mirror of
https://github.com/openai/codex.git
synced 2026-03-14 02:33:48 +00:00
Compare commits
5 Commits
latest-alp
...
dev/steve/
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
3d881fb590 | ||
|
|
2fc0ca1ec8 | ||
|
|
c9230a345d | ||
|
|
a0b88262da | ||
|
|
9a44a7e499 |
35
.github/workflows/rust-release-windows.yml
vendored
35
.github/workflows/rust-release-windows.yml
vendored
@@ -181,6 +181,41 @@ jobs:
|
||||
account-name: ${{ secrets.AZURE_TRUSTED_SIGNING_ACCOUNT_NAME }}
|
||||
certificate-profile-name: ${{ secrets.AZURE_TRUSTED_SIGNING_CERTIFICATE_PROFILE_NAME }}
|
||||
|
||||
- name: Setup Python for runtime packaging
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: "3.13"
|
||||
|
||||
- name: Build Python runtime wheel
|
||||
shell: bash
|
||||
working-directory: ${{ github.workspace }}
|
||||
env:
|
||||
RELEASE_TAG: ${{ github.ref_name }}
|
||||
TARGET: ${{ matrix.target }}
|
||||
run: |
|
||||
set -euo pipefail
|
||||
|
||||
version="${RELEASE_TAG#rust-v}"
|
||||
staging_dir="${RUNNER_TEMP}/codex-cli-bin-${TARGET}"
|
||||
out_dir="${GITHUB_WORKSPACE}/dist-python/${TARGET}"
|
||||
|
||||
python -m pip install --upgrade pip
|
||||
python -m pip install build hatchling
|
||||
|
||||
python sdk/python/scripts/update_sdk_artifacts.py \
|
||||
stage-runtime \
|
||||
"${staging_dir}" \
|
||||
"codex-rs/target/${TARGET}/release/codex.exe" \
|
||||
--runtime-version "${version}"
|
||||
|
||||
python -m build --wheel --outdir "${out_dir}" "${staging_dir}"
|
||||
|
||||
- name: Upload Python runtime wheel
|
||||
uses: actions/upload-artifact@v7
|
||||
with:
|
||||
name: python-runtime-${{ matrix.target }}
|
||||
path: dist-python/${{ matrix.target }}/*
|
||||
|
||||
- name: Stage artifacts
|
||||
shell: bash
|
||||
run: |
|
||||
|
||||
220
.github/workflows/rust-release.yml
vendored
220
.github/workflows/rust-release.yml
vendored
@@ -302,6 +302,41 @@ jobs:
|
||||
apple-notarization-key-id: ${{ secrets.APPLE_NOTARIZATION_KEY_ID }}
|
||||
apple-notarization-issuer-id: ${{ secrets.APPLE_NOTARIZATION_ISSUER_ID }}
|
||||
|
||||
- name: Setup Python for runtime packaging
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: "3.13"
|
||||
|
||||
- name: Build Python runtime wheel
|
||||
shell: bash
|
||||
working-directory: ${{ github.workspace }}
|
||||
env:
|
||||
RELEASE_TAG: ${{ github.ref_name }}
|
||||
TARGET: ${{ matrix.target }}
|
||||
run: |
|
||||
set -euo pipefail
|
||||
|
||||
version="${RELEASE_TAG#rust-v}"
|
||||
staging_dir="${RUNNER_TEMP}/codex-cli-bin-${TARGET}"
|
||||
out_dir="${GITHUB_WORKSPACE}/dist-python/${TARGET}"
|
||||
|
||||
python -m pip install --upgrade pip
|
||||
python -m pip install build hatchling
|
||||
|
||||
python sdk/python/scripts/update_sdk_artifacts.py \
|
||||
stage-runtime \
|
||||
"${staging_dir}" \
|
||||
"codex-rs/target/${TARGET}/release/codex" \
|
||||
--runtime-version "${version}"
|
||||
|
||||
python -m build --wheel --outdir "${out_dir}" "${staging_dir}"
|
||||
|
||||
- name: Upload Python runtime wheel
|
||||
uses: actions/upload-artifact@v7
|
||||
with:
|
||||
name: python-runtime-${{ matrix.target }}
|
||||
path: dist-python/${{ matrix.target }}/*
|
||||
|
||||
- name: Stage artifacts
|
||||
shell: bash
|
||||
run: |
|
||||
@@ -384,6 +419,7 @@ jobs:
|
||||
needs:
|
||||
- build
|
||||
- build-windows
|
||||
- build-python-sdk
|
||||
- shell-tool-mcp
|
||||
name: release
|
||||
runs-on: ubuntu-latest
|
||||
@@ -533,6 +569,190 @@ jobs:
|
||||
exit 1
|
||||
fi
|
||||
|
||||
build-python-sdk:
|
||||
name: build-python-sdk
|
||||
needs:
|
||||
- tag-check
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v6
|
||||
|
||||
- name: Setup Python
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: "3.13"
|
||||
|
||||
- name: Build Python SDK artifacts
|
||||
shell: bash
|
||||
env:
|
||||
RELEASE_TAG: ${{ github.ref_name }}
|
||||
run: |
|
||||
set -euo pipefail
|
||||
|
||||
version="${RELEASE_TAG#rust-v}"
|
||||
core_staging_dir="${RUNNER_TEMP}/codex-app-server-sdk-core"
|
||||
core_out_dir="${GITHUB_WORKSPACE}/dist-python/sdk-core"
|
||||
bundled_staging_dir="${RUNNER_TEMP}/codex-app-server-sdk"
|
||||
bundled_out_dir="${GITHUB_WORKSPACE}/dist-python/sdk"
|
||||
|
||||
python -m pip install --upgrade pip
|
||||
python -m pip install \
|
||||
build \
|
||||
hatchling \
|
||||
"datamodel-code-generator==0.31.2" \
|
||||
"ruff==0.11.13"
|
||||
|
||||
python sdk/python/scripts/update_sdk_artifacts.py generate-types
|
||||
|
||||
python sdk/python/scripts/update_sdk_artifacts.py \
|
||||
stage-sdk-core \
|
||||
"${core_staging_dir}" \
|
||||
--sdk-version "${version}"
|
||||
|
||||
python -m build --outdir "${core_out_dir}" "${core_staging_dir}"
|
||||
|
||||
python sdk/python/scripts/update_sdk_artifacts.py \
|
||||
stage-sdk \
|
||||
"${bundled_staging_dir}" \
|
||||
--sdk-version "${version}" \
|
||||
--runtime-version "${version}"
|
||||
|
||||
python -m build --outdir "${bundled_out_dir}" "${bundled_staging_dir}"
|
||||
|
||||
- name: Upload Python core SDK artifacts
|
||||
uses: actions/upload-artifact@v7
|
||||
with:
|
||||
name: python-sdk-core
|
||||
path: dist-python/sdk-core/*
|
||||
|
||||
- name: Upload Python SDK artifacts
|
||||
uses: actions/upload-artifact@v7
|
||||
with:
|
||||
name: python-sdk
|
||||
path: dist-python/sdk/*
|
||||
|
||||
publish-pypi-runtime:
|
||||
name: publish-pypi-runtime
|
||||
needs:
|
||||
- build
|
||||
- build-windows
|
||||
- release
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
id-token: write
|
||||
contents: read
|
||||
environment:
|
||||
name: pypi
|
||||
|
||||
steps:
