Merge 074ef26b96 into sapling-pr-archive-bolinfest

This commit is contained in:
Michael Bolin
2025-08-18 17:35:45 -07:00
committed by GitHub
13 changed files with 561 additions and 220 deletions

20
codex-rs/Cargo.lock generated
View File

@@ -176,9 +176,9 @@ dependencies = [
[[package]]
name = "anyhow"
version = "1.0.98"
version = "1.0.99"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e16d2d3311acee920a9eb8d33b8cbc1787ce4a264e85f964c2404b969bdcd487"
checksum = "b0674a1ddeecb70197781e945de4b3b8ffb61fa939a5597bcf48503737663100"
[[package]]
name = "arbitrary"
@@ -538,9 +538,9 @@ checksum = "6e4de3bc4ea267985becf712dc6d9eed8b04c953b3fcfb339ebc87acd9804901"
[[package]]
name = "clap"
version = "4.5.43"
version = "4.5.45"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "50fd97c9dc2399518aa331917ac6f274280ec5eb34e555dd291899745c48ec6f"
checksum = "1fc0e74a703892159f5ae7d3aac52c8e6c392f5ae5f359c70b5881d60aaac318"
dependencies = [
"clap_builder",
"clap_derive",
@@ -548,9 +548,9 @@ dependencies = [
[[package]]
name = "clap_builder"
version = "4.5.43"
version = "4.5.44"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c35b5830294e1fa0462034af85cc95225a4cb07092c088c55bda3147cfcd8f65"
checksum = "b3e7f4214277f3c7aa526a59dd3fbe306a370daee1f8b7b8c987069cd8e888a8"
dependencies = [
"anstream",
"anstyle",
@@ -570,9 +570,9 @@ dependencies = [
[[package]]
name = "clap_derive"
version = "4.5.41"
version = "4.5.45"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ef4f52386a59ca4c860f7393bcf8abd8dfd91ecccc0f774635ff68e92eeef491"
checksum = "14cb31bb0a7d536caef2639baa7fad459e15c3144efefa6dbd1c84562c4739f6"
dependencies = [
"heck",
"proc-macro2",
@@ -2639,9 +2639,9 @@ checksum = "03087c2bad5e1034e8cace5926dec053fb3790248370865f5117a7d0213354c8"
[[package]]
name = "libc"
version = "0.2.174"
version = "0.2.175"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1171693293099992e19cddea4e8b849964e9846f4acee11b3948bcc337be8776"
checksum = "6a82ae493e598baaea5209805c49bbf2ea7de956d50d7da0da1164f9c6d28543"
[[package]]
name = "libfuzzer-sys"

View File

@@ -14,14 +14,14 @@ use std::path::PathBuf;
pub async fn login_with_chatgpt(codex_home: PathBuf) -> std::io::Result<()> {
let opts = ServerOptions::new(codex_home, CLIENT_ID.to_string());
let server = run_login_server(opts, None)?;
let server = run_login_server(opts)?;
eprintln!(
"Starting local login server on http://localhost:{}.\nIf your browser did not open, navigate to this URL to authenticate:\n\n{}",
server.actual_port, server.auth_url,
);
server.block_until_done()
server.block_until_done().await
}
pub async fn run_login_with_chatgpt(cli_config_overrides: CliConfigOverrides) -> ! {

View File

@@ -25,7 +25,7 @@ env-flags = "0.1.1"
eventsource-stream = "0.2.3"
fs2 = "0.4.3"
futures = "0.3"
libc = "0.2.174"
libc = "0.2.175"
mcp-types = { path = "../mcp-types" }
mime_guess = "2.0"
os_info = "3.12.0"

View File

@@ -33,6 +33,7 @@ use crate::error::CodexErr;
use crate::error::Result;
use crate::error::UsageLimitReachedError;
use crate::flags::CODEX_RS_SSE_FIXTURE;
use crate::model_family::ModelFamily;
use crate::model_provider_info::ModelProviderInfo;
use crate::model_provider_info::WireApi;
use crate::models::ResponseItem;
@@ -311,6 +312,30 @@ impl ModelClient {
pub fn get_provider(&self) -> ModelProviderInfo {
self.provider.clone()
}
/// Returns the currently configured model slug.
pub fn get_model(&self) -> String {
self.config.model.clone()
}
/// Returns the currently configured model family.
pub fn get_model_family(&self) -> ModelFamily {
self.config.model_family.clone()
}
/// Returns the current reasoning effort setting.
pub fn get_reasoning_effort(&self) -> ReasoningEffortConfig {
self.effort
}
/// Returns the current reasoning summary setting.
pub fn get_reasoning_summary(&self) -> ReasoningSummaryConfig {
self.summary
}
pub fn get_auth(&self) -> Option<CodexAuth> {
self.auth.clone()
}
}
#[derive(Debug, Deserialize, Serialize)]

