Switchyard lets LLM applications route traffic across models and providers while preserving native OpenAI and Anthropic API compatibility - enabling flexible model selection, benchmarking, and cost/performance optimization.
Rust
# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
"""Tests for miner.py — PR-first candidate mining."""
from __future__ import annotations
import json
from unittest.mock import MagicMock, patch
import pytest
from craft_taskgen.miner import (
Candidate,
analyze_pr,
get_diff_stats,
get_prs,
mine_repo,
)
# ---------------------------------------------------------------------------
# get_diff_stats
# ---------------------------------------------------------------------------
def test_get_diff_stats_uses_two_sha_diff(tmp_path):
"""get_diff_stats must find merge-base then diff against it (not raw base_sha)."""
with patch("craft_taskgen.miner.git") as mock_git:
mock_git.side_effect = [
"mergebase0\n", # first call: merge-base base000 head111
"5\t2\tsrc/foo.py\n3\t1\ttests/test_foo.py\n", # second call: diff --numstat mergebase0 head111
]
result = get_diff_stats(tmp_path, "base000", "head111")
calls = mock_git.call_args_list
assert calls[0][0] == (tmp_path, "merge-base", "base000", "head111")
assert calls[1][0] == (tmp_path, "diff", "--numstat", "mergebase0", "head111")
assert len(result["files"]) == 2
assert result["files"][0] == {"path": "src/foo.py", "added": 5, "deleted": 2}
# ---------------------------------------------------------------------------
# Candidate.base_sha
# ---------------------------------------------------------------------------
def test_candidate_requires_base_sha():
"""Candidate must not accept construction without base_sha."""
with pytest.raises(TypeError):
Candidate(sha="abc123", subject="fix: thing", author="alice", date="2025-01-01T00:00:00Z")
def test_candidate_with_base_sha():
"""Candidate constructs normally when base_sha and merge_base_sha are provided."""
c = Candidate(
sha="abc123",
base_sha="def456",
merge_base_sha="ghi789",
subject="fix: thing",
author="alice",
date="2025-01-01T00:00:00Z",
)
assert c.base_sha == "def456"
assert c.merge_base_sha == "ghi789"
# ---------------------------------------------------------------------------
# analyze_pr
# ---------------------------------------------------------------------------
def test_analyze_pr_populates_both_shas(tmp_path):
"""analyze_pr must set sha, base_sha, and merge_base_sha from the PR dict and diff stats."""
pr = {
"sha": "merge111",
"base_sha": "base000",
"subject": "feat: add widget",
"author": "alice",
"date": "2025-06-01T10:00:00Z",
"pr_number": 42,
}
with patch("craft_taskgen.miner.get_diff_stats") as mock_stats:
mock_stats.return_value = {
"files": [
{"path": "src/widget.py", "added": 50, "deleted": 5},
{"path": "src/core.py", "added": 30, "deleted": 2},
{"path": "tests/test_widget.py", "added": 40, "deleted": 0},
],
"merge_base_sha": "mergebase000",
}
candidate = analyze_pr(tmp_path, pr)
mock_stats.assert_called_once_with(tmp_path, "base000", "merge111")
assert candidate.sha == "merge111"
assert candidate.base_sha == "base000"
assert candidate.merge_base_sha == "mergebase000"
assert candidate.has_test_patch is True
# ---------------------------------------------------------------------------
# get_prs
# ---------------------------------------------------------------------------
_FAKE_PR_PAGE = [
{
"number": 42,
"title": "feat: add widget",
"user": {"login": "alice"},
"merged_at": "2025-06-01T10:00:00Z",
"merge_commit_sha": "merge111aaa",
"base": {"sha": "base000bbb"},
},
{
"number": 41,
"title": "fix: rename typo",
"user": {"login": "bob"},
"merged_at": "2025-05-15T08:00:00Z",
"merge_commit_sha": "merge222ccc",
"base": {"sha": "base333ddd"},
},
{
"number": 40,
"title": "fix: something",
"user": {"login": "bob"},
"merged_at": None, # not merged — should be skipped
"merge_commit_sha": None,
"base": {"sha": "base444eee"},
},
]
_OLD_PAGE = [
{
"number": 10,
"title": "old fix",
"user": {"login": "alice"},
"merged_at": "2024-12-01T00:00:00Z",
"merge_commit_sha": "oldsha",
"base": {"sha": "oldbase"},
}
]
def _gh_responses(*pages: list) -> list[MagicMock]:
"""Build subprocess.run mocks for paged gh api calls, ending with empty page."""