|
||||
# Do not publish musl runtime wheels to PyPI yet. GNU and musl builds for
|
||||
# the same architecture currently infer the same wheel tag, which can make
|
||||
# the published Linux runtime nondeterministic.
|
||||
- name: Download macOS arm64 runtime wheel
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-runtime-aarch64-apple-darwin
|
||||
path: dist-pypi/runtime
|
||||
|
||||
- name: Download macOS x64 runtime wheel
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-runtime-x86_64-apple-darwin
|
||||
path: dist-pypi/runtime
|
||||
|
||||
- name: Download Linux GNU arm64 runtime wheel
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-runtime-aarch64-unknown-linux-gnu
|
||||
path: dist-pypi/runtime
|
||||
|
||||
- name: Download Linux GNU x64 runtime wheel
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-runtime-x86_64-unknown-linux-gnu
|
||||
path: dist-pypi/runtime
|
||||
|
||||
- name: Download Windows arm64 runtime wheel
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-runtime-aarch64-pc-windows-msvc
|
||||
path: dist-pypi/runtime
|
||||
|
||||
- name: Download Windows x64 runtime wheel
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-runtime-x86_64-pc-windows-msvc
|
||||
path: dist-pypi/runtime
|
||||
|
||||
- name: List runtime wheels
|
||||
shell: bash
|
||||
run: ls -R dist-pypi/runtime
|
||||
|
||||
- name: Publish Python runtime wheels to PyPI
|
||||
uses: pypa/gh-action-pypi-publish@release/v1
|
||||
with:
|
||||
packages-dir: dist-pypi/runtime/
|
||||
|
||||
publish-pypi-sdk-core:
|
||||
name: publish-pypi-sdk-core
|
||||
needs:
|
||||
- build-python-sdk
|
||||
- publish-pypi-runtime
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
id-token: write
|
||||
contents: read
|
||||
environment:
|
||||
name: pypi
|
||||
|
||||
steps:
|
||||
- name: Download Python core SDK artifacts
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-sdk-core
|
||||
path: dist-pypi/sdk-core
|
||||
|
||||
- name: List core SDK artifacts
|
||||
shell: bash
|
||||
run: ls -R dist-pypi/sdk-core
|
||||
|
||||
- name: Publish Python core SDK to PyPI
|
||||
uses: pypa/gh-action-pypi-publish@release/v1
|
||||
with:
|
||||
packages-dir: dist-pypi/sdk-core/
|
||||
|
||||
publish-pypi-sdk:
|
||||
name: publish-pypi-sdk
|
||||
needs:
|
||||
- build-python-sdk
|
||||
- publish-pypi-sdk-core
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
id-token: write
|
||||
contents: read
|
||||
environment:
|
||||
name: pypi
|
||||
|
||||
steps:
|
||||
- name: Download bundled Python SDK artifacts
|
||||
uses: actions/download-artifact@v8
|
||||
with:
|
||||
name: python-sdk
|
||||
path: dist-pypi/sdk
|
||||
|
||||
- name: List bundled SDK artifacts
|
||||
shell: bash
|
||||
run: ls -R dist-pypi/sdk
|
||||
|
||||
- name: Publish bundled Python SDK to PyPI
|
||||
uses: pypa/gh-action-pypi-publish@release/v1
|
||||
with:
|
||||
packages-dir: dist-pypi/sdk/
|
||||
|
||||
# Publish to npm using OIDC authentication.
|
||||
# July 31, 2025: https://github.blog/changelog/2025-07-31-npm-trusted-publishing-with-oidc-is-generally-available/
|
||||
# npm docs: https://docs.npmjs.com/trusted-publishers
|
||||
|
||||
@@ -5652,7 +5652,6 @@ pub(crate) async fn run_turn(
|
||||
.await;
|
||||
let mut last_agent_message: Option<String> = None;
|
||||
let mut stop_hook_active = false;
|
||||
let mut pending_stop_hook_message: Option<String> = None;
|
||||
// Although from the perspective of codex.rs, TurnDiffTracker has the lifecycle of a Task which contains
|
||||
// many turns, from the perspective of the user, it is a single turn.
|
||||
let turn_diff_tracker = Arc::new(tokio::sync::Mutex::new(TurnDiffTracker::new()));
|
||||
@@ -5744,14 +5743,11 @@ pub(crate) async fn run_turn(
|
||||
}
|
||||
|
||||
// Construct the input that we will send to the model.
|
||||
let mut sampling_request_input: Vec<ResponseItem> = {
|
||||
let sampling_request_input: Vec<ResponseItem> = {
|
||||
sess.clone_history()
|
||||
.await
|
||||
.for_prompt(&turn_context.model_info.input_modalities)
|
||||
};
|
||||
if let Some(stop_hook_message) = pending_stop_hook_message.take() {
|
||||
sampling_request_input.push(DeveloperInstructions::new(stop_hook_message).into());
|
||||
}
|
||||
|
||||
let sampling_request_input_messages = sampling_request_input
|
||||
.iter()
|
||||
@@ -5848,18 +5844,25 @@ pub(crate) async fn run_turn(
|
||||
.await;
|
||||
}
|
||||
if stop_outcome.should_block {
|
||||
if stop_hook_active {
|
||||
if let Some(continuation_prompt) = stop_outcome.continuation_prompt.clone()
|
||||
{
|
||||
let developer_message: ResponseItem =
|
||||
DeveloperInstructions::new(continuation_prompt).into();
|
||||
sess.record_conversation_items(
|
||||
&turn_context,
|
||||
std::slice::from_ref(&developer_message),
|
||||
)
|
||||
.await;
|
||||
stop_hook_active = true;
|
||||
continue;
|
||||
} else {
|
||||
sess.send_event(
|
||||
&turn_context,
|
||||
EventMsg::Warning(WarningEvent {
|
||||
message: "Stop hook blocked twice in the same turn; ignoring the second block to avoid an infinite loop.".to_string(),
|
||||
message: "Stop hook requested continuation without a prompt; ignoring the block.".to_string(),
|
||||
}),
|
||||
)
|
||||
.await;
|
||||
} else {
|
||||
stop_hook_active = true;
|
||||
pending_stop_hook_message = stop_outcome.block_message_for_model;
|
||||
continue;
|
||||
}
|
||||
}
|
||||
if stop_outcome.should_stop {
|
||||
|
||||
271
codex-rs/core/tests/suite/hooks.rs
Normal file
271
codex-rs/core/tests/suite/hooks.rs
Normal file
@@ -0,0 +1,271 @@
|
||||
use std::fs;
|
||||
use std::path::Path;
|
||||
|
||||
use anyhow::Context;
|
||||
use anyhow::Result;
|
||||
use codex_core::features::Feature;
|
||||
use codex_protocol::models::ContentItem;
|
||||
use codex_protocol::models::ResponseItem;
|
||||
use codex_protocol::protocol::RolloutItem;
|
||||
use codex_protocol::protocol::RolloutLine;
|
||||
use core_test_support::responses::ev_assistant_message;
|
||||
use core_test_support::responses::ev_completed;
|
||||
use core_test_support::responses::ev_response_created;
|
||||
use core_test_support::responses::mount_sse_once;
|
||||
use core_test_support::responses::mount_sse_sequence;
|
||||
use core_test_support::responses::sse;
|
||||
use core_test_support::responses::start_mock_server;
|
||||
use core_test_support::skip_if_no_network;
|
||||
use core_test_support::test_codex::test_codex;
|
||||
use pretty_assertions::assert_eq;
|
||||
|
||||
const FIRST_CONTINUATION_PROMPT: &str = "Retry with exactly the phrase meow meow meow.";
|
||||
const SECOND_CONTINUATION_PROMPT: &str = "Now tighten it to just: meow.";
|
||||
|
||||
fn write_stop_hook(home: &Path, block_prompts: &[&str]) -> Result<()> {
|
||||
let script_path = home.join("stop_hook.py");
|
||||
let log_path = home.join("stop_hook_log.jsonl");
|
||||
let prompts_json =
|
||||
serde_json::to_string(block_prompts).context("serialize stop hook prompts for test")?;
|
||||
let script = format!(
|
||||
r#"import json
|
||||
from pathlib import Path
|
||||
import sys
|
||||
|
||||
log_path = Path(r"{log_path}")
|
||||
block_prompts = {prompts_json}
|
||||
|
||||
payload = json.load(sys.stdin)
|
||||
existing = []
|
||||
if log_path.exists():
|
||||
existing = [line for line in log_path.read_text(encoding="utf-8").splitlines() if line.strip()]
|
||||
|
||||
with log_path.open("a", encoding="utf-8") as handle:
|
||||
handle.write(json.dumps(payload) + "\n")
|
||||
|
||||
invocation_index = len(existing)
|
||||
if invocation_index < len(block_prompts):
|
||||
print(json.dumps({{"decision": "block", "reason": block_prompts[invocation_index]}}))
|
||||
else:
|
||||
print(json.dumps({{"systemMessage": f"stop hook pass {{invocation_index + 1}} complete"}}))
|
||||
"#,
|
||||
log_path = log_path.display(),
|
||||
prompts_json = prompts_json,
|
||||
);
|
||||
let hooks = serde_json::json!({
|
||||
"hooks": {
|
||||
"Stop": [{
|
||||
"hooks": [{
|
||||
"type": "command",
|
||||
"command": format!("python3 {}", script_path.display()),
|
||||
"statusMessage": "running stop hook",
|
||||
}]
|
||||
}]
|
||||
}
|
||||
});
|
||||
|
||||
fs::write(&script_path, script).context("write stop hook script")?;
|
||||
fs::write(home.join("hooks.json"), hooks.to_string()).context("write hooks.json")?;
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn rollout_developer_texts(text: &str) -> Result<Vec<String>> {
|
||||
let mut texts = Vec::new();
|
||||
for line in text.lines() {
|
||||
let trimmed = line.trim();
|
||||
if trimmed.is_empty() {
|
||||
continue;
|
||||
}
|
||||
let rollout: RolloutLine = serde_json::from_str(trimmed).context("parse rollout line")?;
|
||||
if let RolloutItem::ResponseItem(ResponseItem::Message { role, content, .. }) = rollout.item
|
||||
&& role == "developer"
|
||||
{
|
||||
for item in content {
|
||||
if let ContentItem::InputText { text } = item {
|
||||
texts.push(text);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
Ok(texts)
|
||||
}
|
||||
|
||||
fn read_stop_hook_inputs(home: &Path) -> Result<Vec<serde_json::Value>> {
|
||||
fs::read_to_string(home.join("stop_hook_log.jsonl"))
|
||||
.context("read stop hook log")?
