mirror of
https://github.com/openai/codex.git
synced 2026-11-15 21:50:46 +00:00
Merge branch 'main' into gverma/skills-data-model
This commit is contained in:
@@ -0,0 +1,154 @@
|
||||
use chrono::DateTime;
|
||||
use chrono::Utc;
|
||||
use codex_api::AuthProvider as ApiAuthProvider;
|
||||
use codex_api::TransportError;
|
||||
use codex_api::error::ApiError;
|
||||
use codex_api::rate_limits::parse_rate_limit;
|
||||
use http::HeaderMap;
|
||||
use serde::Deserialize;
|
||||
|
||||
use crate::auth::CodexAuth;
|
||||
use crate::error::CodexErr;
|
||||
use crate::error::RetryLimitReachedError;
|
||||
use crate::error::UnexpectedResponseError;
|
||||
use crate::error::UsageLimitReachedError;
|
||||
use crate::model_provider_info::ModelProviderInfo;
|
||||
use crate::token_data::PlanType;
|
||||
|
||||
pub(crate) fn map_api_error(err: ApiError) -> CodexErr {
|
||||
match err {
|
||||
ApiError::ContextWindowExceeded => CodexErr::ContextWindowExceeded,
|
||||
ApiError::QuotaExceeded => CodexErr::QuotaExceeded,
|
||||
ApiError::UsageNotIncluded => CodexErr::UsageNotIncluded,
|
||||
ApiError::Retryable { message, delay } => CodexErr::Stream(message, delay),
|
||||
ApiError::Stream(msg) => CodexErr::Stream(msg, None),
|
||||
ApiError::Api { status, message } => CodexErr::UnexpectedStatus(UnexpectedResponseError {
|
||||
status,
|
||||
body: message,
|
||||
request_id: None,
|
||||
}),
|
||||
ApiError::Transport(transport) => match transport {
|
||||
TransportError::Http {
|
||||
status,
|
||||
headers,
|
||||
body,
|
||||
} => {
|
||||
if status == http::StatusCode::INTERNAL_SERVER_ERROR {
|
||||
CodexErr::InternalServerError
|
||||
} else if status == http::StatusCode::TOO_MANY_REQUESTS {
|
||||
if let Some(body) = body
|
||||
&& let Ok(err) = serde_json::from_str::<UsageErrorResponse>(&body)
|
||||
{
|
||||
if err.error.error_type.as_deref() == Some("usage_limit_reached") {
|
||||
let rate_limits = headers.as_ref().and_then(parse_rate_limit);
|
||||
let resets_at = err
|
||||
.error
|
||||
.resets_at
|
||||
.and_then(|seconds| DateTime::<Utc>::from_timestamp(seconds, 0));
|
||||
return CodexErr::UsageLimitReached(UsageLimitReachedError {
|
||||
plan_type: err.error.plan_type,
|
||||
resets_at,
|
||||
rate_limits,
|
||||
});
|
||||
} else if err.error.error_type.as_deref() == Some("usage_not_included") {
|
||||
return CodexErr::UsageNotIncluded;
|
||||
}
|
||||
}
|
||||
|
||||
CodexErr::RetryLimit(RetryLimitReachedError {
|
||||
status,
|
||||
request_id: extract_request_id(headers.as_ref()),
|
||||
})
|
||||
} else {
|
||||
CodexErr::UnexpectedStatus(UnexpectedResponseError {
|
||||
status,
|
||||
body: body.unwrap_or_default(),
|
||||
request_id: extract_request_id(headers.as_ref()),
|
||||
})
|
||||
}
|
||||
}
|
||||
TransportError::RetryLimit => CodexErr::RetryLimit(RetryLimitReachedError {
|
||||
status: http::StatusCode::INTERNAL_SERVER_ERROR,
|
||||
request_id: None,
|
||||
}),
|
||||
TransportError::Timeout => CodexErr::Timeout,
|
||||
TransportError::Network(msg) | TransportError::Build(msg) => {
|
||||
CodexErr::Stream(msg, None)
|
||||
}
|
||||
},
|
||||
ApiError::RateLimit(msg) => CodexErr::Stream(msg, None),
|
||||
}
|
||||
}
|
||||
|
||||
fn extract_request_id(headers: Option<&HeaderMap>) -> Option<String> {
|
||||
headers.and_then(|map| {
|
||||
["cf-ray", "x-request-id", "x-oai-request-id"]
|
||||
.iter()
|
||||
.find_map(|name| {
|
||||
map.get(*name)
|
||||
.and_then(|v| v.to_str().ok())
|
||||
.map(str::to_string)
|
||||
})
|
||||
})
|
||||
}
|
||||
|
||||
pub(crate) async fn auth_provider_from_auth(
|
||||
auth: Option<CodexAuth>,
|
||||
provider: &ModelProviderInfo,
|
||||
) -> crate::error::Result<CoreAuthProvider> {
|
||||
if let Some(api_key) = provider.api_key()? {
|
||||
return Ok(CoreAuthProvider {
|
||||
token: Some(api_key),
|
||||
account_id: None,
|
||||
});
|
||||
}
|
||||
|
||||
if let Some(token) = provider.experimental_bearer_token.clone() {
|
||||
return Ok(CoreAuthProvider {
|
||||
token: Some(token),
|
||||
account_id: None,
|
||||
});
|
||||
}
|
||||
|
||||
if let Some(auth) = auth {
|
||||
let token = auth.get_token().await?;
|
||||
Ok(CoreAuthProvider {
|
||||
token: Some(token),
|
||||
account_id: auth.get_account_id(),
|
||||
})
|
||||
} else {
|
||||
Ok(CoreAuthProvider {
|
||||
token: None,
|
||||
account_id: None,
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
struct UsageErrorResponse {
|
||||
error: UsageErrorBody,
|
||||
}
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
struct UsageErrorBody {
|
||||
#[serde(rename = "type")]
|
||||
error_type: Option<String>,
|
||||
plan_type: Option<PlanType>,
|
||||
resets_at: Option<i64>,
|
||||
}
|
||||
|
||||
#[derive(Clone, Default)]
|
||||
pub(crate) struct CoreAuthProvider {
|
||||
token: Option<String>,
|
||||
account_id: Option<String>,
|
||||
}
|
||||
|
||||
impl ApiAuthProvider for CoreAuthProvider {
|
||||
fn bearer_token(&self) -> Option<String> {
|
||||
self.token.clone()
|
||||
}
|
||||
|
||||
fn account_id(&self) -> Option<String> {
|
||||
self.account_id.clone()
|
||||
}
|
||||
}
|
||||
@@ -1,981 +0,0 @@
|
||||
use std::time::Duration;
|
||||
|
||||
use crate::ModelProviderInfo;
|
||||
use crate::client_common::Prompt;
|
||||
use crate::client_common::ResponseEvent;
|
||||
use crate::client_common::ResponseStream;
|
||||
use crate::default_client::CodexHttpClient;
|
||||
use crate::error::CodexErr;
|
||||
use crate::error::ConnectionFailedError;
|
||||
use crate::error::ResponseStreamFailed;
|
||||
use crate::error::Result;
|
||||
use crate::error::RetryLimitReachedError;
|
||||
use crate::error::UnexpectedResponseError;
|
||||
use crate::model_family::ModelFamily;
|
||||
use crate::tools::spec::create_tools_json_for_chat_completions_api;
|
||||
use crate::util::backoff;
|
||||
use bytes::Bytes;
|
||||
use codex_otel::otel_event_manager::OtelEventManager;
|
||||
use codex_protocol::models::ContentItem;
|
||||
use codex_protocol::models::FunctionCallOutputContentItem;
|
||||
use codex_protocol::models::ReasoningItemContent;
|
||||
use codex_protocol::models::ResponseItem;
|
||||
use codex_protocol::protocol::SessionSource;
|
||||
use codex_protocol::protocol::SubAgentSource;
|
||||
use eventsource_stream::Eventsource;
|
||||
use futures::Stream;
|
||||
use futures::StreamExt;
|
||||
use futures::TryStreamExt;
|
||||
use reqwest::StatusCode;
|
||||
use serde_json::json;
|
||||
use std::pin::Pin;
|
||||
use std::task::Context;
|
||||
use std::task::Poll;
|
||||
use tokio::sync::mpsc;
|
||||
use tokio::time::timeout;
|
||||
use tracing::debug;
|
||||
use tracing::trace;
|
||||
|
||||
/// Implementation for the classic Chat Completions API.
|
||||
pub(crate) async fn stream_chat_completions(
|
||||
prompt: &Prompt,
|
||||
model_family: &ModelFamily,
|
||||
client: &CodexHttpClient,
|
||||
provider: &ModelProviderInfo,
|
||||
otel_event_manager: &OtelEventManager,
|
||||
session_source: &SessionSource,
|
||||
) -> Result<ResponseStream> {
|
||||
if prompt.output_schema.is_some() {
|
||||
return Err(CodexErr::UnsupportedOperation(
|
||||
"output_schema is not supported for Chat Completions API".to_string(),
|
||||
));
|
||||
}
|
||||
|
||||
// Build messages array
|
||||
let mut messages = Vec::<serde_json::Value>::new();
|
||||
|
||||
let full_instructions = prompt.get_full_instructions(model_family);
|
||||
messages.push(json!({"role": "system", "content": full_instructions}));
|
||||
|
||||
let input = prompt.get_formatted_input();
|
||||
|
||||
// Pre-scan: map Reasoning blocks to the adjacent assistant anchor after the last user.
|
||||
// - If the last emitted message is a user message, drop all reasoning.
|
||||
// - Otherwise, for each Reasoning item after the last user message, attach it
|
||||
// to the immediate previous assistant message (stop turns) or the immediate
|
||||
// next assistant anchor (tool-call turns: function/local shell call, or assistant message).
|
||||
let mut reasoning_by_anchor_index: std::collections::HashMap<usize, String> =
|
||||
std::collections::HashMap::new();
|
||||
|
||||
// Determine the last role that would be emitted to Chat Completions.
|
||||
let mut last_emitted_role: Option<&str> = None;
|
||||
for item in &input {
|
||||
match item {
|
||||
ResponseItem::Message { role, .. } => last_emitted_role = Some(role.as_str()),
|
||||
ResponseItem::FunctionCall { .. } | ResponseItem::LocalShellCall { .. } => {
|
||||
last_emitted_role = Some("assistant")
|
||||
}
|
||||
ResponseItem::FunctionCallOutput { .. } => last_emitted_role = Some("tool"),
|
||||
ResponseItem::Reasoning { .. } | ResponseItem::Other => {}
|
||||
ResponseItem::CustomToolCall { .. } => {}
|
||||
ResponseItem::CustomToolCallOutput { .. } => {}
|
||||
ResponseItem::WebSearchCall { .. } => {}
|
||||
ResponseItem::GhostSnapshot { .. } => {}
|
||||
ResponseItem::CompactionSummary { .. } => {}
|
||||
}
|
||||
}
|
||||
|
||||
// Find the last user message index in the input.
|
||||
let mut last_user_index: Option<usize> = None;
|
||||
for (idx, item) in input.iter().enumerate() {
|
||||
if let ResponseItem::Message { role, .. } = item
|
||||
&& role == "user"
|
||||
{
|
||||
last_user_index = Some(idx);
|
||||
}
|
||||
}
|
||||
|
||||
// Attach reasoning only if the conversation does not end with a user message.
|
||||
if !matches!(last_emitted_role, Some("user")) {
|
||||
for (idx, item) in input.iter().enumerate() {
|
||||
// Only consider reasoning that appears after the last user message.
|
||||
if let Some(u_idx) = last_user_index
|
||||
&& idx <= u_idx
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
if let ResponseItem::Reasoning {
|
||||
content: Some(items),
|
||||
..
|
||||
} = item
|
||||
{
|
||||
let mut text = String::new();
|
||||
for entry in items {
|
||||
match entry {
|
||||
ReasoningItemContent::ReasoningText { text: segment }
|
||||
| ReasoningItemContent::Text { text: segment } => text.push_str(segment),
|
||||
}
|
||||
}
|
||||
if text.trim().is_empty() {
|
||||
continue;
|
||||
}
|
||||
|
||||
// Prefer immediate previous assistant message (stop turns)
|
||||
let mut attached = false;
|
||||
if idx > 0
|
||||
&& let ResponseItem::Message { role, .. } = &input[idx - 1]
|
||||
&& role == "assistant"
|
||||
{
|
||||
reasoning_by_anchor_index
|
||||
.entry(idx - 1)
|
||||
.and_modify(|v| v.push_str(&text))
|
||||
.or_insert(text.clone());
|
||||
attached = true;
|
||||
}
|
||||
|
||||
// Otherwise, attach to immediate next assistant anchor (tool-calls or assistant message)
|
||||
if !attached && idx + 1 < input.len() {
|
||||
match &input[idx + 1] {
|
||||
ResponseItem::FunctionCall { .. } | ResponseItem::LocalShellCall { .. } => {
|
||||
reasoning_by_anchor_index
|
||||
.entry(idx + 1)
|
||||
.and_modify(|v| v.push_str(&text))
|
||||
.or_insert(text.clone());
|
||||
}
|
||||
ResponseItem::Message { role, .. } if role == "assistant" => {
|
||||
reasoning_by_anchor_index
|
||||
.entry(idx + 1)
|
||||
.and_modify(|v| v.push_str(&text))
|
||||
.or_insert(text.clone());
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Track last assistant text we emitted to avoid duplicate assistant messages
|
||||
// in the outbound Chat Completions payload (can happen if a final
|
||||
// aggregated assistant message was recorded alongside an earlier partial).
|
||||
let mut last_assistant_text: Option<String> = None;
|
||||
|
||||
for (idx, item) in input.iter().enumerate() {
|
||||
match item {
|
||||
ResponseItem::Message { role, content, .. } => {
|
||||
// Build content either as a plain string (typical for assistant text)
|
||||
// or as an array of content items when images are present (user/tool multimodal).
|
||||
let mut text = String::new();
|
||||
let mut items: Vec<serde_json::Value> = Vec::new();
|
||||
let mut saw_image = false;
|
||||
|
||||
for c in content {
|
||||
match c {
|
||||
ContentItem::InputText { text: t }
|
||||
| ContentItem::OutputText { text: t } => {
|
||||
text.push_str(t);
|
||||
items.push(json!({"type":"text","text": t}));
|
||||
}
|
||||
ContentItem::InputImage { image_url } => {
|
||||
saw_image = true;
|
||||
items.push(json!({"type":"image_url","image_url": {"url": image_url}}));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Skip exact-duplicate assistant messages.
|
||||
if role == "assistant" {
|
||||
if let Some(prev) = &last_assistant_text
|
||||
&& prev == &text
|
||||
{
|
||||
continue;
|
||||
}
|
||||
last_assistant_text = Some(text.clone());
|
||||
}
|
||||
|
||||
// For assistant messages, always send a plain string for compatibility.
|
||||
// For user messages, if an image is present, send an array of content items.
|
||||
let content_value = if role == "assistant" {
|
||||
json!(text)
|
||||
} else if saw_image {
|
||||
json!(items)
|
||||
} else {
|
||||
json!(text)
|
||||
};
|
||||
|
||||
let mut msg = json!({"role": role, "content": content_value});
|
||||
if role == "assistant"
|
||||
&& let Some(reasoning) = reasoning_by_anchor_index.get(&idx)
|
||||
&& let Some(obj) = msg.as_object_mut()
|
||||
{
|
||||
obj.insert("reasoning".to_string(), json!(reasoning));
|
||||
}
|
||||
messages.push(msg);
|
||||
}
|
||||
ResponseItem::FunctionCall {
|
||||
name,
|
||||
arguments,
|
||||
call_id,
|
||||
..
|
||||
} => {
|
||||
let mut msg = json!({
|
||||
"role": "assistant",
|
||||
"content": null,
|
||||
"tool_calls": [{
|
||||
"id": call_id,
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": name,
|
||||
"arguments": arguments,
|
||||
}
|
||||
}]
|
||||
});
|
||||
if let Some(reasoning) = reasoning_by_anchor_index.get(&idx)
|
||||
&& let Some(obj) = msg.as_object_mut()
|
||||
{
|
||||
obj.insert("reasoning".to_string(), json!(reasoning));
|
||||
}
|
||||
messages.push(msg);
|
||||
}
|
||||
ResponseItem::LocalShellCall {
|
||||
id,
|
||||
call_id: _,
|
||||
status,
|
||||
action,
|
||||
} => {
|
||||
// Confirm with API team.
|
||||
let mut msg = json!({
|
||||
"role": "assistant",
|
||||
"content": null,
|
||||
"tool_calls": [{
|
||||
"id": id.clone().unwrap_or_else(|| "".to_string()),
|
||||
"type": "local_shell_call",
|
||||
"status": status,
|
||||
"action": action,
|
||||
}]
|
||||
});
|
||||
if let Some(reasoning) = reasoning_by_anchor_index.get(&idx)
|
||||
&& let Some(obj) = msg.as_object_mut()
|
||||
{
|
||||
obj.insert("reasoning".to_string(), json!(reasoning));
|
||||
}
|
||||
messages.push(msg);
|
||||
}
|
||||
ResponseItem::FunctionCallOutput { call_id, output } => {
|
||||
// Prefer structured content items when available (e.g., images)
|
||||
// otherwise fall back to the legacy plain-string content.
