hgext/fsmonitor/pywatchman/pybser.py
author Kyle Lippincott <spectral@google.com>
Wed, 31 Mar 2021 12:46:54 -0700
changeset 46872 8bca353b1ebc
parent 43988 c8a51a90929b
child 48875 6000f5b25c9b
permissions -rw-r--r--
match: convert O(n) to O(log n) in exactmatcher.visitchildrenset When using narrow, during rebase this is called (at least) once per directory in the set of files in the commit being rebased. Every time it's called, we did the set arithmetic (now extracted and cached), which was probably pretty cheap but not necessary to repeat each time, looped over every item in the matcher and kept things that started with the directory we were querying. With very large narrowspecs, and a commit that touched a file in a large number of directories, this was slow. In a pathological repo, the rebase of a single commit (that touched over 17k files, I believe in approximately as many directories) with a narrowspec that had >32k entries took 8,246s of profiled time, with 5,007s of that spent in visitchildrenset (transitively). With this change, the time spent in visitchildrenset is less than 34s (which is where my profile cut off). Most of the remaining time was network access due to our custom remotefilelog-based setup not properly prefetching. Differential Revision: https://phab.mercurial-scm.org/D10294

# Copyright 2015 Facebook, Inc.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
#  * Redistributions of source code must retain the above copyright notice,
#    this list of conditions and the following disclaimer.
#
#  * Redistributions in binary form must reproduce the above copyright notice,
#    this list of conditions and the following disclaimer in the documentation
#    and/or other materials provided with the distribution.
#
#  * Neither the name Facebook nor the names of its contributors may be used to
#    endorse or promote products derived from this software without specific
#    prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

# no unicode literals
from __future__ import absolute_import, division, print_function

import binascii
import collections
import ctypes
import struct
import sys

from . import compat


BSER_ARRAY = b"\x00"
BSER_OBJECT = b"\x01"
BSER_BYTESTRING = b"\x02"
BSER_INT8 = b"\x03"
BSER_INT16 = b"\x04"
BSER_INT32 = b"\x05"
BSER_INT64 = b"\x06"
BSER_REAL = b"\x07"
BSER_TRUE = b"\x08"
BSER_FALSE = b"\x09"
BSER_NULL = b"\x0a"
BSER_TEMPLATE = b"\x0b"
BSER_SKIP = b"\x0c"
BSER_UTF8STRING = b"\x0d"

if compat.PYTHON3:
    STRING_TYPES = (str, bytes)
    unicode = str

    def tobytes(i):
        return str(i).encode("ascii")

    long = int
else:
    STRING_TYPES = (unicode, str)
    tobytes = bytes

# Leave room for the serialization header, which includes
# our overall length.  To make things simpler, we'll use an
# int32 for the header
EMPTY_HEADER = b"\x00\x01\x05\x00\x00\x00\x00"
EMPTY_HEADER_V2 = b"\x00\x02\x00\x00\x00\x00\x05\x00\x00\x00\x00"


def _int_size(x):
    """Return the smallest size int that can store the value"""
    if -0x80 <= x <= 0x7F:
        return 1
    elif -0x8000 <= x <= 0x7FFF:
        return 2
    elif -0x80000000 <= x <= 0x7FFFFFFF:
        return 4
    elif long(-0x8000000000000000) <= x <= long(0x7FFFFFFFFFFFFFFF):
        return 8
    else:
        raise RuntimeError("Cannot represent value: " + str(x))


def _buf_pos(buf, pos):
    ret = buf[pos]
    # Normalize the return type to bytes
    if compat.PYTHON3 and not isinstance(ret, bytes):
        ret = bytes((ret,))
    return ret


class _bser_buffer(object):
    def __init__(self, version):
        self.bser_version = version
        self.buf = ctypes.create_string_buffer(8192)
        if self.bser_version == 1:
            struct.pack_into(
                tobytes(len(EMPTY_HEADER)) + b"s", self.buf, 0, EMPTY_HEADER
            )
            self.wpos = len(EMPTY_HEADER)
        else:
            assert self.bser_version == 2
            struct.pack_into(
                tobytes(len(EMPTY_HEADER_V2)) + b"s",
                self.buf,
                0,
                EMPTY_HEADER_V2,
            )
            self.wpos = len(EMPTY_HEADER_V2)

