ygg 0.1.19__py3-none-any.whl → 0.1.21__py3-none-any.whl

yggdrasil/ser/callable_serde.py

@@ -1,645 +0,0 @@
-# yggdrasil/ser.py
-from __future__ import annotations
-
-import ast
-import base64
-import builtins
-import inspect
-import json
-import os
-import sys
-import textwrap
-import zlib
-from dataclasses import dataclass, field
-from pathlib import Path
-from typing import Any, Callable, Dict, Iterable, Optional, Tuple
-
-import dill
-
-PyVer = Tuple[int, int]
-
-
-def _pyver() -> PyVer:
-    v = sys.version_info
-    return v.major, v.minor
-
-
-def _b64e(b: bytes) -> str:
-    return base64.b64encode(b).decode("ascii")
-
-
-def _b64d(s: str) -> bytes:
-    return base64.b64decode(s.encode("ascii"))
-
-
-def _is_python_function(obj: Any) -> bool:
-    return inspect.isfunction(obj) or inspect.ismethod(obj) or isinstance(obj, type(lambda: 0))
-
-
-def _safe_dill_dumps(obj: Any) -> Optional[bytes]:
-    try:
-        return dill.dumps(obj, recurse=True)
-    except Exception:
-        return None
-
-
-def _safe_dill_loads(b: Optional[bytes]) -> Any:
-    if b is None:
-        raise ValueError("No dill bytes to load")
-    return dill.loads(b)
-
-
-def _infer_package_root(fn: Callable[..., Any]) -> Tuple[str, ...]:
-    """
-    Return the *topmost* package directory for the module defining `fn`.
-
-    Walk upwards from the module file directory while an __init__.py exists.
-    Example:
-      /repo/my_pkg/sub/mod.py  -> returns /repo/my_pkg
-    If no __init__.py is found, falls back to the module's parent directory.
-    """
-    try:
-        mod = inspect.getmodule(fn)
-        file = getattr(mod, "__file__", None)
-        if not file:
-            return tuple()
-
-        p = Path(file).resolve()
-        if p.suffix.lower() != ".py":
-            return tuple()
-
-        cur = p.parent  # start at directory containing the module file
-
-        # If this directory isn't a package at all, just return it
-        if not (cur / "__init__.py").exists():
-            return (str(cur),)
-
-        # Climb while parent is still a package (has __init__.py)
-        # Stop at the highest directory that is still a package.
-        top = cur
-        while True:
-            parent = top.parent
-            if parent == top:
-                break
-            if (parent / "__init__.py").exists():
-                top = parent
-                continue
-            break
-
-        return (str(top),)
-    except Exception:
-        return tuple()
-
-
-def _capture_exec_env(fn: Callable[..., Any]) -> Dict[str, Dict[str, str]]:
-    """
-    Capture a minimal environment for exec(source) reconstruction.
-
-    - globals: names referenced by fn.__code__.co_names that exist in fn.__globals__ (excluding builtins)
-    - freevars: closure cell contents for fn.__code__.co_freevars
-
-    Values are stored as base64(dill.dumps(value)) (best-effort).
-    """
-    code = fn.__code__
-    fn_globals = getattr(fn, "__globals__", {}) or {}
-
-    referenced = set(code.co_names or ())
-    builtins_set = set(dir(builtins))
-
-    g_payload: Dict[str, str] = {}
-    for name in referenced:
-        if name in builtins_set:
-            continue
-        if name in fn_globals:
-            dumped = _safe_dill_dumps(fn_globals[name])
-            if dumped is not None:
-                g_payload[name] = _b64e(dumped)
-
-    fv_payload: Dict[str, str] = {}
-    freevars = code.co_freevars or ()
-    closure = fn.__closure__ or ()
-    if freevars and closure and len(freevars) == len(closure):
-        for name, cell in zip(freevars, closure):
-            try:
-                val = cell.cell_contents
-            except ValueError:
-                continue
-            dumped = _safe_dill_dumps(val)
-            if dumped is not None:
-                fv_payload[name] = _b64e(dumped)
-
-    return {"globals": g_payload, "freevars": fv_payload}
-
-
-def _capture_module_imports(fn: Callable[..., Any]) -> str:
-    """
-    Capture top-level imports from the module file where `fn` is defined.
