feat: consolidate legacy workflows into gyxx-flow

2026-07-28 14:51:15 +08:00
commit c23b62a8c8
374 changed files with 132990 additions and 0 deletions
@@ -0,0 +1,252 @@
+from __future__ import annotations
+
+import json
+import sys
+from pathlib import Path
+
+import pytest
+
+from gyxx_flow.adapters import (
+    BrowserCookieStore,
+    RuntimeIntegrationCatalog,
+    RuntimeIntegrationError,
+    RuntimeServicePolicy,
+)
+from gyxx_flow.adapters.bootstrap import _rewrite_browser_arguments
+from gyxx_flow.adapters.native import ModuleCommandAdapter, ModuleSourceRoots
+from gyxx_flow.catalog import WorkflowEntry
+from gyxx_flow.core.context import RunContext
+from gyxx_flow.script_catalog import ScriptCatalog, ScriptEntry
+
+
+def _catalog(tmp_path: Path) -> ScriptCatalog:
+    root = tmp_path / "runtime"
+    root.mkdir()
+    first = root / "first.py"
+    second = root / "nested" / "second.py"
+    second.parent.mkdir()
+    first.write_text("print('first')\n", encoding="utf-8")
+    second.write_text("print('second')\n", encoding="utf-8")
+    return ScriptCatalog(
+        (
+            ScriptEntry("demo:first.py", "demo", "first.py", "python", first),
+            ScriptEntry(
+                "demo:nested/second.py",
+                "demo",
+                "nested/second.py",
+                "python",
+                second,
+            ),
+        )
+    )
+
+
+def _write_config(path: Path) -> None:
+    path.write_text(
+        json.dumps(
+            {
+                "schema_version": 1,
+                "cdp_host": "127.0.0.1",
+                "scripts": {
+                    "demo:first.py": 22001,
+                    "demo:nested/second.py": 22002,
+                },
+                "services": {
+                    "feishu": "legacy",
+                    "postgres": "cloud",
+                    "hermes": "local",
+                    "hermes_url": "http://127.0.0.1:8642/v1/chat/completions",
+                },
+            }
+        ),
+        encoding="utf-8",
+    )
+
+
+def test_project_registry_covers_every_script_with_unique_stable_port() -> None:
+    project_root = Path(__file__).resolve().parents[1]
+    scripts = ScriptCatalog.discover_default()
+    first = RuntimeIntegrationCatalog.load(
+        project_root / "config" / "runtime-bindings.json",
+        scripts=scripts,
+        data_root=project_root / "var",
+    )
+    second = RuntimeIntegrationCatalog.load(
+        project_root / "config" / "runtime-bindings.json",
+        scripts=scripts,
+        data_root=project_root / "var",
+    )
+
+    assert len(scripts.script_ids) == 131
+    bindings = [first.binding_for(script_id) for script_id in scripts.script_ids]
+    assert len({binding.cdp_port for binding in bindings}) == 131
+    assert all(22000 <= binding.cdp_port <= 22999 for binding in bindings)
+    assert all(binding.cdp_url.startswith("http://127.0.0.1:") for binding in bindings)
+    assert bindings == [second.binding_for(item.script_id) for item in bindings]
+
+
+def test_bindings_isolate_browser_state_and_relocate_with_data_root(tmp_path: Path) -> None:
+    scripts = _catalog(tmp_path)
+    config = tmp_path / "bindings.json"
+    _write_config(config)
+    data_root = tmp_path / "portable-data"
+    catalog = RuntimeIntegrationCatalog.load(config, scripts=scripts, data_root=data_root)
+
+    first = catalog.binding_for("demo:first.py")
+    second = catalog.binding_for("demo:nested/second.py")
+    assert first.profile_dir != second.profile_dir
+    assert first.cookie_file != second.cookie_file
+    assert first.storage_state_file != second.storage_state_file
+    for path in (
+        first.profile_dir,
+        first.cookie_file,
+        first.storage_state_file,
+        second.profile_dir,
+    ):
+        assert path.is_relative_to(data_root.resolve())
+    assert not first.profile_dir.exists()
+
+
+def test_parent_and_nested_child_receive_different_runtime_environment(tmp_path: Path) -> None:
+    scripts = _catalog(tmp_path)
+    config = tmp_path / "bindings.json"
+    _write_config(config)
+    catalog = RuntimeIntegrationCatalog.load(config, scripts=scripts, data_root=tmp_path / "data")
+
+    parent = catalog.environment_for("demo:first.py", {"KEEP": "yes"})
+    child = catalog.environment_for("demo:nested/second.py", parent)
+
+    assert child["KEEP"] == "yes"
+    assert parent["GYXX_SCRIPT_ID"] == "demo:first.py"
+    assert child["GYXX_SCRIPT_ID"] == "demo:nested/second.py"
+    assert parent["GYXX_BROWSER_CDP_PORT"] != child["GYXX_BROWSER_CDP_PORT"]
+    assert parent["GYXX_BROWSER_PROFILE_DIR"] != child["GYXX_BROWSER_PROFILE_DIR"]
+    assert parent["GYXX_BROWSER_COOKIE_FILE"] != child["GYXX_BROWSER_COOKIE_FILE"]
+
+
+def test_module_command_adapter_injects_entry_script_binding(tmp_path: Path) -> None:
+    scripts = _catalog(tmp_path)
+    config = tmp_path / "bindings.json"
