hgdemandimport: apply lazy module loading to sys.meta_path finders
Python's `sys.meta_path` finders are the primary objects whose job it
is to find a module at import time. When `import` is called, Python
iterates objects in this list and calls `o.find_spec(...)` to find
a `ModuleSpec` (or None if the module couldn't be found by that
finder). If no meta path finder can find a module, import fails.
One of the default meta path finders is `PathFinder`. Its job is to
import modules from the filesystem and is probably the most important
importer. This finder looks at `sys.path` and `sys.path_hooks` to do
its job.
The `ModuleSpec` returned by `MetaPathImporter.find_spec()` has a
`loader` attribute, which defines the concrete module loader to use.
`sys.path_hooks` is a hook point for teaching `PathFinder` to
instantiate custom loader types.
Previously, we injected a custom `sys.path_hook` that told `PathFinder`
to wrap the default loaders with a loader that creates a module object
that is lazy.
This approach worked. But its main limitation was that it only applied
to the `PathFinder` meta path importer. There are other meta path
importers that are registered. And in the case of PyOxidizer loading
modules from memory, `PathFinder` doesn't come into play since
PyOxidizer's own meta path importer was handling all imports.
This commit changes our approach to lazy module loading by proxying
all meta path importers. Specifically, we overload the `find_spec()`
method to swap in a wrapped loader on the `ModuleSpec` before it
is returned. The end result of this is all meta path importers should
be lazy.
As much as I would have loved to utilize .__class__ manipulation to
achieve this, some meta path importers are implemented in C/Rust
in such a way that they cannot be monkeypatched. This is why we
use __getattribute__ to define a proxy.
Also, this change could theoretically open us up to regressions in
meta path importers whose loader is creating module objects which
can't be monkeypatched. But I'm not aware of any of these in the
wild. So I think we'll be safe.
According to hyperfine, this change yields a decent startup time win of
5-6ms:
```
Benchmark #1: ~/.pyenv/versions/3.6.10/bin/python ./hg version
Time (mean ± σ): 86.8 ms ± 0.5 ms [User: 78.0 ms, System: 8.7 ms]
Range (min … max): 86.0 ms … 89.1 ms 50 runs
Time (mean ± σ): 81.1 ms ± 2.7 ms [User: 74.5 ms, System: 6.5 ms]
Range (min … max): 77.8 ms … 90.5 ms 50 runs
Benchmark #2: ~/.pyenv/versions/3.7.6/bin/python ./hg version
Time (mean ± σ): 78.9 ms ± 0.6 ms [User: 70.2 ms, System: 8.7 ms]
Range (min … max): 78.1 ms … 81.2 ms 50 runs
Time (mean ± σ): 73.4 ms ± 0.6 ms [User: 65.3 ms, System: 8.0 ms]
Range (min … max): 72.4 ms … 75.7 ms 50 runs
Benchmark #3: ~/.pyenv/versions/3.8.1/bin/python ./hg version
Time (mean ± σ): 78.1 ms ± 0.6 ms [User: 70.2 ms, System: 7.9 ms]
Range (min … max): 77.4 ms … 80.9 ms 50 runs
Time (mean ± σ): 72.1 ms ± 0.4 ms [User: 64.4 ms, System: 7.6 ms]
Range (min … max): 71.4 ms … 74.1 ms 50 runs
```
Differential Revision: https://phab.mercurial-scm.org/D7954
# error.py - Mercurial exceptions
#
# Copyright 2005-2008 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
"""Mercurial exceptions.
This allows us to catch exceptions at higher levels without forcing
imports.
"""
from __future__ import absolute_import
# Do not import anything but pycompat here, please
from . import pycompat
def _tobytes(exc):
"""Byte-stringify exception in the same way as BaseException_str()"""
if not exc.args:
return b''
if len(exc.args) == 1:
return pycompat.bytestr(exc.args[0])
return b'(%s)' % b', '.join(b"'%s'" % pycompat.bytestr(a) for a in exc.args)
class Hint(object):
"""Mix-in to provide a hint of an error
This should come first in the inheritance list to consume a hint and
pass remaining arguments to the exception class.
"""
def __init__(self, *args, **kw):
self.hint = kw.pop('hint', None)
super(Hint, self).__init__(*args, **kw)
class StorageError(Hint, Exception):
"""Raised when an error occurs in a storage layer.
Usually subclassed by a storage-specific exception.