|
||||
let content_value = if let Some(items) = &output.content_items {
|
||||
let mapped: Vec<serde_json::Value> = items
|
||||
.iter()
|
||||
.map(|it| match it {
|
||||
FunctionCallOutputContentItem::InputText { text } => {
|
||||
json!({"type":"text","text": text})
|
||||
}
|
||||
FunctionCallOutputContentItem::InputImage { image_url } => {
|
||||
json!({"type":"image_url","image_url": {"url": image_url}})
|
||||
}
|
||||
})
|
||||
.collect();
|
||||
json!(mapped)
|
||||
} else {
|
||||
json!(output.content)
|
||||
};
|
||||
|
||||
messages.push(json!({
|
||||
"role": "tool",
|
||||
"tool_call_id": call_id,
|
||||
"content": content_value,
|
||||
}));
|
||||
}
|
||||
ResponseItem::CustomToolCall {
|
||||
id,
|
||||
call_id: _,
|
||||
name,
|
||||
input,
|
||||
status: _,
|
||||
} => {
|
||||
messages.push(json!({
|
||||
"role": "assistant",
|
||||
"content": null,
|
||||
"tool_calls": [{
|
||||
"id": id,
|
||||
"type": "custom",
|
||||
"custom": {
|
||||
"name": name,
|
||||
"input": input,
|
||||
}
|
||||
}]
|
||||
}));
|
||||
}
|
||||
ResponseItem::CustomToolCallOutput { call_id, output } => {
|
||||
messages.push(json!({
|
||||
"role": "tool",
|
||||
"tool_call_id": call_id,
|
||||
"content": output,
|
||||
}));
|
||||
}
|
||||
ResponseItem::GhostSnapshot { .. } => {
|
||||
// Ghost snapshots annotate history but are not sent to the model.
|
||||
continue;
|
||||
}
|
||||
ResponseItem::Reasoning { .. }
|
||||
| ResponseItem::WebSearchCall { .. }
|
||||
| ResponseItem::Other
|
||||
| ResponseItem::CompactionSummary { .. } => {
|
||||
// Omit these items from the conversation history.
|
||||
continue;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
let tools_json = create_tools_json_for_chat_completions_api(&prompt.tools)?;
|
||||
let payload = json!({
|
||||
"model": model_family.slug,
|
||||
"messages": messages,
|
||||
"stream": true,
|
||||
"tools": tools_json,
|
||||
});
|
||||
|
||||
debug!(
|
||||
"POST to {}: {}",
|
||||
provider.get_full_url(&None),
|
||||
payload.to_string()
|
||||
);
|
||||
|
||||
let mut attempt = 0;
|
||||
let max_retries = provider.request_max_retries();
|
||||
loop {
|
||||
attempt += 1;
|
||||
|
||||
let mut req_builder = provider.create_request_builder(client, &None).await?;
|
||||
|
||||
// Include subagent header only for subagent sessions.
|
||||
if let SessionSource::SubAgent(sub) = session_source.clone() {
|
||||
let subagent = if let SubAgentSource::Other(label) = sub {
|
||||
label
|
||||
} else {
|
||||
serde_json::to_value(&sub)
|
||||
.ok()
|
||||
.and_then(|v| v.as_str().map(std::string::ToString::to_string))
|
||||
.unwrap_or_else(|| "other".to_string())
|
||||
};
|
||||
req_builder = req_builder.header("x-openai-subagent", subagent);
|
||||
}
|
||||
|
||||
let res = otel_event_manager
|
||||
.log_request(attempt, || {
|
||||
req_builder
|
||||
.header(reqwest::header::ACCEPT, "text/event-stream")
|
||||
.json(&payload)
|
||||
.send()
|
||||
})
|
||||
.await;
|
||||
|
||||
match res {
|
||||
Ok(resp) if resp.status().is_success() => {
|
||||
let (tx_event, rx_event) = mpsc::channel::<Result<ResponseEvent>>(1600);
|
||||
let stream = resp.bytes_stream().map_err(|e| {
|
||||
CodexErr::ResponseStreamFailed(ResponseStreamFailed {
|
||||
source: e,
|
||||
request_id: None,
|
||||
})
|
||||
});
|
||||
tokio::spawn(process_chat_sse(
|
||||
stream,
|
||||
tx_event,
|
||||
provider.stream_idle_timeout(),
|
||||
otel_event_manager.clone(),
|
||||
));
|
||||
return Ok(ResponseStream { rx_event });
|
||||
}
|
||||
Ok(res) => {
|
||||
let status = res.status();
|
||||
if !(status == StatusCode::TOO_MANY_REQUESTS || status.is_server_error()) {
|
||||
let body = (res.text().await).unwrap_or_default();
|
||||
return Err(CodexErr::UnexpectedStatus(UnexpectedResponseError {
|
||||
status,
|
||||
body,
|
||||
request_id: None,
|
||||
}));
|
||||
}
|
||||
|
||||
if attempt > max_retries {
|
||||
return Err(CodexErr::RetryLimit(RetryLimitReachedError {
|
||||
status,
|
||||
request_id: None,
|
||||
}));
|
||||
}
|
||||
|
||||
let retry_after_secs = res
|
||||
.headers()
|
||||
.get(reqwest::header::RETRY_AFTER)
|
||||
.and_then(|v| v.to_str().ok())
|
||||
.and_then(|s| s.parse::<u64>().ok());
|
||||
|
||||
let delay = retry_after_secs
|
||||
.map(|s| Duration::from_millis(s * 1_000))
|
||||
.unwrap_or_else(|| backoff(attempt));
|
||||
tokio::time::sleep(delay).await;
|
||||
}
|
||||
Err(e) => {
|
||||
if attempt > max_retries {
|
||||
return Err(CodexErr::ConnectionFailed(ConnectionFailedError {
|
||||
source: e,
|
||||
}));
|
||||
}
|
||||
let delay = backoff(attempt);
|
||||
tokio::time::sleep(delay).await;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
async fn append_assistant_text(
|
||||
tx_event: &mpsc::Sender<Result<ResponseEvent>>,
|
||||
assistant_item: &mut Option<ResponseItem>,
|
||||
text: String,
|
||||
) {
|
||||
if assistant_item.is_none() {
|
||||
let item = ResponseItem::Message {
|
||||
id: None,
|
||||
role: "assistant".to_string(),
|
||||
content: vec![],
|
||||
};
|
||||
*assistant_item = Some(item.clone());
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::OutputItemAdded(item)))
|
||||
.await;
|
||||
}
|
||||
|
||||
if let Some(ResponseItem::Message { content, .. }) = assistant_item {
|
||||
content.push(ContentItem::OutputText { text: text.clone() });
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::OutputTextDelta(text.clone())))
|
||||
.await;
|
||||
}
|
||||
}
|
||||
|
||||
async fn append_reasoning_text(
|
||||
tx_event: &mpsc::Sender<Result<ResponseEvent>>,
|
||||
reasoning_item: &mut Option<ResponseItem>,
|
||||
text: String,
|
||||
) {
|
||||
if reasoning_item.is_none() {
|
||||
let item = ResponseItem::Reasoning {
|
||||
id: String::new(),
|
||||
summary: Vec::new(),
|
||||
content: Some(vec![]),
|
||||
encrypted_content: None,
|
||||
};
|
||||
*reasoning_item = Some(item.clone());
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::OutputItemAdded(item)))
|
||||
.await;
|
||||
}
|
||||
|
||||
if let Some(ResponseItem::Reasoning {
|
||||
content: Some(content),
|
||||
..
|
||||
}) = reasoning_item
|
||||
{
|
||||
let content_index = content.len() as i64;
|
||||
content.push(ReasoningItemContent::ReasoningText { text: text.clone() });
|
||||
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::ReasoningContentDelta {
|
||||
delta: text.clone(),
|
||||
content_index,
|
||||
}))
|
||||
.await;
|
||||
}
|
||||
}
|
||||
/// Lightweight SSE processor for the Chat Completions streaming format. The
|
||||
/// output is mapped onto Codex's internal [`ResponseEvent`] so that the rest
|
||||
/// of the pipeline can stay agnostic of the underlying wire format.
|
||||
async fn process_chat_sse<S>(
|
||||
stream: S,
|
||||
tx_event: mpsc::Sender<Result<ResponseEvent>>,
|
||||
idle_timeout: Duration,
|
||||
otel_event_manager: OtelEventManager,
|
||||
) where
|
||||
S: Stream<Item = Result<Bytes>> + Unpin,
|
||||
{
|
||||
let mut stream = stream.eventsource();
|
||||
|
||||
// State to accumulate a function call across streaming chunks.
|
||||
// OpenAI may split the `arguments` string over multiple `delta` events
|
||||
// until the chunk whose `finish_reason` is `tool_calls` is emitted. We
|
||||
// keep collecting the pieces here and forward a single
|
||||
// `ResponseItem::FunctionCall` once the call is complete.
|
||||
#[derive(Default)]
|
||||
struct FunctionCallState {
|
||||
name: Option<String>,
|
||||
arguments: String,
|
||||
call_id: Option<String>,
|
||||
active: bool,
|
||||
}
|
||||
|
||||
let mut fn_call_state = FunctionCallState::default();
|
||||
let mut assistant_item: Option<ResponseItem> = None;
|
||||
let mut reasoning_item: Option<ResponseItem> = None;
|
||||
|
||||
loop {
|
||||
let start = std::time::Instant::now();
|
||||
let response = timeout(idle_timeout, stream.next()).await;
|
||||
let duration = start.elapsed();
|
||||
otel_event_manager.log_sse_event(&response, duration);
|
||||
|
||||
let sse = match response {
|
||||
Ok(Some(Ok(ev))) => ev,
|
||||
Ok(Some(Err(e))) => {
|
||||
let _ = tx_event
|
||||
.send(Err(CodexErr::Stream(e.to_string(), None)))
|
||||
.await;
|
||||
return;
|
||||
}
|
||||
Ok(None) => {
|
||||
// Stream closed gracefully – emit Completed with dummy id.
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::Completed {
|
||||
response_id: String::new(),
|
||||
token_usage: None,
|
||||
}))
|
||||
.await;
|
||||
return;
|
||||
}
|
||||
Err(_) => {
|
||||
let _ = tx_event
|
||||
.send(Err(CodexErr::Stream(
|
||||
"idle timeout waiting for SSE".into(),
|
||||
None,
|
||||
)))
|
||||
.await;
|
||||
return;
|
||||
}
|
||||
};
|
||||
|
||||
// OpenAI Chat streaming sends a literal string "[DONE]" when finished.
|
||||
if sse.data.trim() == "[DONE]" {
|
||||
// Emit any finalized items before closing so downstream consumers receive
|
||||
// terminal events for both assistant content and raw reasoning.
|
||||
if let Some(item) = assistant_item {
|
||||
let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
|
||||
}
|
||||
|
||||
if let Some(item) = reasoning_item {
|
||||
let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
|
||||
}
|
||||
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::Completed {
|
||||
response_id: String::new(),
|
||||
token_usage: None,
|
||||
}))
|
||||
.await;
|
||||
return;
|
||||
}
|
||||
|
||||
// Parse JSON chunk
|
||||
let chunk: serde_json::Value = match serde_json::from_str(&sse.data) {
|
||||
Ok(v) => v,
|
||||
Err(_) => continue,
|
||||
};
|
||||
trace!("chat_completions received SSE chunk: {chunk:?}");
|
||||
|
||||
let choice_opt = chunk.get("choices").and_then(|c| c.get(0));
|
||||
|
||||
if let Some(choice) = choice_opt {
|
||||
// Handle assistant content tokens as streaming deltas.
|
||||
if let Some(content) = choice
|
||||
.get("delta")
|
||||
.and_then(|d| d.get("content"))
|
||||
.and_then(|c| c.as_str())
|
||||
&& !content.is_empty()
|
||||
{
|
||||
append_assistant_text(&tx_event, &mut assistant_item, content.to_string()).await;
|
||||
}
|
||||
|
||||
// Forward any reasoning/thinking deltas if present.
|
||||
// Some providers stream `reasoning` as a plain string while others
|
||||
// nest the text under an object (e.g. `{ "reasoning": { "text": "…" } }`).
|
||||
if let Some(reasoning_val) = choice.get("delta").and_then(|d| d.get("reasoning")) {
|
||||
let mut maybe_text = reasoning_val
|
||||
.as_str()
|
||||
.map(str::to_string)
|
||||
.filter(|s| !s.is_empty());
|
||||
|
||||
if maybe_text.is_none() && reasoning_val.is_object() {
|
||||
if let Some(s) = reasoning_val
|
||||
.get("text")
|
||||
.and_then(|t| t.as_str())
|
||||
.filter(|s| !s.is_empty())
|
||||
{
|
||||
maybe_text = Some(s.to_string());
|
||||
} else if let Some(s) = reasoning_val
|
||||
.get("content")
|
||||
.and_then(|t| t.as_str())
|
||||
.filter(|s| !s.is_empty())
|
||||
{
|
||||
maybe_text = Some(s.to_string());
|
||||
}
|
||||
}
|
||||
|
||||
if let Some(reasoning) = maybe_text {
|
||||
// Accumulate so we can emit a terminal Reasoning item at the end.
|
||||
append_reasoning_text(&tx_event, &mut reasoning_item, reasoning).await;
|
||||
}
|
||||
}
|
||||
|
||||
// Some providers only include reasoning on the final message object.
|
||||
if let Some(message_reasoning) = choice.get("message").and_then(|m| m.get("reasoning"))
|
||||
{
|
||||
// Accept either a plain string or an object with { text | content }
|
||||
if let Some(s) = message_reasoning.as_str() {
|
||||
if !s.is_empty() {
|
||||
append_reasoning_text(&tx_event, &mut reasoning_item, s.to_string()).await;
|
||||
}
|
||||
} else if let Some(obj) = message_reasoning.as_object()
|
||||
&& let Some(s) = obj
|
||||
.get("text")
|
||||
.and_then(|v| v.as_str())
|
||||
.or_else(|| obj.get("content").and_then(|v| v.as_str()))
|
||||
&& !s.is_empty()
|
||||
{
|
||||
append_reasoning_text(&tx_event, &mut reasoning_item, s.to_string()).await;
|
||||
}
|
||||
}
|
||||
|
||||
// Handle streaming function / tool calls.
|
||||
if let Some(tool_calls) = choice
|
||||
.get("delta")
|
||||
.and_then(|d| d.get("tool_calls"))
|
||||
.and_then(|tc| tc.as_array())
|
||||
&& let Some(tool_call) = tool_calls.first()
|
||||
{
|
||||
// Mark that we have an active function call in progress.
|
||||
fn_call_state.active = true;
|
||||
|
||||
// Extract call_id if present.
|
||||
if let Some(id) = tool_call.get("id").and_then(|v| v.as_str()) {
|
||||
fn_call_state.call_id.get_or_insert_with(|| id.to_string());
|
||||
}
|
||||
|
||||
// Extract function details if present.
|
||||
if let Some(function) = tool_call.get("function") {
|
||||
if let Some(name) = function.get("name").and_then(|n| n.as_str()) {
|
||||
fn_call_state.name.get_or_insert_with(|| name.to_string());
|
||||
}
|
||||
|
||||
if let Some(args_fragment) = function.get("arguments").and_then(|a| a.as_str())
|
||||
{
|
||||
fn_call_state.arguments.push_str(args_fragment);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Emit end-of-turn when finish_reason signals completion.
|
||||
if let Some(finish_reason) = choice.get("finish_reason").and_then(|v| v.as_str())
|
||||
&& !finish_reason.is_empty()
|
||||
{
|
||||
match finish_reason {
|
||||
"tool_calls" if fn_call_state.active => {
|
||||
// First, flush the terminal raw reasoning so UIs can finalize
|
||||
// the reasoning stream before any exec/tool events begin.
|
||||
if let Some(item) = reasoning_item.take() {
|
||||
let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
|
||||
}
|
||||
|
||||
// Then emit the FunctionCall response item.
|
||||
let item = ResponseItem::FunctionCall {
|
||||
id: None,
|
||||
name: fn_call_state.name.clone().unwrap_or_else(|| "".to_string()),
|
||||
arguments: fn_call_state.arguments.clone(),
|
||||
call_id: fn_call_state.call_id.clone().unwrap_or_else(String::new),
|
||||
};
|
||||
|
||||
let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
|
||||
}
|
||||
"stop" => {
|
||||
// Regular turn without tool-call. Emit the final assistant message
|
||||
// as a single OutputItemDone so non-delta consumers see the result.
|
||||
if let Some(item) = assistant_item.take() {
|
||||
let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
|
||||
}
|
||||
// Also emit a terminal Reasoning item so UIs can finalize raw reasoning.
|
||||
if let Some(item) = reasoning_item.take() {
|
||||
let _ = tx_event.send(Ok(ResponseEvent::OutputItemDone(item))).await;
|
||||
}
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
|
||||
// Emit Completed regardless of reason so the agent can advance.
|
||||
let _ = tx_event
|
||||
.send(Ok(ResponseEvent::Completed {
|
||||
response_id: String::new(),
|
||||
token_usage: None,
|
||||
}))
|
||||
.await;
|
||||
|
||||
// Prepare for potential next turn (should not happen in same stream).
|
||||
// fn_call_state = FunctionCallState::default();
|
||||
|
||||
return; // End processing for this SSE stream.
