宜博 / lemongraph

Commit b7324fcb670831962235f8c03f079f3e7d71f24d

Authored by bboonnee 2026-07-31 11:22:18 +0800

U-006: c_work_graph_execute dynamic-workflow wrapper (skeleton)

write_workflow_script(): writes an orchestration script to
<target>/.claude/workflows/<name>.js -- versioned with the target project's
own git history, not graphlemon's (03-C §3.5 step 1).

execute_work_graph(): `claude -p 'Run /<name> with args {...}' --output-format
json --json-schema <schema> --permission-mode auto`, no `--bare` (A-1 spike:
skips credential discovery, breaks headless invocation), never
bypassPermissions. Returns structured_output. No budget-usage bookkeeping
or CLASSIFIER_BLOCKED handling yet -- U-012's job ("骨架" here).

Real-call coverage reuses spikes/a1-dynamic-workflows/workflow-under-test.js
verbatim (read from that file, not retyped) through the new wrapper --
"直接复用已验证的代码路径" per development-plan.md U-006, not just citing the
old spike as prior art. Confirmed passing for real: same result as the
original spike (sum=29 for [2,3,4]).

8 new mocked stdlib unittest cases + 1 gated real-agent test. 53 total
mocked tests across U-001-006, all passing; both gated real-agent tests
(U-004/U-005's planner call, U-006's reused spike workflow) re-verified
manually this session.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

Inline Side-by-side

Showing 3 changed files with 205 additions and 0 deletions

src/test_work_graph_execute.py 0 → 100644

View file @b7324fc

		1	+"""U-006 tests: c_work_graph_execute's dynamic-workflow wrapper, mocked at the
		2	+subprocess boundary (CLAUDE.md Testing constraint). Real end-to-end coverage
		3	+is in src/test_work_graph_execute_e2e.py (RUN_REAL_AGENT_E2E=1), reusing the
		4	+exact spike script the A-1 assumption was validated against.
		5	+
		6	+Run: python3 -m unittest src.test_work_graph_execute -v
		7	+"""
		8	+
		9	+import json
		10	+import os
		11	+import subprocess
		12	+import tempfile
		13	+import unittest
		14	+from unittest.mock import patch
		15	+
		16	+from src import work_graph_execute
		17	+
		18	+
		19	+def _completed_process(returncode=0, stdout_obj=None, stderr=""):
		20	+ stdout = json.dumps(stdout_obj) if stdout_obj is not None else ""
		21	+ return subprocess.CompletedProcess(args=["claude"], returncode=returncode, stdout=stdout, stderr=stderr)
		22	+
		23	+
		24	+class WriteWorkflowScriptTest(unittest.TestCase):
		25	+ def test_writes_into_target_claude_workflows_not_graphlemon_repo(self):
		26	+ with tempfile.TemporaryDirectory() as tmp:
		27	+ path = work_graph_execute.write_workflow_script(tmp, "run-123", "export const meta = {}")
		28	+ self.assertEqual(path, os.path.join(tmp, ".claude", "workflows", "run-123.js"))
		29	+ with open(path, encoding="utf-8") as f:
		30	+ self.assertEqual(f.read(), "export const meta = {}")
		31	+
		32	+
		33	+class ExecuteWorkGraphTest(unittest.TestCase):
		34	+ SCHEMA = {"type": "object", "required": ["sum"], "properties": {"sum": {"type": "number"}}}
		35	+
		36	+ def test_happy_path_returns_structured_output(self):
		37	+ with patch("src.work_graph_execute.subprocess.run") as run:
		38	+ run.return_value = _completed_process(
		39	+ stdout_obj={"is_error": False, "structured_output": {"sum": 29}}
		40	+ )
		41	+ output = work_graph_execute.execute_work_graph(
		42	+ "/tmp/some-project", "run-123", {"numbers": [2, 3, 4]}, self.SCHEMA
		43	+ )
		44	+ self.assertEqual(output, {"sum": 29})
		45	+
		46	+ def test_invokes_claude_with_expected_args_and_cwd(self):
		47	+ with patch("src.work_graph_execute.subprocess.run") as run:
		48	+ run.return_value = _completed_process(stdout_obj={"is_error": False, "structured_output": {"sum": 1}})
		49	+ work_graph_execute.execute_work_graph("/tmp/some-project", "run-123", {"n": 1}, self.SCHEMA)
		50	+
		51	+ args, kwargs = run.call_args
		52	+ argv = args[0]
		53	+ self.assertEqual(argv[0], "claude")
		54	+ self.assertIn("-p", argv)
		55	+ self.assertIn("Run /run-123 with args {\"n\": 1}", argv)
		56	+ self.assertIn("--json-schema", argv)
		57	+ self.assertIn(json.dumps(self.SCHEMA), argv)
		58	+ self.assertIn("--permission-mode", argv)
		59	+ self.assertIn("auto", argv)
		60	+ self.assertNotIn("--bare", argv)
		61	+ self.assertNotIn("--dangerously-skip-permissions", argv)
		62	+ self.assertNotIn("bypassPermissions", argv)
		63	+ self.assertEqual(kwargs["cwd"], "/tmp/some-project")
		64	+ self.assertEqual(kwargs["stdin"], subprocess.DEVNULL)
		65	+
		66	+ def test_nonzero_exit_raises(self):
		67	+ with patch("src.work_graph_execute.subprocess.run") as run:
		68	+ run.return_value = _completed_process(returncode=1, stdout_obj=None, stderr="boom")
		69	+ with self.assertRaises(work_graph_execute.WorkGraphExecuteError):
		70	+ work_graph_execute.execute_work_graph("/tmp/some-project", "run-123", {}, self.SCHEMA)
		71	+
		72	+ def test_is_error_true_raises(self):
		73	+ with patch("src.work_graph_execute.subprocess.run") as run:
		74	+ run.return_value = _completed_process(stdout_obj={"is_error": True})
		75	+ with self.assertRaises(work_graph_execute.WorkGraphExecuteError):
		76	+ work_graph_execute.execute_work_graph("/tmp/some-project", "run-123", {}, self.SCHEMA)
		77	+
		78	+ def test_missing_structured_output_raises(self):
		79	+ with patch("src.work_graph_execute.subprocess.run") as run:
		80	+ run.return_value = _completed_process(stdout_obj={"is_error": False})
		81	+ with self.assertRaises(work_graph_execute.WorkGraphExecuteError):
		82	+ work_graph_execute.execute_work_graph("/tmp/some-project", "run-123", {}, self.SCHEMA)
		83	+
		84	+
		85	+if __name__ == "__main__":
		86	+ unittest.main()

