codex-app-server/chv-guest/generate_lockfile.py

#!/usr/bin/env python3
"""在构建宿主上生成 CHV 客机镜像的供应链 lockfile。

事后校验见 check_lockfile.py（校验 BUILDINFO.versions）；
本脚本是生成端：读 BUILDINFO.json，以其中的 versions
（build.sh 已从客机读回的 dpkg-query / pip freeze，
Playwright=1 时另有 chromium 标识）为唯一版本来源，
逐项给 contents 清单配版本，写出 lockfile.json：

  {"tag": ..., "image_sha256": ... | null,
   "generated_at": UTC ISO 时间,
   "versions": {"dpkg": ["name version", ...],
                "pip": ["name version", ...],
                "declared": ["name version", ...]},
   "unresolved": ["查不到版本的名", ...]}

版本来源映射（绝不编造版本号，查不到的进 unresolved）：
  - dpkg 包名 → versions.dpkg（"name version" 行）
  - pip 包名 → versions.pip（"name==version" 行）
  - codex / codex-code-mode-host / codex-resources/bwrap
    → BUILDINFO.codex_version（随 codex 一起发布）
  - /app/host.py → BUILDINFO.repo_commit
  - chromium → versions.chromium（ms-playwright 缓存目录名）

注意：旧版逐项查构建宿主的 dpkg/pip，方法不成立——宿主
版本与客机版本没有必然关系，且 /app/host.py、codex 三件
套在宿主上永远查不到，--verify 永远过不了。本版只读
BUILDINFO.versions，不再现场查宿主。

用法（均在构建宿主上跑）：
  python3 generate_lockfile.py BUILDINFO.json --output lockfile.json
  python3 generate_lockfile.py --verify lockfile.json \
      --buildinfo BUILDINFO.json
"""

import json
import sys
from datetime import datetime, timezone

# 非包名的内容项：版本来自 BUILDINFO 的声明字段，不是 dpkg/pip。
DECLARED_SOURCES = {
    "codex": "codex_version",
    "codex-code-mode-host": "codex_version",
    "codex-resources/bwrap": "codex_version",
    "/app/host.py": "repo_commit",
}


def _index_dpkg(lines):
    """dpkg 行 "name version" → {小写名: "name version"}。"""
    index = {}
    for line in lines or []:
        parts = str(line).split(None, 1)
        if len(parts) == 2:
            index[parts[0].lower()] = "%s %s" % (parts[0], parts[1])
    return index


def _index_pip(lines):
    """pip freeze 行 "name==version" → {小写名: "name version"}。"""
    index = {}
    for line in lines or []:
        if "==" in str(line):
            name, version = str(line).split("==", 1)
            index[name.lower()] = "%s %s" % (name, version.strip())
    return index


def _index_chromium(lines):
    """chromium 标识行（如 chromium-1200）→ {"chromium": 行}。"""
    index = {}
    for line in lines or []:
        text = str(line).strip()
        if text:
            index["chromium"] = "chromium %s" % text
    return index


def generate(buildinfo_path, output_path):
    with open(buildinfo_path) as fh:
        buildinfo = json.load(fh)
    names = buildinfo.get("contents") or []
    if not isinstance(names, list):
        print("BUILDINFO.contents 不是列表", file=sys.stderr)
        return 2
    versions = buildinfo.get("versions") or {}
    dpkg_index = _index_dpkg(versions.get("dpkg"))
    pip_index = _index_pip(versions.get("pip"))
    chromium_index = _index_chromium(versions.get("chromium"))
    dpkg, pip, declared, unresolved = [], [], [], []
    for name in names:
        key = name.lower()
        if name in DECLARED_SOURCES:
            version = buildinfo.get(DECLARED_SOURCES[name])
            if version:
                declared.append("%s %s" % (name, version))
            else:
                unresolved.append(name)
            continue
        if key == "chromium":
            entry = chromium_index.get("chromium")
            if entry:
                declared.append(entry)
            else:
                unresolved.append(name)
            continue
        if key in dpkg_index:
            dpkg.append(dpkg_index[key])
            continue
        if key in pip_index:
            pip.append(pip_index[key])
            continue
        unresolved.append(name)
    lockfile = {
        "tag": buildinfo.get("tag"),
        "image_sha256": buildinfo.get("image_sha256")
        or buildinfo.get("sha256"),
        "generated_at": datetime.now(timezone.utc).isoformat(),
        "versions": {"dpkg": sorted(dpkg), "pip": sorted(pip),
                     "declared": sorted(declared)},
        "unresolved": sorted(unresolved),
    }
    with open(output_path, "w") as fh:
        json.dump(lockfile, fh, indent=2, sort_keys=True)
        fh.write("\n")
    if unresolved:
        print("警告：%d 项查不到版本，已列入 unresolved：%s"
              % (len(unresolved), ", ".join(sorted(unresolved))),
              file=sys.stderr)
    return 0


def verify(lockfile_path, buildinfo_path):
    with open(lockfile_path) as fh:
        lockfile = json.load(fh)
    if buildinfo_path:
        with open(buildinfo_path) as fh:
            buildinfo = json.load(fh)
        if lockfile.get("tag") != buildinfo.get("tag"):
            print("tag 不一致：lockfile %s != BUILDINFO %s"
                  % (lockfile.get("tag"), buildinfo.get("tag")),
                  file=sys.stderr)
            return 1
    versions = lockfile.get("versions") or {}
    for kind in ("dpkg", "pip"):
        entries = versions.get(kind)
        if not entries:
            print("versions.%s 为空，供应链清单缺失" % kind,
                  file=sys.stderr)
            return 1
        for entry in entries:
            parts = str(entry).split()
            if len(parts) < 2 or not parts[1]:
                print("版本为空的条目：%s %r" % (kind, entry),
                      file=sys.stderr)
                return 1
    unresolved = lockfile.get("unresolved") or []
    if unresolved:
        print("清单不完整，unresolved 非空：%s" % sorted(unresolved),
              file=sys.stderr)
        return 1
    print("lockfile 校验通过：tag=%s dpkg=%d pip=%d declared=%d"
          % (lockfile.get("tag"), len(versions["dpkg"]),
             len(versions["pip"]),
             len(versions.get("declared") or [])))
    return 0


def main(argv):
    if "--verify" in argv:
        lockfile_path = argv[argv.index("--verify") + 1]
        buildinfo_path = None
        if "--buildinfo" in argv:
            buildinfo_path = argv[argv.index("--buildinfo") + 1]
        return verify(lockfile_path, buildinfo_path)
    if len(argv) < 2:
        print("用法：generate_lockfile.py BUILDINFO.json "
              "--output lockfile.json | --verify lockfile.json "
              "[--buildinfo BUILDINFO.json]", file=sys.stderr)
        return 2
    output_path = None
    if "--output" in argv:
        output_path = argv[argv.index("--output") + 1]
    if not output_path:
        print("缺少 --output", file=sys.stderr)
        return 2
    return generate(argv[1], output_path)


if __name__ == "__main__":
    raise SystemExit(main(sys.argv))