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Optional client-side aggregation helper
|
||||
///
|
||||
/// Stream adapter that merges the incremental `OutputItemDone` chunks coming from
|
||||
/// [`process_chat_sse`] into a *running* assistant message, **suppressing the
|
||||
/// per-token deltas**. The stream stays silent while the model is thinking
|
||||
/// and only emits two events per turn:
|
||||
///
|
||||
/// 1. `ResponseEvent::OutputItemDone` with the *complete* assistant message
|
||||
/// (fully concatenated).
|
||||
/// 2. The original `ResponseEvent::Completed` right after it.
|
||||
///
|
||||
/// This mirrors the behaviour the TypeScript CLI exposes to its higher layers.
|
||||
///
|
||||
/// The adapter is intentionally *lossless*: callers who do **not** opt in via
|
||||
/// [`AggregateStreamExt::aggregate()`] keep receiving the original unmodified
|
||||
/// events.
|
||||
#[derive(Copy, Clone, Eq, PartialEq)]
|
||||
enum AggregateMode {
|
||||
AggregatedOnly,
|
||||
Streaming,
|
||||
}
|
||||
pub(crate) struct AggregatedChatStream<S> {
|
||||
inner: S,
|
||||
cumulative: String,
|
||||
cumulative_reasoning: String,
|
||||
pending: std::collections::VecDeque<ResponseEvent>,
|
||||
mode: AggregateMode,
|
||||
}
|
||||
|
||||
impl<S> Stream for AggregatedChatStream<S>
|
||||
where
|
||||
S: Stream<Item = Result<ResponseEvent>> + Unpin,
|
||||
{
|
||||
type Item = Result<ResponseEvent>;
|
||||
|
||||
fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
|
||||
let this = self.get_mut();
|
||||
|
||||
// First, flush any buffered events from the previous call.
|
||||
if let Some(ev) = this.pending.pop_front() {
|
||||
return Poll::Ready(Some(Ok(ev)));
|
||||
}
|
||||
|
||||
loop {
|
||||
match Pin::new(&mut this.inner).poll_next(cx) {
|
||||
Poll::Pending => return Poll::Pending,
|
||||
Poll::Ready(None) => return Poll::Ready(None),
|
||||
Poll::Ready(Some(Err(e))) => return Poll::Ready(Some(Err(e))),
|
||||
Poll::Ready(Some(Ok(ResponseEvent::OutputItemDone(item)))) => {
|
||||
// If this is an incremental assistant message chunk, accumulate but
|
||||
// do NOT emit yet. Forward any other item (e.g. FunctionCall) right
|
||||
// away so downstream consumers see it.
|
||||
|
||||
let is_assistant_message = matches!(
|
||||
&item,
|
||||
codex_protocol::models::ResponseItem::Message { role, .. } if role == "assistant"
|
||||
);
|
||||
|
||||
if is_assistant_message {
|
||||
match this.mode {
|
||||
AggregateMode::AggregatedOnly => {
|
||||
// Only use the final assistant message if we have not
|
||||
// seen any deltas; otherwise, deltas already built the
|
||||
// cumulative text and this would duplicate it.
|
||||
if this.cumulative.is_empty()
|
||||
&& let codex_protocol::models::ResponseItem::Message {
|
||||
content,
|
||||
..
|
||||
} = &item
|
||||
&& let Some(text) = content.iter().find_map(|c| match c {
|
||||
codex_protocol::models::ContentItem::OutputText {
|
||||
text,
|
||||
} => Some(text),
|
||||
_ => None,
|
||||
})
|
||||
{
|
||||
this.cumulative.push_str(text);
|
||||
}
|
||||
// Swallow assistant message here; emit on Completed.
|
||||
continue;
|
||||
}
|
||||
AggregateMode::Streaming => {
|
||||
// In streaming mode, if we have not seen any deltas, forward
|
||||
// the final assistant message directly. If deltas were seen,
|
||||
// suppress the final message to avoid duplication.
|
||||
if this.cumulative.is_empty() {
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::OutputItemDone(
|
||||
item,
|
||||
))));
|
||||
} else {
|
||||
continue;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Not an assistant message – forward immediately.
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::OutputItemDone(item))));
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::RateLimits(snapshot)))) => {
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::RateLimits(snapshot))));
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::Completed {
|
||||
response_id,
|
||||
token_usage,
|
||||
}))) => {
|
||||
// Build any aggregated items in the correct order: Reasoning first, then Message.
|
||||
let mut emitted_any = false;
|
||||
|
||||
if !this.cumulative_reasoning.is_empty()
|
||||
&& matches!(this.mode, AggregateMode::AggregatedOnly)
|
||||
{
|
||||
let aggregated_reasoning =
|
||||
codex_protocol::models::ResponseItem::Reasoning {
|
||||
id: String::new(),
|
||||
summary: Vec::new(),
|
||||
content: Some(vec![
|
||||
codex_protocol::models::ReasoningItemContent::ReasoningText {
|
||||
text: std::mem::take(&mut this.cumulative_reasoning),
|
||||
},
|
||||
]),
|
||||
encrypted_content: None,
|
||||
};
|
||||
this.pending
|
||||
.push_back(ResponseEvent::OutputItemDone(aggregated_reasoning));
|
||||
emitted_any = true;
|
||||
}
|
||||
|
||||
// Always emit the final aggregated assistant message when any
|
||||
// content deltas have been observed. In AggregatedOnly mode this
|
||||
// is the sole assistant output; in Streaming mode this finalizes
|
||||
// the streamed deltas into a terminal OutputItemDone so callers
|
||||
// can persist/render the message once per turn.
|
||||
if !this.cumulative.is_empty() {
|
||||
let aggregated_message = codex_protocol::models::ResponseItem::Message {
|
||||
id: None,
|
||||
role: "assistant".to_string(),
|
||||
content: vec![codex_protocol::models::ContentItem::OutputText {
|
||||
text: std::mem::take(&mut this.cumulative),
|
||||
}],
|
||||
};
|
||||
this.pending
|
||||
.push_back(ResponseEvent::OutputItemDone(aggregated_message));
|
||||
emitted_any = true;
|
||||
}
|
||||
|
||||
// Always emit Completed last when anything was aggregated.
|
||||
if emitted_any {
|
||||
this.pending.push_back(ResponseEvent::Completed {
|
||||
response_id: response_id.clone(),
|
||||
token_usage: token_usage.clone(),
|
||||
});
|
||||
// Return the first pending event now.
|
||||
if let Some(ev) = this.pending.pop_front() {
|
||||
return Poll::Ready(Some(Ok(ev)));
|
||||
}
|
||||
}
|
||||
|
||||
// Nothing aggregated – forward Completed directly.
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::Completed {
|
||||
response_id,
|
||||
token_usage,
|
||||
})));
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::Created))) => {
|
||||
// These events are exclusive to the Responses API and
|
||||
// will never appear in a Chat Completions stream.
|
||||
continue;
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::OutputTextDelta(delta)))) => {
|
||||
// Always accumulate deltas so we can emit a final OutputItemDone at Completed.
|
||||
this.cumulative.push_str(&delta);
|
||||
if matches!(this.mode, AggregateMode::Streaming) {
|
||||
// In streaming mode, also forward the delta immediately.
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::OutputTextDelta(delta))));
|
||||
} else {
|
||||
continue;
|
||||
}
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::ReasoningContentDelta {
|
||||
delta,
|
||||
content_index,
|
||||
}))) => {
|
||||
// Always accumulate reasoning deltas so we can emit a final Reasoning item at Completed.
|
||||
this.cumulative_reasoning.push_str(&delta);
|
||||
if matches!(this.mode, AggregateMode::Streaming) {
|
||||
// In streaming mode, also forward the delta immediately.
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::ReasoningContentDelta {
|
||||
delta,
|
||||
content_index,
|
||||
})));
|
||||
} else {
|
||||
continue;
|
||||
}
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::ReasoningSummaryDelta { .. }))) => {
|
||||
continue;
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::ReasoningSummaryPartAdded { .. }))) => {
|
||||
continue;
|
||||
}
|
||||
Poll::Ready(Some(Ok(ResponseEvent::OutputItemAdded(item)))) => {
|
||||
return Poll::Ready(Some(Ok(ResponseEvent::OutputItemAdded(item))));
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Extension trait that activates aggregation on any stream of [`ResponseEvent`].
|
||||
pub(crate) trait AggregateStreamExt: Stream<Item = Result<ResponseEvent>> + Sized {
|
||||
/// Returns a new stream that emits **only** the final assistant message
|
||||
/// per turn instead of every incremental delta. The produced
|
||||
/// `ResponseEvent` sequence for a typical text turn looks like:
|
||||
///
|
||||
/// ```ignore
|
||||
/// OutputItemDone(<full message>)
|
||||
/// Completed
|
||||
/// ```
|
||||
///
|
||||
/// No other `OutputItemDone` events will be seen by the caller.
|
||||
///
|
||||
/// Usage:
|
||||
///
|
||||
/// ```ignore
|
||||
/// let agg_stream = client.stream(&prompt).await?.aggregate();
|
||||
/// while let Some(event) = agg_stream.next().await {
|
||||
/// // event now contains cumulative text
|
||||
/// }
|
||||
/// ```
|
||||
fn aggregate(self) -> AggregatedChatStream<Self> {
|
||||
AggregatedChatStream::new(self, AggregateMode::AggregatedOnly)
|
||||
}
|
||||
}
|
||||
|
||||
impl<T> AggregateStreamExt for T where T: Stream<Item = Result<ResponseEvent>> + Sized {}
|
||||
|
||||
impl<S> AggregatedChatStream<S> {
|
||||
fn new(inner: S, mode: AggregateMode) -> Self {
|
||||
AggregatedChatStream {
|
||||
inner,
|
||||
cumulative: String::new(),
|
||||
cumulative_reasoning: String::new(),
|
||||
pending: std::collections::VecDeque::new(),
|
||||
mode,
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn streaming_mode(inner: S) -> Self {
|
||||
Self::new(inner, AggregateMode::Streaming)
|
||||
}
|
||||
}
|
||||
+324
-1431
File diff suppressed because it is too large
Load Diff
@@ -1,16 +1,11 @@
|
||||
use crate::client_common::tools::ToolSpec;
|
||||
use crate::error::Result;
|
||||
use crate::model_family::ModelFamily;
|
||||
use crate::protocol::RateLimitSnapshot;
|
||||
use crate::protocol::TokenUsage;
|
||||
pub use codex_api::common::ResponseEvent;
|
||||
use codex_apply_patch::APPLY_PATCH_TOOL_INSTRUCTIONS;
|
||||
use codex_protocol::config_types::ReasoningEffort as ReasoningEffortConfig;
|
||||
use codex_protocol::config_types::ReasoningSummary as ReasoningSummaryConfig;
|
||||
use codex_protocol::config_types::Verbosity as VerbosityConfig;
|
||||
use codex_protocol::models::ResponseItem;
|
||||
use futures::Stream;
|
||||
use serde::Deserialize;
|
||||
use serde::Serialize;
|
||||
use serde_json::Value;
|
||||
use std::borrow::Cow;
|
||||
use std::collections::HashSet;
|
||||
@@ -184,104 +179,6 @@ fn strip_total_output_header(output: &str) -> Option<(&str, u32)> {
|
||||
Some((remainder, total_lines))
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub enum ResponseEvent {
|
||||
Created,
|
||||
OutputItemDone(ResponseItem),
|
||||
OutputItemAdded(ResponseItem),
|
||||
Completed {
|
||||
response_id: String,
|
||||
token_usage: Option<TokenUsage>,
|
||||
},
|
||||
OutputTextDelta(String),
|
||||
ReasoningSummaryDelta {
|
||||
delta: String,
|
||||
summary_index: i64,
|
||||
},
|
||||
ReasoningContentDelta {
|
||||
delta: String,
|
||||
content_index: i64,
|
||||
},
|
||||
ReasoningSummaryPartAdded {
|
||||
summary_index: i64,
|
||||
},
|
||||
RateLimits(RateLimitSnapshot),
|
||||
}
|
||||
|
||||
#[derive(Debug, Serialize)]
|
||||
pub(crate) struct Reasoning {
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub(crate) effort: Option<ReasoningEffortConfig>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub(crate) summary: Option<ReasoningSummaryConfig>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Serialize, Default, Clone)]
|
||||
#[serde(rename_all = "snake_case")]
|
||||
pub(crate) enum TextFormatType {
|
||||
#[default]
|
||||
JsonSchema,
|
||||
}
|
||||
|
||||
#[derive(Debug, Serialize, Default, Clone)]
|
||||
pub(crate) struct TextFormat {
|
||||
pub(crate) r#type: TextFormatType,
|
||||
pub(crate) strict: bool,
|
||||
pub(crate) schema: Value,
|
||||
pub(crate) name: String,
|
||||
}
|
||||
|
||||
/// Controls under the `text` field in the Responses API for GPT-5.
|
||||
#[derive(Debug, Serialize, Default, Clone)]
|
||||
pub(crate) struct TextControls {
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub(crate) verbosity: Option<OpenAiVerbosity>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub(crate) format: Option<TextFormat>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Serialize, Default, Clone)]
|
||||
#[serde(rename_all = "lowercase")]
|
||||
pub(crate) enum OpenAiVerbosity {
|
||||
Low,
|
||||
#[default]
|
||||
Medium,
|
||||
High,
|
||||
}
|
||||
|
||||
impl From<VerbosityConfig> for OpenAiVerbosity {
|
||||
fn from(v: VerbosityConfig) -> Self {
|
||||
match v {
|
||||
VerbosityConfig::Low => OpenAiVerbosity::Low,
|
||||
VerbosityConfig::Medium => OpenAiVerbosity::Medium,
|
||||
VerbosityConfig::High => OpenAiVerbosity::High,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Request object that is serialized as JSON and POST'ed when using the
|
||||
/// Responses API.
|
||||
#[derive(Debug, Serialize)]
|
||||
pub(crate) struct ResponsesApiRequest<'a> {
|
||||
pub(crate) model: &'a str,
|
||||
pub(crate) instructions: &'a str,
|
||||
// TODO(mbolin): ResponseItem::Other should not be serialized. Currently,
|
||||
// we code defensively to avoid this case, but perhaps we should use a
|
||||
// separate enum for serialization.
|
||||
pub(crate) input: &'a Vec<ResponseItem>,
|
||||
pub(crate) tools: &'a [serde_json::Value],
|
||||
pub(crate) tool_choice: &'static str,
|
||||
pub(crate) parallel_tool_calls: bool,
|
||||
pub(crate) reasoning: Option<Reasoning>,
|
||||
pub(crate) store: bool,
|
||||
pub(crate) stream: bool,
|
||||
pub(crate) include: Vec<String>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub(crate) prompt_cache_key: Option<String>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub(crate) text: Option<TextControls>,
|
||||
}
|
||||
|
||||
pub(crate) mod tools {
|
||||
use crate::tools::spec::JsonSchema;
|
||||
use serde::Deserialize;
|
||||
@@ -341,25 +238,6 @@ pub(crate) mod tools {
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn create_text_param_for_request(
|
||||
verbosity: Option<VerbosityConfig>,
|
||||
output_schema: &Option<Value>,
|
||||
) -> Option<TextControls> {
|
||||
if verbosity.is_none() && output_schema.is_none() {
|
||||
return None;
|
||||
}
|
||||
|
||||
Some(TextControls {
|
||||
verbosity: verbosity.map(std::convert::Into::into),
|
||||
format: output_schema.as_ref().map(|schema| TextFormat {
|
||||
r#type: TextFormatType::JsonSchema,
|
||||
strict: true,
|
||||
schema: schema.clone(),
|
||||
name: "codex_output_schema".to_string(),
|
||||
}),
|
||||
})
|
||||
}
|
||||
|
||||
pub struct ResponseStream {
|
||||
pub(crate) rx_event: mpsc::Receiver<Result<ResponseEvent>>,
|
||||
}
|
||||
@@ -375,6 +253,10 @@ impl Stream for ResponseStream {
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use crate::model_family::find_family_for_model;
|
||||
use codex_api::ResponsesApiRequest;
|
||||
use codex_api::common::OpenAiVerbosity;
|
||||
use codex_api::common::TextControls;
|
||||
use codex_api::create_text_param_for_request;
|
||||
use pretty_assertions::assert_eq;
|
||||
|
||||
use super::*;
|
||||
|
||||
+19
-37
@@ -102,6 +102,7 @@ use crate::protocol::TurnDiffEvent;
|
||||
use crate::protocol::WarningEvent;
|
||||
use crate::rollout::RolloutRecorder;
|
||||
use crate::rollout::RolloutRecorderParams;
|
||||
use crate::rollout::map_session_init_error;
|
||||
use crate::shell;
|
||||
use crate::state::ActiveTurn;
|
||||
use crate::state::SessionServices;
|
||||
@@ -206,7 +207,7 @@ impl Codex {
|
||||
.await
|
||||
.map_err(|e| {
|
||||
error!("Failed to create session: {e:#}");
|
||||
CodexErr::InternalAgentDied
|
||||
map_session_init_error(&e, &config.codex_home)
|
||||
})?;
|
||||
let conversation_id = session.conversation_id;
|
||||
|
||||
@@ -508,7 +509,7 @@ impl Session {
|
||||
|
||||
let rollout_recorder = rollout_recorder.map_err(|e| {
|
||||
error!("failed to initialize rollout recorder: {e:#}");
|
||||
anyhow::anyhow!("failed to initialize rollout recorder: {e:#}")
|
||||
anyhow::Error::from(e)
|
||||
})?;
|
||||
let rollout_path = rollout_recorder.rollout_path.clone();
|
||||
|
||||
@@ -1189,22 +1190,17 @@ impl Session {
|
||||
}
|
||||
}
|
||||
|
||||
/// Record a user input item to conversation history and also persist a
|
||||
/// corresponding UserMessage EventMsg to rollout.