+    _write_config(config)
+    catalog = RuntimeIntegrationCatalog.load(config, scripts=scripts, data_root=tmp_path / "data")
+    runtime_root = tmp_path / "runtime"
+    adapter = ModuleCommandAdapter(
+        ModuleSourceRoots({"demo": runtime_root}),
+        base_env={"KEEP": "yes"},
+        project_root=tmp_path,
+        data_root=tmp_path / "data",
+        integration_catalog=catalog,
+    )
+    entry = WorkflowEntry("demo.workflow", "demo", "manual", "first.py")
+    context = RunContext.create("demo.workflow", "2026-07-27")
+
+    command = adapter.build(entry, context=context)
+
+    assert command.env["KEEP"] == "yes"
+    assert command.env["GYXX_SCRIPT_ID"] == "demo:first.py"
+    assert command.env["GYXX_BROWSER_CDP_PORT"] == "22001"
+
+
+def test_cookie_store_round_trip_and_storage_state_are_atomic(tmp_path: Path) -> None:
+    store = BrowserCookieStore(
+        cookie_file=tmp_path / "cookies" / "cookies.json",
+        storage_state_file=tmp_path / "cookies" / "storage_state.json",
+    )
+    cookies = [{"name": "session", "value": "secret-sentinel", "domain": ".example.test", "path": "/"}]
+    state = {"cookies": cookies, "origins": []}
+
+    assert store.load_cookies() == []
+    assert store.load_storage_state() is None
+    store.save_cookies(cookies)
+    store.save_storage_state(state)
+
+    assert store.load_cookies() == cookies
+    assert store.load_storage_state() == state
+    assert not list((tmp_path / "cookies").glob("*.tmp"))
+    assert "secret-sentinel" not in repr(store)
+
+
+def test_service_policy_preserves_original_feishu_and_cloud_db_and_local_hermes() -> None:
+    policy = RuntimeServicePolicy(
+        hermes_url="http://127.0.0.1:8642/v1/chat/completions"
+    )
+    original = {
+        "LARK_PROFILE": "original-profile",
+        "PG_HOST": "cloud-db.example.test",
+        "PG_PASSWORD": "placeholder",
+        "DB_HOST": "",
+        "CUSTOM": "keep",
+    }
+
+    result = policy.apply(original)
+
+    assert result["LARK_PROFILE"] == "original-profile"
+    assert result["PG_HOST"] == "cloud-db.example.test"
+    assert result["PG_PASSWORD"] == "placeholder"
+    assert result["DB_HOST"] == "cloud-db.example.test"
+    assert result["AUTOFLOW_PG_HOST"] == "cloud-db.example.test"
+    assert result["GYXX_FEISHU_MODE"] == "legacy"
+    assert result["GYXX_POSTGRES_MODE"] == "cloud"
+    assert result["GYXX_HERMES_MODE"] == "local"
+    assert result["HERMES_ANALYZER_URL"].startswith("http://127.0.0.1:")
+    assert "secret-sentinel" not in repr(policy)
+
+
+@pytest.mark.parametrize(
+    ("environment", "message"),
+    [
+        ({"PG_HOST": "127.0.0.1"}, "cloud"),
+        ({"DB_HOST": "localhost"}, "cloud"),
+        ({"AUTOFLOW_PG_HOST": "::1"}, "cloud"),
+        ({"HERMES_ANALYZER_URL": "https://remote.example.test/v1"}, "local"),
+        ({"ANALYZER_API_SERVER_URL": "http://10.0.0.8:8642/v1"}, "local"),
+    ],
+)
+def test_service_policy_rejects_local_database_or_remote_hermes(
+    environment: dict[str, str], message: str
+) -> None:
+    with pytest.raises(RuntimeIntegrationError, match=message):
+        RuntimeServicePolicy().apply(environment)
+
+
+def test_registry_rejects_missing_duplicate_or_unknown_script_allocations(tmp_path: Path) -> None:
+    scripts = _catalog(tmp_path)
+    config = tmp_path / "bindings.json"
+    _write_config(config)
+    payload = json.loads(config.read_text(encoding="utf-8"))
+    payload["scripts"]["demo:nested/second.py"] = 22001
+    config.write_text(json.dumps(payload), encoding="utf-8")
+    with pytest.raises(RuntimeIntegrationError, match="unique"):
+        RuntimeIntegrationCatalog.load(config, scripts=scripts, data_root=tmp_path / "data")
+
+    payload["scripts"].pop("demo:nested/second.py")
+    payload["scripts"]["demo:unknown.py"] = 22003
+    config.write_text(json.dumps(payload), encoding="utf-8")
+    with pytest.raises(RuntimeIntegrationError, match="coverage"):
+        RuntimeIntegrationCatalog.load(config, scripts=scripts, data_root=tmp_path / "data")
+
+
+def test_nested_script_browser_arguments_are_rebound(monkeypatch, tmp_path: Path) -> None:
+    scripts = _catalog(tmp_path)
+    config = tmp_path / "bindings.json"
+    _write_config(config)
+    binding = RuntimeIntegrationCatalog.load(
+        config, scripts=scripts, data_root=tmp_path / "data"
+    ).binding_for("demo:nested/second.py")
+    monkeypatch.setattr(
+        "sys.argv",
+        [
+            "second.py",
+            "--user-data-dir",
+            "parent-profile",
+            "--cdp-url=http://127.0.0.1:22001",
+            "--cdp-port",
+            "22001",
+        ],
+    )
+
+    _rewrite_browser_arguments(binding)
+
+    assert sys.argv[2] == str(binding.profile_dir)
+    assert sys.argv[3] == f"--cdp-url={binding.cdp_url}"
+    assert sys.argv[5] == str(binding.cdp_port)