-
-    We extract only:
-      - `import x`
-      - `import x as y`
-      - `from a.b import c`
-      - `from a.b import c as d`
-
-    Returned as a single string block (may be empty).
-    """
-    try:
-        src_file = inspect.getsourcefile(fn) or inspect.getfile(fn)
-        if not src_file:
-            return ""
-        p = Path(src_file)
-        if not p.exists() or p.suffix.lower() != ".py":
-            return ""
-
-        text = p.read_text(encoding="utf-8")
-        tree = ast.parse(text)
-
-        imports: list[str] = []
-        for node in tree.body:
-            if isinstance(node, (ast.Import, ast.ImportFrom)):
-                seg = ast.get_source_segment(text, node)
-                if seg:
-                    imports.append(seg.strip())
-
-        # de-dupe while preserving order
-        seen = set()
-        uniq: list[str] = []
-        for line in imports:
-            if line not in seen:
-                seen.add(line)
-                uniq.append(line)
-
-        return "\n".join(uniq).strip() + ("\n" if uniq else "")
-    except Exception:
-        return ""
-
-
-def parse_tagged_result(stdout_text: str, result_tag: str) -> Dict[str, Any]:
-    """
-    Extract the last tagged JSON payload printed by a cluster command.
-    Expects lines like:  <<<RESULT>>>{...json...}
-    """
-    last = None
-    for line in stdout_text.splitlines():
-        if line.startswith(result_tag):
-            last = line[len(result_tag) :]
-    if last is None:
-        raise ValueError(f"Result tag {result_tag!r} not found in output")
-    try:
-        return json.loads(last)
-    except Exception as e:
-        raise ValueError("Tagged result was not valid JSON") from e
-
-
-class CommandError(RuntimeError):
-    """
-    Raised when the remote Databricks command returns ok=false.
-
-    Attributes
-    ----------
-    error: str
-        Short error message from remote.
-    traceback: str
-        Remote traceback (string) if provided.
-    raw: dict
-        Full parsed JSON payload from the tagged output line.
-    """
-
-    def __init__(self, error: str, traceback: str = "", raw: Optional[Dict[str, Any]] = None):
-        self.error = error
-        self.traceback = traceback
-        self.raw = raw or {}
-        msg = error if not traceback else f"{error}\n{traceback}"
-        super().__init__(msg.rstrip())
-
-
-class CommandResultParseError(ValueError):
-    """
-    Raised when the command output cannot be parsed or decoded.
-    """
-
-    def __init__(self, message: str, stdout_text: Optional[str] = None):
-        self.stdout_text = stdout_text
-        super().__init__(message)
-
-
-@dataclass
-class CallableSerdeMixin:
-    """
-    Encapsulates a callable + serialization strategy + Databricks command generation.
-
-    - Same Python (major, minor): prefer dill
-    - Different Python (major, minor): exec(imports + source) with captured globals/freevars
-    """
-
-    fn: Callable[..., Any]
-    package_root: Tuple[str, ...] = field(default_factory=tuple)
-    ALLOW_EXEC_SOURCE: bool = True
-
-    @classmethod
-    def from_callable(cls, fn: Callable[..., Any]) -> "CallableSerdeMixin":
-        if isinstance(fn, CallableSerdeMixin):
-            return fn
-        return cls(fn=fn, package_root=_infer_package_root(fn))
-
-    def __call__(self, *args: Any, **kwargs: Any) -> Any:
-        return self.fn(*args, **kwargs)
-
-    # ---------- pickle protocol ----------
-    def __getstate__(self) -> Dict[str, Any]:
-        if _is_python_function(self.fn):
-            src = None
-            try:
-                src = textwrap.dedent(inspect.getsource(self.fn))
-            except Exception:
-                src = None
-
-            payload: Dict[str, Any] = {
-                "__callable__": True,
-                "pyver": list(_pyver()),
-                "name": getattr(self.fn, "__name__", None),
-                "qualname": getattr(self.fn, "__qualname__", None),
-                "module": getattr(self.fn, "__module__", None),
-                "imports": _capture_module_imports(self.fn),  # NEW
-                "source": src,
-                "dill_b64": None,