|
||||
async fn record_input_and_rollout_usermsg(
|
||||
pub(crate) async fn record_response_item_and_emit_turn_item(
|
||||
&self,
|
||||
turn_context: &TurnContext,
|
||||
response_input: &ResponseInputItem,
|
||||
response_item: ResponseItem,
|
||||
) {
|
||||
let response_item: ResponseItem = response_input.clone().into();
|
||||
// Add to conversation history and persist response item to rollout
|
||||
// Add to conversation history and persist response item to rollout.
|
||||
self.record_conversation_items(turn_context, std::slice::from_ref(&response_item))
|
||||
.await;
|
||||
|
||||
// Derive user message events and persist only UserMessage to rollout
|
||||
let turn_item = parse_turn_item(&response_item);
|
||||
|
||||
if let Some(item @ TurnItem::UserMessage(_)) = turn_item {
|
||||
// Derive a turn item and emit lifecycle events if applicable.
|
||||
if let Some(item) = parse_turn_item(&response_item) {
|
||||
self.emit_turn_item_started(turn_context, &item).await;
|
||||
self.emit_turn_item_completed(turn_context, item).await;
|
||||
}
|
||||
@@ -1745,6 +1741,10 @@ mod handlers {
|
||||
|
||||
pub async fn shutdown(sess: &Arc<Session>, sub_id: String) -> bool {
|
||||
sess.abort_all_tasks(TurnAbortReason::Interrupted).await;
|
||||
sess.services
|
||||
.unified_exec_manager
|
||||
.terminate_all_sessions()
|
||||
.await;
|
||||
info!("Shutting down Codex instance");
|
||||
|
||||
// Gracefully flush and shutdown rollout recorder on session end so tests
|
||||
@@ -1876,12 +1876,7 @@ async fn spawn_review_thread(
|
||||
text: review_prompt,
|
||||
}];
|
||||
let tc = Arc::new(review_turn_context);
|
||||
sess.spawn_task(
|
||||
tc.clone(),
|
||||
input,
|
||||
ReviewTask::new(review_request.append_to_original_thread),
|
||||
)
|
||||
.await;
|
||||
sess.spawn_task(tc.clone(), input, ReviewTask::new()).await;
|
||||
|
||||
// Announce entering review mode so UIs can switch modes.
|
||||
sess.send_event(&tc, EventMsg::EnteredReviewMode(review_request))
|
||||
@@ -1917,7 +1912,8 @@ pub(crate) async fn run_task(
|
||||
sess.send_event(&turn_context, event).await;
|
||||
|
||||
let initial_input_for_turn: ResponseInputItem = ResponseInputItem::from(input);
|
||||
sess.record_input_and_rollout_usermsg(turn_context.as_ref(), &initial_input_for_turn)
|
||||
let response_item: ResponseItem = initial_input_for_turn.clone().into();
|
||||
sess.record_response_item_and_emit_turn_item(turn_context.as_ref(), response_item)
|
||||
.await;
|
||||
|
||||
sess.maybe_start_ghost_snapshot(Arc::clone(&turn_context), cancellation_token.child_token())
|
||||
@@ -2882,7 +2878,7 @@ mod tests {
|
||||
let input = vec![UserInput::Text {
|
||||
text: "start review".to_string(),
|
||||
}];
|
||||
sess.spawn_task(Arc::clone(&tc), input, ReviewTask::new(true))
|
||||
sess.spawn_task(Arc::clone(&tc), input, ReviewTask::new())
|
||||
.await;
|
||||
|
||||
sess.abort_all_tasks(TurnAbortReason::Interrupted).await;
|
||||
@@ -2910,6 +2906,8 @@ mod tests {
|
||||
.expect("event");
|
||||
match evt.msg {
|
||||
EventMsg::RawResponseItem(_) => continue,
|
||||
EventMsg::ItemStarted(_) | EventMsg::ItemCompleted(_) => continue,
|
||||
EventMsg::AgentMessage(_) => continue,
|
||||
EventMsg::TurnAborted(e) => {
|
||||
assert_eq!(TurnAbortReason::Interrupted, e.reason);
|
||||
break;
|
||||
@@ -2919,23 +2917,7 @@ mod tests {
|
||||
}
|
||||
|
||||
let history = sess.clone_history().await.get_history();
|
||||
let found = history.iter().any(|item| match item {
|
||||
ResponseItem::Message { role, content, .. } if role == "user" => {
|
||||
content.iter().any(|ci| match ci {
|
||||
ContentItem::InputText { text } => {
|
||||
text.contains("<user_action>")
|
||||
&& text.contains("review")
|
||||
&& text.contains("interrupted")
|
||||
}
|
||||
_ => false,
|
||||
})
|
||||
}
|
||||
_ => false,
|
||||
});
|
||||
assert!(
|
||||
found,
|
||||
"synthetic review interruption not recorded in history"
|
||||
);
|
||||
let _ = history;
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
|
||||
@@ -8,8 +8,8 @@ use crate::codex::get_last_assistant_message_from_turn;
|
||||
use crate::error::CodexErr;
|
||||
use crate::error::Result as CodexResult;
|
||||
use crate::features::Feature;
|
||||
use crate::protocol::AgentMessageEvent;
|
||||
use crate::protocol::CompactedItem;
|
||||
use crate::protocol::ContextCompactedEvent;
|
||||
use crate::protocol::EventMsg;
|
||||
use crate::protocol::TaskStartedEvent;
|
||||
use crate::protocol::TurnContextItem;
|
||||
@@ -175,9 +175,7 @@ async fn run_compact_task_inner(
|
||||
});
|
||||
sess.persist_rollout_items(&[rollout_item]).await;
|
||||
|
||||
let event = EventMsg::AgentMessage(AgentMessageEvent {
|
||||
message: "Compact task completed".to_string(),
|
||||
});
|
||||
let event = EventMsg::ContextCompacted(ContextCompactedEvent {});
|
||||
sess.send_event(&turn_context, event).await;
|
||||
|
||||
let warning = EventMsg::Warning(WarningEvent {
|
||||
|
||||
@@ -4,8 +4,8 @@ use crate::Prompt;
|
||||
use crate::codex::Session;
|
||||
use crate::codex::TurnContext;
|
||||
use crate::error::Result as CodexResult;
|
||||
use crate::protocol::AgentMessageEvent;
|
||||
use crate::protocol::CompactedItem;
|
||||
use crate::protocol::ContextCompactedEvent;
|
||||
use crate::protocol::EventMsg;
|
||||
use crate::protocol::RolloutItem;
|
||||
use crate::protocol::TaskStartedEvent;
|
||||
@@ -74,9 +74,7 @@ async fn run_remote_compact_task_inner_impl(
|
||||
sess.persist_rollout_items(&[RolloutItem::Compacted(compacted_item)])
|
||||
.await;
|
||||
|
||||
let event = EventMsg::AgentMessage(AgentMessageEvent {
|
||||
message: "Compact task completed".to_string(),
|
||||
});
|
||||
let event = EventMsg::ContextCompacted(ContextCompactedEvent {});
|
||||
sess.send_event(turn_context, event).await;
|
||||
|
||||
Ok(())
|
||||
|
||||
@@ -70,7 +70,7 @@ pub const GPT_5_CODEX_MEDIUM_MODEL: &str = "gpt-5.1-codex";
|
||||
/// the context window.
|
||||
pub(crate) const PROJECT_DOC_MAX_BYTES: usize = 32 * 1024; // 32 KiB
|
||||
|
||||
pub(crate) const CONFIG_TOML_FILE: &str = "config.toml";
|
||||
pub const CONFIG_TOML_FILE: &str = "config.toml";
|
||||
|
||||
/// Application configuration loaded from disk and merged with overrides.
|
||||
#[derive(Debug, Clone, PartialEq)]
|
||||
|
||||
@@ -11,15 +11,15 @@ use toml::Value as TomlValue;
|
||||
#[cfg(unix)]
|
||||
const CODEX_MANAGED_CONFIG_SYSTEM_PATH: &str = "/etc/codex/managed_config.toml";
|
||||
|
||||
#[derive(Debug)]
|
||||
pub(crate) struct LoadedConfigLayers {
|
||||
#[derive(Debug, Clone)]
|
||||
pub struct LoadedConfigLayers {
|
||||
pub base: TomlValue,
|
||||
pub managed_config: Option<TomlValue>,
|
||||
pub managed_preferences: Option<TomlValue>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Default)]
|
||||
pub(crate) struct LoaderOverrides {
|
||||
#[derive(Debug, Default, Clone)]
|
||||
pub struct LoaderOverrides {
|
||||
pub managed_config_path: Option<PathBuf>,
|
||||
#[cfg(target_os = "macos")]
|
||||
pub managed_preferences_base64: Option<String>,
|
||||
@@ -47,11 +47,15 @@ pub async fn load_config_as_toml(codex_home: &Path) -> io::Result<TomlValue> {
|
||||
load_config_as_toml_with_overrides(codex_home, LoaderOverrides::default()).await
|
||||
}
|
||||
|
||||
pub async fn load_config_layers(codex_home: &Path) -> io::Result<LoadedConfigLayers> {
|
||||
load_config_layers_with_overrides(codex_home, LoaderOverrides::default()).await
|
||||
}
|
||||
|
||||
fn default_empty_table() -> TomlValue {
|
||||
TomlValue::Table(Default::default())
|
||||
}
|
||||
|
||||
pub(crate) async fn load_config_layers_with_overrides(
|
||||
pub async fn load_config_layers_with_overrides(
|
||||
codex_home: &Path,
|
||||
overrides: LoaderOverrides,
|
||||
) -> io::Result<LoadedConfigLayers> {
|
||||
@@ -130,7 +134,7 @@ async fn read_config_from_path(
|
||||
}
|
||||
|
||||
/// Merge config `overlay` into `base`, giving `overlay` precedence.
|
||||
pub(crate) fn merge_toml_values(base: &mut TomlValue, overlay: &TomlValue) {
|
||||
pub fn merge_toml_values(base: &mut TomlValue, overlay: &TomlValue) {
|
||||
if let TomlValue::Table(overlay_table) = overlay
|
||||
&& let TomlValue::Table(base_table) = base
|
||||
{
|
||||
@@ -147,6 +151,10 @@ pub(crate) fn merge_toml_values(base: &mut TomlValue, overlay: &TomlValue) {
|
||||
}
|
||||
|
||||
fn managed_config_default_path(codex_home: &Path) -> PathBuf {
|
||||
if let Ok(path) = std::env::var("CODEX_MANAGED_CONFIG_PATH") {
|
||||
return PathBuf::from(path);
|
||||
}
|
||||
|
||||
#[cfg(unix)]
|
||||
{
|
||||
let _ = codex_home;
|
||||
|
||||
@@ -258,6 +258,11 @@ fn sanitize_user_agent(candidate: String, fallback: &str) -> String {
|
||||
|
||||
/// Create an HTTP client with default `originator` and `User-Agent` headers set.
|
||||
pub fn create_client() -> CodexHttpClient {
|
||||
let inner = build_reqwest_client();
|
||||
CodexHttpClient::new(inner)
|
||||
}
|
||||
|
||||
pub fn build_reqwest_client() -> reqwest::Client {
|
||||
use reqwest::header::HeaderMap;
|
||||
|
||||
let mut headers = HeaderMap::new();
|
||||
@@ -272,8 +277,7 @@ pub fn create_client() -> CodexHttpClient {
|
||||
builder = builder.no_proxy();
|
||||
}
|
||||
|
||||
let inner = builder.build().unwrap_or_else(|_| reqwest::Client::new());
|
||||
CodexHttpClient::new(inner)
|
||||
builder.build().unwrap_or_else(|_| reqwest::Client::new())
|
||||
}
|
||||
|
||||
fn is_sandboxed() -> bool {
|
||||
|
||||
@@ -5,10 +5,10 @@
|
||||
// the TUI or the tracing stack).
|
||||
#![deny(clippy::print_stdout, clippy::print_stderr)]
|
||||
|
||||
pub mod api_bridge;
|
||||
mod apply_patch;
|
||||
pub mod auth;
|
||||
pub mod bash;
|
||||
mod chat_completions;
|
||||
mod client;
|
||||
mod client_common;
|
||||
pub mod codex;
|
||||
|
||||
@@ -5,11 +5,13 @@
|
||||
//! 2. User-defined entries inside `~/.codex/config.toml` under the `model_providers`
|
||||
//! key. These override or extend the defaults at runtime.
|
||||
|
||||
use crate::CodexAuth;
|
||||
use crate::default_client::CodexHttpClient;
|
||||
use crate::default_client::CodexRequestBuilder;
|
||||
use crate::error::CodexErr;
|
||||
use codex_api::Provider as ApiProvider;
|
||||
use codex_api::WireApi as ApiWireApi;
|
||||
use codex_api::provider::RetryConfig as ApiRetryConfig;
|
||||
use codex_app_server_protocol::AuthMode;
|
||||
use http::HeaderMap;
|
||||
use http::header::HeaderName;
|
||||
use http::header::HeaderValue;
|
||||
use serde::Deserialize;
|
||||
use serde::Serialize;
|
||||
use std::collections::HashMap;
|
||||
@@ -97,148 +99,14 @@ pub struct ModelProviderInfo {
|
||||
}
|
||||
|
||||
impl ModelProviderInfo {
|
||||
/// Construct a `POST` RequestBuilder for the given URL using the provided
|
||||
/// [`CodexHttpClient`] applying:
|
||||
/// • provider-specific headers (static + env based)
|
||||
/// • Bearer auth header when an API key is available.
|
||||
/// • Auth token for OAuth.
|
||||
///
|
||||
/// If the provider declares an `env_key` but the variable is missing/empty, returns an [`Err`] identical to the
|
||||
/// one produced by [`ModelProviderInfo::api_key`].
|
||||
pub async fn create_request_builder<'a>(
|
||||
&'a self,
|
||||
client: &'a CodexHttpClient,
|
||||
auth: &Option<CodexAuth>,
|
||||
) -> crate::error::Result<CodexRequestBuilder> {
|
||||
let effective_auth = self.effective_auth(auth)?;
|
||||
|
||||
let url = self.get_full_url(&effective_auth);
|
||||
|
||||
let mut builder = client.post(url);
|
||||
|
||||
if let Some(auth) = effective_auth.as_ref() {
|
||||
builder = builder.bearer_auth(auth.get_token().await?);
|
||||
}
|
||||
|
||||
Ok(self.apply_http_headers(builder))
|
||||
}
|
||||
|
||||
pub async fn create_compact_request_builder<'a>(
|
||||
&'a self,
|
||||
client: &'a CodexHttpClient,
|
||||
auth: &Option<CodexAuth>,
|
||||
) -> crate::error::Result<CodexRequestBuilder> {
|
||||
if self.wire_api != WireApi::Responses {
|
||||
return Err(CodexErr::UnsupportedOperation(
|
||||
"Compaction endpoint requires Responses API providers".to_string(),
|
||||
));
|
||||
}
|
||||
let effective_auth = self.effective_auth(auth)?;
|
||||
|
||||
let url = self.get_compact_url(&effective_auth).ok_or_else(|| {
|
||||
CodexErr::UnsupportedOperation(
|
||||
"Compaction endpoint requires Responses API providers".to_string(),
|
||||
)
|
||||
})?;
|
||||
|
||||
let mut builder = client.post(url);
|
||||
|
||||
if let Some(auth) = effective_auth.as_ref() {
|
||||
builder = builder.bearer_auth(auth.get_token().await?);
|
||||
}
|
||||
|
||||
Ok(self.apply_http_headers(builder))
|
||||
}
|
||||
|
||||
fn effective_auth(&self, auth: &Option<CodexAuth>) -> crate::error::Result<Option<CodexAuth>> {
|
||||
if let Some(secret_key) = &self.experimental_bearer_token {
|
||||
return Ok(Some(CodexAuth::from_api_key(secret_key)));
|
||||
}
|
||||
|
||||
match self.api_key() {
|
||||
Ok(Some(key)) => Ok(Some(CodexAuth::from_api_key(&key))),
|
||||
Ok(None) => Ok(auth.clone()),
|
||||
Err(err) => {
|
||||
if auth.is_some() {
|
||||
Ok(auth.clone())
|
||||
} else {
|
||||
Err(err)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn get_query_string(&self) -> String {
|
||||
self.query_params
|
||||
.as_ref()
|
||||
.map_or_else(String::new, |params| {
|
||||
let full_params = params
|
||||
.iter()
|
||||
.map(|(k, v)| format!("{k}={v}"))
|
||||
.collect::<Vec<_>>()
|
||||
.join("&");
|
||||
format!("?{full_params}")
|
||||
})
|
||||
}
|
||||
|
||||
pub(crate) fn get_full_url(&self, auth: &Option<CodexAuth>) -> String {
|
||||
let default_base_url = if matches!(
|
||||
auth,
|
||||
Some(CodexAuth {
|
||||
mode: AuthMode::ChatGPT,
|
||||
..
