edit prediction cli: Skip worktree scan (#44658)

Release Notes:

- N/A

Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>

crates/edit_prediction_cli/src/load_project.rs

@@ -11,7 +11,7 @@ use futures::{
     lock::{Mutex, OwnedMutexGuard},
 };
 use gpui::{AsyncApp, Entity};
-use language::{Anchor, Buffer, ToOffset, ToPoint};
+use language::{Anchor, Buffer, LanguageNotFound, ToOffset, ToPoint};
 use project::buffer_store::BufferStoreEvent;
 use project::{Project, ProjectPath};
 use std::{
@@ -77,6 +77,19 @@ async fn cursor_position(
     project: &Entity<Project>,
     cx: &mut AsyncApp,
 ) -> (Entity<Buffer>, Anchor) {
+    let language_registry = project
+        .read_with(cx, |project, _| project.languages().clone())
+        .unwrap();
+    let result = language_registry
+        .load_language_for_file_path(&example.cursor_path)
+        .await;
+
+    if let Err(error) = result
+        && !error.is::<LanguageNotFound>()
+    {
+        panic!("Failed to load language for file path: {}", error);
+    }
+
     let worktree = project
         .read_with(cx, |project, cx| {
             project.visible_worktrees(cx).next().unwrap()
@@ -115,7 +128,8 @@ async fn cursor_position(
         let mut matches = text.match_indices(&cursor_excerpt);
         let (excerpt_offset, _) = matches.next().unwrap_or_else(|| {
             panic!(
-                "\nExcerpt:\n\n{cursor_excerpt}\nBuffer text:\n{text}\n.Cursor excerpt did not exist in buffer."
+                "\nExcerpt:\n\n{cursor_excerpt}\nBuffer text:\n{text}\n.Example: {}\nCursor excerpt did not exist in buffer.",
+                example.name
             );
         });
         assert!(matches.next().is_none(), "More than one cursor position match found for {}", &example.name);
@@ -151,6 +165,12 @@ async fn setup_project(
         })
         .unwrap();
 
+    project
+        .update(cx, |project, cx| {
+            project.disable_worktree_scanner(cx);
+        })
+        .unwrap();
+
     let worktree = project
         .update(cx, |project, cx| {
             project.create_worktree(&example.worktree_path(), true, cx)

crates/edit_prediction_cli/src/main.rs

@@ -45,7 +45,7 @@ enum Command {
     /// Parse markdown examples and output a combined .jsonl file
     ParseExample,
     /// Create git worktrees for each example and load file contents
-    LoadBuffer,
+    LoadProject,
     /// Retrieve context for input examples.
     Context,
     /// Generate a prompt string for a specific model
@@ -144,15 +144,19 @@ fn main() {
                 _ => (),
             };
 
-            for data in examples.chunks_mut(args.max_parallelism) {
+            let chunks = examples.chunks_mut(args.max_parallelism);
+            let total_chunks = chunks.len();
+            for (batch_ix, data) in chunks.enumerate() {
                 let mut futures = Vec::new();
+                eprintln!("Processing batch: {}/{}", batch_ix + 1, total_chunks);
+
                 for example in data.iter_mut() {
                     let cx = cx.clone();
                     let app_state = app_state.clone();
                     futures.push(async {
                         match &command {
                             Command::ParseExample => {}
-                            Command::LoadBuffer => {
+                            Command::LoadProject => {
                                 run_load_project(example, app_state.clone(), cx).await;
                             }
                             Command::Context => {

crates/edit_prediction_cli/src/retrieve_context.rs

@@ -8,7 +8,7 @@ use collections::HashSet;
 use edit_prediction::{DebugEvent, EditPredictionStore};
 use futures::{FutureExt as _, StreamExt as _, channel::mpsc};
 use gpui::{AsyncApp, Entity, Task};
-use language::{Buffer, LanguageNotFound};
+use language::Buffer;
 use project::Project;
 use std::{sync::Arc, time::Duration};
 
@@ -71,19 +71,6 @@ async fn wait_for_language_server_to_start(
     buffer: &Entity<Buffer>,
     cx: &mut AsyncApp,
 ) {
-    let language_registry = project
-        .read_with(cx, |project, _| project.languages().clone())
-        .unwrap();
-    let result = language_registry
-        .load_language_for_file_path(&example.cursor_path)
-        .await;
-
-    if let Err(error) = result
-        && !error.is::<LanguageNotFound>()
-    {
-        panic!("Failed to load language for file path: {}", error);
-    }
-
     let Some(language_id) = buffer
         .read_with(cx, |buffer, _cx| {
             buffer.language().map(|language| language.id())