View File

@@ -989,7 +989,7 @@ async fn submission_loop(
rx_sub: Receiver<Submission>,
) {
// Wrap once to avoid cloning TurnContext for each task.
let turn_context = Arc::new(turn_context);
let mut turn_context = Arc::new(turn_context);
// To break out of this loop, send Op::Shutdown.
while let Ok(sub) = rx_sub.recv().await {
debug!(?sub, "Submission");
@@ -997,6 +997,83 @@ async fn submission_loop(
Op::Interrupt => {
sess.interrupt_task();
}
Op::OverrideTurnContext {
cwd,
approval_policy,
sandbox_policy,
model,
effort,
summary,
} => {
// Recalculate the persistent turn context with provided overrides.
let prev = Arc::clone(&turn_context);
let provider = prev.client.get_provider();
// Effective model + family
let (effective_model, effective_family) = if let Some(m) = model {
let fam =
find_family_for_model(&m).unwrap_or_else(|| config.model_family.clone());
(m, fam)
} else {
(prev.client.get_model(), prev.client.get_model_family())
};
// Effective reasoning settings
let effective_effort = effort.unwrap_or(prev.client.get_reasoning_effort());
let effective_summary = summary.unwrap_or(prev.client.get_reasoning_summary());
let auth = prev.client.get_auth();
// Build updated config for the client
let mut updated_config = (*config).clone();
updated_config.model = effective_model.clone();
updated_config.model_family = effective_family.clone();
let client = ModelClient::new(
Arc::new(updated_config),
auth,
provider,
effective_effort,
effective_summary,
sess.session_id,
);
let new_approval_policy = approval_policy.unwrap_or(prev.approval_policy);
let new_sandbox_policy = sandbox_policy
.clone()
.unwrap_or(prev.sandbox_policy.clone());
let new_cwd = cwd.clone().unwrap_or_else(|| prev.cwd.clone());
let tools_config = ToolsConfig::new(
&effective_family,
new_approval_policy,
new_sandbox_policy.clone(),
config.include_plan_tool,
config.include_apply_patch_tool,
);
let new_turn_context = TurnContext {
client,
tools_config,
user_instructions: prev.user_instructions.clone(),
base_instructions: prev.base_instructions.clone(),
approval_policy: new_approval_policy,
sandbox_policy: new_sandbox_policy.clone(),
shell_environment_policy: prev.shell_environment_policy.clone(),
cwd: new_cwd.clone(),
disable_response_storage: prev.disable_response_storage,
};
// Install the new persistent context for subsequent tasks/turns.
turn_context = Arc::new(new_turn_context);
if cwd.is_some() || approval_policy.is_some() || sandbox_policy.is_some() {
sess.record_conversation_items(&[ResponseItem::from(EnvironmentContext::new(
new_cwd,
new_approval_policy,
new_sandbox_policy,
))])
.await;
}
}
Op::UserInput { items } => {
// attempt to inject input into current task
if let Err(items) = sess.inject_input(items) {
@@ -1057,7 +1134,7 @@ async fn submission_loop(
cwd,
disable_response_storage: turn_context.disable_response_storage,
};
// TODO: record the new environment context in the conversation history
// no current task, spawn a new one with the perturn context
let task =
AgentTask::spawn(sess.clone(), Arc::new(fresh_turn_context), sub.id, items);