|
||||
})
|
||||
) {
|
||||
"https://chatgpt.com/backend-api/codex"
|
||||
} else {
|
||||
"https://api.openai.com/v1"
|
||||
};
|
||||
let query_string = self.get_query_string();
|
||||
let base_url = self
|
||||
.base_url
|
||||
.clone()
|
||||
.unwrap_or(default_base_url.to_string());
|
||||
|
||||
match self.wire_api {
|
||||
WireApi::Responses => format!("{base_url}/responses{query_string}"),
|
||||
WireApi::Chat => format!("{base_url}/chat/completions{query_string}"),
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn get_compact_url(&self, auth: &Option<CodexAuth>) -> Option<String> {
|
||||
if self.wire_api != WireApi::Responses {
|
||||
return None;
|
||||
}
|
||||
let full = self.get_full_url(auth);
|
||||
if let Some((path, query)) = full.split_once('?') {
|
||||
Some(format!("{path}/compact?{query}"))
|
||||
} else {
|
||||
Some(format!("{full}/compact"))
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) fn is_azure_responses_endpoint(&self) -> bool {
|
||||
if self.wire_api != WireApi::Responses {
|
||||
return false;
|
||||
}
|
||||
|
||||
if self.name.eq_ignore_ascii_case("azure") {
|
||||
return true;
|
||||
}
|
||||
|
||||
self.base_url
|
||||
.as_ref()
|
||||
.map(|base| matches_azure_responses_base_url(base))
|
||||
.unwrap_or(false)
|
||||
}
|
||||
|
||||
/// Apply provider-specific HTTP headers (both static and environment-based)
|
||||
/// onto an existing [`CodexRequestBuilder`] and return the updated
|
||||
/// builder.
|
||||
fn apply_http_headers(&self, mut builder: CodexRequestBuilder) -> CodexRequestBuilder {
|
||||
#[allow(dead_code)]
|
||||
fn build_header_map(&self) -> crate::error::Result<HeaderMap> {
|
||||
let mut headers = HeaderMap::new();
|
||||
if let Some(extra) = &self.http_headers {
|
||||
for (k, v) in extra {
|
||||
builder = builder.header(k, v);
|
||||
if let (Ok(name), Ok(value)) = (HeaderName::try_from(k), HeaderValue::try_from(v)) {
|
||||
headers.insert(name, value);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -246,12 +114,52 @@ impl ModelProviderInfo {
|
||||
for (header, env_var) in env_headers {
|
||||
if let Ok(val) = std::env::var(env_var)
|
||||
&& !val.trim().is_empty()
|
||||
&& let (Ok(name), Ok(value)) =
|
||||
(HeaderName::try_from(header), HeaderValue::try_from(val))
|
||||
{
|
||||
builder = builder.header(header, val);
|
||||
headers.insert(name, value);
|
||||
}
|
||||
}
|
||||
}
|
||||
builder
|
||||
|
||||
Ok(headers)
|
||||
}
|
||||
|
||||
pub(crate) fn to_api_provider(
|
||||
&self,
|
||||
auth_mode: Option<AuthMode>,
|
||||
) -> crate::error::Result<ApiProvider> {
|
||||
let default_base_url = if matches!(auth_mode, Some(AuthMode::ChatGPT)) {
|
||||
"https://chatgpt.com/backend-api/codex"
|
||||
} else {
|
||||
"https://api.openai.com/v1"
|
||||
};
|
||||
let base_url = self
|
||||
.base_url
|
||||
.clone()
|
||||
.unwrap_or_else(|| default_base_url.to_string());
|
||||
|
||||
let headers = self.build_header_map()?;
|
||||
let retry = ApiRetryConfig {
|
||||
max_attempts: self.request_max_retries(),
|
||||
base_delay: Duration::from_millis(200),
|
||||
retry_429: false,
|
||||
retry_5xx: true,
|
||||
retry_transport: true,
|
||||
};
|
||||
|
||||
Ok(ApiProvider {
|
||||
name: self.name.clone(),
|
||||
base_url,
|
||||
query_params: self.query_params.clone(),
|
||||
wire: match self.wire_api {
|
||||
WireApi::Responses => ApiWireApi::Responses,
|
||||
WireApi::Chat => ApiWireApi::Chat,
|
||||
},
|
||||
headers,
|
||||
retry,
|
||||
stream_idle_timeout: self.stream_idle_timeout(),
|
||||
})
|
||||
}
|
||||
|
||||
/// If `env_key` is Some, returns the API key for this provider if present
|
||||
@@ -409,18 +317,6 @@ pub fn create_oss_provider_with_base_url(base_url: &str, wire_api: WireApi) -> M
|
||||
}
|
||||
}
|
||||
|
||||
fn matches_azure_responses_base_url(base_url: &str) -> bool {
|
||||
let base = base_url.to_ascii_lowercase();
|
||||
const AZURE_MARKERS: [&str; 5] = [
|
||||
"openai.azure.",
|
||||
"cognitiveservices.azure.",
|
||||
"aoai.azure.",
|
||||
"azure-api.",
|
||||
"azurefd.",
|
||||
];
|
||||
AZURE_MARKERS.iter().any(|marker| base.contains(marker))
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
@@ -517,8 +413,16 @@ env_http_headers = { "X-Example-Env-Header" = "EXAMPLE_ENV_VAR" }
|
||||
|
||||
#[test]
|
||||
fn detects_azure_responses_base_urls() {
|
||||
fn provider_for(base_url: &str) -> ModelProviderInfo {
|
||||
ModelProviderInfo {
|
||||
let positive_cases = [
|
||||
"https://foo.openai.azure.com/openai",
|
||||
"https://foo.openai.azure.us/openai/deployments/bar",
|
||||
"https://foo.cognitiveservices.azure.cn/openai",
|
||||
"https://foo.aoai.azure.com/openai",
|
||||
"https://foo.openai.azure-api.net/openai",
|
||||
"https://foo.z01.azurefd.net/",
|
||||
];
|
||||
for base_url in positive_cases {
|
||||
let provider = ModelProviderInfo {
|
||||
name: "test".into(),
|
||||
base_url: Some(base_url.into()),
|
||||
env_key: None,
|
||||
@@ -532,21 +436,10 @@ env_http_headers = { "X-Example-Env-Header" = "EXAMPLE_ENV_VAR" }
|
||||
stream_max_retries: None,
|
||||
stream_idle_timeout_ms: None,
|
||||
requires_openai_auth: false,
|
||||
}
|
||||
}
|
||||
|
||||
let positive_cases = [
|
||||
"https://foo.openai.azure.com/openai",
|
||||
"https://foo.openai.azure.us/openai/deployments/bar",
|
||||
"https://foo.cognitiveservices.azure.cn/openai",
|
||||
"https://foo.aoai.azure.com/openai",
|
||||
"https://foo.openai.azure-api.net/openai",
|
||||
"https://foo.z01.azurefd.net/",
|
||||
];
|
||||
for base_url in positive_cases {
|
||||
let provider = provider_for(base_url);
|
||||
};
|
||||
let api = provider.to_api_provider(None).expect("api provider");
|
||||
assert!(
|
||||
provider.is_azure_responses_endpoint(),
|
||||
api.is_azure_responses_endpoint(),
|
||||
"expected {base_url} to be detected as Azure"
|
||||
);
|
||||
}
|
||||
@@ -566,7 +459,8 @@ env_http_headers = { "X-Example-Env-Header" = "EXAMPLE_ENV_VAR" }
|
||||
stream_idle_timeout_ms: None,
|
||||
requires_openai_auth: false,
|
||||
};
|
||||
assert!(named_provider.is_azure_responses_endpoint());
|
||||
let named_api = named_provider.to_api_provider(None).expect("api provider");
|
||||
assert!(named_api.is_azure_responses_endpoint());
|
||||
|
||||
let negative_cases = [
|
||||
"https://api.openai.com/v1",
|
||||
@@ -574,9 +468,24 @@ env_http_headers = { "X-Example-Env-Header" = "EXAMPLE_ENV_VAR" }
|
||||
"https://myproxy.azurewebsites.net/openai",
|
||||
];
|
||||
for base_url in negative_cases {
|
||||
let provider = provider_for(base_url);
|
||||
let provider = ModelProviderInfo {
|
||||
name: "test".into(),
|
||||
base_url: Some(base_url.into()),
|
||||
env_key: None,
|
||||
env_key_instructions: None,
|
||||
experimental_bearer_token: None,
|
||||
wire_api: WireApi::Responses,
|
||||
query_params: None,
|
||||
http_headers: None,
|
||||
env_http_headers: None,
|
||||
request_max_retries: None,
|
||||
stream_max_retries: None,
|
||||
stream_idle_timeout_ms: None,
|
||||
requires_openai_auth: false,
|
||||
};
|
||||
let api = provider.to_api_provider(None).expect("api provider");
|
||||
assert!(
|
||||
!provider.is_azure_responses_endpoint(),
|
||||
!api.is_azure_responses_endpoint(),
|
||||
"expected {base_url} not to be detected as Azure"
|
||||
);
|
||||
}
|
||||
|
||||
@@ -1,4 +1,5 @@
|
||||
use crate::protocol::ReviewFinding;
|
||||
use crate::protocol::ReviewOutputEvent;
|
||||
|
||||
// Note: We keep this module UI-agnostic. It returns plain strings that
|
||||
// higher layers (e.g., TUI) may style as needed.
|
||||
@@ -10,6 +11,8 @@ fn format_location(item: &ReviewFinding) -> String {
|
||||
format!("{path}:{start}-{end}")
|
||||
}
|
||||
|
||||
const REVIEW_FALLBACK_MESSAGE: &str = "Reviewer failed to output a response.";
|
||||
|
||||
/// Format a full review findings block as plain text lines.
|
||||
///
|
||||
/// - When `selection` is `Some`, each item line includes a checkbox marker:
|
||||
@@ -53,3 +56,27 @@ pub fn format_review_findings_block(
|
||||
|
||||
lines.join("\n")
|
||||
}
|
||||
|
||||
/// Render a human-readable review summary suitable for a user-facing message.
|
||||
///
|
||||
/// Returns either the explanation, the formatted findings block, or both
|
||||
/// separated by a blank line. If neither is present, emits a fallback message.
|
||||
pub fn render_review_output_text(output: &ReviewOutputEvent) -> String {
|
||||
let mut sections = Vec::new();
|
||||
let explanation = output.overall_explanation.trim();
|
||||
if !explanation.is_empty() {
|
||||
sections.push(explanation.to_string());
|
||||
}
|
||||
if !output.findings.is_empty() {
|
||||
let findings = format_review_findings_block(&output.findings, None);
|
||||
let trimmed = findings.trim();
|
||||
if !trimmed.is_empty() {
|
||||
sections.push(trimmed.to_string());
|
||||
}
|
||||
}
|
||||
if sections.is_empty() {
|
||||
REVIEW_FALLBACK_MESSAGE.to_string()
|
||||
} else {
|
||||
sections.join("\n\n")
|
||||
}
|
||||
}
|
||||
|
||||
@@ -0,0 +1,49 @@
|
||||
use std::io::ErrorKind;
|
||||
use std::path::Path;
|
||||
|
||||
use crate::error::CodexErr;
|
||||
use crate::rollout::SESSIONS_SUBDIR;
|
||||
|
||||
pub(crate) fn map_session_init_error(err: &anyhow::Error, codex_home: &Path) -> CodexErr {
|
||||
if let Some(mapped) = err
|
||||
.chain()
|
||||
.filter_map(|cause| cause.downcast_ref::<std::io::Error>())
|
||||
.find_map(|io_err| map_rollout_io_error(io_err, codex_home))
|
||||
{
|
||||
return mapped;
|
||||
}
|
||||
|
||||
CodexErr::Fatal(format!("Failed to initialize session: {err:#}"))
|
||||
}
|
||||
|
||||
fn map_rollout_io_error(io_err: &std::io::Error, codex_home: &Path) -> Option<CodexErr> {
|
||||
let sessions_dir = codex_home.join(SESSIONS_SUBDIR);
|
||||
let hint = match io_err.kind() {
|
||||
ErrorKind::PermissionDenied => format!(
|
||||
"Codex cannot access session files at {} (permission denied). If sessions were created using sudo, fix ownership: sudo chown -R $(whoami) {}",
|
||||
sessions_dir.display(),
|
||||
codex_home.display()
|
||||
),
|
||||
ErrorKind::NotFound => format!(
|
||||
"Session storage missing at {}. Create the directory or choose a different Codex home.",
|
||||
sessions_dir.display()
|
||||
),
|
||||
ErrorKind::AlreadyExists => format!(
|
||||
"Session storage path {} is blocked by an existing file. Remove or rename it so Codex can create sessions.",
|
||||
sessions_dir.display()
|
||||
),
|
||||
ErrorKind::InvalidData | ErrorKind::InvalidInput => format!(
|
||||
"Session data under {} looks corrupt or unreadable. Clearing the sessions directory may help (this will remove saved conversations).",
|
||||
sessions_dir.display()
|
||||
),
|
||||
ErrorKind::IsADirectory | ErrorKind::NotADirectory => format!(
|
||||
"Session storage path {} has an unexpected type. Ensure it is a directory Codex can use for session files.",
|
||||
sessions_dir.display()
|
||||
),
|
||||
_ => return None,
|
||||
};
|
||||
|
||||
Some(CodexErr::Fatal(format!(
|
||||
"{hint} (underlying error: {io_err})"
|
||||
)))
|
||||
}
|
||||
@@ -7,11 +7,13 @@ pub const ARCHIVED_SESSIONS_SUBDIR: &str = "archived_sessions";
|
||||
pub const INTERACTIVE_SESSION_SOURCES: &[SessionSource] =
|
||||
&[SessionSource::Cli, SessionSource::VSCode];
|
||||
|
||||
pub(crate) mod error;
|
||||
pub mod list;
|
||||
pub(crate) mod policy;
|
||||
pub mod recorder;
|
||||
|
||||
pub use codex_protocol::protocol::SessionMeta;
|
||||
pub(crate) use error::map_session_init_error;
|
||||
pub use list::find_conversation_path_by_id_str;
|
||||
pub use recorder::RolloutRecorder;
|
||||
pub use recorder::RolloutRecorderParams;
|
||||
|
||||
@@ -42,6 +42,7 @@ pub(crate) fn should_persist_event_msg(ev: &EventMsg) -> bool {
|
||||
| EventMsg::AgentReasoning(_)
|
||||
| EventMsg::AgentReasoningRawContent(_)
|
||||
| EventMsg::TokenCount(_)
|
||||
| EventMsg::ContextCompacted(_)
|
||||
| EventMsg::EnteredReviewMode(_)
|
||||
| EventMsg::ExitedReviewMode(_)
|
||||
| EventMsg::UndoCompleted(_)
|
||||
|
||||
@@ -14,6 +14,7 @@ use crate::protocol::SandboxPolicy;
|
||||
use askama::Template;
|
||||
use codex_otel::otel_event_manager::OtelEventManager;
|
||||
use codex_protocol::ConversationId;
|
||||
use codex_protocol::config_types::ReasoningEffort as ReasoningEffortConfig;
|
||||
use codex_protocol::models::ContentItem;
|
||||
use codex_protocol::models::ResponseItem;
|
||||
use codex_protocol::protocol::SandboxCommandAssessment;
|
||||
@@ -23,7 +24,8 @@ use serde_json::json;
|
||||
use tokio::time::timeout;
|
||||
use tracing::warn;
|
||||
|
||||
const SANDBOX_ASSESSMENT_TIMEOUT: Duration = Duration::from_secs(5);
|
||||
const SANDBOX_ASSESSMENT_TIMEOUT: Duration = Duration::from_secs(15);
|
||||
const SANDBOX_ASSESSMENT_REASONING_EFFORT: ReasoningEffortConfig = ReasoningEffortConfig::Medium;
|
||||
|
||||
#[derive(Template)]
|
||||
#[template(path = "sandboxing/assessment_prompt.md", escape = "none")]
|
||||
@@ -130,7 +132,7 @@ pub(crate) async fn assess_command(
|
||||
Some(auth_manager),
|
||||
child_otel,
|
||||
provider,
|
||||
config.model_reasoning_effort,
|
||||
Some(SANDBOX_ASSESSMENT_REASONING_EFFORT),
|
||||
config.model_reasoning_summary,
|
||||
conversation_id,
|
||||
session_source,
|
||||
|
||||
@@ -17,6 +17,7 @@ use crate::codex::Session;
|
||||
use crate::codex::TurnContext;
|
||||
use crate::codex_delegate::run_codex_conversation_one_shot;
|
||||
use crate::review_format::format_review_findings_block;
|
||||
use crate::review_format::render_review_output_text;
|
||||
use crate::state::TaskKind;
|
||||
use codex_protocol::user_input::UserInput;
|
||||
|
||||
@@ -24,15 +25,11 @@ use super::SessionTask;
|
||||
use super::SessionTaskContext;
|
||||
|
||||
#[derive(Clone, Copy)]
|
||||
pub(crate) struct ReviewTask {
|
||||
append_to_original_thread: bool,
|
||||
}
|
||||
pub(crate) struct ReviewTask;
|
||||
|
||||
impl ReviewTask {
|
||||
pub(crate) fn new(append_to_original_thread: bool) -> Self {
|
||||
Self {
|
||||
append_to_original_thread,
|
||||
}
|
||||
pub(crate) fn new() -> Self {
|
||||
Self
|
||||
}
|
||||
}
|
||||
|
||||
@@ -62,25 +59,13 @@ impl SessionTask for ReviewTask {
|
||||
None => None,
|
||||
};
|
||||
if !cancellation_token.is_cancelled() {
|
||||
exit_review_mode(
|
||||
session.clone_session(),
|
||||
output.clone(),
|
||||
ctx.clone(),
|
||||
self.append_to_original_thread,
|
||||
)
|
||||
.await;
|
||||
exit_review_mode(session.clone_session(), output.clone(), ctx.clone()).await;
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
async fn abort(&self, session: Arc<SessionTaskContext>, ctx: Arc<TurnContext>) {
|
||||
exit_review_mode(
|
||||
session.clone_session(),
|
||||
None,
|
||||
ctx,
|
||||
self.append_to_original_thread,
|
||||
)
|
||||
.await;
|
||||
exit_review_mode(session.clone_session(), None, ctx).await;
|
||||
}
|
||||
}
|
||||
|
||||
@@ -197,39 +182,57 @@ pub(crate) async fn exit_review_mode(
|
||||
session: Arc<Session>,
|
||||
review_output: Option<ReviewOutputEvent>,
|
||||
ctx: Arc<TurnContext>,
|
||||
append_to_original_thread: bool,
|
||||
) {
|
||||
if append_to_original_thread {
|
||||
let user_message = if let Some(out) = review_output.clone() {
|
||||
let mut findings_str = String::new();
|
||||
let text = out.overall_explanation.trim();
|
||||
if !text.is_empty() {
|
||||
findings_str.push_str(text);
|
||||
}
|
||||
if !out.findings.is_empty() {
|
||||
let block = format_review_findings_block(&out.findings, None);
|
||||
findings_str.push_str(&format!("\n{block}"));
|
||||
}
|
||||
crate::client_common::REVIEW_EXIT_SUCCESS_TMPL.replace("{results}", &findings_str)
|
||||
} else {
|
||||
crate::client_common::REVIEW_EXIT_INTERRUPTED_TMPL.to_string()
|
||||
};
|
||||
const REVIEW_USER_MESSAGE_ID: &str = "review:rollout:user";
|
||||
const REVIEW_ASSISTANT_MESSAGE_ID: &str = "review:rollout:assistant";
|
||||
let (user_message, assistant_message) = if let Some(out) = review_output.clone() {
|
||||
let mut findings_str = String::new();
|
||||
let text = out.overall_explanation.trim();
|
||||
if !text.is_empty() {
|
||||
findings_str.push_str(text);
|
||||
}
|
||||
if !out.findings.is_empty() {
|
||||
let block = format_review_findings_block(&out.findings, None);
|
||||
findings_str.push_str(&format!("\n{block}"));
|
||||
}
|
||||
let rendered =
|
||||
crate::client_common::REVIEW_EXIT_SUCCESS_TMPL.replace("{results}", &findings_str);
|
||||
let assistant_message = render_review_output_text(&out);
|
||||
(rendered, assistant_message)
|
||||
} else {
|
||||
let rendered = crate::client_common::REVIEW_EXIT_INTERRUPTED_TMPL.to_string();
|
||||
let assistant_message =
|
||||
"Review was interrupted. Please re-run /review and wait for it to complete."
