truenas-truecloud-patch/patch/sitecustomize.py

"""
TrueCloud provider patch — sitecustomize.py
Installed into Python site-packages on every boot by apply.sh.

Hooks two middlewared module imports using the find_spec / exec_module API
(required for Python 3.12+, which ships with TrueNAS SCALE 25.x / Debian 13):

  middlewared.rclone.remote.b2
      Adds get_restic_config() so the native restic B2 backend works.
      Restic repo URL: b2:<bucket>/<folder>
      Auth: B2_ACCOUNT_ID, B2_ACCOUNT_KEY

  middlewared.plugins.cloud_backup.restic
      Fixes the URL builder for providers with no hostname component.
      Stock code: f"{rclone_type}:{url}/{remote_path}" → "b2:/bucket/path" (broken)
      Patched:    "b2:bucket/path" when url == ""

Both patches are no-ops if the module already provides the functionality
(i.e. a future TrueNAS version adds native support). All errors are caught
and written to stderr so middlewared always starts regardless of patch state.
"""

import sys


# ── Import hook ───────────────────────────────────────────────────────────────

class _Finder:
    """
    find_spec-based meta path finder (Python 3.4+, required for 3.12+).
    Intercepts specific module imports, loads them normally, then patches.
    """

    _targets = frozenset({
        "middlewared.rclone.remote.b2",
        "middlewared.plugins.cloud_backup.restic",
    })

    def __init__(self):
        self._loading = set()   # guards against re-entrant imports
        self._done = set()      # modules already patched

    def find_spec(self, fullname, path, target=None):  # noqa: ARG002
        if (
            fullname in self._targets
            and fullname not in self._done
            and fullname not in self._loading
        ):
            import importlib.machinery
            import importlib.util

            # Find the real file spec HERE, before Python adds the module to
            # sys.modules.  If we deferred this to exec_module, find_spec would
            # short-circuit via sys.modules[fullname].__spec__ (our own spec) and
            # exec_module would call itself recursively forever.
            self._loading.add(fullname)
            try:
                real_spec = importlib.util.find_spec(fullname)
            finally:
                self._loading.discard(fullname)

            if real_spec is None:
                return None  # module doesn't exist; don't intercept

            return importlib.machinery.ModuleSpec(
                fullname,
                _Loader(self, fullname, real_spec),
                origin=real_spec.origin,
                is_package=real_spec.submodule_search_locations is not None,
            )
        return None

    def _mark_done(self, fullname):
        self._done.add(fullname)
        if self._done >= self._targets:
            try:
                sys.meta_path.remove(self)
            except ValueError:
                pass


class _Loader:
    def __init__(self, finder, fullname, real_spec):
        self._finder = finder
        self._fullname = fullname
        self._real_spec = real_spec

    def create_module(self, spec):  # noqa: ARG002
        return None  # use Python's default module creation

    def exec_module(self, module):
        fullname = self._fullname
        real_spec = self._real_spec

        try:
            real_spec.loader.exec_module(module)
            # Fix module metadata so it looks like a normal import.
            module.__spec__ = real_spec
            module.__loader__ = real_spec.loader
            if real_spec.origin:
                module.__file__ = real_spec.origin
        finally:
            # Mark done even on failure so a broken module doesn't cause
            # infinite retry loops on subsequent import attempts.
            self._finder._mark_done(fullname)

        # exec_module succeeded — apply our patch.
        try:
            _PATCHES[fullname](module)
        except Exception as exc:
            sys.stderr.write(
                f"[truecloud-patch] patch failed for {fullname}: {exc}\n"
            )
            _record_status(fullname, ok=False, detail=str(exc))


# ── Patch functions ───────────────────────────────────────────────────────────

def _patch_b2(module):
    cls = module.B2RcloneRemote

    if hasattr(cls, "get_restic_config"):
        # A future TrueNAS version already added native B2 restic support.
        return

    def get_restic_config(self, task):  # noqa: ARG001
        p = task["credentials"]["provider"]
        return "", {
            "B2_ACCOUNT_ID": p["account"],
            "B2_ACCOUNT_KEY": p["key"],
        }

    cls.get_restic_config = get_restic_config
    cls.restic = True
    sys.stderr.write("[truecloud-patch] B2 restic support enabled\n")
    _record_status("middlewared.rclone.remote.b2", ok=True)


def _patch_restic(module):
    if getattr(module.get_restic_config, "_truecloud_patched", False):
        return

    import dataclasses
    import re

    _orig = module.get_restic_config

    # Matches "scheme:/path" — the broken form the stock URL builder produces
    # when a provider has no hostname component (url == "").
    # Does NOT match "scheme://path" (Storj and similar legitimately use ://).
    _broken_url = re.compile(r'^(\w[\w+.-]*):/(?!/)(.+)$')

    def get_restic_config(cloud_backup):
        # Call the original — it handles transfer_setting, cache, RESTIC_PASSWORD,
        # env construction, and everything else we don't own.
        result = _orig(cloud_backup)

        # Scan the built command for the -r <repo> argument and fix the URL if
        # it contains a stray leading slash: "b2:/bucket/path" → "b2:bucket/path".
        cmd = list(result.cmd)
        for i, part in enumerate(cmd):
            if i and cmd[i - 1] == "-r":
                m = _broken_url.match(part)
                if m:
                    cmd[i] = f"{m.group(1)}:{m.group(2)}"
                    # dataclasses.replace passes all other fields through, so
                    # new ResticConfig fields added in future TrueNAS versions
                    # are preserved automatically.
                    return dataclasses.replace(result, cmd=cmd)
                break  # -r arg present and already correct
        return result  # URL was fine; return original unchanged

    get_restic_config._truecloud_patched = True
    module.get_restic_config = get_restic_config
    sys.stderr.write("[truecloud-patch] restic URL fix applied\n")
    _record_status("middlewared.plugins.cloud_backup.restic", ok=True)


_PATCHES = {
    "middlewared.rclone.remote.b2": _patch_b2,
    "middlewared.plugins.cloud_backup.restic": _patch_restic,
}

_STATUS_FILE = "/data/truecloud-patch/hook_status.json"
_hook_status: dict = {}


def _record_status(fullname: str, ok: bool, detail: str = "") -> None:
    import json
    import os
    import time

    _hook_status[fullname] = {"ok": ok, "detail": detail}
    if len(_hook_status) < len(_PATCHES):
        return  # wait until all patches have reported

    payload = {
        "patched_at": time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()),
        "patches": _hook_status,
    }
    try:
        tmp = _STATUS_FILE + ".tmp"
        with open(tmp, "w", encoding="utf-8") as fh:
            json.dump(payload, fh, indent=2)
        os.replace(tmp, _STATUS_FILE)  # atomic on POSIX
    except OSError:
        pass  # non-fatal — status file is informational only


# ── Entry point ───────────────────────────────────────────────────────────────

def _install():
    import os
    if os.path.exists("/data/truecloud-patch/disabled"):
        return  # kill switch: touch /data/truecloud-patch/disabled to bypass this hook
    import importlib.util
    if importlib.util.find_spec("middlewared") is None:
        return  # not a middlewared Python process; nothing to do
    sys.meta_path.append(_Finder())


try:
    _install()
except Exception:
    pass  # never raise from sitecustomize.py — it would prevent Python from starting