|
||||
.lines()
|
||||
.filter(|line| !line.trim().is_empty())
|
||||
.map(|line| serde_json::from_str(line).context("parse stop hook log line"))
|
||||
.collect()
|
||||
}
|
||||
|
||||
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
||||
async fn stop_hook_can_block_multiple_times_in_same_turn() -> Result<()> {
|
||||
skip_if_no_network!(Ok(()));
|
||||
|
||||
let server = start_mock_server().await;
|
||||
let responses = mount_sse_sequence(
|
||||
&server,
|
||||
vec![
|
||||
sse(vec![
|
||||
ev_response_created("resp-1"),
|
||||
ev_assistant_message("msg-1", "draft one"),
|
||||
ev_completed("resp-1"),
|
||||
]),
|
||||
sse(vec![
|
||||
ev_response_created("resp-2"),
|
||||
ev_assistant_message("msg-2", "draft two"),
|
||||
ev_completed("resp-2"),
|
||||
]),
|
||||
sse(vec![
|
||||
ev_response_created("resp-3"),
|
||||
ev_assistant_message("msg-3", "final draft"),
|
||||
ev_completed("resp-3"),
|
||||
]),
|
||||
],
|
||||
)
|
||||
.await;
|
||||
|
||||
let mut builder = test_codex()
|
||||
.with_pre_build_hook(|home| {
|
||||
if let Err(error) = write_stop_hook(
|
||||
home,
|
||||
&[FIRST_CONTINUATION_PROMPT, SECOND_CONTINUATION_PROMPT],
|
||||
) {
|
||||
panic!("failed to write stop hook test fixture: {error}");
|
||||
}
|
||||
})
|
||||
.with_config(|config| {
|
||||
config
|
||||
.features
|
||||
.enable(Feature::CodexHooks)
|
||||
.expect("test config should allow feature update");
|
||||
});
|
||||
let test = builder.build(&server).await?;
|
||||
|
||||
test.submit_turn("hello from the sea").await?;
|
||||
|
||||
let requests = responses.requests();
|
||||
assert_eq!(requests.len(), 3);
|
||||
assert!(
|
||||
requests[1]
|
||||
.message_input_texts("developer")
|
||||
.contains(&FIRST_CONTINUATION_PROMPT.to_string()),
|
||||
"second request should include the first continuation prompt",
|
||||
);
|
||||
assert!(
|
||||
requests[2]
|
||||
.message_input_texts("developer")
|
||||
.contains(&FIRST_CONTINUATION_PROMPT.to_string()),
|
||||
"third request should retain the first continuation prompt from history",
|
||||
);
|
||||
assert!(
|
||||
requests[2]
|
||||
.message_input_texts("developer")
|
||||
.contains(&SECOND_CONTINUATION_PROMPT.to_string()),
|
||||
"third request should include the second continuation prompt",
|
||||
);
|
||||
|
||||
let hook_inputs = read_stop_hook_inputs(test.codex_home_path())?;
|
||||
assert_eq!(hook_inputs.len(), 3);
|
||||
assert_eq!(
|
||||
hook_inputs
|
||||
.iter()
|
||||
.map(|input| input["stop_hook_active"]
|
||||
.as_bool()
|
||||
.expect("stop_hook_active bool"))
|
||||
.collect::<Vec<_>>(),
|
||||
vec![false, true, true],
|
||||
);
|
||||
|
||||
let rollout_path = test.codex.rollout_path().expect("rollout path");
|
||||
let rollout_text = fs::read_to_string(&rollout_path)?;
|
||||
let developer_texts = rollout_developer_texts(&rollout_text)?;
|
||||
assert!(
|
||||
developer_texts.contains(&FIRST_CONTINUATION_PROMPT.to_string()),
|
||||
"rollout should persist the first continuation prompt",
|
||||
);
|
||||
assert!(
|
||||
developer_texts.contains(&SECOND_CONTINUATION_PROMPT.to_string()),
|
||||
"rollout should persist the second continuation prompt",
|
||||
);
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
||||
async fn resumed_thread_keeps_stop_continuation_prompt_in_history() -> Result<()> {
|
||||
skip_if_no_network!(Ok(()));
|
||||
|
||||
let server = start_mock_server().await;
|
||||
let initial_responses = mount_sse_sequence(
|
||||
&server,
|
||||
vec![
|
||||
sse(vec![
|
||||
ev_response_created("resp-1"),
|
||||
ev_assistant_message("msg-1", "initial draft"),
|
||||
ev_completed("resp-1"),
|
||||
]),
|
||||
sse(vec![
|
||||
ev_response_created("resp-2"),
|
||||
ev_assistant_message("msg-2", "revised draft"),
|
||||
ev_completed("resp-2"),
|
||||
]),
|
||||
],
|
||||
)
|
||||
.await;
|
||||
|
||||
let mut initial_builder = test_codex()
|
||||
.with_pre_build_hook(|home| {
|
||||
if let Err(error) = write_stop_hook(home, &[FIRST_CONTINUATION_PROMPT]) {
|
||||
panic!("failed to write stop hook test fixture: {error}");
|
||||
}
|
||||
})
|
||||
.with_config(|config| {
|
||||
config
|
||||
.features
|
||||
.enable(Feature::CodexHooks)
|
||||
.expect("test config should allow feature update");
|
||||
});
|
||||
let initial = initial_builder.build(&server).await?;
|
||||
let home = initial.home.clone();
|
||||
let rollout_path = initial
|
||||
.session_configured
|
||||
.rollout_path
|
||||
.clone()
|
||||
.expect("rollout path");
|
||||
|
||||
initial.submit_turn("tell me something").await?;
|
||||
|
||||
assert_eq!(initial_responses.requests().len(), 2);
|
||||
|
||||
let resumed_response = mount_sse_once(
|
||||
&server,
|
||||
sse(vec![
|
||||
ev_response_created("resp-3"),
|
||||
ev_assistant_message("msg-3", "fresh turn after resume"),
|
||||
ev_completed("resp-3"),
|
||||
]),
|
||||
)
|
||||
.await;
|
||||
|
||||
let mut resume_builder = test_codex().with_config(|config| {
|
||||
config
|
||||
.features
|
||||
.enable(Feature::CodexHooks)
|
||||
.expect("test config should allow feature update");
|
||||
});
|
||||
let resumed = resume_builder.resume(&server, home, rollout_path).await?;
|
||||
|
||||
resumed.submit_turn("and now continue").await?;
|
||||
|
||||
let resumed_request = resumed_response.single_request();
|
||||
assert!(
|
||||
resumed_request
|
||||
.message_input_texts("developer")
|
||||
.contains(&FIRST_CONTINUATION_PROMPT.to_string()),
|
||||
"resumed request should keep the persisted continuation prompt in history",
|
||||
);
|
||||
|
||||
Ok(())
|
||||
}
|
||||
@@ -77,6 +77,8 @@ mod exec_policy;
|
||||
mod fork_thread;
|
||||
mod grep_files;
|
||||
mod hierarchical_agents;
|
||||
#[cfg(not(target_os = "windows"))]
|
||||
mod hooks;
|
||||
mod image_rollout;
|
||||
mod items;
|
||||
mod js_repl;
|
||||
|
||||
@@ -24,6 +24,7 @@
|
||||
},
|
||||
"reason": {
|
||||
"default": null,
|
||||
"description": "Claude requires `reason` when `decision` is `block`; we enforce that semantic rule during output parsing rather than in the JSON schema.",
|
||||
"type": "string"
|
||||
},
|
||||
"stopReason": {
|
||||
|
||||
@@ -17,6 +17,7 @@ pub(crate) struct StopOutput {
|
||||
pub universal: UniversalOutput,
|
||||
pub should_block: bool,
|
||||
pub reason: Option<String>,
|
||||
pub invalid_block_reason: Option<String>,
|
||||
}
|
||||
|
||||
use crate::schema::HookUniversalOutputWire;
|
||||
@@ -37,10 +38,21 @@ pub(crate) fn parse_session_start(stdout: &str) -> Option<SessionStartOutput> {
|
||||
|
||||
pub(crate) fn parse_stop(stdout: &str) -> Option<StopOutput> {
|
||||
let wire: StopCommandOutputWire = parse_json(stdout)?;
|
||||
let should_block = matches!(wire.decision, Some(StopDecisionWire::Block));