|
||||
.to_string();
|
||||
(rendered, assistant_message)
|
||||
};
|
||||
|
||||
session
|
||||
.record_conversation_items(
|
||||
&ctx,
|
||||
&[ResponseItem::Message {
|
||||
id: None,
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText { text: user_message }],
|
||||
}],
|
||||
)
|
||||
.await;
|
||||
}
|
||||
session
|
||||
.record_conversation_items(
|
||||
&ctx,
|
||||
&[ResponseItem::Message {
|
||||
id: Some(REVIEW_USER_MESSAGE_ID.to_string()),
|
||||
role: "user".to_string(),
|
||||
content: vec![ContentItem::InputText { text: user_message }],
|
||||
}],
|
||||
)
|
||||
.await;
|
||||
session
|
||||
.send_event(
|
||||
ctx.as_ref(),
|
||||
EventMsg::ExitedReviewMode(ExitedReviewModeEvent { review_output }),
|
||||
)
|
||||
.await;
|
||||
session
|
||||
.record_response_item_and_emit_turn_item(
|
||||
ctx.as_ref(),
|
||||
ResponseItem::Message {
|
||||
id: Some(REVIEW_ASSISTANT_MESSAGE_ID.to_string()),
|
||||
role: "assistant".to_string(),
|
||||
content: vec![ContentItem::OutputText {
|
||||
text: assistant_message,
|
||||
}],
|
||||
},
|
||||
)
|
||||
.await;
|
||||
}
|
||||
|
||||
@@ -81,6 +81,7 @@ impl SessionTask for UserShellCommandTask {
|
||||
turn_context.as_ref(),
|
||||
EventMsg::ExecCommandBegin(ExecCommandBeginEvent {
|
||||
call_id: call_id.clone(),
|
||||
process_id: None,
|
||||
turn_id: turn_context.sub_id.clone(),
|
||||
command: command.clone(),
|
||||
cwd: cwd.clone(),
|
||||
@@ -139,6 +140,7 @@ impl SessionTask for UserShellCommandTask {
|
||||
turn_context.as_ref(),
|
||||
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
|
||||
call_id,
|
||||
process_id: None,
|
||||
turn_id: turn_context.sub_id.clone(),
|
||||
command: command.clone(),
|
||||
cwd: cwd.clone(),
|
||||
@@ -161,6 +163,7 @@ impl SessionTask for UserShellCommandTask {
|
||||
turn_context.as_ref(),
|
||||
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
|
||||
call_id: call_id.clone(),
|
||||
process_id: None,
|
||||
turn_id: turn_context.sub_id.clone(),
|
||||
command: command.clone(),
|
||||
cwd: cwd.clone(),
|
||||
@@ -205,6 +208,7 @@ impl SessionTask for UserShellCommandTask {
|
||||
turn_context.as_ref(),
|
||||
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
|
||||
call_id,
|
||||
process_id: None,
|
||||
turn_id: turn_context.sub_id.clone(),
|
||||
command,
|
||||
cwd,
|
||||
|
||||
@@ -65,12 +65,14 @@ pub(crate) async fn emit_exec_command_begin(
|
||||
parsed_cmd: &[ParsedCommand],
|
||||
source: ExecCommandSource,
|
||||
interaction_input: Option<String>,
|
||||
process_id: Option<&str>,
|
||||
) {
|
||||
ctx.session
|
||||
.send_event(
|
||||
ctx.turn,
|
||||
EventMsg::ExecCommandBegin(ExecCommandBeginEvent {
|
||||
call_id: ctx.call_id.to_string(),
|
||||
process_id: process_id.map(str::to_owned),
|
||||
turn_id: ctx.turn.sub_id.clone(),
|
||||
command: command.to_vec(),
|
||||
cwd: cwd.to_path_buf(),
|
||||
@@ -100,6 +102,7 @@ pub(crate) enum ToolEmitter {
|
||||
source: ExecCommandSource,
|
||||
interaction_input: Option<String>,
|
||||
parsed_cmd: Vec<ParsedCommand>,
|
||||
process_id: Option<String>,
|
||||
},
|
||||
}
|
||||
|
||||
@@ -132,6 +135,7 @@ impl ToolEmitter {
|
||||
cwd: PathBuf,
|
||||
source: ExecCommandSource,
|
||||
interaction_input: Option<String>,
|
||||
process_id: Option<String>,
|
||||
) -> Self {
|
||||
let parsed_cmd = parse_command(command);
|
||||
Self::UnifiedExec {
|
||||
@@ -140,6 +144,7 @@ impl ToolEmitter {
|
||||
source,
|
||||
interaction_input,
|
||||
parsed_cmd,
|
||||
process_id,
|
||||
}
|
||||
}
|
||||
|
||||
@@ -157,7 +162,7 @@ impl ToolEmitter {
|
||||
) => {
|
||||
emit_exec_stage(
|
||||
ctx,
|
||||
ExecCommandInput::new(command, cwd.as_path(), parsed_cmd, *source, None),
|
||||
ExecCommandInput::new(command, cwd.as_path(), parsed_cmd, *source, None, None),
|
||||
stage,
|
||||
)
|
||||
.await;
|
||||
@@ -229,6 +234,7 @@ impl ToolEmitter {
|
||||
source,
|
||||
interaction_input,
|
||||
parsed_cmd,
|
||||
process_id,
|
||||
},
|
||||
stage,
|
||||
) => {
|
||||
@@ -240,6 +246,7 @@ impl ToolEmitter {
|
||||
parsed_cmd,
|
||||
*source,
|
||||
interaction_input.as_deref(),
|
||||
process_id.as_deref(),
|
||||
),
|
||||
stage,
|
||||
)
|
||||
@@ -319,6 +326,7 @@ struct ExecCommandInput<'a> {
|
||||
parsed_cmd: &'a [ParsedCommand],
|
||||
source: ExecCommandSource,
|
||||
interaction_input: Option<&'a str>,
|
||||
process_id: Option<&'a str>,
|
||||
}
|
||||
|
||||
impl<'a> ExecCommandInput<'a> {
|
||||
@@ -328,6 +336,7 @@ impl<'a> ExecCommandInput<'a> {
|
||||
parsed_cmd: &'a [ParsedCommand],
|
||||
source: ExecCommandSource,
|
||||
interaction_input: Option<&'a str>,
|
||||
process_id: Option<&'a str>,
|
||||
) -> Self {
|
||||
Self {
|
||||
command,
|
||||
@@ -335,6 +344,7 @@ impl<'a> ExecCommandInput<'a> {
|
||||
parsed_cmd,
|
||||
source,
|
||||
interaction_input,
|
||||
process_id,
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -362,6 +372,7 @@ async fn emit_exec_stage(
|
||||
exec_input.parsed_cmd,
|
||||
exec_input.source,
|
||||
exec_input.interaction_input.map(str::to_owned),
|
||||
exec_input.process_id,
|
||||
)
|
||||
.await;
|
||||
}
|
||||
@@ -402,6 +413,7 @@ async fn emit_exec_end(
|
||||
ctx.turn,
|
||||
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
|
||||
call_id: ctx.call_id.to_string(),
|
||||
process_id: exec_input.process_id.map(str::to_owned),
|
||||
turn_id: ctx.turn.sub_id.clone(),
|
||||
command: exec_input.command.to_vec(),
|
||||
cwd: exec_input.cwd.to_path_buf(),
|
||||
|
||||
@@ -46,6 +46,7 @@ struct ExecCommandArgs {
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
struct WriteStdinArgs {
|
||||
// The model is trained on `session_id`.
|
||||
session_id: i32,
|
||||
#[serde(default)]
|
||||
chars: String,
|
||||
@@ -128,6 +129,7 @@ impl ToolHandler for UnifiedExecHandler {
|
||||
"failed to parse exec_command arguments: {err:?}"
|
||||
))
|
||||
})?;
|
||||
let process_id = manager.allocate_process_id().await;
|
||||
|
||||
let command = get_command(&args);
|
||||
let ExecCommandArgs {
|
||||
@@ -168,6 +170,7 @@ impl ToolHandler for UnifiedExecHandler {
|
||||
cwd.clone(),
|
||||
ExecCommandSource::UnifiedExecStartup,
|
||||
None,
|
||||
Some(process_id.clone()),
|
||||
);
|
||||
emitter.emit(event_ctx, ToolEventStage::Begin).await;
|
||||
|
||||
@@ -175,6 +178,7 @@ impl ToolHandler for UnifiedExecHandler {
|
||||
.exec_command(
|
||||
ExecCommandRequest {
|
||||
command,
|
||||
process_id,
|
||||
yield_time_ms,
|
||||
max_output_tokens,
|
||||
workdir,
|
||||
@@ -197,7 +201,7 @@ impl ToolHandler for UnifiedExecHandler {
|
||||
manager
|
||||
.write_stdin(WriteStdinRequest {
|
||||
call_id: &call_id,
|
||||
session_id: args.session_id,
|
||||
process_id: &args.session_id.to_string(),
|
||||
input: &args.chars,
|
||||
yield_time_ms: args.yield_time_ms,
|
||||
max_output_tokens: args.max_output_tokens,
|
||||
@@ -255,8 +259,9 @@ fn format_response(response: &UnifiedExecResponse) -> String {
|
||||
sections.push(format!("Process exited with code {exit_code}"));
|
||||
}
|
||||
|
||||
if let Some(session_id) = response.session_id {
|
||||
sections.push(format!("Process running with session ID {session_id}"));
|
||||
if let Some(process_id) = &response.process_id {
|
||||
// Training still uses "session ID".
|
||||
sections.push(format!("Process running with session ID {process_id}"));
|
||||
}
|
||||
|
||||
if let Some(original_token_count) = response.original_token_count {
|
||||
|
||||
@@ -5,8 +5,9 @@ use thiserror::Error;
|
||||
pub(crate) enum UnifiedExecError {
|
||||
#[error("Failed to create unified exec session: {message}")]
|
||||
CreateSession { message: String },
|
||||
#[error("Unknown session id {session_id}")]
|
||||
UnknownSessionId { session_id: i32 },
|
||||
// Called "session" in the model's training.
|
||||
#[error("Unknown session id {process_id}")]
|
||||
UnknownSessionId { process_id: String },
|
||||
#[error("failed to write to stdin")]
|
||||
WriteToStdin,
|
||||
#[error("missing command line for unified exec request")]
|
||||
|
||||
@@ -22,9 +22,9 @@
|
||||
//! - `session_manager.rs`: orchestration (approvals, sandboxing, reuse) and request handling.
|
||||
|
||||
use std::collections::HashMap;
|
||||
use std::collections::HashSet;
|
||||
use std::path::PathBuf;
|
||||
use std::sync::Arc;
|
||||
use std::sync::atomic::AtomicI32;
|
||||
use std::time::Duration;
|
||||
|
||||
use rand::Rng;
|
||||
@@ -67,6 +67,7 @@ impl UnifiedExecContext {
|
||||
#[derive(Debug)]
|
||||
pub(crate) struct ExecCommandRequest {
|
||||
pub command: Vec<String>,
|
||||
pub process_id: String,
|
||||
pub yield_time_ms: u64,
|
||||
pub max_output_tokens: Option<usize>,
|
||||
pub workdir: Option<PathBuf>,
|
||||
@@ -77,7 +78,7 @@ pub(crate) struct ExecCommandRequest {
|
||||
#[derive(Debug)]
|
||||
pub(crate) struct WriteStdinRequest<'a> {
|
||||
pub call_id: &'a str,
|
||||
pub session_id: i32,
|
||||
pub process_id: &'a str,
|
||||
pub input: &'a str,
|
||||
pub yield_time_ms: u64,
|
||||
pub max_output_tokens: Option<usize>,
|
||||
@@ -89,7 +90,7 @@ pub(crate) struct UnifiedExecResponse {
|
||||
pub chunk_id: String,
|
||||
pub wall_time: Duration,
|
||||
pub output: String,
|
||||
pub session_id: Option<i32>,
|
||||
pub process_id: Option<String>,
|
||||
pub exit_code: Option<i32>,
|
||||
pub original_token_count: Option<usize>,
|
||||
pub session_command: Option<Vec<String>>,
|
||||
@@ -97,15 +98,34 @@ pub(crate) struct UnifiedExecResponse {
|
||||
|
||||
#[derive(Default)]
|
||||
pub(crate) struct UnifiedExecSessionManager {
|
||||
next_session_id: AtomicI32,
|
||||
sessions: Mutex<HashMap<i32, SessionEntry>>,
|
||||
session_store: Mutex<SessionStore>,
|
||||
}
|
||||
|
||||
// Required for mutex sharing.