    def ensure_size(self, size):
        while ctypes.sizeof(self.buf) - self.wpos < size:
            ctypes.resize(self.buf, ctypes.sizeof(self.buf) * 2)

    def append_long(self, val):
        size = _int_size(val)
        to_write = size + 1
        self.ensure_size(to_write)
        if size == 1:
            struct.pack_into(b"=cb", self.buf, self.wpos, BSER_INT8, val)
        elif size == 2:
            struct.pack_into(b"=ch", self.buf, self.wpos, BSER_INT16, val)
        elif size == 4:
            struct.pack_into(b"=ci", self.buf, self.wpos, BSER_INT32, val)
        elif size == 8:
            struct.pack_into(b"=cq", self.buf, self.wpos, BSER_INT64, val)
        else:
            raise RuntimeError("Cannot represent this long value")
        self.wpos += to_write

    def append_string(self, s):
        if isinstance(s, unicode):
            s = s.encode("utf-8")
        s_len = len(s)
        size = _int_size(s_len)
        to_write = 2 + size + s_len
        self.ensure_size(to_write)
        if size == 1:
            struct.pack_into(
                b"=ccb" + tobytes(s_len) + b"s",
                self.buf,
                self.wpos,
                BSER_BYTESTRING,
                BSER_INT8,
                s_len,
                s,
            )
        elif size == 2:
            struct.pack_into(
                b"=cch" + tobytes(s_len) + b"s",
                self.buf,
                self.wpos,
                BSER_BYTESTRING,
                BSER_INT16,
                s_len,
                s,
            )
        elif size == 4:
            struct.pack_into(
                b"=cci" + tobytes(s_len) + b"s",
                self.buf,
                self.wpos,
                BSER_BYTESTRING,
                BSER_INT32,
                s_len,
                s,
            )
        elif size == 8:
            struct.pack_into(
                b"=ccq" + tobytes(s_len) + b"s",
                self.buf,
                self.wpos,
                BSER_BYTESTRING,
                BSER_INT64,
                s_len,
                s,
            )
        else:
            raise RuntimeError("Cannot represent this string value")
        self.wpos += to_write

