cdsl-cad/backend/scripts/remove_generation_spec_artifacts.py

#!/usr/bin/env python3
"""Remove retired semantic-plan artifacts from the local workspace.

Run without arguments to list every file and metadata document that would
change. Run again with --apply to make the irreversible cleanup.
"""

from __future__ import annotations

import argparse
import json
from pathlib import Path
from typing import Any


RETIRED_FILENAMES = {
    "generation-spec.json",
    "generation-provenance.json",
    "acceptance-report.json",
}
RETIRED_DIAGNOSTIC_PREFIX = "generation_spec_"
RETIRED_TOOL_NAMES = {
    "create_generation_spec",
    "author_cdsl_from_generation_spec",
    "compile_generation_spec",
    "patch_generation_spec",
    "generate_flange_sleeve_model",
}
RETIRED_METADATA_KEYS = {
    "generation_spec_path",
    "generation_provenance_path",
    "acceptance_path",
    "approximations",
    "current_design_intent_id",
    "design_intents",
    "design_intent_id",
    "design_intent_path",
    "design_intent_structures",
    "design_intent_assumptions",
    "design_intent_capability_gaps",
    "capability_translations",
    "evidence",
}


def _strip_metadata(value: Any) -> bool:
    changed = False
    if isinstance(value, list):
        for item in value:
            changed = _strip_metadata(item) or changed
        return changed
    if not isinstance(value, dict):
        return False
    for key in list(value):
        if key in RETIRED_METADATA_KEYS or key.startswith("generation_spec_"):
            value.pop(key)
            changed = True
    operation = value.get("operation")
    if isinstance(operation, dict) and "spec" in str(operation.get("type") or "").casefold():
        value["operation"] = {}
        changed = True
    for key, child in list(value.items()):
        if key == "source" and isinstance(child, str) and "generation_spec" in child.casefold():
            value[key] = "legacy_cdsl"
            changed = True
        elif key in {"message", "text"} and isinstance(child, str) and "generationspec" in child.casefold():
            value[key] = "已清理已弃用的规格流程诊断；当前任务统一使用直接 CDSL 生成。"
            changed = True
        else:
            changed = _strip_metadata(child) or changed
    return changed


def _retired_diagnostic(path: Path) -> bool:
    if path.name.startswith(RETIRED_DIAGNOSTIC_PREFIX):
        return True
    try:
        payload = json.loads(path.read_text(encoding="utf-8"))
    except (OSError, json.JSONDecodeError):
        return False
    if not isinstance(payload, dict):
        return False
    tool_name = str(payload.get("tool_name") or "")
    arguments = str(payload.get("arguments") or "")
    return tool_name in RETIRED_TOOL_NAMES or "generation-spec" in arguments or "generation_spec" in arguments


def cleanup_plan(data_root: Path) -> tuple[list[Path], list[Path]]:
    files: list[Path] = []
    metadata: list[Path] = []
    task_root = data_root / "tasks"
    conversation_root = data_root / "conversations"
    if task_root.is_dir():
        for path in task_root.rglob("*"):
            if path.is_file() and (
                path.name in RETIRED_FILENAMES
                or (path.parent.name == "planning" and path.name.startswith("design-intent-"))
            ):
                files.append(path)
        for path in task_root.rglob("*.json"):
            try:
                payload = json.loads(path.read_text(encoding="utf-8"))
            except (OSError, json.JSONDecodeError):
                continue
            original = json.dumps(payload, sort_keys=True, ensure_ascii=False)
            _strip_metadata(payload)
            if json.dumps(payload, sort_keys=True, ensure_ascii=False) != original:
                metadata.append(path)
    if conversation_root.is_dir():
        for path in conversation_root.rglob("*.json"):
            if _retired_diagnostic(path):
                files.append(path)
                continue
            try:
                payload = json.loads(path.read_text(encoding="utf-8"))
            except (OSError, json.JSONDecodeError):
                continue
            original = json.dumps(payload, sort_keys=True, ensure_ascii=False)
            _strip_metadata(payload)
            if json.dumps(payload, sort_keys=True, ensure_ascii=False) != original:
                metadata.append(path)
    file_set = set(files)
    return sorted(file_set), sorted(set(metadata) - file_set)


def apply_cleanup(data_root: Path) -> tuple[list[Path], list[Path]]:
    files, metadata = cleanup_plan(data_root)
    for path in files:
        path.unlink()
    for path in metadata:
        payload = json.loads(path.read_text(encoding="utf-8"))
        if _strip_metadata(payload):
            path.write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
    return files, metadata


def main() -> int:
    parser = argparse.ArgumentParser(description="Remove retired GenerationSpec and semantic-plan artifacts")
    parser.add_argument("--data-root", type=Path, default=Path(__file__).resolve().parents[1] / "data")
    parser.add_argument("--apply", action="store_true", help="Delete the listed artifacts and rewrite metadata")
    args = parser.parse_args()
    files, metadata = cleanup_plan(args.data_root)
    action = "Will remove" if not args.apply else "Removing"
    for path in files:
        print(f"{action} artifact: {path}")
    for path in metadata:
        print(f"{action} retired metadata: {path}")
    if not files and not metadata:
        print("No retired GenerationSpec or semantic-plan artifacts found.")
    if not args.apply:
        print("Dry run only. Re-run with --apply to make these changes.")
        return 0
    apply_cleanup(args.data_root)
    print(f"Removed {len(files)} artifacts and rewrote {len(metadata)} metadata documents.")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())