-                "env": _capture_exec_env(self.fn),
-            }
-
-            dumped = _safe_dill_dumps(self.fn)
-            if dumped is not None:
-                payload["dill_b64"] = _b64e(dumped)
-
-            return {
-                "fn": payload,
-                "package_root": tuple(self.package_root),
-                "ALLOW_EXEC_SOURCE": bool(self.ALLOW_EXEC_SOURCE),
-            }
-
-        dumped = _safe_dill_dumps(self.fn)
-        if dumped is None:
-            raise ValueError("Callable object could not be dill-serialized")
-        return {
-            "fn": {"__callable__": False, "dill_b64": _b64e(dumped)},
-            "package_root": tuple(self.package_root),
-            "ALLOW_EXEC_SOURCE": bool(self.ALLOW_EXEC_SOURCE),
-        }
-
-    def __setstate__(self, state: Dict[str, Any]) -> None:
-        self.package_root = tuple(state.get("package_root") or ())
-        self.ALLOW_EXEC_SOURCE = bool(state.get("ALLOW_EXEC_SOURCE", True))
-
-        fn_payload = state["fn"]
-
-        if isinstance(fn_payload, dict) and fn_payload.get("__callable__") is False:
-            self.fn = _safe_dill_loads(_b64d(fn_payload["dill_b64"]))
-            return
-
-        if not isinstance(fn_payload, dict) or fn_payload.get("__callable__") is not True:
-            raise ValueError("Invalid callable payload")
-
-        src_pyver = tuple(fn_payload.get("pyver") or ())
-        cur_pyver = _pyver()
-
-        if src_pyver == cur_pyver and fn_payload.get("dill_b64"):
-            try:
-                self.fn = _safe_dill_loads(_b64d(fn_payload["dill_b64"]))
-                if callable(self.fn):
-                    return
-            except Exception:
-                pass
-
-        if not self.ALLOW_EXEC_SOURCE:
-            raise ValueError("Exec-based restore disabled and dill path unavailable/failed")
-
-        imports = fn_payload.get("imports") or ""
-        source = fn_payload.get("source")
-        name = fn_payload.get("name")
-        env = fn_payload.get("env") or {}
-
-        if not source:
-            raise ValueError("No source available for exec-based restore")
-
-        ns: Dict[str, Any] = {}
-
-        # preload captured names
-        for bucket in ("globals", "freevars"):
-            items = (env.get(bucket) or {})
-            for k, b64 in items.items():
-                try:
-                    ns[k] = _safe_dill_loads(_b64d(b64))
-                except Exception:
-                    pass
-
-        # NEW: run module imports first
-        if imports.strip():
-            exec(imports, ns, ns)
-
-        exec(source, ns, ns)
-
-        if name and name in ns and callable(ns[name]):
-            self.fn = ns[name]
-            return
-
-        cands = [v for v in ns.values() if callable(v)]
-        if cands:
-            self.fn = cands[-1]
-            return
-
-        raise ValueError("exec(source) succeeded but no callable could be recovered")
-
-    # ---------- Databricks command generation ----------
-    def to_command(
-        self,
-        *,
-        args: Tuple[Any, ...] = (),
-        kwargs: Optional[Dict[str, Any]] = None,
-        env_keys: Iterable[str] = (),
-        env_variables: Optional[Dict[str, str]] = None,
-        use_dill: bool,
-        byte_limit: int = 0,
-        result_tag: str = "<<<RESULT>>>",
-    ) -> str:
-        if kwargs is None:
-            kwargs = {}
-        if env_variables is None:
-            env_variables = {}
-
-        # capture env vars from client process
-        client_env: Dict[str, str] = {}
-        for k in env_keys:
-            v = os.environ.get(k)
-            if v is not None:
-                client_env[k] = v
-        client_env.update(env_variables)
-
-        if not _is_python_function(self.fn):
-            raise ValueError("to_command supports Python functions/methods/lambdas only")
-
-        src = None
-        try:
-            src = textwrap.dedent(inspect.getsource(self.fn))
-        except Exception:
-            src = None
-
-        imports = _capture_module_imports(self.fn)
-
-        callable_payload: Dict[str, Any] = {
-            "__callable__": True,
-            "pyver": list(_pyver()),