|
||||
let invalid_block_reason = if should_block
|
||||
&& match wire.reason.as_deref() {
|
||||
Some(reason) => reason.trim().is_empty(),
|
||||
None => true,
|
||||
} {
|
||||
Some(invalid_block_message())
|
||||
} else {
|
||||
None
|
||||
};
|
||||
Some(StopOutput {
|
||||
universal: UniversalOutput::from(wire.universal),
|
||||
should_block: matches!(wire.decision, Some(StopDecisionWire::Block)),
|
||||
should_block: should_block && invalid_block_reason.is_none(),
|
||||
reason: wire.reason,
|
||||
invalid_block_reason,
|
||||
})
|
||||
}
|
||||
|
||||
@@ -69,3 +81,7 @@ where
|
||||
}
|
||||
serde_json::from_value(value).ok()
|
||||
}
|
||||
|
||||
fn invalid_block_message() -> String {
|
||||
"Stop hook returned decision:block without a non-empty reason".to_string()
|
||||
}
|
||||
|
||||
@@ -34,16 +34,16 @@ pub struct StopOutcome {
|
||||
pub stop_reason: Option<String>,
|
||||
pub should_block: bool,
|
||||
pub block_reason: Option<String>,
|
||||
pub block_message_for_model: Option<String>,
|
||||
pub continuation_prompt: Option<String>,
|
||||
}
|
||||
|
||||
#[derive(Debug, PartialEq, Eq)]
|
||||
#[derive(Debug, Default, PartialEq, Eq)]
|
||||
struct StopHandlerData {
|
||||
should_stop: bool,
|
||||
stop_reason: Option<String>,
|
||||
should_block: bool,
|
||||
block_reason: Option<String>,
|
||||
block_message_for_model: Option<String>,
|
||||
continuation_prompt: Option<String>,
|
||||
}
|
||||
|
||||
pub(crate) fn preview(
|
||||
@@ -69,7 +69,7 @@ pub(crate) async fn run(
|
||||
stop_reason: None,
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
continuation_prompt: None,
|
||||
};
|
||||
}
|
||||
|
||||
@@ -102,34 +102,15 @@ pub(crate) async fn run(
|
||||
)
|
||||
.await;
|
||||
|
||||
let should_stop = results.iter().any(|result| result.data.should_stop);
|
||||
let stop_reason = results
|
||||
.iter()
|
||||
.find_map(|result| result.data.stop_reason.clone());
|
||||
|
||||
let should_block = !should_stop && results.iter().any(|result| result.data.should_block);
|
||||
let block_reason = if should_block {
|
||||
results
|
||||
.iter()
|
||||
.find_map(|result| result.data.block_reason.clone())
|
||||
} else {
|
||||
None
|
||||
};
|
||||
let block_message_for_model = if should_block {
|
||||
results
|
||||
.iter()
|
||||
.find_map(|result| result.data.block_message_for_model.clone())
|
||||
} else {
|
||||
None
|
||||
};
|
||||
let aggregate = aggregate_results(results.iter().map(|result| &result.data));
|
||||
|
||||
StopOutcome {
|
||||
hook_events: results.into_iter().map(|result| result.completed).collect(),
|
||||
should_stop,
|
||||
stop_reason,
|
||||
should_block,
|
||||
block_reason,
|
||||
block_message_for_model,
|
||||
should_stop: aggregate.should_stop,
|
||||
stop_reason: aggregate.stop_reason,
|
||||
should_block: aggregate.should_block,
|
||||
block_reason: aggregate.block_reason,
|
||||
continuation_prompt: aggregate.continuation_prompt,
|
||||
}
|
||||
}
|
||||
|
||||
@@ -144,7 +125,7 @@ fn parse_completed(
|
||||
let mut stop_reason = None;
|
||||
let mut should_block = false;
|
||||
let mut block_reason = None;
|
||||
let mut block_message_for_model = None;
|
||||
let mut continuation_prompt = None;
|
||||
|
||||
match run_result.error.as_deref() {
|
||||
Some(error) => {
|
||||
@@ -176,12 +157,18 @@ fn parse_completed(
|
||||
text: stop_reason_text,
|
||||
});
|
||||
}
|
||||
} else if let Some(invalid_block_reason) = parsed.invalid_block_reason {
|
||||
status = HookRunStatus::Failed;
|
||||
entries.push(HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: invalid_block_reason,
|
||||
});
|
||||
} else if parsed.should_block {
|
||||
if let Some(reason) = parsed.reason.as_deref().and_then(trimmed_non_empty) {
|
||||
status = HookRunStatus::Blocked;
|
||||
should_block = true;
|
||||
block_reason = Some(reason.clone());
|
||||
block_message_for_model = Some(reason.clone());
|
||||
continuation_prompt = Some(reason.clone());
|
||||
entries.push(HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Feedback,
|
||||
text: reason,
|
||||
@@ -190,8 +177,9 @@ fn parse_completed(
|
||||
status = HookRunStatus::Failed;
|
||||
entries.push(HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: "hook returned decision \"block\" without a non-empty reason"
|
||||
.to_string(),
|
||||
text:
|
||||
"Stop hook returned decision:block without a non-empty reason"
|
||||
.to_string(),
|
||||
});
|
||||
}
|
||||
}
|
||||
@@ -208,7 +196,7 @@ fn parse_completed(
|
||||
status = HookRunStatus::Blocked;
|
||||
should_block = true;
|
||||
block_reason = Some(reason.clone());
|
||||
block_message_for_model = Some(reason.clone());
|
||||
continuation_prompt = Some(reason.clone());
|
||||
entries.push(HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Feedback,
|
||||
text: reason,
|
||||
@@ -217,7 +205,9 @@ fn parse_completed(
|
||||
status = HookRunStatus::Failed;
|
||||
entries.push(HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: "hook exited with code 2 without stderr feedback".to_string(),
|
||||
text:
|
||||
"Stop hook exited with code 2 but did not write a continuation prompt to stderr"
|
||||
.to_string(),
|
||||
});
|
||||
}
|
||||
}
|
||||
@@ -250,11 +240,57 @@ fn parse_completed(
|
||||
stop_reason,
|
||||
should_block,
|
||||
block_reason,
|
||||
block_message_for_model,
|
||||
continuation_prompt,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
fn aggregate_results<'a>(
|
||||
results: impl IntoIterator<Item = &'a StopHandlerData>,
|
||||
) -> StopHandlerData {
|
||||
let results = results.into_iter().collect::<Vec<_>>();
|
||||
let should_stop = results.iter().any(|result| result.should_stop);
|
||||
let stop_reason = results.iter().find_map(|result| result.stop_reason.clone());
|
||||
let should_block = !should_stop && results.iter().any(|result| result.should_block);
|
||||
let block_reason = if should_block {
|
||||
join_block_text(results.iter().copied(), |result| {
|
||||
result.block_reason.as_deref()
|
||||
})
|
||||
} else {
|
||||
None
|
||||
};
|
||||
let continuation_prompt = if should_block {
|
||||
join_block_text(results.iter().copied(), |result| {
|
||||
result.continuation_prompt.as_deref()
|
||||
})
|
||||
} else {
|
||||
None
|
||||
};
|
||||
|
||||
StopHandlerData {
|
||||
should_stop,
|
||||
stop_reason,
|
||||
should_block,
|
||||
block_reason,
|
||||
continuation_prompt,
|
||||
}
|
||||
}
|
||||
|
||||
fn join_block_text<'a>(
|
||||
results: impl IntoIterator<Item = &'a StopHandlerData>,
|
||||
select: impl Fn(&'a StopHandlerData) -> Option<&'a str>,
|
||||
) -> Option<String> {
|
||||
let parts = results
|
||||
.into_iter()
|
||||
.filter_map(select)
|
||||
.map(str::to_owned)
|
||||