View File

@@ -1,9 +1,13 @@
use codex_core::ConversationManager;
use codex_core::ModelProviderInfo;
use codex_core::built_in_model_providers;
use codex_core::protocol::AskForApproval;
use codex_core::protocol::EventMsg;
use codex_core::protocol::InputItem;
use codex_core::protocol::Op;
use codex_core::protocol::SandboxPolicy;
use codex_core::protocol_config_types::ReasoningEffort;
use codex_core::protocol_config_types::ReasoningSummary;
use codex_login::CodexAuth;
use core_test_support::load_default_config_for_test;
use core_test_support::load_sse_fixture_with_id;
@@ -129,3 +133,230 @@ async fn prefixes_context_and_instructions_once_and_consistently_across_requests
);
assert_eq!(body2["input"], expected_body2);
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn overrides_turn_context_but_keeps_cached_prefix_and_key_constant() {
use pretty_assertions::assert_eq;
let server = MockServer::start().await;
let sse = sse_completed("resp");
let template = ResponseTemplate::new(200)
.insert_header("content-type", "text/event-stream")
.set_body_raw(sse, "text/event-stream");
// Expect two POSTs to /v1/responses
Mock::given(method("POST"))
.and(path("/v1/responses"))
.respond_with(template)
.expect(2)
.mount(&server)
.await;
let model_provider = ModelProviderInfo {
base_url: Some(format!("{}/v1", server.uri())),
..built_in_model_providers()["openai"].clone()
};
let cwd = TempDir::new().unwrap();
let codex_home = TempDir::new().unwrap();
let mut config = load_default_config_for_test(&codex_home);
config.cwd = cwd.path().to_path_buf();
config.model_provider = model_provider;
config.user_instructions = Some("be consistent and helpful".to_string());
let conversation_manager = ConversationManager::default();
let codex = conversation_manager
.new_conversation_with_auth(config, Some(CodexAuth::from_api_key("Test API Key")))
.await
.expect("create new conversation")
.conversation;
// First turn
codex
.submit(Op::UserInput {
items: vec![InputItem::Text {
text: "hello 1".into(),
}],
})
.await
.unwrap();
wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await;
// Change everything about the turn context.
let new_cwd = TempDir::new().unwrap();
let writable = TempDir::new().unwrap();
codex
.submit(Op::OverrideTurnContext {
cwd: Some(new_cwd.path().to_path_buf()),
approval_policy: Some(AskForApproval::Never),
sandbox_policy: Some(SandboxPolicy::WorkspaceWrite {
writable_roots: vec![writable.path().to_path_buf()],
network_access: true,
exclude_tmpdir_env_var: true,
exclude_slash_tmp: true,
}),
model: Some("o3".to_string()),
effort: Some(ReasoningEffort::High),
summary: Some(ReasoningSummary::Detailed),
})
.await
.unwrap();
// Second turn after overrides
codex
.submit(Op::UserInput {
items: vec![InputItem::Text {
text: "hello 2".into(),
}],
})
.await
.unwrap();
wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await;
// Verify we issued exactly two requests, and the cached prefix stayed identical.
let requests = server.received_requests().await.unwrap();
assert_eq!(requests.len(), 2, "expected two POST requests");
let body1 = requests[0].body_json::<serde_json::Value>().unwrap();
let body2 = requests[1].body_json::<serde_json::Value>().unwrap();
// prompt_cache_key should remain constant across overrides
assert_eq!(
body1["prompt_cache_key"], body2["prompt_cache_key"],
"prompt_cache_key should not change across overrides"
);
// The entire prefix from the first request should be identical and reused
// as the prefix of the second request, ensuring cache hit potential.
let expected_user_message_2 = serde_json::json!({
"type": "message",
"id": serde_json::Value::Null,
"role": "user",
"content": [ { "type": "input_text", "text": "hello 2" } ]
});
// After overriding the turn context, the environment context should be emitted again
// reflecting the new cwd, approval policy and sandbox settings.
let expected_env_text_2 = format!(
"<environment_context>\nCurrent working directory: {}\nApproval policy: never\nSandbox mode: workspace-write\nNetwork access: enabled\n</environment_context>",
new_cwd.path().to_string_lossy()
);
let expected_env_msg_2 = serde_json::json!({
"type": "message",
"id": serde_json::Value::Null,
"role": "user",
"content": [ { "type": "input_text", "text": expected_env_text_2 } ]
});
let expected_body2 = serde_json::json!(
[
body1["input"].as_array().unwrap().as_slice(),
[expected_env_msg_2, expected_user_message_2].as_slice(),
]
.concat()
);
assert_eq!(body2["input"], expected_body2);
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn per_turn_overrides_keep_cached_prefix_and_key_constant() {
use pretty_assertions::assert_eq;
let server = MockServer::start().await;
let sse = sse_completed("resp");
let template = ResponseTemplate::new(200)
.insert_header("content-type", "text/event-stream")
.set_body_raw(sse, "text/event-stream");
// Expect two POSTs to /v1/responses
Mock::given(method("POST"))
.and(path("/v1/responses"))
.respond_with(template)
.expect(2)
.mount(&server)
.await;
let model_provider = ModelProviderInfo {
base_url: Some(format!("{}/v1", server.uri())),
..built_in_model_providers()["openai"].clone()
};
let cwd = TempDir::new().unwrap();
let codex_home = TempDir::new().unwrap();
let mut config = load_default_config_for_test(&codex_home);
config.cwd = cwd.path().to_path_buf();
config.model_provider = model_provider;
config.user_instructions = Some("be consistent and helpful".to_string());
let conversation_manager = ConversationManager::default();
let codex = conversation_manager
.new_conversation_with_auth(config, Some(CodexAuth::from_api_key("Test API Key")))
.await
.expect("create new conversation")
.conversation;
// First turn
codex
.submit(Op::UserInput {
items: vec![InputItem::Text {
text: "hello 1".into(),
}],
})
.await
.unwrap();
wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await;
// Second turn using per-turn overrides via UserTurn
let new_cwd = TempDir::new().unwrap();
let writable = TempDir::new().unwrap();
codex
.submit(Op::UserTurn {
items: vec![InputItem::Text {
text: "hello 2".into(),
}],
cwd: new_cwd.path().to_path_buf(),
approval_policy: AskForApproval::Never,
sandbox_policy: SandboxPolicy::WorkspaceWrite {
writable_roots: vec![writable.path().to_path_buf()],
network_access: true,
exclude_tmpdir_env_var: true,
exclude_slash_tmp: true,
},
model: "o3".to_string(),
effort: ReasoningEffort::High,
summary: ReasoningSummary::Detailed,
})
.await
.unwrap();
wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await;
// Verify we issued exactly two requests, and the cached prefix stayed identical.
let requests = server.received_requests().await.unwrap();
assert_eq!(requests.len(), 2, "expected two POST requests");
let body1 = requests[0].body_json::<serde_json::Value>().unwrap();
let body2 = requests[1].body_json::<serde_json::Value>().unwrap();
// prompt_cache_key should remain constant across per-turn overrides
assert_eq!(
body1["prompt_cache_key"], body2["prompt_cache_key"],
"prompt_cache_key should not change across per-turn overrides"
);
// The entire prefix from the first request should be identical and reused
// as the prefix of the second request.
let expected_user_message_2 = serde_json::json!({
"type": "message",
"id": serde_json::Value::Null,
"role": "user",
"content": [ { "type": "input_text", "text": "hello 2" } ]
});
let expected_body2 = serde_json::json!(
[
body1["input"].as_array().unwrap().as_slice(),
[expected_user_message_2].as_slice(),
]
.concat()
);
assert_eq!(body2["input"], expected_body2);
}