src/test_work_graph_execute_e2e.py 0 → 100644

View file @b7324fc

		1	+"""U-006's real-call coverage: reuses the exact script the A-1 spike validated
		2	+(spikes/a1-dynamic-workflows/workflow-under-test.js) through the new
		3	+write_workflow_script/execute_work_graph wrapper, not a re-typed copy --
		4	+"直接复用...已验证的代码路径" per development-plan.md U-006. Gated behind
		5	+RUN_REAL_AGENT_E2E=1 (CLAUDE.md Testing constraint).
		6	+
		7	+Run: RUN_REAL_AGENT_E2E=1 python3 -m unittest src.test_work_graph_execute_e2e -v
		8	+"""
		9	+
		10	+import os
		11	+import subprocess
		12	+import tempfile
		13	+import unittest
		14	+
		15	+from src import work_graph_execute
		16	+from src.test_support import skip_unless_env
		17	+
		18	+_SPIKE_SCRIPT = os.path.join(
		19	+ os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
		20	+ "spikes", "a1-dynamic-workflows", "workflow-under-test.js",
		21	+)
		22	+_SCHEMA = {
		23	+ "type": "object",
		24	+ "required": ["input_numbers", "squares", "sum"],
		25	+ "properties": {
		26	+ "input_numbers": {"type": "array"},
		27	+ "squares": {"type": "array"},
		28	+ "sum": {"type": "number"},
		29	+ },
		30	+}
		31	+
		32	+
		33	+@skip_unless_env("RUN_REAL_AGENT_E2E")
		34	+class RealWorkGraphExecuteTest(unittest.TestCase):
		35	+ def test_reused_a1_spike_script_runs_through_the_new_wrapper(self):
		36	+ with tempfile.TemporaryDirectory() as tmp:
		37	+ subprocess.run(["git", "init"], cwd=tmp, check=True, capture_output=True)
		38	+ with open(_SPIKE_SCRIPT, encoding="utf-8") as f:
		39	+ script_content = f.read()
		40	+
		41	+ work_graph_execute.write_workflow_script(tmp, "graphlemon-a1-spike", script_content)
		42	+ output = work_graph_execute.execute_work_graph(
		43	+ tmp, "graphlemon-a1-spike", {"numbers": [2, 3, 4]}, _SCHEMA, timeout=120
		44	+ )
		45	+
		46	+ self.assertEqual(output["input_numbers"], [2, 3, 4])
		47	+ self.assertEqual(output["squares"], [4, 9, 16])
		48	+ self.assertEqual(output["sum"], 29)
		49	+
		50	+
		51	+if __name__ == "__main__":
		52	+ unittest.main()

