fix(indexer): fan out when index code chunks

crates/tabby-index/src/indexer.rs

@@ -1,4 +1,4 @@
-use std::collections::HashSet;
+use std::{collections::HashSet, sync::mpsc::sync_channel};
 
 use anyhow::{bail, Result};
 use async_stream::stream;
@@ -76,35 +76,49 @@ impl<T: ToIndexId> TantivyDocBuilder<T> {
         let doc_id = id.clone();
         let doc_attributes = self.builder.build_attributes(&document).await;
         let s = stream! {
-            let mut failed_count: u64 = 0;
+            let (tx, rx) = sync_channel(32);
+
             for await chunk_doc in self.build_chunks(cloned_id, source_id.clone(), updated_at, document).await {
-                match chunk_doc.await {
-                    Ok(Ok(doc)) => {
-                        yield tokio::spawn(async move { Some(doc) });
-                    }
-                    Ok(Err(e)) => {
-                        warn!("Failed to build chunk for document '{}': {}", doc_id, e);
-                        failed_count += 1;
-                    }
-                    Err(e) => {
-                        warn!("Failed to call build chunk '{}': {}", doc_id, e);
-                        failed_count += 1;
+                let tx = tx.clone();
+                let doc_id = doc_id.clone();
+                yield tokio::spawn(async move {
+                    match chunk_doc.await {
+                        Ok(Ok(doc)) => {
+                           Some(doc)
+                        }
+                        Ok(Err(e)) => {
+                            warn!("Failed to build chunk for document '{}': {}", doc_id, e);
+                            tx.send(1).unwrap();
+                            None
+                        }
+                        Err(e) => {
+                            warn!("Failed to call build chunk '{}': {}", doc_id, e);
+                            tx.send(1).unwrap();
+                            None
+                        }
                     }
-                }
+                });
             };
 
+            // drop tx to signal the end of the stream
+            // the cloned is dropped in its own thread
+            drop(tx);
+
             let mut doc = doc! {
                 schema.field_id => doc_id,
                 schema.field_source_id => source_id,
                 schema.field_corpus => self.corpus,
                 schema.field_attributes => doc_attributes,
                 schema.field_updated_at => updated_at,
             };
-            if failed_count > 0 {
-                doc.add_u64(schema.field_failed_chunks_count, failed_count);
-            }
 
-            yield tokio::spawn(async move { Some(doc) });
+            yield tokio::spawn(async move {
+                let failed_count = rx.iter().count();
+                if failed_count > 0 {
+                    doc.add_u64(schema.field_failed_chunks_count, failed_count as u64);
+                }
+                Some(doc)
+             });
         };
 
         (id, s)

crates/tabby-index/src/indexer_tests.rs

@@ -195,14 +195,18 @@ mod builder_tests {
                 std::thread::available_parallelism().unwrap().get() * 2,
                 32,
             ))
+            .map(|handler| handler.unwrap())
             .collect::<Vec<_>>()
             .await
+            .into_iter()
+            .flatten()
+            .collect::<Vec<_>>()
         });
 
         // the chunks should be failed as no embedding is provided
         // the last element is the document itself
         assert_eq!(res.len(), 1);
-        let doc = res.last().unwrap().as_ref().unwrap().as_ref().unwrap();
+        let doc = res.last().unwrap();
 
         let schema = IndexSchema::instance();
         let failed_count = doc
@@ -250,16 +254,20 @@ mod builder_tests {
                 std::thread::available_parallelism().unwrap().get() * 2,
                 32,
             ))
+            .map(|handler| handler.unwrap())
             .collect::<Vec<_>>()
             .await
+            .into_iter()
+            .flatten()
+            .collect::<Vec<_>>()
         });
 
         // The last element is the document itself,
         // while the preceding elements are the chunks.
         // Given that the embedding is empty,
         // all chunks should be considered failed and skipped.
         assert_eq!(res.len(), 1);
-        let doc = res.last().unwrap().as_ref().unwrap().as_ref().unwrap();
+        let doc = res.last().unwrap();
 
         let schema = IndexSchema::instance();
         let failed_count = doc