View File

@@ -20,7 +20,7 @@ clap = { version = "4", features = ["derive"] }
codex-common = { path = "../common", features = ["cli"] }
codex-core = { path = "../core" }
landlock = "0.4.1"
libc = "0.2.172"
libc = "0.2.175"
seccompiler = "0.5.0"
[target.'cfg(target_os = "linux")'.dev-dependencies]

View File

@@ -3,11 +3,7 @@ use std::io::{self};
use std::path::Path;
use std::path::PathBuf;
use std::sync::Arc;
use std::sync::atomic::AtomicBool;
use std::sync::atomic::Ordering;
use std::sync::mpsc;
use std::thread;
use std::time::Duration;
use crate::AuthDotJson;
use crate::get_auth_file;
@@ -32,7 +28,6 @@ pub struct ServerOptions {
pub port: u16,
pub open_browser: bool,
pub force_state: Option<String>,
pub login_timeout: Option<Duration>,
}
impl ServerOptions {
@@ -44,7 +39,6 @@ impl ServerOptions {
port: DEFAULT_PORT,
open_browser: true,
force_state: None,
login_timeout: None,
}
}
}
@@ -52,52 +46,38 @@ impl ServerOptions {
pub struct LoginServer {
pub auth_url: String,
pub actual_port: u16,
pub server_handle: thread::JoinHandle<io::Result<()>>,
pub shutdown_flag: Arc<AtomicBool>,
pub server: Arc<Server>,
server_handle: tokio::task::JoinHandle<io::Result<()>>,
shutdown_handle: ShutdownHandle,
}
impl LoginServer {
pub fn block_until_done(self) -> io::Result<()> {
#[expect(clippy::expect_used)]
pub async fn block_until_done(self) -> io::Result<()> {
self.server_handle
.join()
.expect("can't join on the server thread")
.await
.map_err(|err| io::Error::other(format!("login server thread panicked: {err:?}")))?
}
pub fn cancel(&self) {
shutdown(&self.shutdown_flag, &self.server);
self.shutdown_handle.shutdown();
}
pub fn cancel_handle(&self) -> ShutdownHandle {
ShutdownHandle {
shutdown_flag: self.shutdown_flag.clone(),
server: self.server.clone(),
}
self.shutdown_handle.clone()
}
}
#[derive(Clone)]
#[derive(Clone, Debug)]
pub struct ShutdownHandle {
shutdown_flag: Arc<AtomicBool>,
server: Arc<Server>,
shutdown_notify: Arc<tokio::sync::Notify>,
}
impl ShutdownHandle {
pub fn cancel(&self) {
shutdown(&self.shutdown_flag, &self.server);
pub fn shutdown(&self) {
self.shutdown_notify.notify_waiters();
}
}
pub fn shutdown(shutdown_flag: &AtomicBool, server: &Server) {
shutdown_flag.store(true, Ordering::SeqCst);
server.unblock();
}
pub fn run_login_server(
opts: ServerOptions,
shutdown_flag: Option<Arc<AtomicBool>>,
) -> io::Result<LoginServer> {
pub fn run_login_server(opts: ServerOptions) -> io::Result<LoginServer> {
let pkce = generate_pkce();
let state = opts.force_state.clone().unwrap_or_else(generate_state);
@@ -119,75 +99,71 @@ pub fn run_login_server(
if opts.open_browser {
let _ = webbrowser::open(&auth_url);
}
let shutdown_flag = shutdown_flag.unwrap_or_else(|| Arc::new(AtomicBool::new(false)));
let shutdown_flag_clone = shutdown_flag.clone();
let timeout_flag = Arc::new(AtomicBool::new(false));
// Channel used to signal completion to timeout watcher.
let (done_tx, done_rx) = mpsc::channel::<()>();
// Map blocking reads from server.recv() to an async channel.
let (tx, mut rx) = tokio::sync::mpsc::channel::<Request>(16);
let _server_handle = {
let server = server.clone();
thread::spawn(move || -> io::Result<()> {
while let Ok(request) = server.recv() {
tx.blocking_send(request).map_err(|e| {
eprintln!("Failed to send request to channel: {e}");
io::Error::other("Failed to send request to channel")
})?;
}
Ok(())
})
};
if let Some(timeout) = opts.login_timeout {
spawn_timeout_watcher(
done_rx,
timeout,
shutdown_flag.clone(),
timeout_flag.clone(),
server.clone(),
);
}
let shutdown_notify = Arc::new(tokio::sync::Notify::new());
let server_handle = {
let shutdown_notify = shutdown_notify.clone();
let server = server.clone();
tokio::spawn(async move {
let result = loop {
tokio::select! {
_ = shutdown_notify.notified() => {
break Err(io::Error::other("Login was not completed"));
}
maybe_req = rx.recv() => {
let Some(req) = maybe_req else {
break Err(io::Error::other("Login was not completed"));
};
let server_for_thread = server.clone();
let server_handle = thread::spawn(move || {
while !shutdown_flag.load(Ordering::SeqCst) {
let req = match server_for_thread.recv() {
Ok(r) => r,
