Files
gyxx-flow/src/gyxx_flow/acceptance.py
T

302 lines
10 KiB
Python

"""Machine-readable acceptance evidence derived from repository state."""
from __future__ import annotations
import hashlib
import json
import re
from dataclasses import asdict, dataclass
from pathlib import Path
from gyxx_flow.catalog import CatalogError, WorkflowCatalog
from gyxx_flow.core.config import Settings
from gyxx_flow.script_catalog import ScriptCatalog, ScriptCatalogError
from gyxx_flow.security import scan_repository
_CHECKLIST = re.compile(r"^\s*-\s*\[([ xX])\]\s+(P\d+\.\d+)\b", re.MULTILINE)
@dataclass(frozen=True, slots=True)
class PlanItem:
item_id: str
completed: bool
@dataclass(frozen=True, slots=True)
class AcceptanceReport:
items: tuple[PlanItem, ...]
checks: dict[str, bool]
@property
def is_complete(self) -> bool:
return bool(self.items) and all(item.completed for item in self.items) and all(
self.checks.values()
)
def as_dict(self) -> dict[str, object]:
completed = sum(item.completed for item in self.items)
return {
"schema_version": 1,
"is_complete": self.is_complete,
"summary": {
"total": len(self.items),
"completed": completed,
"pending": len(self.items) - completed,
},
"items": [asdict(item) for item in self.items],
"checks": dict(sorted(self.checks.items())),
}
def parse_plan_checklist(path: Path) -> tuple[PlanItem, ...]:
text = Path(path).read_text(encoding="utf-8")
items = tuple(
PlanItem(item_id=match.group(2), completed=match.group(1).casefold() == "x")
for match in _CHECKLIST.finditer(text)
)
if len({item.item_id for item in items}) != len(items):
raise ValueError("plan contains duplicate acceptance item IDs")
return items
def build_acceptance_report(settings: Settings) -> AcceptanceReport:
project_root = settings.project_root
items = parse_plan_checklist(project_root / "docs" / "plan.md")
checks = {
"catalog_22_tasks": _catalog_has_22_tasks(project_root),
"scheduled_graphs_explicit": _scheduled_graphs_are_explicit(project_root),
"python_scheduler_only": _python_scheduler_is_the_only_scheduler(project_root),
"runtime_service_policy": _runtime_service_policy_is_configured(project_root),
"native_entrypoints_local": _native_entrypoints_are_local(project_root),
"runtime_sources_decoupled": _runtime_sources_are_decoupled(project_root),
"public_command_registry_complete": _public_command_registry_is_complete(
project_root
),
"source_manifests_verified": _source_manifests_are_verified(project_root),
"secret_scan_clean": not scan_repository(project_root),
}
return AcceptanceReport(items, checks)
def _catalog_has_22_tasks(project_root: Path) -> bool:
try:
catalog = WorkflowCatalog.load(project_root / "config")
except Exception:
return False
scheduled = catalog.scheduled_workflows()
scheduled_ids = {workflow.workflow_id for workflow in scheduled}
schedule_ids = {schedule.workflow_id for schedule in catalog.schedules}
return bool(scheduled) and scheduled_ids == schedule_ids
def _scheduled_graphs_are_explicit(project_root: Path) -> bool:
try:
catalog = WorkflowCatalog.load(project_root / "config")
except Exception:
return False
return all(workflow.steps for workflow in catalog.scheduled_workflows())
def _python_scheduler_is_the_only_scheduler(project_root: Path) -> bool:
scheduler_service = project_root / "src" / "gyxx_flow" / "scheduler_service.py"
retired_windows_scheduler = project_root / "src" / "gyxx_flow" / "scheduler.py"
systemd_unit = project_root / "deploy" / "gyxx-flow.service"
legacy_installer = project_root / "deploy" / "windows-service" / "install.ps1"
try:
unit = systemd_unit.read_text(encoding="utf-8").casefold()
except OSError:
return False
legacy_is_safe = True
if legacy_installer.exists():
try:
installer = legacy_installer.read_text(encoding="utf-8").casefold()
except OSError:
return False
legacy_is_safe = (
"schedule run" in installer
and "schtasks" not in installer
and "new-scheduledtask" not in installer
)
return (
scheduler_service.is_file()
and not retired_windows_scheduler.exists()
and "execstart=/opt/gyxx-flow/.venv/bin/python -m gyxx_flow schedule run"
in unit
and "environment=gyxx_data_root=/var/lib/gyxx-flow" in unit
and "oncalendar=" not in unit
and legacy_is_safe
)
def _runtime_service_policy_is_configured(project_root: Path) -> bool:
try:
payload = json.loads(