-            "name": getattr(self.fn, "__name__", None),
-            "qualname": getattr(self.fn, "__qualname__", None),
-            "module": getattr(self.fn, "__module__", None),
-            "imports": imports,  # NEW
-            "source": src,
-            "dill_b64": None,
-            "env": _capture_exec_env(self.fn),
-        }
-        dumped_fn = _safe_dill_dumps(self.fn)
-        if dumped_fn is not None:
-            callable_payload["dill_b64"] = _b64e(dumped_fn)
-
-        # args/kwargs transport
-        dumped_args = _safe_dill_dumps(args)
-        dumped_kwargs = _safe_dill_dumps(kwargs)
-        if dumped_args is None or dumped_kwargs is None:
-            raise ValueError("Failed to dill-serialize args/kwargs")
-
-        args_pack: Dict[str, Any] = {"kind": "dill", "b64": _b64e(dumped_args)}
-        kwargs_pack: Dict[str, Any] = {"kind": "dill", "b64": _b64e(dumped_kwargs)}
-
-        if not byte_limit:
-            byte_limit = 512 * 1024
-
-        payload = {
-            "callable": callable_payload,
-            "use_dill": bool(use_dill),
-            "args": args_pack,
-            "kwargs": kwargs_pack,
-            "env": client_env,
-            "result_tag": result_tag,
-            "byte_limit": byte_limit
-        }
-
-        return f"""
-# --- generated by yggdrasil.ser.CallableSerdeMixin.to_command ---
-import base64, json, os, traceback, zlib
-import dill
-from yggdrasil.databricks import *
-
-_payload = {payload!r}
-
-def _b64d(s: str) -> bytes:
-    return base64.b64decode(s.encode("ascii"))
-
-def _try_dill_load(b64: str):
-    return dill.loads(_b64d(b64))
-
-def _hydrate_env(env_dict):
-    ns = {{}}
-    for bucket in ("globals", "freevars"):
-        items = (env_dict or {{}}).get(bucket, {{}}) or {{}}
-        for name, b64 in items.items():
-            try:
-                ns[name] = _try_dill_load(b64)
-            except Exception:
-                pass
-    return ns
-
-def _load_args(pack):
-    if pack["kind"] == "dill":
-        return dill.loads(_b64d(pack["b64"]))
-    return tuple(json.loads(pack["text"]))
-
-def _load_kwargs(pack):
-    if pack["kind"] == "dill":
-        return dill.loads(_b64d(pack["b64"]))
-    return dict(json.loads(pack["text"]))
-
-def _load_callable(cpack, use_dill: bool):
-    # Prefer dill when requested/available
-    if use_dill and cpack.get("dill_b64"):
-        fn = _try_dill_load(cpack["dill_b64"])
-        if callable(fn):
-            return fn
-
-    src = cpack.get("source")
-    name = cpack.get("name")
-    imports = cpack.get("imports") or ""
-    if not src:
-        raise ValueError("No source available for exec-based restore")
-
-    ns = {{}}
-    ns.update(_hydrate_env(cpack.get("env") or {{}}))
-
-    # NEW: exec module imports first
-    if imports.strip():
-        exec(imports, ns, ns)
-
-    exec(src, ns, ns)
-
-    if name and name in ns and callable(ns[name]):
-        return ns[name]
-
-    cands = [v for v in ns.values() if callable(v)]
-    if cands:
-        return cands[-1]
-
-    raise ValueError("exec(source) ran but no callable was recovered")
-
-def _emit(tag: str, obj: dict):
-    print(tag + json.dumps(obj, ensure_ascii=False, separators=(",", ":")))
-
-
-def _zlib_level(n: int, limit: int) -> int:
-    # ratio of size to limit
-    r = n / max(1, limit)
-
-    # fast for huge payloads, stronger only when slightly over
-    if r >= 32:
-        return 1
-    if r >= 16:
-        return 2
-    if r >= 8:
-        return 3
-    if r >= 4:
-        return 4
-    if r >= 2:
-        return 5
-    if r >= 1.25:
-        return 6
-    # barely over: squeeze a bit more
-    return 7
-
-# apply env vars
-for k, v in (_payload.get("env") or {{}}).items():
-    if v is not None:
-        os.environ[str(k)] = str(v)
-
-tag = _payload.get("result_tag", "<<<RESULT>>>")
-byte_limit = int((_payload.get("byte_limit") or 2_000_000))  # ~2MB serialized bytes
-
-try:
-    use_dill = bool(_payload.get("use_dill", False))
-    fn = _load_callable(_payload["callable"], use_dill=use_dill)
-    args = _load_args(_payload["args"])
-    kwargs = _load_kwargs(_payload["kwargs"])
-
-    out = fn(*args, **kwargs)
-    out_raw = dill.dumps(out, recurse=True)