    def append_recursive(self, val):
        if isinstance(val, bool):
            needed = 1
            self.ensure_size(needed)
            if val:
                to_encode = BSER_TRUE
            else:
                to_encode = BSER_FALSE
            struct.pack_into(b"=c", self.buf, self.wpos, to_encode)
            self.wpos += needed
        elif val is None:
            needed = 1
            self.ensure_size(needed)
            struct.pack_into(b"=c", self.buf, self.wpos, BSER_NULL)
            self.wpos += needed
        elif isinstance(val, (int, long)):
            self.append_long(val)
        elif isinstance(val, STRING_TYPES):
            self.append_string(val)
        elif isinstance(val, float):
            needed = 9
            self.ensure_size(needed)
            struct.pack_into(b"=cd", self.buf, self.wpos, BSER_REAL, val)
            self.wpos += needed
        elif isinstance(val, collections.Mapping) and isinstance(
            val, collections.Sized
        ):
            val_len = len(val)
            size = _int_size(val_len)
            needed = 2 + size
            self.ensure_size(needed)
            if size == 1:
                struct.pack_into(
                    b"=ccb",
                    self.buf,
                    self.wpos,
                    BSER_OBJECT,
                    BSER_INT8,
                    val_len,
                )
            elif size == 2:
                struct.pack_into(
                    b"=cch",
                    self.buf,
                    self.wpos,
                    BSER_OBJECT,
                    BSER_INT16,
                    val_len,
                )
            elif size == 4:
                struct.pack_into(
                    b"=cci",
                    self.buf,
                    self.wpos,
                    BSER_OBJECT,
                    BSER_INT32,
                    val_len,
                )
            elif size == 8:
                struct.pack_into(
                    b"=ccq",
                    self.buf,
                    self.wpos,
                    BSER_OBJECT,
                    BSER_INT64,
                    val_len,
                )
            else:
                raise RuntimeError("Cannot represent this mapping value")
            self.wpos += needed
            if compat.PYTHON3:
                iteritems = val.items()
            else:
                iteritems = val.iteritems()  # noqa: B301 Checked version above
            for k, v in iteritems:
                self.append_string(k)
                self.append_recursive(v)
        elif isinstance(val, collections.Iterable) and isinstance(
            val, collections.Sized
        ):
            val_len = len(val)
            size = _int_size(val_len)
            needed = 2 + size
            self.ensure_size(needed)
            if size == 1:
                struct.pack_into(
                    b"=ccb", self.buf, self.wpos, BSER_ARRAY, BSER_INT8, val_len
                )
            elif size == 2:
                struct.pack_into(
                    b"=cch",
                    self.buf,
                    self.wpos,
                    BSER_ARRAY,
                    BSER_INT16,
                    val_len,
                )
            elif size == 4:
                struct.pack_into(
                    b"=cci",
                    self.buf,
                    self.wpos,
                    BSER_ARRAY,
                    BSER_INT32,
                    val_len,
                )
            elif size == 8:
                struct.pack_into(
                    b"=ccq",
                    self.buf,
                    self.wpos,
                    BSER_ARRAY,
                    BSER_INT64,
                    val_len,
                )
            else:
                raise RuntimeError("Cannot represent this sequence value")
            self.wpos += needed
            for v in val:
                self.append_recursive(v)
        else:
            raise RuntimeError("Cannot represent unknown value type")


def dumps(obj, version=1, capabilities=0):
    bser_buf = _bser_buffer(version=version)
    bser_buf.append_recursive(obj)
    # Now fill in the overall length
    if version == 1:
        obj_len = bser_buf.wpos - len(EMPTY_HEADER)
        struct.pack_into(b"=i", bser_buf.buf, 3, obj_len)
    else:
        obj_len = bser_buf.wpos - len(EMPTY_HEADER_V2)
        struct.pack_into(b"=i", bser_buf.buf, 2, capabilities)
        struct.pack_into(b"=i", bser_buf.buf, 7, obj_len)
    return bser_buf.buf.raw[: bser_buf.wpos]


# This is a quack-alike with the bserObjectType in bser.c
# It provides by getattr accessors and getitem for both index
# and name.
class _BunserDict(object):
    __slots__ = ("_keys", "_values")

    def __init__(self, keys, values):
        self._keys = keys
        self._values = values

    def __getattr__(self, name):
        return self.__getitem__(name)

    def __getitem__(self, key):
        if isinstance(key, (int, long)):
            return self._values[key]
        elif key.startswith("st_"):
            # hack^Wfeature to allow mercurial to use "st_size" to
            # reference "size"
            key = key[3:]
        try:
            return self._values[self._keys.index(key)]
        except ValueError:
            raise KeyError("_BunserDict has no key %s" % key)

    def __len__(self):
        return len(self._keys)


class Bunser(object):
    def __init__(self, mutable=True, value_encoding=None, value_errors=None):
        self.mutable = mutable
        self.value_encoding = value_encoding

        if value_encoding is None:
            self.value_errors = None
        elif value_errors is None:
            self.value_errors = "strict"
        else:
            self.value_errors = value_errors

    @staticmethod
    def unser_int(buf, pos):
        try:
            int_type = _buf_pos(buf, pos)
        except IndexError:
            raise ValueError("Invalid bser int encoding, pos out of range")
        if int_type == BSER_INT8:
            needed = 2
            fmt = b"=b"
        elif int_type == BSER_INT16:
            needed = 3
            fmt = b"=h"
        elif int_type == BSER_INT32:
            needed = 5
            fmt = b"=i"
        elif int_type == BSER_INT64:
            needed = 9
            fmt = b"=q"
        else:
            raise ValueError(
                "Invalid bser int encoding 0x%s at position %s"
                % (binascii.hexlify(int_type).decode("ascii"), pos)
            )
        int_val = struct.unpack_from(fmt, buf, pos + 1)[0]
        return (int_val, pos + needed)

