Warn on empty result for single file answer contexts (#704)

Author: calyptobai

server/bleep/src/indexes/file.rs

@@ -156,7 +156,7 @@ impl Indexable for File {
 }
 
 impl Indexer<File> {
-    pub async fn file_body(&self, file_disk_path: &str) -> Result<ContentDocument> {
+    pub async fn file_body(&self, file_disk_path: &str) -> Result<Option<ContentDocument>> {
         // Mostly taken from `by_path`, below.
         //
         // TODO: This can be unified with `by_path` below, but we first need to decide on a unified
@@ -274,7 +274,7 @@ impl Indexer<File> {
         &self,
         repo_ref: &RepoRef,
         relative_path: &str,
-    ) -> Result<ContentDocument> {
+    ) -> Result<Option<ContentDocument>> {
         let reader = self.reader.read().await;
         let searcher = reader.searcher();
 
@@ -300,7 +300,7 @@ impl Indexer<File> {
         &self,
         query: Box<dyn tantivy::query::Query>,
         searcher: tantivy::Searcher,
-    ) -> Result<ContentDocument> {
+    ) -> Result<Option<ContentDocument>> {
         let file_source = &self.source;
 
         let collector = TopDocs::with_limit(1);
@@ -310,14 +310,16 @@ impl Indexer<File> {
 
         match search_results.as_slice() {
             // no paths matched, the input path was not well formed
-            [] => Err(anyhow::Error::msg("no path found")),
+            [] => Ok(None),
 
             // exactly one path, good
             [(_, doc_addr)] => {
                 let retrieved_doc = searcher
                     .doc(*doc_addr)
                     .expect("failed to get document by address");
-                Ok(ContentReader.read_document(file_source, retrieved_doc))
+                Ok(Some(
+                    ContentReader.read_document(file_source, retrieved_doc),
+                ))
             }
 
             // more than one path matched, this can occur when top docs is no

server/bleep/src/webserver/answer.rs

@@ -475,20 +475,6 @@ impl Conversation {
                     paths = semantic_paths;
                 }
 
-                let prompt = Some("§alias, path".to_owned())
-                    .into_iter()
-                    .chain(paths.iter().map(|p| format!("{}, {p}", self.path_alias(p))))
-                    .collect::<Vec<_>>()
-                    .join("\n");
-
-                ctx.track_query(
-                    EventData::input_stage("path search")
-                        .with_payload("query", query)
-                        .with_payload("is_semantic", is_semantic)
-                        .with_payload("results", &paths)
-                        .with_payload("raw_prompt", prompt),
-                );
-
                 let formatted_paths = paths
                     .iter()
                     .map(|p| SeenPath {
@@ -497,7 +483,17 @@ impl Conversation {
                     })
                     .collect::<Vec<_>>();
 
-                serde_json::to_string(&formatted_paths).unwrap()
+                let prompt = serde_json::to_string(&formatted_paths).unwrap();
+
+                ctx.track_query(
+                    EventData::input_stage("path search")
+                        .with_payload("query", query)
+                        .with_payload("is_semantic", is_semantic)
+                        .with_payload("results", &paths)
+                        .with_payload("raw_prompt", &prompt),
+                );
+
+                prompt
             }
 
             Action::Code { query } => {
@@ -665,6 +661,7 @@ impl Conversation {
                     .by_path(repo_ref, &path)
                     .await
                     .with_context(|| format!("failed to read path: {path}"))?
+                    .with_context(|| format!("path does not exist in the index: {path}"))?
                     .content
                     .lines()
                     .enumerate()
@@ -878,27 +875,34 @@ impl Conversation {
             let alias = path_aliases[0];
             let path = self.paths[alias].clone();
 
-            let file_contents = ctx
+            let doc = ctx
                 .app
                 .indexes
                 .file
                 .by_path(&self.repo_ref, &path)
                 .await
-                .with_context(|| format!("failed to read path: {}", path))?
-                .content;
-
-            let bpe =
-                tiktoken_rs::get_bpe_from_model("gpt-4").context("invalid model requested")?;
-
-            let trimmed_file_contents = limit_tokens(&file_contents, bpe, 4000);
-
-            vec![CodeChunk {
-                alias: alias as u32,
-                path,
-                start_line: 1,
-                end_line: trimmed_file_contents.lines().count() as u32 + 1,
-                snippet: trimmed_file_contents.to_owned(),
-            }]
+                .with_context(|| format!("failed to read path: {}", path))?;
+
+            match doc {
+                Some(doc) => {
+                    let bpe = tiktoken_rs::get_bpe_from_model("gpt-4")
+                        .context("invalid model requested")?;
+
+                    let trimmed_file_contents = limit_tokens(&doc.content, bpe, 4000);
+
+                    vec![CodeChunk {
+                        alias: alias as u32,
+                        path,
+                        start_line: 1,
+                        end_line: trimmed_file_contents.lines().count() as u32 + 1,
+                        snippet: trimmed_file_contents.to_owned(),
+                    }]
+                }
+                None => {
+                    warn!("only path alias did not return any results");
+                    vec![]
+                }
+            }
         } else {
             self.code_chunks
                 .iter()
@@ -1291,6 +1295,7 @@ impl Conversation {
                     .by_path(repo_ref, &path)
                     .await
                     .unwrap()
+                    .unwrap_or_else(|| panic!("path did not exist in the index: {path}"))
                     .content;
 
                 chunks

server/bleep/src/webserver/file.rs

@@ -38,7 +38,8 @@ pub(super) async fn handle<'a>(
             params.path.to_str().context("invalid file path")?,
         )
         .await
-        .map_err(Error::internal)?;
+        .map_err(Error::internal)?
+        .ok_or_else(|| Error::user("file not found").with_status(StatusCode::NOT_FOUND))?;
 
     Ok(json(FileResponse {
         contents: split_by_lines(&doc.content, &doc.line_end_indices, &params)?.to_string(),

server/bleep/src/webserver/hoverable.rs

@@ -31,7 +31,8 @@ pub(super) async fn handle(
     let repo_ref = &payload.repo_ref.parse::<RepoRef>().map_err(Error::user)?;
 
     let document = match indexes.file.by_path(repo_ref, &payload.relative_path).await {
-        Ok(doc) => doc,
+        Ok(Some(doc)) => doc,
+        Ok(None) => return Err(Error::user("file not found").with_status(StatusCode::NOT_FOUND)),
         Err(e) => return Err(Error::user(e)),
     };
 

server/bleep/src/webserver/intelligence.rs

@@ -59,7 +59,8 @@ pub(super) async fn handle(
         .file
         .by_path(&repo_ref, &payload.relative_path)
         .await
-        .map_err(Error::user)?;
+        .map_err(Error::user)?
+        .ok_or_else(|| Error::user("path not found").with_status(StatusCode::NOT_FOUND))?;
     let lang = source_document.lang.as_deref();
     let all_docs = {
         let associated_langs = match lang.map(TSLanguage::from_id) {