-
-    if len(out_raw) > byte_limit:
-        lvl = _zlib_level(len(out_raw), byte_limit)
-        out_comp = zlib.compress(out_raw, level=lvl)
-        out_b64 = base64.b64encode(out_comp).decode("ascii")
-        _emit(tag, {{"ok": True, "encoding": f"dill+zlib{{lvl}}+b64", "payload": out_b64}})
-    else:
-        out_b64 = base64.b64encode(out_raw).decode("ascii")
-        _emit(tag, {{"ok": True, "encoding": "dill+b64", "payload": out_b64}})
-
-except Exception as e:
-    tb = traceback.format_exc()
-    err = {{"cls": e.__class__.__name__, "msg": str(e), "tb": tb}}
-    _emit(tag, {{"ok": False, "err": err}})
-
-# --- end generated command ---
-""".lstrip()
-
-    @staticmethod
-    def parse_command_result(
-        stdout_text: str,
-        *,
-        result_tag: str = "<<<RESULT>>>",
-        decode: bool = True,
-    ) -> Any:
-        """
-        Parse Databricks command stdout and return decoded result or raise.
-
-        Protocol:
-          prints lines like:  <<<RESULT>>>{...json...}
-          last tagged line wins
-
-        If ok=false -> raise DatabricksCommandError
-        If ok=true and decode=True:
-          - encoding == "dill+b64": dill.loads(base64(payload))
-          - encoding == "repr": returns payload as-is
-        If decode=False: returns the raw parsed dict.
-        """
-        last: Optional[str] = None
-        for line in stdout_text.splitlines():
-            if line.startswith(result_tag):
-                last = line[len(result_tag):]
-            else:
-                print(line)
-
-        if last is None:
-            raise CommandResultParseError(
-                f"Result tag {result_tag!r} not found in command output",
-                stdout_text=stdout_text,
-            )
-
-        try:
-            msg: Dict[str, Any] = json.loads(last)
-        except Exception as e:
-            raise CommandResultParseError(
-                "Tagged result is not valid JSON",
-                stdout_text=stdout_text,
-            ) from e
-
-        if not decode:
-            return msg
-
-        ok = bool(msg.get("ok", False))
-        if not ok:
-            error = msg.get("err", {})
-            error_class = str(error.get("cls", "RuntimeError"))
-            error_message = str(error.get("msg", "Remote execution failed"))
-            error_traceback = str(error.get("tb", ""))
-
-            base = CommandError(
-                error=error_message,
-                traceback=error_traceback,
-                raw=msg,
-            )
-
-            if error_class == "ModuleNotFound":
-                raise ModuleNotFoundError(error_message) from base
-
-            raise base
-
-        encoding = msg.get("encoding")
-        payload = msg.get("payload")
-
-        if encoding == "dill+b64":
-            if not isinstance(payload, str):
-                raise CommandResultParseError(
-                    "Expected base64 string payload for dill+b64 encoding",
-                    stdout_text=stdout_text,
-                )
-            try:
-                data = base64.b64decode(payload.encode("ascii"))
-                return dill.loads(data)
-            except Exception as e:
-                raise CommandResultParseError(
-                    "Failed to decode dill+b64 payload",
-                    stdout_text=stdout_text,
-                ) from e
-
-        if encoding == "repr":
-            return payload
-
-        if encoding.startswith("dill+zlib") and encoding.endswith("+b64"):
-            if not isinstance(payload, str):
-                raise CommandResultParseError("Expected base64 string payload for zlib encoding",
-                                              stdout_text=stdout_text)
-            try:
-                data = base64.b64decode(payload.encode("ascii"))
-                raw = zlib.decompress(data)
-                return dill.loads(raw)
-            except Exception as e:
-                raise CommandResultParseError("Failed to decode dill+zlib+b64 payload", stdout_text=stdout_text) from e
-
-        raise CommandResultParseError(
-            f"Unknown result encoding: {encoding!r}",
-            stdout_text=stdout_text,
-        )