>

crates/tabby-common/src/index/code/mod.rs

@@ -10,8 +10,9 @@ use super::{corpus, IndexSchema};
 use crate::api::code::CodeSearchQuery;
 
 pub mod fields {
+    pub const ATTRIBUTE_COMMIT: &str = "commit";
+
     pub const CHUNK_GIT_URL: &str = "chunk_git_url";
-    pub const CHUNK_COMMIT: &str = "chunk_commit";
     pub const CHUNK_FILEPATH: &str = "chunk_filepath";
     pub const CHUNK_LANGUAGE: &str = "chunk_language";
     pub const CHUNK_BODY: &str = "chunk_body";

crates/tabby-index/src/code/index.rs

@@ -154,7 +154,7 @@ fn require_updates(indexer: Arc<Indexer>, id: &str) -> bool {
 
 fn should_backfill(indexer: Arc<Indexer>, id: &str) -> bool {
     // v0.23.0 add the commit field to the code document.
-    !indexer.has_attribute_field(id, code::fields::CHUNK_COMMIT)
+    !indexer.has_attribute_field(id, code::fields::ATTRIBUTE_COMMIT)
 }
 
 fn is_valid_file(file: &SourceCode) -> bool {

crates/tabby-index/src/code/mod.rs

@@ -62,8 +62,10 @@ impl CodeBuilder {
 
 #[async_trait]
 impl IndexAttributeBuilder<SourceCode> for CodeBuilder {
-    async fn build_attributes(&self, _source_code: &SourceCode) -> serde_json::Value {
-        json!({})
+    async fn build_attributes(&self, source_code: &SourceCode) -> serde_json::Value {
+        json!({
+            code::fields::ATTRIBUTE_COMMIT: source_code.commit,
+        })
     }
 
     async fn build_chunk_attributes<'a>(
@@ -102,7 +104,6 @@ impl IndexAttributeBuilder<SourceCode> for CodeBuilder {
                 let attributes = json!({
                     code::fields::CHUNK_FILEPATH: source_code.filepath,
                     code::fields::CHUNK_GIT_URL: source_code.git_url,
-                    code::fields::CHUNK_COMMIT: source_code.commit,
                     code::fields::CHUNK_LANGUAGE: source_code.language,
                     code::fields::CHUNK_BODY: body,
                     code::fields::CHUNK_START_LINE: start_line,

crates/tabby/src/services/code.rs

@@ -10,7 +10,7 @@ use tabby_common::{
     index::{
         self,
         code::{self, tokenize_code},
-        IndexSchema,
+        corpus, IndexSchema,
     },
 };
 use tabby_inference::Embedding;
@@ -76,16 +76,37 @@ impl CodeSearchImpl {
                 .await?
         };
 
-        Ok(merge_code_responses_by_rank(
-            &params,
-            docs_from_embedding,
-            docs_from_bm25,
-        ))
+        let mut merged_codes =
+            merge_code_responses_by_rank(&params, docs_from_embedding, docs_from_bm25);
+        add_doc_attribute(reader, &mut merged_codes).await;
+
+        Ok(merged_codes)
     }
 }
 
 const RANK_CONSTANT: f32 = 60.0;
 
+async fn add_doc_attribute(reader: &IndexReader, searched_code: &mut CodeSearchResponse) {
+    let schema = IndexSchema::instance();
+    for hit in searched_code.hits.iter_mut() {
+        let query = schema.doc_query(corpus::CODE, &hit.doc.file_id);
+        let doc = reader
+            .searcher()
+            .search(&query, &TopDocs::with_limit(1))
+            .unwrap();
+        if doc.len() == 0 {
+            continue;
+        }
+        let doc = reader.searcher().doc(doc[0].1).unwrap();
+        hit.doc.commit = get_json_text_field_optional(
+            &doc,
+            schema.field_attributes,
+            code::fields::ATTRIBUTE_COMMIT,
+        )
+        .map(|s| s.to_owned());
+    }
+}
+
 fn merge_code_responses_by_rank(
     params: &CodeSearchParams,
     embedding_resp: Vec<(f32, TantivyDocument)>,
@@ -187,12 +208,7 @@ fn create_hit(scores: CodeSearchScores, doc: TantivyDocument) -> CodeSearchHit {
         .to_owned(),
         // commit is introduced in v0.23, but it is also a required field
         // so we need to handle the case where it's not present
-        commit: get_json_text_field_optional(
-            &doc,
-            schema.field_chunk_attributes,
-            code::fields::CHUNK_COMMIT,
-        )
-        .map(|s| s.to_owned()),
+        commit: None,
         language: get_json_text_field(
             &doc,
             schema.field_chunk_attributes,