Err(e) => {
// If we've been asked to shut down, break gracefully so that
// we can report timeout or cancellation status uniformly.
if shutdown_flag.load(Ordering::SeqCst) {
break;
} else {
return Err(io::Error::other(e));
let url_raw = req.url().to_string();
let response =
process_request(&url_raw, &opts, &redirect_uri, &pkce, actual_port, &state).await;
let is_login_complete = matches!(response, HandledRequest::ResponseAndExit(_));
match response {
HandledRequest::Response(r) | HandledRequest::ResponseAndExit(r) => {
let _ = tokio::task::spawn_blocking(move || req.respond(r)).await;
}
HandledRequest::RedirectWithHeader(header) => {
let redirect = Response::empty(302).with_header(header);
let _ = tokio::task::spawn_blocking(move || req.respond(redirect)).await;
}
}
if is_login_complete {
break Ok(());
}
}
}
};
let response = process_request(&req, &opts, &redirect_uri, &pkce, actual_port, &state);
let is_login_complete = matches!(response, HandledRequest::ResponseAndExit(_));
match response {
HandledRequest::Response(r) | HandledRequest::ResponseAndExit(r) => {
let _ = req.respond(r);
}
HandledRequest::RedirectWithHeader(header) => {
let redirect = Response::empty(302).with_header(header);
let _ = req.respond(redirect);
}
}
if is_login_complete {
shutdown_flag.store(true, Ordering::SeqCst);
// Login has succeeded, so disarm the timeout watcher.
let _ = done_tx.send(());
return Ok(());
}
}
// Login has failed or timed out, so disarm the timeout watcher.
let _ = done_tx.send(());
if timeout_flag.load(Ordering::SeqCst) {
Err(io::Error::other("Login timed out"))
} else {
Err(io::Error::other("Login was not completed"))
}
});
// Ensure that the server is unblocked so the thread dedicated to
// running `server.recv()` in a loop exits cleanly.
server.unblock();
result
})
};
Ok(LoginServer {
auth_url: auth_url.clone(),
auth_url,
actual_port,
server_handle,
shutdown_flag: shutdown_flag_clone,
server,
shutdown_handle: ShutdownHandle { shutdown_notify },
})
}
@@ -197,15 +173,14 @@ enum HandledRequest {
ResponseAndExit(Response<Cursor<Vec<u8>>>),
}
fn process_request(
req: &Request,
async fn process_request(
url_raw: &str,
opts: &ServerOptions,
redirect_uri: &str,
pkce: &PkceCodes,
actual_port: u16,
state: &str,
) -> HandledRequest {
let url_raw = req.url().to_string();
let parsed_url = match url::Url::parse(&format!("http://localhost{url_raw}")) {
Ok(u) => u,
Err(e) => {
@@ -236,18 +211,22 @@ fn process_request(
};
match exchange_code_for_tokens(&opts.issuer, &opts.client_id, redirect_uri, pkce, &code)
.await
{
Ok(tokens) => {
// Obtain API key via token-exchange and persist
let api_key =
obtain_api_key(&opts.issuer, &opts.client_id, &tokens.id_token).ok();
if let Err(err) = persist_tokens(
let api_key = obtain_api_key(&opts.issuer, &opts.client_id, &tokens.id_token)
.await
.ok();
if let Err(err) = persist_tokens_async(
&opts.codex_home,
api_key.clone(),
tokens.id_token.clone(),
Some(tokens.access_token.clone()),
Some(tokens.refresh_token.clone()),
) {
)
.await
{
eprintln!("Persist error: {err}");
return HandledRequest::Response(
Response::from_string(format!("Unable to persist auth file: {err}"))
@@ -292,29 +271,6 @@ fn process_request(
}
}
/// Spawns a detached thread that waits for either a completion signal on `done_rx`
/// or the specified `timeout` to elapse. If the timeout elapses first it marks
/// the `shutdown_flag`, records `timeout_flag`, and unblocks the HTTP server so
/// that the main server loop can exit promptly.
fn spawn_timeout_watcher(
done_rx: mpsc::Receiver<()>,
timeout: Duration,
shutdown_flag: Arc<AtomicBool>,
timeout_flag: Arc<AtomicBool>,
server: Arc<Server>,
) {
thread::spawn(move || {
if done_rx.recv_timeout(timeout).is_err()
&& shutdown_flag
.compare_exchange(false, true, Ordering::SeqCst, Ordering::SeqCst)
.is_ok()
{
timeout_flag.store(true, Ordering::SeqCst);
server.unblock();
}
});
}
fn build_authorize_url(
issuer: &str,
client_id: &str,
@@ -353,7 +309,7 @@ struct ExchangedTokens {
refresh_token: String,
}
fn exchange_code_for_tokens(
async fn exchange_code_for_tokens(
issuer: &str,
client_id: &str,
redirect_uri: &str,
@@ -367,7 +323,7 @@ fn exchange_code_for_tokens(