(project_root / "config" / "runtime-bindings.json").read_text(
encoding="utf-8"
)
)
services = payload["services"]
except (OSError, json.JSONDecodeError, KeyError, TypeError):
return False
local_urls = (
services.get("hermes_url", ""),
services.get("hermes_collector_url", ""),
services.get("hermes_analyzer_gateway_url", ""),
services.get("hermes_collector_gateway_url", ""),
)
return (
services.get("postgres") == "cloud"
and not services.get("postgres_host")
and not services.get("postgres_database")
and not services.get("postgres_user")
and services.get("hermes") == "local"
and all(url.startswith("http://127.0.0.1:") for url in local_urls)
)
def _native_entrypoints_are_local(project_root: Path) -> bool:
try:
catalog = WorkflowCatalog.load(project_root / "config")
for workflow in catalog.workflows:
if workflow.trigger == "unavailable":
continue
target = (
project_root
/ "src"
/ "gyxx_flow"
/ "modules"
/ workflow.module
/ workflow.entry
).resolve(strict=True)
if not target.is_file() or not target.is_relative_to(project_root):
return False
config_text = (project_root / "config" / "workflows.json").read_text(
encoding="utf-8"
)
return "legacy" not in config_text.casefold()
except (OSError, ValueError, CatalogError):
return False
def _public_command_registry_is_complete(project_root: Path) -> bool:
try:
workflows = WorkflowCatalog.load(project_root / "config").workflows
commands = ScriptCatalog.discover_default()
except (OSError, ValueError, ScriptCatalogError):
return False
runnable = tuple(item for item in workflows if item.trigger != "unavailable")
expected_modules = {item.module for item in runnable}
scripts = commands.scripts
if (
not scripts
or len(commands.command_ids) != len(set(commands.command_ids))
or {item.module for item in scripts} != expected_modules
):
return False
resolved_root = project_root.resolve()
for script in scripts:
if not script.path.is_file() or not script.path.is_relative_to(resolved_root):
return False
try:
for workflow in runnable:
entries = (
tuple(step.entry for step in workflow.steps)
if workflow.steps
else (workflow.entry,)
)
for entry in entries:
command = commands.get(f"{workflow.module}:{entry}")
if command.module != workflow.module or command.entry != entry:
return False
except ScriptCatalogError:
return False
return True
def _runtime_sources_are_decoupled(project_root: Path) -> bool:
module_root = project_root / "src" / "gyxx_flow" / "modules"
forbidden = (
"d:\\yingxiaoyunying",
"d:\\shop-data-flow",
"d:\\product-collector-analyze-flow",
"e:\\auto-flow",
"gyxx_legacy_",
"deferredlegacycommandstep",
)
try:
for path in module_root.rglob("*"):
if not path.is_file() or path.suffix.casefold() in {".pyc", ".pyo"}:
continue
if path.is_symlink() or not path.resolve().is_relative_to(project_root):
return False
content = path.read_bytes()
if b"\x00" in content:
continue
text: str | None = None
for encoding in ("utf-8-sig", "gb18030"):
try:
text = content.decode(encoding)
break
except UnicodeDecodeError:
continue
if text is None:
continue
normalized = text.casefold().replace("/", "\\")
while "\\\\" in normalized:
normalized = normalized.replace("\\\\", "\\")
if any(value in normalized for value in forbidden):
return False
except OSError:
return False
return True
def _source_manifests_are_verified(project_root: Path) -> bool:
modules = (
"content_marketing",
"product_commerce",
"shop_intelligence",
"supply_chain",
)
try:
for module in modules:
payload = json.loads(
(
project_root
/ "config"
/ "source-manifests"
/ f"{module}.json"
).read_text(encoding="utf-8-sig")
)
if payload.get("schema_version") != 1 or payload.get("module") != module:
return False
files = payload.get("files")
if not isinstance(files, list) or not files:
return False
for item in files:
target = project_root.joinpath(*Path(item["target_relative_path"]).parts)
resolved = target.resolve(strict=True)
if (
target.is_symlink()
or not resolved.is_relative_to(project_root)
or not resolved.is_file()
or _sha256(resolved) != item["target_sha256"]
):
return False
except (OSError, json.JSONDecodeError, KeyError, TypeError, ValueError):
return False
return True
def _sha256(path: Path) -> str:
digest = hashlib.sha256()
with path.open("rb") as stream:
for chunk in iter(lambda: stream.read(1024 * 1024), b""):
digest.update(chunk)
return digest.hexdigest()