.collect::<Vec<_>>();
|
||||
if parts.is_empty() {
|
||||
return None;
|
||||
}
|
||||
Some(parts.join("\n\n"))
|
||||
}
|
||||
|
||||
fn trimmed_non_empty(text: &str) -> Option<String> {
|
||||
let trimmed = text.trim();
|
||||
if !trimmed.is_empty() {
|
||||
@@ -292,7 +328,7 @@ fn serialization_failure_outcome(
|
||||
stop_reason: None,
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
continuation_prompt: None,
|
||||
}
|
||||
}
|
||||
|
||||
@@ -307,10 +343,55 @@ mod tests {
|
||||
use pretty_assertions::assert_eq;
|
||||
|
||||
use super::StopHandlerData;
|
||||
use super::aggregate_results;
|
||||
use super::parse_completed;
|
||||
use crate::engine::ConfiguredHandler;
|
||||
use crate::engine::command_runner::CommandRunResult;
|
||||
|
||||
#[test]
|
||||
fn block_decision_with_reason_sets_continuation_prompt() {
|
||||
let parsed = parse_completed(
|
||||
&handler(),
|
||||
run_result(
|
||||
Some(0),
|
||||
r#"{"decision":"block","reason":"retry with tests"}"#,
|
||||
"",
|
||||
),
|
||||
Some("turn-1".to_string()),
|
||||
);
|
||||
|
||||
assert_eq!(
|
||||
parsed.data,
|
||||
StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: true,
|
||||
block_reason: Some("retry with tests".to_string()),
|
||||
continuation_prompt: Some("retry with tests".to_string()),
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Blocked);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn block_decision_without_reason_is_invalid() {
|
||||
let parsed = parse_completed(
|
||||
&handler(),
|
||||
run_result(Some(0), r#"{"decision":"block"}"#, ""),
|
||||
Some("turn-1".to_string()),
|
||||
);
|
||||
|
||||
assert_eq!(parsed.data, StopHandlerData::default());
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Failed);
|
||||
assert_eq!(
|
||||
parsed.completed.run.entries,
|
||||
vec![HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: "Stop hook returned decision:block without a non-empty reason".to_string(),
|
||||
}]
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn continue_false_overrides_block_decision() {
|
||||
let parsed = parse_completed(
|
||||
@@ -330,7 +411,7 @@ mod tests {
|
||||
stop_reason: Some("done".to_string()),
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
continuation_prompt: None,
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Stopped);
|
||||
@@ -351,36 +432,25 @@ mod tests {
|
||||
stop_reason: None,
|
||||
should_block: true,
|
||||
block_reason: Some("retry with tests".to_string()),
|
||||
block_message_for_model: Some("retry with tests".to_string()),
|
||||
continuation_prompt: Some("retry with tests".to_string()),
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Blocked);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn block_decision_without_reason_fails_instead_of_blocking() {
|
||||
let parsed = parse_completed(
|
||||
&handler(),
|
||||
run_result(Some(0), r#"{"decision":"block"}"#, ""),
|
||||
Some("turn-1".to_string()),
|
||||
);
|
||||
fn exit_code_two_without_stderr_does_not_block() {
|
||||
let parsed = parse_completed(&handler(), run_result(Some(2), "", " "), None);
|
||||
|
||||
assert_eq!(
|
||||
parsed.data,
|
||||
StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.data, StopHandlerData::default());
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Failed);
|
||||
assert_eq!(
|
||||
parsed.completed.run.entries,
|
||||
vec![HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: "hook returned decision \"block\" without a non-empty reason".to_string(),
|
||||
text:
|
||||
"Stop hook exited with code 2 but did not write a continuation prompt to stderr"
|
||||
.to_string(),
|
||||
}]
|
||||
);
|
||||
}
|
||||
@@ -393,50 +463,13 @@ mod tests {
|
||||
Some("turn-1".to_string()),
|
||||
);
|
||||
|
||||
assert_eq!(
|
||||
parsed.data,
|
||||
StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.data, StopHandlerData::default());
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Failed);
|
||||
assert_eq!(
|
||||
parsed.completed.run.entries,
|
||||
vec![HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: "hook returned decision \"block\" without a non-empty reason".to_string(),
|
||||
}]
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn exit_code_two_without_stderr_feedback_fails_instead_of_blocking() {
|
||||
let parsed = parse_completed(
|
||||
&handler(),
|
||||
run_result(Some(2), "ignored stdout", " "),
|
||||
Some("turn-1".to_string()),
|
||||
);
|
||||
|
||||
assert_eq!(
|
||||
parsed.data,
|
||||
StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Failed);
|
||||
assert_eq!(
|
||||
parsed.completed.run.entries,
|
||||
vec![HookOutputEntry {
|
||||
kind: HookOutputEntryKind::Error,
|
||||
text: "hook exited with code 2 without stderr feedback".to_string(),
|
||||
text: "Stop hook returned decision:block without a non-empty reason".to_string(),
|
||||
}]
|
||||
);
|
||||
}
|
||||
@@ -449,16 +482,7 @@ mod tests {
|
||||
Some("turn-1".to_string()),
|
||||
);
|
||||
|
||||
assert_eq!(
|
||||
parsed.data,
|
||||
StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: false,
|
||||
block_reason: None,
|
||||
block_message_for_model: None,
|
||||
}
|
||||
);
|
||||
assert_eq!(parsed.data, StopHandlerData::default());
|
||||
assert_eq!(parsed.completed.run.status, HookRunStatus::Failed);
|
||||
assert_eq!(
|
||||
parsed.completed.run.entries,
|
||||
@@ -469,6 +493,37 @@ mod tests {
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn aggregate_results_concatenates_blocking_reasons_in_declaration_order() {
|
||||
let aggregate = aggregate_results([
|
||||
&StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: true,
|
||||
block_reason: Some("first".to_string()),
|
||||
continuation_prompt: Some("first".to_string()),
|
||||
},
|
||||
&StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: true,
|
||||
block_reason: Some("second".to_string()),
|
||||
continuation_prompt: Some("second".to_string()),
|
||||
},
|
||||
]);
|
||||
|
||||
assert_eq!(
|
||||
aggregate,
|
||||
StopHandlerData {
|
||||
should_stop: false,
|
||||
stop_reason: None,
|
||||
should_block: true,
|
||||
block_reason: Some("first\n\nsecond".to_string()),
|
||||
continuation_prompt: Some("first\n\nsecond".to_string()),
|
||||
}
|
||||
);
|
||||
}
|
||||
|
||||
fn handler() -> ConfiguredHandler {
|
||||
ConfiguredHandler {
|
||||
event_name: HookEventName::Stop,
|
||||
|
||||
@@ -96,6 +96,8 @@ pub(crate) struct StopCommandOutputWire {
|
||||
pub universal: HookUniversalOutputWire,
|
||||
#[serde(default)]
|
||||
pub decision: Option<StopDecisionWire>,
|
||||
/// Claude requires `reason` when `decision` is `block`; we enforce that
|
||||
/// semantic rule during output parsing rather than in the JSON schema.