"""
__bytes__ = _tobytes
class RevlogError(StorageError):
__bytes__ = _tobytes
class SidedataHashError(RevlogError):
def __init__(self, key, expected, got):
self.sidedatakey = key
self.expecteddigest = expected
self.actualdigest = got
class FilteredIndexError(IndexError):
__bytes__ = _tobytes
class LookupError(RevlogError, KeyError):
def __init__(self, name, index, message):
self.name = name
self.index = index
# this can't be called 'message' because at least some installs of
# Python 2.6+ complain about the 'message' property being deprecated
self.lookupmessage = message
if isinstance(name, bytes) and len(name) == 20:
from .node import short
name = short(name)
RevlogError.__init__(self, b'%s@%s: %s' % (index, name, message))
def __bytes__(self):
return RevlogError.__bytes__(self)
def __str__(self):
return RevlogError.__str__(self)
class AmbiguousPrefixLookupError(LookupError):
pass
class FilteredLookupError(LookupError):
pass
class ManifestLookupError(LookupError):
pass
class CommandError(Exception):
"""Exception raised on errors in parsing the command line."""
__bytes__ = _tobytes
class InterventionRequired(Hint, Exception):
"""Exception raised when a command requires human intervention."""
__bytes__ = _tobytes
class Abort(Hint, Exception):
"""Raised if a command needs to print an error and exit."""
__bytes__ = _tobytes
if pycompat.ispy3:
def __str__(self):
# the output would be unreadable if the message was translated,
# but do not replace it with encoding.strfromlocal(), which
# may raise another exception.
return pycompat.sysstr(self.__bytes__())
class HookLoadError(Abort):
"""raised when loading a hook fails, aborting an operation
Exists to allow more specialized catching."""
class HookAbort(Abort):
"""raised when a validation hook fails, aborting an operation
Exists to allow more specialized catching."""
class ConfigError(Abort):
"""Exception raised when parsing config files"""
class UpdateAbort(Abort):
"""Raised when an update is aborted for destination issue"""
class MergeDestAbort(Abort):
"""Raised when an update is aborted for destination issues"""
class NoMergeDestAbort(MergeDestAbort):
"""Raised when an update is aborted because there is nothing to merge"""
class ManyMergeDestAbort(MergeDestAbort):
"""Raised when an update is aborted because destination is ambiguous"""
class ResponseExpected(Abort):
"""Raised when an EOF is received for a prompt"""
def __init__(self):
from .i18n import _
Abort.__init__(self, _(b'response expected'))
class OutOfBandError(Hint, Exception):
"""Exception raised when a remote repo reports failure"""
__bytes__ = _tobytes
class ParseError(Hint, Exception):
"""Raised when parsing config files and {rev,file}sets (msg[, pos])"""
__bytes__ = _tobytes
class PatchError(Exception):
__bytes__ = _tobytes
class UnknownIdentifier(ParseError):
"""Exception raised when a {rev,file}set references an unknown identifier"""
def __init__(self, function, symbols):
from .i18n import _
ParseError.__init__(self, _(b"unknown identifier: %s") % function)
self.function = function
self.symbols = symbols
class RepoError(Hint, Exception):
__bytes__ = _tobytes
class RepoLookupError(RepoError):
pass
class FilteredRepoLookupError(RepoLookupError):
pass
class CapabilityError(RepoError):
pass
class RequirementError(RepoError):
"""Exception raised if .hg/requires has an unknown entry."""
class StdioError(IOError):
"""Raised if I/O to stdout or stderr fails"""
def __init__(self, err):
IOError.__init__(self, err.errno, err.strerror)
# no __bytes__() because error message is derived from the standard IOError
class UnsupportedMergeRecords(Abort):
def __init__(self, recordtypes):
from .i18n import _
self.recordtypes = sorted(recordtypes)
s = b' '.join(self.recordtypes)
Abort.__init__(
self,
_(b'unsupported merge state records: %s') % s,
hint=_(
b'see https://mercurial-scm.org/wiki/MergeStateRecords for '
b'more information'
),
)
class UnknownVersion(Abort):
"""generic exception for aborting from an encounter with an unknown version
"""
def __init__(self, msg, hint=None, version=None):
self.version = version
super(UnknownVersion, self).__init__(msg, hint=hint)
class LockError(IOError):
def __init__(self, errno, strerror, filename, desc):
IOError.__init__(self, errno, strerror, filename)
self.desc = desc
# no __bytes__() because error message is derived from the standard IOError
class LockHeld(LockError):
def __init__(self, errno, filename, desc, locker):
LockError.__init__(self, errno, b'Lock held', filename, desc)
self.locker = locker
class LockUnavailable(LockError):
pass
# LockError is for errors while acquiring the lock -- this is unrelated
class LockInheritanceContractViolation(RuntimeError):
__bytes__ = _tobytes
class ResponseError(Exception):
"""Raised to print an error with part of output and exit."""