refresh_token: String,
}
let client = reqwest::blocking::Client::new();
let client = reqwest::Client::new();
let resp = client
.post(format!("{issuer}/oauth/token"))
.header("Content-Type", "application/x-www-form-urlencoded")
@@ -379,6 +335,7 @@ fn exchange_code_for_tokens(
urlencoding::encode(&pkce.code_verifier)
))
.send()
.await
.map_err(io::Error::other)?;
if !resp.status().is_success() {
@@ -388,7 +345,7 @@ fn exchange_code_for_tokens(
)));
}
let tokens: TokenResponse = resp.json().map_err(io::Error::other)?;
let tokens: TokenResponse = resp.json().await.map_err(io::Error::other)?;
Ok(ExchangedTokens {
id_token: tokens.id_token,
access_token: tokens.access_token,
@@ -396,43 +353,49 @@ fn exchange_code_for_tokens(
})
}
fn persist_tokens(
async fn persist_tokens_async(
codex_home: &Path,
api_key: Option<String>,
id_token: String,
access_token: Option<String>,
refresh_token: Option<String>,
) -> io::Result<()> {
let auth_file = get_auth_file(codex_home);
if let Some(parent) = auth_file.parent() {
if !parent.exists() {
std::fs::create_dir_all(parent).map_err(io::Error::other)?;
// Reuse existing synchronous logic but run it off the async runtime.
let codex_home = codex_home.to_path_buf();
tokio::task::spawn_blocking(move || {
let auth_file = get_auth_file(&codex_home);
if let Some(parent) = auth_file.parent() {
if !parent.exists() {
std::fs::create_dir_all(parent).map_err(io::Error::other)?;
}
}
}
let mut auth = read_or_default(&auth_file);
if let Some(key) = api_key {
auth.openai_api_key = Some(key);
}
let tokens = auth
.tokens
.get_or_insert_with(crate::token_data::TokenData::default);
tokens.id_token = crate::token_data::parse_id_token(&id_token).map_err(io::Error::other)?;
// Persist chatgpt_account_id if present in claims
if let Some(acc) = jwt_auth_claims(&id_token)
.get("chatgpt_account_id")
.and_then(|v| v.as_str())
{
tokens.account_id = Some(acc.to_string());
}
if let Some(at) = access_token {
tokens.access_token = at;
}
if let Some(rt) = refresh_token {
tokens.refresh_token = rt;
}
auth.last_refresh = Some(Utc::now());
super::write_auth_json(&auth_file, &auth)
let mut auth = read_or_default(&auth_file);
if let Some(key) = api_key {
auth.openai_api_key = Some(key);
}
let tokens = auth
.tokens
.get_or_insert_with(crate::token_data::TokenData::default);
tokens.id_token = crate::token_data::parse_id_token(&id_token).map_err(io::Error::other)?;
// Persist chatgpt_account_id if present in claims
if let Some(acc) = jwt_auth_claims(&id_token)
.get("chatgpt_account_id")
.and_then(|v| v.as_str())
{
tokens.account_id = Some(acc.to_string());
}
if let Some(at) = access_token {
tokens.access_token = at;
}
if let Some(rt) = refresh_token {
tokens.refresh_token = rt;
}
auth.last_refresh = Some(Utc::now());
super::write_auth_json(&auth_file, &auth)
})
.await
.map_err(|e| io::Error::other(format!("persist task failed: {e}")))?
}
fn read_or_default(path: &Path) -> AuthDotJson {
@@ -525,13 +488,13 @@ fn jwt_auth_claims(jwt: &str) -> serde_json::Map<String, serde_json::Value> {
serde_json::Map::new()
}
fn obtain_api_key(issuer: &str, client_id: &str, id_token: &str) -> io::Result<String> {
async fn obtain_api_key(issuer: &str, client_id: &str, id_token: &str) -> io::Result<String> {
// Token exchange for an API key access token
#[derive(serde::Deserialize)]
struct ExchangeResp {
access_token: String,
}
let client = reqwest::blocking::Client::new();
let client = reqwest::Client::new();
let resp = client
.post(format!("{issuer}/oauth/token"))
.header("Content-Type", "application/x-www-form-urlencoded")
@@ -544,6 +507,7 @@ fn obtain_api_key(issuer: &str, client_id: &str, id_token: &str) -> io::Result<S
urlencoding::encode("urn:ietf:params:oauth:token-type:id_token")
))
.send()
.await
.map_err(io::Error::other)?;
if !resp.status().is_success() {
return Err(io::Error::other(format!(
@@ -551,6 +515,6 @@ fn obtain_api_key(issuer: &str, client_id: &str, id_token: &str) -> io::Result<S
resp.status()
)));
}
let body: ExchangeResp = resp.json().map_err(io::Error::other)?;
let body: ExchangeResp = resp.json().await.map_err(io::Error::other)?;
Ok(body.access_token)
}

