github/esphome
2
0
Fork 0
mirror of https://github.com/esphome/esphome.git synced 2026-06-24 12:33:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

[ci] Split integration tests into 3 buckets when count is more than 10 (#16152)

Browse Source
This commit is contained in:
J. Nick Koston
2026-04-30 19:10:53 -05:00
committed by GitHub
parent 61261b4a59
commit 2fbfb4c385
3 changed files with 156 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

66
script/determine-jobs.py
View File

@@ -6,8 +6,7 @@ what files have changed. It outputs JSON with the following structure:
{
"integration_tests": true/false,
"integration_tests_run_all": true/false,
"integration_test_files": ["tests/integration/test_foo.py", ...],
"integration_test_buckets": [{"name": "1/3", "tests": ["tests/integration/test_foo.py", ...]}, ...],
"clang_tidy": true/false,
"clang_format": true/false,
"python_linters": true/false,
@@ -81,6 +80,62 @@ CLANG_TIDY_SPLIT_THRESHOLD = 65
# Isolated components count as 10x, groupable components count as 1x
COMPONENT_TEST_BATCH_SIZE = 40
# Integration test bucketing: when more than the threshold tests are scheduled,
# fan out across this many parallel jobs. Below the threshold, a single job runs.
INTEGRATION_TESTS_SPLIT_THRESHOLD = 10
INTEGRATION_TESTS_SPLIT_BUCKETS = 3
def _split_list(items: list[str], n: int) -> list[list[str]]:
"""Split a list into n roughly-equal contiguous parts (matches script/clang-tidy)."""
k, m = divmod(len(items), n)
return [items[i * k + min(i, m) : (i + 1) * k + min(i + 1, m)] for i in range(n)]
def _all_integration_test_files() -> list[str]:
"""Return all integration test file paths, sorted, relative to repo root."""
return sorted(
str(p.relative_to(root_path))
for p in (Path(root_path) / "tests" / "integration").glob("test_*.py")
)
def _compute_integration_test_buckets(
integration_run_all: bool,
integration_test_files: list[str],
) -> tuple[bool, list[dict[str, Any]]]:
"""Compute (run_integration, buckets) from the determine_integration_tests result.
Pure function for unit testing — no I/O beyond `_all_integration_test_files`
when `integration_run_all` is set.
`buckets` is a list of `{name, tests}` dicts where `tests` is a JSON-friendly
list of file paths so the workflow can build a bash array via jq, avoiding
shell word-splitting / glob hazards.
"""
if integration_run_all:
files = _all_integration_test_files()
else:
files = sorted(integration_test_files)
# Empty list (e.g. run_all expansion with no files on disk) would otherwise
# cause the workflow to invoke pytest with no path argument and collect
# tests outside tests/integration/. Suppress the run instead.
if not files:
return False, []
if len(files) > INTEGRATION_TESTS_SPLIT_THRESHOLD:
parts = [
part for part in _split_list(files, INTEGRATION_TESTS_SPLIT_BUCKETS) if part
]
buckets = [
{"name": f"{i + 1}/{len(parts)}", "tests": part}
for i, part in enumerate(parts)
]
else:
buckets = [{"name": "1/1", "tests": files}]
return True, buckets
class Platform(StrEnum):
"""Platform identifiers for memory impact analysis."""
@@ -812,7 +867,9 @@ def main() -> None:
integration_run_all, integration_test_files = determine_integration_tests(
args.branch
)
run_integration = integration_run_all or bool(integration_test_files)
run_integration, integration_test_buckets = _compute_integration_test_buckets(
integration_run_all, integration_test_files
)
run_clang_tidy = should_run_clang_tidy(args.branch)
run_clang_format = should_run_clang_format(args.branch)
run_python_linters = should_run_python_linters(args.branch)
@@ -944,8 +1001,7 @@ def main() -> None:
output: dict[str, Any] = {
"integration_tests": run_integration,
"integration_tests_run_all": integration_run_all,
"integration_test_files": integration_test_files,
"integration_test_buckets": integration_test_buckets,
"clang_tidy": run_clang_tidy,
"clang_tidy_mode": clang_tidy_mode,
"clang_format": run_clang_format,