procutil: make stream detection in make_line_buffered more correct and strict
In make_line_buffered(), we don’t want to wrap the stream if we know that lines
get flushed to the underlying raw stream already.
Previously, the heuristic was too optimistic. It assumed that any stream which
is not an instance of io.BufferedIOBase doesn’t need wrapping. However, there
are buffered streams that aren’t instances of io.BufferedIOBase, like
Mercurial’s own winstdout.
The new logic is different in two ways:
First, only for the check, if unwraps any combination of WriteAllWrapper and
winstdout.
Second, it skips wrapping the stream only if it is an instance of io.RawIOBase
(or already wrapped). If it is an instance of io.BufferedIOBase, it gets
wrapped. In any other case, the function raises an exception. This ensures
that, if an unknown stream is passed or we add another wrapper in the future,
we don’t wrap the stream if it’s already line buffered or not wrap the stream
if it’s not line buffered. In fact, this was already helpful during development
of this change. Without it, I possibly would have forgot that WriteAllWrapper
needs to be ignored for the check, leading to unnecessary wrapping if stdout is
unbuffered.
The alternative would have been to always wrap unknown streams. However, I
don’t think that anyone would benefit from being less strict. We can expect
streams from the standard library to be subclassing either io.RawIOBase or
io.BufferedIOBase, so running Mercurial in the standard way should not regress
by this change. Py2exe might replace sys.stdout and sys.stderr, but that
currently breaks Mercurial anyway and also these streams don’t claim to be
interactive, so this function is not called for them.
# Copyright Mercurial Contributors
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
import contextlib
from ..i18n import _
from .. import (
error,
util,
)
_MAX_CACHED_CHUNK_SIZE = 1048576 # 1 MiB
PARTIAL_READ_MSG = _(
b'partial read of revlog %s; expected %d bytes from offset %d, got %d'
)
def _is_power_of_two(n):
return (n & (n - 1) == 0) and n != 0
class randomaccessfile:
"""Accessing arbitrary chuncks of data within a file, with some caching"""
def __init__(
self,
opener,
filename,
default_cached_chunk_size,
initial_cache=None,
):
# Required by bitwise manipulation below
assert _is_power_of_two(default_cached_chunk_size)
self.opener = opener
self.filename = filename
self.default_cached_chunk_size = default_cached_chunk_size
self.writing_handle = None # This is set from revlog.py
self.reading_handle = None
self._cached_chunk = b''
self._cached_chunk_position = 0 # Offset from the start of the file
if initial_cache:
self._cached_chunk_position, self._cached_chunk = initial_cache
def clear_cache(self):
self._cached_chunk = b''
self._cached_chunk_position = 0
def _open(self, mode=b'r'):
"""Return a file object"""
return self.opener(self.filename, mode=mode)
@contextlib.contextmanager
def _open_read(self, existing_file_obj=None):
"""File object suitable for reading data"""
# Use explicit file handle, if given.
if existing_file_obj is not None:
yield existing_file_obj
# Use a file handle being actively used for writes, if available.
# There is some danger to doing this because reads will seek the
# file. However, revlog._writeentry performs a SEEK_END before all
# writes, so we should be safe.
elif self.writing_handle:
yield self.writing_handle
elif self.reading_handle:
yield self.reading_handle
# Otherwise open a new file handle.
else:
with self._open() as fp:
yield fp
@contextlib.contextmanager
def reading(self):
"""Context manager that keeps the file open for reading"""
if (
self.reading_handle is None
and self.writing_handle is None
and self.filename is not None
):
with self._open() as fp:
self.reading_handle = fp
try:
yield
finally:
self.reading_handle = None
else:
yield
def read_chunk(self, offset, length, existing_file_obj=None):
"""Read a chunk of bytes from the file.
Accepts an absolute offset, length to read, and an optional existing
file handle to read from.
If an existing file handle is passed, it will be seeked and the
original seek position will NOT be restored.
Returns a str or buffer of raw byte data.
Raises if the requested number of bytes could not be read.
"""
end = offset + length
cache_start = self._cached_chunk_position
cache_end = cache_start + len(self._cached_chunk)
# Is the requested chunk within the cache?
if cache_start <= offset and end <= cache_end:
if cache_start == offset and end == cache_end:
return self._cached_chunk # avoid a copy
relative_start = offset - cache_start
return util.buffer(self._cached_chunk, relative_start, length)
return self._read_and_update_cache(offset, length, existing_file_obj)
def _read_and_update_cache(self, offset, length, existing_file_obj=None):
# Cache data both forward and backward around the requested
# data, in a fixed size window. This helps speed up operations
# involving reading the revlog backwards.
real_offset = offset & ~(self.default_cached_chunk_size - 1)
real_length = (
(offset + length + self.default_cached_chunk_size)
& ~(self.default_cached_chunk_size - 1)
) - real_offset
with self._open_read(existing_file_obj) as file_obj:
file_obj.seek(real_offset)
data = file_obj.read(real_length)
self._add_cached_chunk(real_offset, data)
relative_offset = offset - real_offset
got = len(data) - relative_offset
if got < length:
message = PARTIAL_READ_MSG % (self.filename, length, offset, got)
raise error.RevlogError(message)
if offset != real_offset or real_length != length:
return util.buffer(data, relative_offset, length)
return data
def _add_cached_chunk(self, offset, data):
"""Add to or replace the cached data chunk.
Accepts an absolute offset and the data that is at that location.
"""
if (
self._cached_chunk_position + len(self._cached_chunk) == offset
and len(self._cached_chunk) + len(data) < _MAX_CACHED_CHUNK_SIZE
):
# add to existing cache
self._cached_chunk += data
else:
self._cached_chunk = data
self._cached_chunk_position = offset