View File

@@ -73,8 +73,8 @@ fn start_mock_issuer() -> (SocketAddr, thread::JoinHandle<()>) {
(addr, handle)
}
#[test]
fn end_to_end_login_flow_persists_auth_json() {
#[tokio::test]
async fn end_to_end_login_flow_persists_auth_json() {
if std::env::var(CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR).is_ok() {
println!(
"Skipping test because it cannot execute when network is disabled in a Codex sandbox."
@@ -100,22 +100,21 @@ fn end_to_end_login_flow_persists_auth_json() {
port: 0,
open_browser: false,
force_state: Some(state),
login_timeout: None,
};
let server = run_login_server(opts, None).unwrap();
let server = run_login_server(opts).unwrap();
let login_port = server.actual_port;
// Simulate browser callback, and follow redirect to /success
let client = reqwest::blocking::Client::builder()
let client = reqwest::Client::builder()
.redirect(reqwest::redirect::Policy::limited(5))
.build()
.unwrap();
let url = format!("http://127.0.0.1:{login_port}/auth/callback?code=abc&state=test_state_123");
let resp = client.get(&url).send().unwrap();
let resp = client.get(&url).send().await.unwrap();
assert!(resp.status().is_success());
// Wait for server shutdown
server.block_until_done().unwrap();
server.block_until_done().await.unwrap();
// Validate auth.json
let auth_path = codex_home.join("auth.json");
@@ -133,8 +132,8 @@ fn end_to_end_login_flow_persists_auth_json() {
drop(issuer_handle);
}
#[test]
fn creates_missing_codex_home_dir() {
#[tokio::test]
async fn creates_missing_codex_home_dir() {
if std::env::var(CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR).is_ok() {
println!(
"Skipping test because it cannot execute when network is disabled in a Codex sandbox."
@@ -159,17 +158,16 @@ fn creates_missing_codex_home_dir() {
port: 0,
open_browser: false,
force_state: Some(state),
login_timeout: None,
};
let server = run_login_server(opts, None).unwrap();
let server = run_login_server(opts).unwrap();
let login_port = server.actual_port;
let client = reqwest::blocking::Client::new();
let client = reqwest::Client::new();
let url = format!("http://127.0.0.1:{login_port}/auth/callback?code=abc&state=state2");
let resp = client.get(&url).send().unwrap();
let resp = client.get(&url).send().await.unwrap();
assert!(resp.status().is_success());
server.block_until_done().unwrap();
server.block_until_done().await.unwrap();
let auth_path = codex_home.join("auth.json");
assert!(

View File

@@ -66,7 +66,7 @@ struct ActiveLogin {
impl ActiveLogin {
fn drop(&self) {
self.shutdown_handle.cancel();
self.shutdown_handle.shutdown();
}
}
@@ -146,7 +146,6 @@ impl CodexMessageProcessor {
let opts = LoginServerOptions {
open_browser: false,
login_timeout: Some(LOGIN_CHATGPT_TIMEOUT),
..LoginServerOptions::new(config.codex_home.clone(), CLIENT_ID.to_string())
};
@@ -155,9 +154,10 @@ impl CodexMessageProcessor {
Error(JSONRPCErrorError),
}
let reply = match run_login_server(opts, None) {
let reply = match run_login_server(opts) {
Ok(server) => {
let login_id = Uuid::new_v4();
let shutdown_handle = server.cancel_handle();
// Replace active login if present.
{
@@ -166,7 +166,7 @@ impl CodexMessageProcessor {
existing.drop();
}
*guard = Some(ActiveLogin {
shutdown_handle: server.cancel_handle(),
shutdown_handle: shutdown_handle.clone(),
login_id,
});
}
@@ -180,15 +180,19 @@ impl CodexMessageProcessor {
let outgoing_clone = self.outgoing.clone();
let active_login = self.active_login.clone();
tokio::spawn(async move {
let result =
tokio::task::spawn_blocking(move || server.block_until_done()).await;
let (success, error_msg) = match result {
let (success, error_msg) = match tokio::time::timeout(
LOGIN_CHATGPT_TIMEOUT,
server.block_until_done(),
)
.await
{
Ok(Ok(())) => (true, None),
Ok(Err(err)) => (false, Some(format!("Login server error: {err}"))),
Err(join_err) => (
false,
Some(format!("failed to join login server thread: {join_err}")),
),
Err(_elapsed) => {
// Timeout: cancel server and report
shutdown_handle.shutdown();
(false, Some("Login timed out".to_string()))
}
};
let notification = LoginChatGptCompleteNotification {
login_id,