|
||||
#[serde(default)]
|
||||
pub reason: Option<String>,
|
||||
}
|
||||
|
||||
@@ -11,9 +11,16 @@ cd sdk/python
|
||||
python -m pip install -e .
|
||||
```
|
||||
|
||||
Published SDK builds pin an exact `codex-cli-bin` runtime dependency. For local
|
||||
repo development, pass `AppServerConfig(codex_bin=...)` to point at a local
|
||||
build explicitly.
|
||||
This checked-in package is the runtime-free core distribution:
|
||||
`codex-app-server-sdk-core`.
|
||||
|
||||
Published releases expose two Python package names:
|
||||
|
||||
- `codex-app-server-sdk-core`: the actual Python SDK code, without bundled binaries
|
||||
- `codex-app-server-sdk`: a bundled metapackage that depends on `codex-app-server-sdk-core` and `codex-cli-bin`
|
||||
|
||||
For local repo development, pass `AppServerConfig(codex_bin=...)` to point at
|
||||
a local build explicitly.
|
||||
|
||||
## Quickstart
|
||||
|
||||
@@ -48,9 +55,9 @@ python examples/01_quickstart_constructor/async.py
|
||||
|
||||
The repo no longer checks `codex` binaries into `sdk/python`.
|
||||
|
||||
Published SDK builds are pinned to an exact `codex-cli-bin` package version,
|
||||
and that runtime package carries the platform-specific binary for the target
|
||||
wheel.
|
||||
Published `codex-app-server-sdk` builds depend on an exact
|
||||
`codex-app-server-sdk-core` version plus an exact `codex-cli-bin` version, and
|
||||
that runtime package carries the platform-specific binary for the target wheel.
|
||||
|
||||
For local repo development, the checked-in `sdk/python-runtime` package is only
|
||||
a template for staged release artifacts. Editable installs should use an
|
||||
@@ -61,9 +68,14 @@ explicit `codex_bin` override instead.
|
||||
```bash
|
||||
cd sdk/python
|
||||
python scripts/update_sdk_artifacts.py generate-types
|
||||
python scripts/update_sdk_artifacts.py \
|
||||
stage-sdk-core \
|
||||
/tmp/codex-python-release/codex-app-server-sdk-core \
|
||||
--sdk-version 1.2.3
|
||||
python scripts/update_sdk_artifacts.py \
|
||||
stage-sdk \
|
||||
/tmp/codex-python-release/codex-app-server-sdk \
|
||||
--sdk-version 1.2.3 \
|
||||
--runtime-version 1.2.3
|
||||
python scripts/update_sdk_artifacts.py \
|
||||
stage-runtime \
|
||||
@@ -75,17 +87,22 @@ python scripts/update_sdk_artifacts.py \
|
||||
This supports the CI release flow:
|
||||
|
||||
- run `generate-types` before packaging
|
||||
- stage `codex-app-server-sdk` once with an exact `codex-cli-bin==...` dependency
|
||||
- stage `codex-app-server-sdk-core` with the release tag version as `--sdk-version`
|
||||
- stage `codex-app-server-sdk` as a bundled metapackage pinned to exact `codex-app-server-sdk-core==...` and `codex-cli-bin==...`
|
||||
- stage `codex-cli-bin` on each supported platform runner with the same pinned runtime version
|
||||
- build and publish `codex-cli-bin` as platform wheels only; do not publish an sdist
|
||||
- publish `codex-app-server-sdk-core` to PyPI after the runtime wheels land
|
||||
- publish `codex-app-server-sdk` to PyPI last, using the same release version
|
||||
|
||||
## Compatibility and versioning
|
||||
|
||||
- Package: `codex-app-server-sdk`
|
||||
- Core package: `codex-app-server-sdk-core`
|
||||
- Bundled package: `codex-app-server-sdk`
|
||||
- Runtime package: `codex-cli-bin`
|
||||
- Current SDK version in this repo: `0.2.0`
|
||||
- Python: `>=3.10`
|
||||
- Target protocol: Codex `app-server` JSON-RPC v2
|
||||
- Release policy: published Python package versions should match the `rust-v...` release tag
|
||||
- Recommendation: keep SDK and `codex` CLI reasonably up to date together
|
||||
|
||||
## Notes
|
||||
|
||||
@@ -38,16 +38,23 @@ Common causes:
|
||||
- local auth/session is missing
|
||||
- incompatible/old app-server
|
||||
|
||||
Maintainers stage releases by building the SDK once and the runtime once per
|
||||
platform with the same pinned runtime version. Publish `codex-cli-bin` as
|
||||
platform wheels only; do not publish an sdist:
|
||||
Maintainers stage releases by building the core SDK once, the bundled SDK
|
||||
metapackage once, and the runtime once per platform with the same pinned
|
||||
runtime version. Publish `codex-cli-bin` as platform wheels only; do not
|
||||
publish an sdist. Published Python package versions should match the
|
||||
`rust-v...` release tag:
|
||||
|
||||
```bash
|
||||
cd sdk/python
|
||||
python scripts/update_sdk_artifacts.py generate-types
|
||||
python scripts/update_sdk_artifacts.py \
|
||||
stage-sdk-core \
|
||||
/tmp/codex-python-release/codex-app-server-sdk-core \
|
||||
--sdk-version 1.2.3
|
||||
python scripts/update_sdk_artifacts.py \
|
||||
stage-sdk \
|
||||
/tmp/codex-python-release/codex-app-server-sdk \
|
||||
--sdk-version 1.2.3 \
|
||||
--runtime-version 1.2.3
|
||||
python scripts/update_sdk_artifacts.py \
|
||||
stage-runtime \
|
||||
|
||||
@@ -3,9 +3,9 @@ requires = ["hatchling>=1.24.0"]
|
||||
build-backend = "hatchling.build"
|
||||
|
||||
[project]
|
||||
name = "codex-app-server-sdk"
|
||||
name = "codex-app-server-sdk-core"
|
||||
version = "0.2.0"
|
||||
description = "Python SDK for Codex app-server v2"
|
||||
description = "Core Python SDK for Codex app-server v2"
|
||||
readme = "README.md"
|
||||
requires-python = ">=3.10"
|
||||
license = { text = "Apache-2.0" }
|
||||
|
||||
@@ -15,8 +15,13 @@ import types
|
||||
import typing
|
||||
from dataclasses import dataclass
|
||||
from pathlib import Path
|
||||
from textwrap import dedent
|
||||
from typing import Any, Callable, Sequence, get_args, get_origin
|
||||
|
||||
CORE_SDK_PKG_NAME = "codex-app-server-sdk-core"
|
||||
BUNDLED_SDK_PKG_NAME = "codex-app-server-sdk"
|
||||
RUNTIME_PKG_NAME = "codex-cli-bin"
|
||||
|
||||
|
||||
def repo_root() -> Path:
|
||||
return Path(__file__).resolve().parents[3]
|
||||
@@ -110,23 +115,7 @@ def _rewrite_project_version(pyproject_text: str, version: str) -> str:
|
||||
return updated
|
||||
|
||||
|
||||
def _rewrite_sdk_runtime_dependency(pyproject_text: str, runtime_version: str) -> str:
|
||||
match = re.search(r"^dependencies = \[(.*?)\]$", pyproject_text, flags=re.MULTILINE)
|
||||
if match is None:
|
||||
raise RuntimeError(
|
||||
"Could not find dependencies array in sdk/python/pyproject.toml"
|
||||
)
|
||||
|
||||
raw_items = [item.strip() for item in match.group(1).split(",") if item.strip()]
|
||||
raw_items = [item for item in raw_items if "codex-cli-bin" not in item]
|
||||
raw_items.append(f'"codex-cli-bin=={runtime_version}"')
|
||||
replacement = "dependencies = [\n " + ",\n ".join(raw_items) + ",\n]"
|
||||
return pyproject_text[: match.start()] + replacement + pyproject_text[match.end() :]
|
||||
|
||||
|
||||