__bytes__ = _tobytes
class UnknownCommand(Exception):
"""Exception raised if command is not in the command table."""
__bytes__ = _tobytes
class AmbiguousCommand(Exception):
"""Exception raised if command shortcut matches more than one command."""
__bytes__ = _tobytes
# derived from KeyboardInterrupt to simplify some breakout code
class SignalInterrupt(KeyboardInterrupt):
"""Exception raised on SIGTERM and SIGHUP."""
class SignatureError(Exception):
__bytes__ = _tobytes
class PushRaced(RuntimeError):
"""An exception raised during unbundling that indicate a push race"""
__bytes__ = _tobytes
class ProgrammingError(Hint, RuntimeError):
"""Raised if a mercurial (core or extension) developer made a mistake"""
def __init__(self, msg, *args, **kwargs):
# On Python 3, turn the message back into a string since this is
# an internal-only error that won't be printed except in a
# stack traces.
msg = pycompat.sysstr(msg)
super(ProgrammingError, self).__init__(msg, *args, **kwargs)
__bytes__ = _tobytes
class WdirUnsupported(Exception):
"""An exception which is raised when 'wdir()' is not supported"""
__bytes__ = _tobytes
# bundle2 related errors
class BundleValueError(ValueError):
"""error raised when bundle2 cannot be processed"""
__bytes__ = _tobytes
class BundleUnknownFeatureError(BundleValueError):
def __init__(self, parttype=None, params=(), values=()):
self.parttype = parttype
self.params = params
self.values = values
if self.parttype is None:
msg = b'Stream Parameter'
else:
msg = parttype
entries = self.params
if self.params and self.values:
assert len(self.params) == len(self.values)
entries = []
for idx, par in enumerate(self.params):
val = self.values[idx]
if val is None:
entries.append(val)
else:
entries.append(b"%s=%r" % (par, pycompat.maybebytestr(val)))
if entries:
msg = b'%s - %s' % (msg, b', '.join(entries))
ValueError.__init__(self, msg)
class ReadOnlyPartError(RuntimeError):
"""error raised when code tries to alter a part being generated"""
__bytes__ = _tobytes
class PushkeyFailed(Abort):
"""error raised when a pushkey part failed to update a value"""
def __init__(
self, partid, namespace=None, key=None, new=None, old=None, ret=None
):
self.partid = partid
self.namespace = namespace
self.key = key
self.new = new
self.old = old
self.ret = ret
# no i18n expected to be processed into a better message
Abort.__init__(
self, b'failed to update value for "%s/%s"' % (namespace, key)
)
class CensoredNodeError(StorageError):
"""error raised when content verification fails on a censored node
Also contains the tombstone data substituted for the uncensored data.
"""
def __init__(self, filename, node, tombstone):
from .node import short
StorageError.__init__(self, b'%s:%s' % (filename, short(node)))
self.tombstone = tombstone
class CensoredBaseError(StorageError):
"""error raised when a delta is rejected because its base is censored
A delta based on a censored revision must be formed as single patch
operation which replaces the entire base with new content. This ensures
the delta may be applied by clones which have not censored the base.
"""
class InvalidBundleSpecification(Exception):
"""error raised when a bundle specification is invalid.
This is used for syntax errors as opposed to support errors.
"""
__bytes__ = _tobytes
class UnsupportedBundleSpecification(Exception):
"""error raised when a bundle specification is not supported."""
__bytes__ = _tobytes
class CorruptedState(Exception):
"""error raised when a command is not able to read its state from file"""
__bytes__ = _tobytes
class PeerTransportError(Abort):
"""Transport-level I/O error when communicating with a peer repo."""
class InMemoryMergeConflictsError(Exception):
"""Exception raised when merge conflicts arose during an in-memory merge."""
__bytes__ = _tobytes
class WireprotoCommandError(Exception):
"""Represents an error during execution of a wire protocol command.
Should only be thrown by wire protocol version 2 commands.
The error is a formatter string and an optional iterable of arguments.
"""
def __init__(self, message, args=None):
self.message = message
self.messageargs = args