src/work_graph_execute.py 0 → 100644

View file @b7324fc

		1	+"""c_work_graph_execute (03-C §3.5): dynamic-workflow invocation skeleton.
		2	+
		3	+U-006 scope: write an orchestration script into the target project's
		4	+`.claude/workflows/<name>.js` and invoke it via `claude -p 'Run /<name> with
		5	+args {...}'`, reusing the exact code path the A-1 spike validated
		6	+(spikes/a1-dynamic-workflows/, confirmed 2026-07-29). No budget-usage
		7	+bookkeeping or CLASSIFIER_BLOCKED handling yet -- that's U-012's job
		8	+("骨架" here, per development-plan.md).
		9	+"""
		10	+
		11	+import json
		12	+import os
		13	+import subprocess
		14	+
		15	+
		16	+class WorkGraphExecuteError(Exception):
		17	+ """Raised when the claude subprocess exits non-zero, reports is_error, or
		18	+ produces no structured_output."""
		19	+
		20	+
		21	+def write_workflow_script(target_folder_path, name, script_content):
		22	+ """Writes `<target>/.claude/workflows/<name>.js` -- versioned with the
		23	+ target project's own git history (03-C §3.5 step 1), not graphlemon's."""
		24	+ workflows_dir = os.path.join(target_folder_path, ".claude", "workflows")
		25	+ os.makedirs(workflows_dir, exist_ok=True)
		26	+ path = os.path.join(workflows_dir, name + ".js")
		27	+ with open(path, "w", encoding="utf-8") as f:
		28	+ f.write(script_content)
		29	+ return path
		30	+
		31	+
		32	+def execute_work_graph(target_folder_path, name, args, output_schema, timeout=600):
		33	+ """Runs the workflow already written to `.claude/workflows/<name>.js`
		34	+ (write_workflow_script) via `claude -p 'Run /<name> with args {...}'`.
		35	+
		36	+ No `--bare` (A-1 spike: skips credential/login discovery, unrelated to
		37	+ workflow presence). `--permission-mode auto`, never `bypassPermissions`/
		38	+ `--dangerously-skip-permissions` (05 §5.3.1). Returns the parsed
		39	+ `structured_output` dict.
		40	+ """
		41	+ prompt = "Run /{} with args {}".format(name, json.dumps(args))
		42	+ result = subprocess.run(
		43	+ [
		44	+ "claude", "-p", prompt,
		45	+ "--output-format", "json",
		46	+ "--json-schema", json.dumps(output_schema),
		47	+ "--permission-mode", "auto",
		48	+ ],
		49	+ cwd=target_folder_path,
		50	+ stdin=subprocess.DEVNULL,
		51	+ capture_output=True,
		52	+ text=True,
		53	+ timeout=timeout,
		54	+ )
		55	+ if result.returncode != 0:
		56	+ raise WorkGraphExecuteError(
		57	+ "workflow {} exited {}: {}".format(name, result.returncode, result.stderr)
		58	+ )
		59	+
		60	+ payload = json.loads(result.stdout)
		61	+ if payload.get("is_error"):
		62	+ raise WorkGraphExecuteError("workflow {} reported is_error: {}".format(name, payload))
		63	+
		64	+ structured_output = payload.get("structured_output")
		65	+ if structured_output is None:
		66	+ raise WorkGraphExecuteError("workflow {} produced no structured_output: {}".format(name, payload))
		67	+ return structured_output