def stage_python_sdk_package(
|
||||
staging_dir: Path, sdk_version: str, runtime_version: str
|
||||
) -> Path:
|
||||
def stage_python_core_sdk_package(staging_dir: Path, sdk_version: str) -> Path:
|
||||
_copy_package_tree(sdk_root(), staging_dir)
|
||||
sdk_bin_dir = staging_dir / "src" / "codex_app_server" / "bin"
|
||||
if sdk_bin_dir.exists():
|
||||
@@ -135,11 +124,73 @@ def stage_python_sdk_package(
|
||||
pyproject_path = staging_dir / "pyproject.toml"
|
||||
pyproject_text = pyproject_path.read_text()
|
||||
pyproject_text = _rewrite_project_version(pyproject_text, sdk_version)
|
||||
pyproject_text = _rewrite_sdk_runtime_dependency(pyproject_text, runtime_version)
|
||||
pyproject_path.write_text(pyproject_text)
|
||||
return staging_dir
|
||||
|
||||
|
||||
def stage_python_sdk_package(
|
||||
staging_dir: Path, sdk_version: str, runtime_version: str
|
||||
) -> Path:
|
||||
if staging_dir.exists():
|
||||
if staging_dir.is_dir():
|
||||
shutil.rmtree(staging_dir)
|
||||
else:
|
||||
staging_dir.unlink()
|
||||
|
||||
package_dir = staging_dir / "src" / "codex_app_server_sdk_meta"
|
||||
package_dir.mkdir(parents=True, exist_ok=True)
|
||||
(package_dir / "__init__.py").write_text(
|
||||
'"""Bundled Codex app-server SDK package metadata."""\n'
|
||||
)
|
||||
|
||||
pyproject = dedent(
|
||||
f"""
|
||||
[build-system]
|
||||
requires = ["hatchling>=1.24.0"]
|
||||
build-backend = "hatchling.build"
|
||||
|
||||
[project]
|
||||
name = "{BUNDLED_SDK_PKG_NAME}"
|
||||
version = "{sdk_version}"
|
||||
description = "Bundled Python SDK for Codex app-server v2"
|
||||
readme = "README.md"
|
||||
requires-python = ">=3.10"
|
||||
license = {{ text = "Apache-2.0" }}
|
||||
authors = [{{ name = "OpenClaw Assistant" }}]
|
||||
dependencies = [
|
||||
"{CORE_SDK_PKG_NAME}=={sdk_version}",
|
||||
"{RUNTIME_PKG_NAME}=={runtime_version}",
|
||||
]
|
||||
|
||||
[project.urls]
|
||||
Homepage = "https://github.com/openai/codex"
|
||||
Repository = "https://github.com/openai/codex"
|
||||
Issues = "https://github.com/openai/codex/issues"
|
||||
|
||||
[tool.hatch.build.targets.wheel]
|
||||
packages = ["src/codex_app_server_sdk_meta"]
|
||||
|
||||
[tool.hatch.build.targets.sdist]
|
||||
include = ["src/codex_app_server_sdk_meta/**", "README.md", "pyproject.toml"]
|
||||
"""
|
||||
).lstrip()
|
||||
(staging_dir / "pyproject.toml").write_text(pyproject)
|
||||
(staging_dir / "README.md").write_text(
|
||||
"\n".join(
|
||||
[
|
||||
"# Codex App Server Python SDK",
|
||||
"",
|
||||
"Bundled metapackage for the Codex app-server Python SDK.",
|
||||
f"It depends on `{CORE_SDK_PKG_NAME}` and `{RUNTIME_PKG_NAME}`",
|
||||
"at the same version so a regular install includes both the SDK",
|
||||
"and the packaged Codex runtime binary.",
|
||||
"",
|
||||
]
|
||||
)
|
||||
)
|
||||
return staging_dir
|
||||
|
||||
|
||||
def stage_python_runtime_package(
|
||||
staging_dir: Path, runtime_version: str, binary_path: Path
|
||||
) -> Path:
|
||||
@@ -558,6 +609,7 @@ class PublicFieldSpec:
|
||||
@dataclass(frozen=True)
|
||||
class CliOps:
|
||||
generate_types: Callable[[], None]
|
||||
stage_python_core_sdk_package: Callable[[Path, str], Path]
|
||||
stage_python_sdk_package: Callable[[Path, str, str], Path]
|
||||
stage_python_runtime_package: Callable[[Path, str, Path], Path]
|
||||
current_sdk_version: Callable[[], str]
|
||||
@@ -916,23 +968,37 @@ def build_parser() -> argparse.ArgumentParser:
|
||||
"generate-types", help="Regenerate Python protocol-derived types"
|
||||
)
|
||||
|
||||
stage_sdk_core_parser = subparsers.add_parser(
|
||||
"stage-sdk-core",
|
||||
help="Stage a releasable core SDK package without a bundled runtime",
|
||||
)
|
||||
stage_sdk_core_parser.add_argument(
|
||||
"staging_dir",
|
||||
type=Path,
|
||||
help="Output directory for the staged core SDK package",
|
||||
)
|
||||
stage_sdk_core_parser.add_argument(
|
||||
"--sdk-version",
|
||||
help="Version to write into the staged core SDK package (defaults to sdk/python current version)",
|
||||
)
|
||||
|
||||
stage_sdk_parser = subparsers.add_parser(
|
||||
"stage-sdk",
|
||||
help="Stage a releasable SDK package pinned to a runtime version",
|
||||
help="Stage a releasable bundled SDK metapackage pinned to a runtime version",
|
||||
)
|
||||
stage_sdk_parser.add_argument(
|
||||
"staging_dir",
|
||||
type=Path,
|
||||
help="Output directory for the staged SDK package",
|
||||
help="Output directory for the staged bundled SDK package",
|
||||
)
|
||||
stage_sdk_parser.add_argument(
|
||||
"--runtime-version",
|
||||
required=True,
|
||||
help="Pinned codex-cli-bin version for the staged SDK package",
|
||||
help="Pinned codex-cli-bin version for the staged bundled SDK package",
|
||||
)
|
||||
stage_sdk_parser.add_argument(
|
||||
"--sdk-version",
|
||||
help="Version to write into the staged SDK package (defaults to sdk/python current version)",
|
||||
help="Version to write into the staged bundled SDK package (defaults to sdk/python current version)",
|
||||
)
|
||||
|
||||
stage_runtime_parser = subparsers.add_parser(
|
||||
@@ -964,6 +1030,7 @@ def parse_args(argv: Sequence[str] | None = None) -> argparse.Namespace:
|
||||
def default_cli_ops() -> CliOps:
|
||||
return CliOps(
|
||||
generate_types=generate_types,
|
||||
stage_python_core_sdk_package=stage_python_core_sdk_package,
|
||||
stage_python_sdk_package=stage_python_sdk_package,
|
||||
stage_python_runtime_package=stage_python_runtime_package,
|
||||
current_sdk_version=current_sdk_version,
|
||||
@@ -973,6 +1040,12 @@ def default_cli_ops() -> CliOps:
|
||||
def run_command(args: argparse.Namespace, ops: CliOps) -> None:
|
||||
if args.command == "generate-types":
|
||||
ops.generate_types()
|
||||
elif args.command == "stage-sdk-core":
|
||||
ops.generate_types()
|
||||
ops.stage_python_core_sdk_package(
|
||||
args.staging_dir,
|
||||
args.sdk_version or ops.current_sdk_version(),
|
||||
)
|
||||
elif args.command == "stage-sdk":
|
||||
ops.generate_types()
|
||||
ops.stage_python_sdk_package(
|
||||
|
||||
@@ -47,6 +47,7 @@ from .retry import retry_on_overload
|
||||
|
||||
ModelT = TypeVar("ModelT", bound=BaseModel)
|
||||
ApprovalHandler = Callable[[str, JsonObject | None], JsonObject]
|
||||
BUNDLED_SDK_PKG_NAME = "codex-app-server-sdk"
|
||||
RUNTIME_PKG_NAME = "codex-cli-bin"
|
||||
|
||||
|
||||
@@ -82,9 +83,9 @@ def _installed_codex_path() -> Path:
|
||||
from codex_cli_bin import bundled_codex_path
|
||||
except ImportError as exc:
|
||||
raise FileNotFoundError(
|
||||
"Unable to locate the pinned Codex runtime. Install the published SDK build "
|
||||
f"with its {RUNTIME_PKG_NAME} dependency, or set AppServerConfig.codex_bin "
|
||||
"explicitly."