View File

@@ -76,6 +76,38 @@ pub enum Op {
summary: ReasoningSummaryConfig,
},
/// Override parts of the persistent turn context for subsequent turns.
///
/// All fields are optional; when omitted, the existing value is preserved.
/// This does not enqueue any input it only updates defaults used for
/// future `UserInput` turns.
OverrideTurnContext {
/// Updated `cwd` for sandbox/tool calls.
#[serde(skip_serializing_if = "Option::is_none")]
cwd: Option<PathBuf>,
/// Updated command approval policy.
#[serde(skip_serializing_if = "Option::is_none")]
approval_policy: Option<AskForApproval>,
/// Updated sandbox policy for tool calls.
#[serde(skip_serializing_if = "Option::is_none")]
sandbox_policy: Option<SandboxPolicy>,
/// Updated model slug. When set, the model family is derived
/// automatically.
#[serde(skip_serializing_if = "Option::is_none")]
model: Option<String>,
/// Updated reasoning effort (honored only for reasoning-capable models).
#[serde(skip_serializing_if = "Option::is_none")]
effort: Option<ReasoningEffortConfig>,
/// Updated reasoning summary preference (honored only for reasoning-capable models).
#[serde(skip_serializing_if = "Option::is_none")]
summary: Option<ReasoningSummaryConfig>,
},
/// Approve a command execution
ExecApproval {
/// The id of the submission we are approving

View File

@@ -230,6 +230,11 @@ impl TextArea {
code: KeyCode::Backspace,
modifiers: KeyModifiers::NONE,
..
}
| KeyEvent {
code: KeyCode::Char('h'),
modifiers: KeyModifiers::CONTROL,
..
} => self.delete_backward(1),
KeyEvent {
code: KeyCode::Delete,

View File

@@ -1,5 +1,6 @@
use codex_login::CLIENT_ID;
use codex_login::ServerOptions;
use codex_login::ShutdownHandle;
use codex_login::run_login_server;
use crossterm::event::KeyCode;
use crossterm::event::KeyEvent;
@@ -24,10 +25,6 @@ use crate::onboarding::onboarding_screen::KeyboardHandler;
use crate::onboarding::onboarding_screen::StepStateProvider;
use crate::shimmer::shimmer_spans;
use std::path::PathBuf;
use std::sync::Arc;
use std::sync::atomic::AtomicBool;
use std::sync::atomic::Ordering;
use std::thread::JoinHandle;
use super::onboarding_screen::StepState;
// no additional imports
@@ -46,13 +43,14 @@ pub(crate) enum SignInState {
/// Used to manage the lifecycle of SpawnedLogin and ensure it gets cleaned up.
pub(crate) struct ContinueInBrowserState {
auth_url: String,
shutdown_flag: Option<Arc<AtomicBool>>,
_login_wait_handle: Option<JoinHandle<()>>,
shutdown_handle: Option<ShutdownHandle>,
_login_wait_handle: Option<tokio::task::JoinHandle<()>>,
}
impl Drop for ContinueInBrowserState {
fn drop(&mut self) {
if let Some(flag) = &self.shutdown_flag {
flag.store(true, Ordering::SeqCst);
if let Some(flag) = &self.shutdown_handle {
flag.shutdown();
}
}
}
@@ -283,16 +281,21 @@ impl AuthModeWidget {
fn start_chatgpt_login(&mut self) {
self.error = None;
let opts = ServerOptions::new(self.codex_home.clone(), CLIENT_ID.to_string());
let server = run_login_server(opts, None);
let server = run_login_server(opts);
match server {
Ok(child) => {
let auth_url = child.auth_url.clone();
let shutdown_flag = child.shutdown_flag.clone();
let shutdown_handle = child.cancel_handle();
let event_tx = self.event_tx.clone();
let join_handle = tokio::spawn(async move {
spawn_completion_poller(child, event_tx).await;
});
self.sign_in_state =
SignInState::ChatGptContinueInBrowser(ContinueInBrowserState {
auth_url,
shutdown_flag: Some(shutdown_flag),
_login_wait_handle: Some(self.spawn_completion_poller(child)),
shutdown_handle: Some(shutdown_handle),
_login_wait_handle: Some(join_handle),
});
self.event_tx.send(AppEvent::RequestRedraw);
}
@@ -313,19 +316,21 @@ impl AuthModeWidget {
}
self.event_tx.send(AppEvent::RequestRedraw);
}
}
fn spawn_completion_poller(&self, child: codex_login::LoginServer) -> JoinHandle<()> {
let event_tx = self.event_tx.clone();
std::thread::spawn(move || {
if let Ok(()) = child.block_until_done() {
event_tx.send(AppEvent::OnboardingAuthComplete(Ok(())));
} else {
event_tx.send(AppEvent::OnboardingAuthComplete(Err(
"login failed".to_string()
)));
}
})
}
async fn spawn_completion_poller(
child: codex_login::LoginServer,
event_tx: AppEventSender,
) -> tokio::task::JoinHandle<()> {
tokio::spawn(async move {
if let Ok(()) = child.block_until_done().await {
event_tx.send(AppEvent::OnboardingAuthComplete(Ok(())));
} else {
event_tx.send(AppEvent::OnboardingAuthComplete(Err(
"login failed".to_string()
)));
}
})
}
impl StepStateProvider for AuthModeWidget {