|
||||
#[derive(Default)]
|
||||
pub(crate) struct SessionStore {
|
||||
sessions: HashMap<String, SessionEntry>,
|
||||
reserved_sessions_id: HashSet<String>,
|
||||
}
|
||||
|
||||
impl SessionStore {
|
||||
fn remove(&mut self, session_id: &str) -> Option<SessionEntry> {
|
||||
self.reserved_sessions_id.remove(session_id);
|
||||
self.sessions.remove(session_id)
|
||||
}
|
||||
|
||||
pub(crate) fn clear(&mut self) {
|
||||
self.reserved_sessions_id.clear();
|
||||
self.sessions.clear();
|
||||
}
|
||||
}
|
||||
|
||||
struct SessionEntry {
|
||||
session: session::UnifiedExecSession,
|
||||
session: UnifiedExecSession,
|
||||
session_ref: Arc<Session>,
|
||||
turn_ref: Arc<TurnContext>,
|
||||
call_id: String,
|
||||
process_id: String,
|
||||
command: Vec<String>,
|
||||
cwd: PathBuf,
|
||||
started_at: tokio::time::Instant,
|
||||
@@ -159,6 +179,11 @@ mod tests {
|
||||
) -> Result<UnifiedExecResponse, UnifiedExecError> {
|
||||
let context =
|
||||
UnifiedExecContext::new(Arc::clone(session), Arc::clone(turn), "call".to_string());
|
||||
let process_id = session
|
||||
.services
|
||||
.unified_exec_manager
|
||||
.allocate_process_id()
|
||||
.await;
|
||||
|
||||
session
|
||||
.services
|
||||
@@ -166,6 +191,7 @@ mod tests {
|
||||
.exec_command(
|
||||
ExecCommandRequest {
|
||||
command: vec!["bash".to_string(), "-lc".to_string(), cmd.to_string()],
|
||||
process_id,
|
||||
yield_time_ms,
|
||||
max_output_tokens: None,
|
||||
workdir: None,
|
||||
@@ -179,7 +205,7 @@ mod tests {
|
||||
|
||||
async fn write_stdin(
|
||||
session: &Arc<Session>,
|
||||
session_id: i32,
|
||||
process_id: &str,
|
||||
input: &str,
|
||||
yield_time_ms: u64,
|
||||
) -> Result<UnifiedExecResponse, UnifiedExecError> {
|
||||
@@ -188,7 +214,7 @@ mod tests {
|
||||
.unified_exec_manager
|
||||
.write_stdin(WriteStdinRequest {
|
||||
call_id: "write-stdin",
|
||||
session_id,
|
||||
process_id,
|
||||
input,
|
||||
yield_time_ms,
|
||||
max_output_tokens: None,
|
||||
@@ -221,11 +247,15 @@ mod tests {
|
||||
let (session, turn) = test_session_and_turn();
|
||||
|
||||
let open_shell = exec_command(&session, &turn, "bash -i", 2_500).await?;
|
||||
let session_id = open_shell.session_id.expect("expected session_id");
|
||||
let process_id = open_shell
|
||||
.process_id
|
||||
.as_ref()
|
||||
.expect("expected process_id")
|
||||
.as_str();
|
||||
|
||||
write_stdin(
|
||||
&session,
|
||||
session_id,
|
||||
process_id,
|
||||
"export CODEX_INTERACTIVE_SHELL_VAR=codex\n",
|
||||
2_500,
|
||||
)
|
||||
@@ -233,7 +263,7 @@ mod tests {
|
||||
|
||||
let out_2 = write_stdin(
|
||||
&session,
|
||||
session_id,
|
||||
process_id,
|
||||
"echo $CODEX_INTERACTIVE_SHELL_VAR\n",
|
||||
2_500,
|
||||
)
|
||||
@@ -253,11 +283,15 @@ mod tests {
|
||||
let (session, turn) = test_session_and_turn();
|
||||
|
||||
let shell_a = exec_command(&session, &turn, "bash -i", 2_500).await?;
|
||||
let session_a = shell_a.session_id.expect("expected session id");
|
||||
let session_a = shell_a
|
||||
.process_id
|
||||
.as_ref()
|
||||
.expect("expected process id")
|
||||
.clone();
|
||||
|
||||
write_stdin(
|
||||
&session,
|
||||
session_a,
|
||||
session_a.as_str(),
|
||||
"export CODEX_INTERACTIVE_SHELL_VAR=codex\n",
|
||||
2_500,
|
||||
)
|
||||
@@ -265,9 +299,10 @@ mod tests {
|
||||
|
||||
let out_2 =
|
||||
exec_command(&session, &turn, "echo $CODEX_INTERACTIVE_SHELL_VAR", 2_500).await?;
|
||||
tokio::time::sleep(Duration::from_secs(2)).await;
|
||||
assert!(
|
||||
out_2.session_id.is_none(),
|
||||
"short command should not retain a session"
|
||||
out_2.process_id.is_none(),
|
||||
"short command should not report a process id if it exits quickly"
|
||||
);
|
||||
assert!(
|
||||
!out_2.output.contains("codex"),
|
||||
@@ -276,7 +311,11 @@ mod tests {
|
||||
|
||||
let out_3 = write_stdin(
|
||||
&session,
|
||||
session_a,
|
||||
shell_a
|
||||
.process_id
|
||||
.as_ref()
|
||||
.expect("expected process id")
|
||||
.as_str(),
|
||||
"echo $CODEX_INTERACTIVE_SHELL_VAR\n",
|
||||
2_500,
|
||||
)
|
||||
@@ -296,11 +335,15 @@ mod tests {
|
||||
let (session, turn) = test_session_and_turn();
|
||||
|
||||
let open_shell = exec_command(&session, &turn, "bash -i", 2_500).await?;
|
||||
let session_id = open_shell.session_id.expect("expected session id");
|
||||
let process_id = open_shell
|
||||
.process_id
|
||||
.as_ref()
|
||||
.expect("expected process id")
|
||||
.as_str();
|
||||
|
||||
write_stdin(
|
||||
&session,
|
||||
session_id,
|
||||
process_id,
|
||||
"export CODEX_INTERACTIVE_SHELL_VAR=codex\n",
|
||||
2_500,
|
||||
)
|
||||
@@ -308,7 +351,7 @@ mod tests {
|
||||
|
||||
let out_2 = write_stdin(
|
||||
&session,
|
||||
session_id,
|
||||
process_id,
|
||||
"sleep 5 && echo $CODEX_INTERACTIVE_SHELL_VAR\n",
|
||||
10,
|
||||
)
|
||||
@@ -320,7 +363,7 @@ mod tests {
|
||||
|
||||
tokio::time::sleep(Duration::from_secs(7)).await;
|
||||
|
||||
let out_3 = write_stdin(&session, session_id, "", 100).await?;
|
||||
let out_3 = write_stdin(&session, process_id, "", 100).await?;
|
||||
|
||||
assert!(
|
||||
out_3.output.contains("codex"),
|
||||
@@ -337,7 +380,7 @@ mod tests {
|
||||
|
||||
let result = exec_command(&session, &turn, "echo codex", 120_000).await?;
|
||||
|
||||
assert!(result.session_id.is_none());
|
||||
assert!(result.process_id.is_some());
|
||||
assert!(result.output.contains("codex"));
|
||||
|
||||
Ok(())
|
||||
@@ -350,8 +393,8 @@ mod tests {
|
||||
let result = exec_command(&session, &turn, "echo codex", 2_500).await?;
|
||||
|
||||
assert!(
|
||||
result.session_id.is_none(),
|
||||
"completed command should not retain session"
|
||||
result.process_id.is_some(),
|
||||
"completed command should report a process id"
|
||||
);
|
||||
assert!(result.output.contains("codex"));
|
||||
|
||||
@@ -359,9 +402,10 @@ mod tests {
|
||||
session
|
||||
.services
|
||||
.unified_exec_manager
|
||||
.sessions
|
||||
.session_store
|
||||
.lock()
|
||||
.await
|
||||
.sessions
|
||||
.is_empty()
|
||||
);
|
||||
|
||||
@@ -375,31 +419,36 @@ mod tests {
|
||||
let (session, turn) = test_session_and_turn();
|
||||
|
||||
let open_shell = exec_command(&session, &turn, "bash -i", 2_500).await?;
|
||||
let session_id = open_shell.session_id.expect("expected session id");
|
||||
let process_id = open_shell
|
||||
.process_id
|
||||
.as_ref()
|
||||
.expect("expected process id")
|
||||
.as_str();
|
||||
|
||||
write_stdin(&session, session_id, "exit\n", 2_500).await?;
|
||||
write_stdin(&session, process_id, "exit\n", 2_500).await?;
|
||||
|
||||
tokio::time::sleep(Duration::from_millis(200)).await;
|
||||
|
||||
let err = write_stdin(&session, session_id, "", 100)
|
||||
let err = write_stdin(&session, process_id, "", 100)
|
||||
.await
|
||||
.expect_err("expected unknown session error");
|
||||
|
||||
match err {
|
||||
UnifiedExecError::UnknownSessionId { session_id: err_id } => {
|
||||
assert_eq!(err_id, session_id);
|
||||
UnifiedExecError::UnknownSessionId { process_id: err_id } => {
|
||||
assert_eq!(err_id, process_id, "process id should match request");
|
||||
}
|
||||
other => panic!("expected UnknownSessionId, got {other:?}"),
|
||||
}
|
||||
|
||||
assert!(
|
||||
!session
|
||||
session
|
||||
.services
|
||||
.unified_exec_manager
|
||||
.sessions
|
||||
.session_store
|
||||
.lock()
|
||||
.await
|
||||
.contains_key(&session_id)
|
||||
.sessions
|
||||
.is_empty()
|
||||
);
|
||||
|
||||
Ok(())
|
||||
|
||||
@@ -1,9 +1,9 @@
|
||||
use rand::Rng;
|
||||
use std::cmp::Reverse;
|
||||
use std::collections::HashMap;
|
||||
use std::collections::HashSet;
|
||||
use std::path::PathBuf;
|
||||
use std::sync::Arc;
|
||||
|
||||
use tokio::sync::Notify;
|
||||
use tokio::sync::mpsc;
|
||||
use tokio::time::Duration;
|
||||
@@ -36,6 +36,7 @@ use crate::truncate::formatted_truncate_text;
|
||||
use super::ExecCommandRequest;
|
||||
use super::MAX_UNIFIED_EXEC_SESSIONS;
|
||||
use super::SessionEntry;
|
||||
use super::SessionStore;
|
||||
use super::UnifiedExecContext;
|
||||
use super::UnifiedExecError;
|
||||
use super::UnifiedExecResponse;
|
||||
@@ -48,6 +49,24 @@ use super::session::OutputBuffer;
|
||||
use super::session::OutputHandles;
|
||||
use super::session::UnifiedExecSession;
|
||||
|
||||
const UNIFIED_EXEC_ENV: [(&str, &str); 8] = [
|
||||
("NO_COLOR", "1"),
|
||||
("TERM", "dumb"),
|
||||
("LANG", "C.UTF-8"),
|
||||
("LC_CTYPE", "C.UTF-8"),
|
||||
("LC_ALL", "C.UTF-8"),
|
||||
("COLORTERM", ""),
|
||||
("PAGER", "cat"),
|
||||
("GIT_PAGER", "cat"),
|
||||
];
|
||||
|
||||
fn apply_unified_exec_env(mut env: HashMap<String, String>) -> HashMap<String, String> {
|
||||
for (key, value) in UNIFIED_EXEC_ENV {
|
||||
env.insert(key.to_string(), value.to_string());
|
||||
}
|
||||
env
|
||||
}
|
||||
|
||||
struct PreparedSessionHandles {
|
||||
writer_tx: mpsc::Sender<Vec<u8>>,
|
||||
output_buffer: OutputBuffer,
|
||||
@@ -57,9 +76,39 @@ struct PreparedSessionHandles {
|
||||
turn_ref: Arc<TurnContext>,
|
||||
command: Vec<String>,
|
||||
cwd: PathBuf,
|
||||
process_id: String,
|
||||
}
|
||||
|
||||
impl UnifiedExecSessionManager {
|
||||
pub(crate) async fn allocate_process_id(&self) -> String {
|
||||
loop {
|
||||
let mut store = self.session_store.lock().await;
|
||||
|
||||
let process_id = if !cfg!(test) && !cfg!(feature = "deterministic_process_ids") {
|
||||
// production mode → random
|
||||
rand::rng().random_range(1_000..100_000).to_string()
|
||||
} else {
|
||||
// test or deterministic mode
|
||||
let next = store
|
||||
.reserved_sessions_id
|
||||
.iter()
|
||||
.filter_map(|s| s.parse::<i32>().ok())
|
||||
.max()
|
||||
.map(|m| std::cmp::max(m, 999) + 1)
|
||||
.unwrap_or(1000);
|
||||
|
||||
next.to_string()
|
||||
};
|
||||
|
||||
if store.reserved_sessions_id.contains(&process_id) {
|
||||
continue;
|
||||
}
|
||||
|
||||
store.reserved_sessions_id.insert(process_id.clone());
|
||||
return process_id;
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) async fn exec_command(
|
||||
&self,
|
||||
request: ExecCommandRequest,
|
||||
@@ -104,14 +153,20 @@ impl UnifiedExecSessionManager {
|
||||
let has_exited = session.has_exited();
|
||||
let exit_code = session.exit_code();
|
||||
let chunk_id = generate_chunk_id();
|
||||
let session_id = if has_exited {
|
||||
let process_id = if has_exited {
|
||||
None
|
||||
} else {
|
||||
// Only store session if not exited.
|
||||
let stored_id = self
|
||||
.store_session(session, context, &request.command, cwd.clone(), start)
|
||||
.await;
|
||||
Some(stored_id)
|
||||
self.store_session(
|
||||
session,
|
||||
context,
|
||||
&request.command,
|
||||
cwd.clone(),
|
||||
start,
|
||||
request.process_id.clone(),
|
||||
)
|
||||
.await;
|
||||
Some(request.process_id.clone())
|
||||
};
|
||||
let original_token_count = approx_token_count(&text);
|
||||
|
||||
@@ -120,18 +175,18 @@ impl UnifiedExecSessionManager {
|
||||
chunk_id,
|
||||
wall_time,
|
||||
output,
|
||||
session_id,
|
||||
process_id: process_id.clone(),
|
||||
exit_code,
|
||||
original_token_count: Some(original_token_count),
|
||||
session_command: Some(request.command.clone()),
|
||||
};
|
||||
|
||||
if response.session_id.is_some() {
|
||||
if !has_exited {
|
||||
Self::emit_waiting_status(&context.session, &context.turn, &request.command).await;
|
||||
}
|
||||
|
||||
// If the command completed during this call, emit an ExecCommandEnd via the emitter.
|
||||
if response.session_id.is_none() {
|
||||
if has_exited {
|
||||
let exit = response.exit_code.unwrap_or(-1);
|
||||
Self::emit_exec_end_from_context(
|
||||
context,
|
||||
@@ -140,6 +195,9 @@ impl UnifiedExecSessionManager {
|
||||
response.output.clone(),
|
||||
exit,
|
||||
response.wall_time,
|
||||
// We always emit the process ID in order to keep consistency between the Begin
|
||||
// event and the End event.