|
||||
"Unable to locate the Codex runtime. Install the published "
|
||||
f"{BUNDLED_SDK_PKG_NAME} package, install {RUNTIME_PKG_NAME} "
|
||||
"alongside the core SDK, or set AppServerConfig.codex_bin explicitly."
|
||||
) from exc
|
||||
|
||||
return bundled_codex_path()
|
||||
|
||||
@@ -117,7 +117,7 @@ def test_python_codegen_schema_annotation_adds_stable_variant_titles() -> None:
|
||||
]
|
||||
assert ask_for_approval_titles == [
|
||||
"AskForApprovalValue",
|
||||
"RejectAskForApproval",
|
||||
"GranularAskForApproval",
|
||||
]
|
||||
|
||||
reasoning_summary_titles = [
|
||||
@@ -238,17 +238,31 @@ def test_stage_runtime_release_replaces_existing_staging_dir(tmp_path: Path) ->
|
||||
assert script.staged_runtime_bin_path(staged).read_text() == "fake codex\n"
|
||||
|
||||
|
||||
def test_stage_sdk_release_injects_exact_runtime_pin(tmp_path: Path) -> None:
|
||||
def test_stage_core_sdk_release_sets_version_without_runtime_pin(tmp_path: Path) -> None:
|
||||
script = _load_update_script_module()
|
||||
staged = script.stage_python_core_sdk_package(tmp_path / "sdk-core-stage", "0.2.1")
|
||||
|
||||
pyproject = (staged / "pyproject.toml").read_text()
|
||||
assert 'version = "0.2.1"' in pyproject
|
||||
assert 'name = "codex-app-server-sdk-core"' in pyproject
|
||||
assert "codex-cli-bin" not in pyproject
|
||||
assert not any((staged / "src" / "codex_app_server").glob("bin/**"))
|
||||
|
||||
|
||||
def test_stage_bundled_sdk_release_injects_exact_core_and_runtime_pins(
|
||||
tmp_path: Path,
|
||||
) -> None:
|
||||
script = _load_update_script_module()
|
||||
staged = script.stage_python_sdk_package(tmp_path / "sdk-stage", "0.2.1", "1.2.3")
|
||||
|
||||
pyproject = (staged / "pyproject.toml").read_text()
|
||||
assert 'name = "codex-app-server-sdk"' in pyproject
|
||||
assert 'version = "0.2.1"' in pyproject
|
||||
assert '"codex-app-server-sdk-core==0.2.1"' in pyproject
|
||||
assert '"codex-cli-bin==1.2.3"' in pyproject
|
||||
assert not any((staged / "src" / "codex_app_server").glob("bin/**"))
|
||||
|
||||
|
||||
def test_stage_sdk_release_replaces_existing_staging_dir(tmp_path: Path) -> None:
|
||||
def test_stage_bundled_sdk_release_replaces_existing_staging_dir(tmp_path: Path) -> None:
|
||||
script = _load_update_script_module()
|
||||
staging_dir = tmp_path / "sdk-stage"
|
||||
old_file = staging_dir / "stale.txt"
|
||||
@@ -261,6 +275,49 @@ def test_stage_sdk_release_replaces_existing_staging_dir(tmp_path: Path) -> None
|
||||
assert not old_file.exists()
|
||||
|
||||
|
||||
def test_stage_sdk_core_runs_type_generation_before_staging(tmp_path: Path) -> None:
|
||||
script = _load_update_script_module()
|
||||
calls: list[str] = []
|
||||
args = script.parse_args(
|
||||
[
|
||||
"stage-sdk-core",
|
||||
str(tmp_path / "sdk-core-stage"),
|
||||
]
|
||||
)
|
||||
|
||||
def fake_generate_types() -> None:
|
||||
calls.append("generate_types")
|
||||
|
||||
def fake_stage_core_sdk_package(_staging_dir: Path, _sdk_version: str) -> Path:
|
||||
calls.append("stage_sdk_core")
|
||||
return tmp_path / "sdk-core-stage"
|
||||
|
||||
def fake_stage_sdk_package(
|
||||
_staging_dir: Path, _sdk_version: str, _runtime_version: str
|
||||
) -> Path:
|
||||
raise AssertionError("bundled sdk staging should not run for stage-sdk-core")
|
||||
|
||||
def fake_stage_runtime_package(
|
||||
_staging_dir: Path, _runtime_version: str, _runtime_binary: Path
|
||||
) -> Path:
|
||||
raise AssertionError("runtime staging should not run for stage-sdk-core")
|
||||
|
||||
def fake_current_sdk_version() -> str:
|
||||
return "0.2.0"
|
||||
|
||||
ops = script.CliOps(
|
||||
generate_types=fake_generate_types,
|
||||
stage_python_core_sdk_package=fake_stage_core_sdk_package,
|
||||
stage_python_sdk_package=fake_stage_sdk_package,
|
||||
stage_python_runtime_package=fake_stage_runtime_package,
|
||||
current_sdk_version=fake_current_sdk_version,
|
||||
)
|
||||
|
||||
script.run_command(args, ops)
|
||||
|
||||
assert calls == ["generate_types", "stage_sdk_core"]
|
||||
|
||||
|
||||
def test_stage_sdk_runs_type_generation_before_staging(tmp_path: Path) -> None:
|
||||
script = _load_update_script_module()
|
||||
calls: list[str] = []
|
||||
@@ -276,6 +333,9 @@ def test_stage_sdk_runs_type_generation_before_staging(tmp_path: Path) -> None:
|
||||
def fake_generate_types() -> None:
|
||||
calls.append("generate_types")
|
||||
|
||||
def fake_stage_core_sdk_package(_staging_dir: Path, _sdk_version: str) -> Path:
|
||||
raise AssertionError("core sdk staging should not run for stage-sdk")
|
||||
|
||||
def fake_stage_sdk_package(
|
||||
_staging_dir: Path, _sdk_version: str, _runtime_version: str
|
||||
) -> Path:
|
||||
@@ -292,6 +352,7 @@ def test_stage_sdk_runs_type_generation_before_staging(tmp_path: Path) -> None:
|
||||
|
||||
ops = script.CliOps(
|
||||
generate_types=fake_generate_types,
|
||||
stage_python_core_sdk_package=fake_stage_core_sdk_package,
|
||||
stage_python_sdk_package=fake_stage_sdk_package,
|
||||
stage_python_runtime_package=fake_stage_runtime_package,
|
||||
current_sdk_version=fake_current_sdk_version,
|
||||
@@ -320,6 +381,9 @@ def test_stage_runtime_stages_binary_without_type_generation(tmp_path: Path) ->
|
||||
def fake_generate_types() -> None:
|
||||
calls.append("generate_types")
|
||||
|
||||
def fake_stage_core_sdk_package(_staging_dir: Path, _sdk_version: str) -> Path:
|
||||
raise AssertionError("core sdk staging should not run for stage-runtime")
|
||||
|
||||
def fake_stage_sdk_package(
|
||||
_staging_dir: Path, _sdk_version: str, _runtime_version: str
|
||||
) -> Path:
|
||||
@@ -336,6 +400,7 @@ def test_stage_runtime_stages_binary_without_type_generation(tmp_path: Path) ->
|
||||
|
||||
ops = script.CliOps(
|
||||
generate_types=fake_generate_types,
|
||||
stage_python_core_sdk_package=fake_stage_core_sdk_package,
|
||||
stage_python_sdk_package=fake_stage_sdk_package,
|
||||
stage_python_runtime_package=fake_stage_runtime_package,
|
||||
current_sdk_version=fake_current_sdk_version,
|
||||
|
||||
Reference in New Issue
Block a user