results = []
for page in pages:
m = MagicMock()
m.returncode = 0
m.stdout = json.dumps(page)
results.append(m)
empty = MagicMock()
empty.returncode = 0
empty.stdout = json.dumps([])
results.append(empty)
return results
def test_get_prs_returns_merged_only():
"""get_prs skips PRs where merged_at is None."""
with patch("subprocess.run", side_effect=_gh_responses(_FAKE_PR_PAGE)):
prs = get_prs("owner/repo")
assert len(prs) == 2 # PR 40 (unmerged) excluded
assert prs[0]["pr_number"] == 42
assert prs[0]["sha"] == "merge111aaa"
assert prs[0]["base_sha"] == "base000bbb"
assert prs[0]["subject"] == "feat: add widget"
assert prs[0]["author"] == "alice"
def test_get_prs_raises_on_gh_failure():
"""get_prs raises RuntimeError if gh api returns non-zero."""
mock_result = MagicMock()
mock_result.returncode = 1
mock_result.stderr = "authentication required"
with patch("subprocess.run", return_value=mock_result):
with pytest.raises(RuntimeError, match="gh api failed"):
get_prs("owner/repo")
def test_get_prs_stops_paging_at_after_date():
"""get_prs stops fetching pages when first merged PR on a page predates --after."""
with patch("subprocess.run", side_effect=_gh_responses(_FAKE_PR_PAGE, _OLD_PAGE)) as mock_run:
prs = get_prs("owner/repo", after="2025-01-01")
# Only 2 merged PRs from page 1 pass the date filter; page 2 (2024) triggers stop
assert len(prs) == 2
assert all(p["date"] >= "2025-01-01" for p in prs)
# Exactly 2 subprocess calls: page 1 + page 2 (which triggered stop; no page 3)
assert mock_run.call_count == 2
# ---------------------------------------------------------------------------
# mine_repo
# ---------------------------------------------------------------------------
def test_mine_repo_returns_candidates_and_pr_count(tmp_path):
"""mine_repo returns (candidates, n_prs_scanned). Candidates have base_sha and merge_base_sha."""
fake_prs = [
{
"sha": "merge111",
"base_sha": "base000",
"subject": "feat: add lots of files",
"author": "alice",
"date": "2025-06-01T10:00:00Z",
"pr_number": 1,
},
{
"sha": "merge222",
"base_sha": "base111",
"subject": "docs: update readme", # will score 0 — no tests
"author": "bob",
"date": "2025-05-01T10:00:00Z",
"pr_number": 2,
},
]
good_stats = {
"files": [
{"path": "src/a.py", "added": 60, "deleted": 0},
{"path": "src/b.py", "added": 40, "deleted": 0},
{"path": "src/c.py", "added": 30, "deleted": 0},
{"path": "tests/test_a.py", "added": 50, "deleted": 0},
],
"merge_base_sha": "mergebase000",
}
no_test_stats = {
"files": [{"path": "README.md", "added": 5, "deleted": 0}],
"merge_base_sha": "mergebase111",
}
def fake_diff_stats(_repo, base_sha, _sha):
return good_stats if base_sha == "base000" else no_test_stats
with (
patch("craft_taskgen.miner.get_prs", return_value=fake_prs),
patch("craft_taskgen.miner.get_diff_stats", side_effect=fake_diff_stats),
):
results, n_prs = mine_repo(tmp_path, "owner/repo", after=None, top_n=5)
assert n_prs == 2 # total PRs fetched
assert len(results) == 1 # only PR 1 scored > 0
assert results[0]["sha"] == "merge111"
assert results[0]["base_sha"] == "base000"
assert results[0]["merge_base_sha"] == "mergebase000"