|
||||
Some(request.process_id),
|
||||
)
|
||||
.await;
|
||||
}
|
||||
@@ -151,7 +209,7 @@ impl UnifiedExecSessionManager {
|
||||
&self,
|
||||
request: WriteStdinRequest<'_>,
|
||||
) -> Result<UnifiedExecResponse, UnifiedExecError> {
|
||||
let session_id = request.session_id;
|
||||
let process_id = request.process_id.to_string();
|
||||
|
||||
let PreparedSessionHandles {
|
||||
writer_tx,
|
||||
@@ -162,13 +220,15 @@ impl UnifiedExecSessionManager {
|
||||
turn_ref,
|
||||
command: session_command,
|
||||
cwd: session_cwd,
|
||||
} = self.prepare_session_handles(session_id).await?;
|
||||
process_id,
|
||||
} = self.prepare_session_handles(process_id.as_str()).await?;
|
||||
|
||||
let interaction_emitter = ToolEmitter::unified_exec(
|
||||
&session_command,
|
||||
session_cwd.clone(),
|
||||
ExecCommandSource::UnifiedExecInteraction,
|
||||
(!request.input.is_empty()).then(|| request.input.to_string()),
|
||||
Some(process_id.clone()),
|
||||
);
|
||||
let make_event_ctx = || {
|
||||
ToolEventCtx::new(
|
||||
@@ -215,17 +275,21 @@ impl UnifiedExecSessionManager {
|
||||
let original_token_count = approx_token_count(&text);
|
||||
let chunk_id = generate_chunk_id();
|
||||
|
||||
let status = self.refresh_session_state(session_id).await;
|
||||
let (session_id, exit_code, completion_entry, event_call_id) = match status {
|
||||
SessionStatus::Alive { exit_code, call_id } => {
|
||||
(Some(session_id), exit_code, None, call_id)
|
||||
}
|
||||
let status = self.refresh_session_state(process_id.as_str()).await;
|
||||
let (process_id, exit_code, completion_entry, event_call_id) = match status {
|
||||
SessionStatus::Alive {
|
||||
exit_code,
|
||||
call_id,
|
||||
process_id,
|
||||
} => (Some(process_id), exit_code, None, call_id),
|
||||
SessionStatus::Exited { exit_code, entry } => {
|
||||
let call_id = entry.call_id.clone();
|
||||
(None, exit_code, Some(*entry), call_id)
|
||||
}
|
||||
SessionStatus::Unknown => {
|
||||
return Err(UnifiedExecError::UnknownSessionId { session_id });
|
||||
return Err(UnifiedExecError::UnknownSessionId {
|
||||
process_id: request.process_id.to_string(),
|
||||
});
|
||||
}
|
||||
};
|
||||
|
||||
@@ -234,7 +298,7 @@ impl UnifiedExecSessionManager {
|
||||
chunk_id,
|
||||
wall_time,
|
||||
output,
|
||||
session_id,
|
||||
process_id,
|
||||
exit_code,
|
||||
original_token_count: Some(original_token_count),
|
||||
session_command: Some(session_command.clone()),
|
||||
@@ -255,7 +319,7 @@ impl UnifiedExecSessionManager {
|
||||
)
|
||||
.await;
|
||||
|
||||
if response.session_id.is_some() {
|
||||
if response.process_id.is_some() {
|
||||
Self::emit_waiting_status(&session_ref, &turn_ref, &session_command).await;
|
||||
}
|
||||
|
||||
@@ -268,16 +332,17 @@ impl UnifiedExecSessionManager {
|
||||
Ok(response)
|
||||
}
|
||||
|
||||
async fn refresh_session_state(&self, session_id: i32) -> SessionStatus {
|
||||
let mut sessions = self.sessions.lock().await;
|
||||
let Some(entry) = sessions.get(&session_id) else {
|
||||
async fn refresh_session_state(&self, process_id: &str) -> SessionStatus {
|
||||
let mut store = self.session_store.lock().await;
|
||||
let Some(entry) = store.sessions.get(process_id) else {
|
||||
return SessionStatus::Unknown;
|
||||
};
|
||||
|
||||
let exit_code = entry.session.exit_code();
|
||||
let process_id = entry.process_id.clone();
|
||||
|
||||
if entry.session.has_exited() {
|
||||
let Some(entry) = sessions.remove(&session_id) else {
|
||||
let Some(entry) = store.remove(&process_id) else {
|
||||
return SessionStatus::Unknown;
|
||||
};
|
||||
SessionStatus::Exited {
|
||||
@@ -288,18 +353,23 @@ impl UnifiedExecSessionManager {
|
||||
SessionStatus::Alive {
|
||||
exit_code,
|
||||
call_id: entry.call_id.clone(),
|
||||
process_id,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
async fn prepare_session_handles(
|
||||
&self,
|
||||
session_id: i32,
|
||||
process_id: &str,
|
||||
) -> Result<PreparedSessionHandles, UnifiedExecError> {
|
||||
let mut sessions = self.sessions.lock().await;
|
||||
let entry = sessions
|
||||
.get_mut(&session_id)
|
||||
.ok_or(UnifiedExecError::UnknownSessionId { session_id })?;
|
||||
let mut store = self.session_store.lock().await;
|
||||
let entry =
|
||||
store
|
||||
.sessions
|
||||
.get_mut(process_id)
|
||||
.ok_or(UnifiedExecError::UnknownSessionId {
|
||||
process_id: process_id.to_string(),
|
||||
})?;
|
||||
entry.last_used = Instant::now();
|
||||
let OutputHandles {
|
||||
output_buffer,
|
||||
@@ -316,6 +386,7 @@ impl UnifiedExecSessionManager {
|
||||
turn_ref: Arc::clone(&entry.turn_ref),
|
||||
command: entry.command.clone(),
|
||||
cwd: entry.cwd.clone(),
|
||||
process_id: entry.process_id.clone(),
|
||||
})
|
||||
}
|
||||
|
||||
@@ -329,6 +400,7 @@ impl UnifiedExecSessionManager {
|
||||
.map_err(|_| UnifiedExecError::WriteToStdin)
|
||||
}
|
||||
|
||||
#[allow(clippy::too_many_arguments)]
|
||||
async fn store_session(
|
||||
&self,
|
||||
session: UnifiedExecSession,
|
||||
@@ -336,24 +408,22 @@ impl UnifiedExecSessionManager {
|
||||
command: &[String],
|
||||
cwd: PathBuf,
|
||||
started_at: Instant,
|
||||
) -> i32 {
|
||||
let session_id = self
|
||||
.next_session_id
|
||||
.fetch_add(1, std::sync::atomic::Ordering::SeqCst);
|
||||
process_id: String,
|
||||
) {
|
||||
let entry = SessionEntry {
|
||||
session,
|
||||
session_ref: Arc::clone(&context.session),
|
||||
turn_ref: Arc::clone(&context.turn),
|
||||
call_id: context.call_id.clone(),
|
||||
process_id: process_id.clone(),
|
||||
command: command.to_vec(),
|
||||
cwd,
|
||||
started_at,
|
||||
last_used: started_at,
|
||||
};
|
||||
let mut sessions = self.sessions.lock().await;
|
||||
Self::prune_sessions_if_needed(&mut sessions);
|
||||
sessions.insert(session_id, entry);
|
||||
session_id
|
||||
let mut store = self.session_store.lock().await;
|
||||
Self::prune_sessions_if_needed(&mut store);
|
||||
store.sessions.insert(process_id, entry);
|
||||
}
|
||||
|
||||
async fn emit_exec_end_from_entry(
|
||||
@@ -381,6 +451,7 @@ impl UnifiedExecSessionManager {
|
||||
entry.cwd,
|
||||
ExecCommandSource::UnifiedExecStartup,
|
||||
None,
|
||||
Some(entry.process_id.clone()),
|
||||
);
|
||||
emitter
|
||||
.emit(event_ctx, ToolEventStage::Success(output))
|
||||
@@ -394,6 +465,7 @@ impl UnifiedExecSessionManager {
|
||||
aggregated_output: String,
|
||||
exit_code: i32,
|
||||
duration: Duration,
|
||||
process_id: Option<String>,
|
||||
) {
|
||||
let output = ExecToolCallOutput {
|
||||
exit_code,
|
||||
@@ -409,8 +481,13 @@ impl UnifiedExecSessionManager {
|
||||
&context.call_id,
|
||||
None,
|
||||
);
|
||||
let emitter =
|
||||
ToolEmitter::unified_exec(command, cwd, ExecCommandSource::UnifiedExecStartup, None);
|
||||
let emitter = ToolEmitter::unified_exec(
|
||||
command,
|
||||
cwd,
|
||||
ExecCommandSource::UnifiedExecStartup,
|
||||
None,
|
||||
process_id,
|
||||
);
|
||||
emitter
|
||||
.emit(event_ctx, ToolEventStage::Success(output))
|
||||
.await;
|
||||
@@ -460,12 +537,13 @@ impl UnifiedExecSessionManager {
|
||||
justification: Option<String>,
|
||||
context: &UnifiedExecContext,
|
||||
) -> Result<UnifiedExecSession, UnifiedExecError> {
|
||||
let env = apply_unified_exec_env(create_env(&context.turn.shell_environment_policy));
|
||||
let mut orchestrator = ToolOrchestrator::new();
|
||||
let mut runtime = UnifiedExecRuntime::new(self);
|
||||
let req = UnifiedExecToolRequest::new(
|
||||
command.to_vec(),
|
||||
cwd,
|
||||
create_env(&context.turn.shell_environment_policy),
|
||||
env,
|
||||
with_escalated_permissions,
|
||||
justification,
|
||||
create_approval_requirement_for_command(
|
||||
@@ -555,48 +633,54 @@ impl UnifiedExecSessionManager {
|
||||
collected
|
||||
}
|
||||
|
||||
fn prune_sessions_if_needed(sessions: &mut HashMap<i32, SessionEntry>) {
|
||||
if sessions.len() < MAX_UNIFIED_EXEC_SESSIONS {
|
||||
fn prune_sessions_if_needed(store: &mut SessionStore) {
|
||||
if store.sessions.len() < MAX_UNIFIED_EXEC_SESSIONS {
|
||||
return;
|
||||
}
|
||||
|
||||
let meta: Vec<(i32, Instant, bool)> = sessions
|
||||
let meta: Vec<(String, Instant, bool)> = store
|
||||
.sessions
|
||||
.iter()
|
||||
.map(|(id, entry)| (*id, entry.last_used, entry.session.has_exited()))
|
||||
.map(|(id, entry)| (id.clone(), entry.last_used, entry.session.has_exited()))
|
||||
.collect();
|
||||
|
||||
if let Some(session_id) = Self::session_id_to_prune_from_meta(&meta) {
|
||||
sessions.remove(&session_id);
|
||||
store.remove(&session_id);
|
||||
}
|
||||
}
|
||||
|
||||
// Centralized pruning policy so we can easily swap strategies later.
|
||||
fn session_id_to_prune_from_meta(meta: &[(i32, Instant, bool)]) -> Option<i32> {
|
||||
fn session_id_to_prune_from_meta(meta: &[(String, Instant, bool)]) -> Option<String> {
|
||||
if meta.is_empty() {
|
||||
return None;
|
||||
}
|
||||
|
||||
let mut by_recency = meta.to_vec();
|
||||
by_recency.sort_by_key(|(_, last_used, _)| Reverse(*last_used));
|
||||
let protected: HashSet<i32> = by_recency
|
||||
let protected: HashSet<String> = by_recency
|
||||
.iter()
|
||||
.take(8)
|
||||
.map(|(session_id, _, _)| *session_id)
|
||||
.map(|(process_id, _, _)| process_id.clone())
|
||||
.collect();
|
||||
|
||||
let mut lru = meta.to_vec();
|
||||
lru.sort_by_key(|(_, last_used, _)| *last_used);
|
||||
|
||||
if let Some((session_id, _, _)) = lru
|
||||
if let Some((process_id, _, _)) = lru
|
||||
.iter()
|
||||
.find(|(session_id, _, exited)| !protected.contains(session_id) && *exited)
|
||||
.find(|(process_id, _, exited)| !protected.contains(process_id) && *exited)
|
||||
{
|
||||
return Some(*session_id);
|
||||
return Some(process_id.clone());
|
||||
}
|
||||
|
||||
lru.into_iter()
|
||||
.find(|(session_id, _, _)| !protected.contains(session_id))
|
||||
.map(|(session_id, _, _)| session_id)
|
||||
.find(|(process_id, _, _)| !protected.contains(process_id))
|
||||
.map(|(process_id, _, _)| process_id)
|
||||
}
|
||||
|
||||
pub(crate) async fn terminate_all_sessions(&self) {
|
||||
let mut sessions = self.session_store.lock().await;
|
||||
sessions.clear();
|
||||
}
|
||||
}
|
||||
|
||||
@@ -604,6 +688,7 @@ enum SessionStatus {
|
||||
Alive {
|
||||
exit_code: Option<i32>,
|
||||
call_id: String,
|
||||
process_id: String,
|
||||
},
|
||||
Exited {
|
||||
exit_code: Option<i32>,
|
||||
@@ -619,67 +704,99 @@ mod tests {
|
||||
use tokio::time::Duration;
|
||||
use tokio::time::Instant;
|
||||
|
||||
#[test]
|
||||
fn unified_exec_env_injects_defaults() {
|
||||
let env = apply_unified_exec_env(HashMap::new());
|
||||
let expected = HashMap::from([
|
||||
("NO_COLOR".to_string(), "1".to_string()),
|
||||
("TERM".to_string(), "dumb".to_string()),
|
||||
("LANG".to_string(), "C.UTF-8".to_string()),
|
||||
("LC_CTYPE".to_string(), "C.UTF-8".to_string()),
|
||||
("LC_ALL".to_string(), "C.UTF-8".to_string()),
|
||||
("COLORTERM".to_string(), String::new()),
|
||||
("PAGER".to_string(), "cat".to_string()),
|
||||
("GIT_PAGER".to_string(), "cat".to_string()),
|
||||
]);
|
||||
|
||||
assert_eq!(env, expected);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn unified_exec_env_overrides_existing_values() {
|
||||
let mut base = HashMap::new();
|
||||
base.insert("NO_COLOR".to_string(), "0".to_string());
|
||||
base.insert("PATH".to_string(), "/usr/bin".to_string());
|
||||
|
||||
let env = apply_unified_exec_env(base);
|
||||
|
||||
assert_eq!(env.get("NO_COLOR"), Some(&"1".to_string()));
|
||||
assert_eq!(env.get("PATH"), Some(&"/usr/bin".to_string()));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn pruning_prefers_exited_sessions_outside_recently_used() {
|
||||
let now = Instant::now();
|
||||
let id = |n: i32| n.to_string();
|
||||
let meta = vec![
|
||||
(1, now - Duration::from_secs(40), false),
|
||||
(2, now - Duration::from_secs(30), true),
|
||||
(3, now - Duration::from_secs(20), false),
|
||||
(4, now - Duration::from_secs(19), false),
|
||||
(5, now - Duration::from_secs(18), false),
|
||||
(6, now - Duration::from_secs(17), false),
|
||||
(7, now - Duration::from_secs(16), false),
|
||||
(8, now - Duration::from_secs(15), false),
|
||||
(9, now - Duration::from_secs(14), false),
|
||||
(10, now - Duration::from_secs(13), false),
|
||||
(id(1), now - Duration::from_secs(40), false),
|
||||
(id(2), now - Duration::from_secs(30), true),
|
||||
(id(3), now - Duration::from_secs(20), false),
|
||||
(id(4), now - Duration::from_secs(19), false),
|
||||
(id(5), now - Duration::from_secs(18), false),
|
||||
(id(6), now - Duration::from_secs(17), false),
|
||||
(id(7), now - Duration::from_secs(16), false),
|
||||
(id(8), now - Duration::from_secs(15), false),
|
||||
(id(9), now - Duration::from_secs(14), false),
|
||||
(id(10), now - Duration::from_secs(13), false),
|
||||
];
|
||||
|
||||
let candidate = UnifiedExecSessionManager::session_id_to_prune_from_meta(&meta);
|
||||
|
||||
assert_eq!(candidate, Some(2));
|
||||
assert_eq!(candidate, Some(id(2)));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn pruning_falls_back_to_lru_when_no_exited() {
|
||||
let now = Instant::now();
|
||||
let id = |n: i32| n.to_string();
|
||||
let meta = vec![
|
||||
(1, now - Duration::from_secs(40), false),
|
||||
(2, now - Duration::from_secs(30), false),
|
||||
(3, now - Duration::from_secs(20), false),
|
||||
(4, now - Duration::from_secs(19), false),
|
||||
(5, now - Duration::from_secs(18), false),
|
||||
(6, now - Duration::from_secs(17), false),
|
||||
(7, now - Duration::from_secs(16), false),
|
||||
(8, now - Duration::from_secs(15), false),
|
||||
(9, now - Duration::from_secs(14), false),
|
||||
(10, now - Duration::from_secs(13), false),
|
||||
(id(1), now - Duration::from_secs(40), false),
|
||||
(id(2), now - Duration::from_secs(30), false),
|
||||
(id(3), now - Duration::from_secs(20), false),
|
||||
(id(4), now - Duration::from_secs(19), false),
|
||||
(id(5), now - Duration::from_secs(18), false),
|
||||
(id(6), now - Duration::from_secs(17), false),
|
||||
(id(7), now - Duration::from_secs(16), false),
|
||||
(id(8), now - Duration::from_secs(15), false),
|
||||
(id(9), now - Duration::from_secs(14), false),
|
||||
(id(10), now - Duration::from_secs(13), false),
|
||||
];
|
||||
|
||||
let candidate = UnifiedExecSessionManager::session_id_to_prune_from_meta(&meta);
|
||||
|
||||
assert_eq!(candidate, Some(1));
|
||||
assert_eq!(candidate, Some(id(1)));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn pruning_protects_recent_sessions_even_if_exited() {
|
||||
let now = Instant::now();
|
||||
let id = |n: i32| n.to_string();
|
||||
let meta = vec![
|
||||
(1, now - Duration::from_secs(40), false),
|
||||
(2, now - Duration::from_secs(30), false),
|
||||
(3, now - Duration::from_secs(20), true),
|
||||
(4, now - Duration::from_secs(19), false),
|
||||
(5, now - Duration::from_secs(18), false),
|
||||
(6, now - Duration::from_secs(17), false),
|
||||
(7, now - Duration::from_secs(16), false),
|
||||
(8, now - Duration::from_secs(15), false),
|
||||
(9, now - Duration::from_secs(14), false),
|
||||
(10, now - Duration::from_secs(13), true),
|
||||
(id(1), now - Duration::from_secs(40), false),
|
||||
(id(2), now - Duration::from_secs(30), false),
|
||||
(id(3), now - Duration::from_secs(20), true),
|
||||
(id(4), now - Duration::from_secs(19), false),
|
||||
(id(5), now - Duration::from_secs(18), false),
|
||||
(id(6), now - Duration::from_secs(17), false),
|
||||
(id(7), now - Duration::from_secs(16), false),
|
||||
(id(8), now - Duration::from_secs(15), false),
|
||||
(id(9), now - Duration::from_secs(14), false),
|
||||
(id(10), now - Duration::from_secs(13), true),
|
||||
];
|
||||
|
||||
let candidate = UnifiedExecSessionManager::session_id_to_prune_from_meta(&meta);
|
||||
|
||||
// (10) is exited but among the last 8; we should drop the LRU outside that set.
|
||||
assert_eq!(candidate, Some(1));
|
||||
assert_eq!(candidate, Some(id(1)));
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user