    def unser_utf8_string(self, buf, pos):
        str_len, pos = self.unser_int(buf, pos + 1)
        str_val = struct.unpack_from(tobytes(str_len) + b"s", buf, pos)[0]
        return (str_val.decode("utf-8"), pos + str_len)

    def unser_bytestring(self, buf, pos):
        str_len, pos = self.unser_int(buf, pos + 1)
        str_val = struct.unpack_from(tobytes(str_len) + b"s", buf, pos)[0]
        if self.value_encoding is not None:
            str_val = str_val.decode(self.value_encoding, self.value_errors)
            # str_len stays the same because that's the length in bytes
        return (str_val, pos + str_len)

    def unser_array(self, buf, pos):
        arr_len, pos = self.unser_int(buf, pos + 1)
        arr = []
        for _ in range(arr_len):
            arr_item, pos = self.loads_recursive(buf, pos)
            arr.append(arr_item)

        if not self.mutable:
            arr = tuple(arr)

        return arr, pos

    def unser_object(self, buf, pos):
        obj_len, pos = self.unser_int(buf, pos + 1)
        if self.mutable:
            obj = {}
        else:
            keys = []
            vals = []

        for _ in range(obj_len):
            key, pos = self.unser_utf8_string(buf, pos)
            val, pos = self.loads_recursive(buf, pos)
            if self.mutable:
                obj[key] = val
            else:
                keys.append(key)
                vals.append(val)

        if not self.mutable:
            obj = _BunserDict(keys, vals)

        return obj, pos

    def unser_template(self, buf, pos):
        val_type = _buf_pos(buf, pos + 1)
        if val_type != BSER_ARRAY:
            raise RuntimeError("Expect ARRAY to follow TEMPLATE")
        # force UTF-8 on keys
        keys_bunser = Bunser(mutable=self.mutable, value_encoding="utf-8")
        keys, pos = keys_bunser.unser_array(buf, pos + 1)
        nitems, pos = self.unser_int(buf, pos)
        arr = []
        for _ in range(nitems):
            if self.mutable:
                obj = {}
            else:
                vals = []

            for keyidx in range(len(keys)):
                if _buf_pos(buf, pos) == BSER_SKIP:
                    pos += 1
                    ele = None
                else:
                    ele, pos = self.loads_recursive(buf, pos)

                if self.mutable:
                    key = keys[keyidx]
                    obj[key] = ele
                else:
                    vals.append(ele)

            if not self.mutable:
                obj = _BunserDict(keys, vals)

            arr.append(obj)
        return arr, pos

    def loads_recursive(self, buf, pos):
        val_type = _buf_pos(buf, pos)
        if (
            val_type == BSER_INT8
            or val_type == BSER_INT16
            or val_type == BSER_INT32
            or val_type == BSER_INT64
        ):
            return self.unser_int(buf, pos)
        elif val_type == BSER_REAL:
            val = struct.unpack_from(b"=d", buf, pos + 1)[0]
            return (val, pos + 9)
        elif val_type == BSER_TRUE:
            return (True, pos + 1)
        elif val_type == BSER_FALSE:
            return (False, pos + 1)
        elif val_type == BSER_NULL:
            return (None, pos + 1)
        elif val_type == BSER_BYTESTRING:
            return self.unser_bytestring(buf, pos)
        elif val_type == BSER_UTF8STRING:
            return self.unser_utf8_string(buf, pos)
        elif val_type == BSER_ARRAY:
            return self.unser_array(buf, pos)
        elif val_type == BSER_OBJECT:
            return self.unser_object(buf, pos)
        elif val_type == BSER_TEMPLATE:
            return self.unser_template(buf, pos)
        else:
            raise ValueError(
                "unhandled bser opcode 0x%s"
                % binascii.hexlify(val_type).decode("ascii")
            )


def _pdu_info_helper(buf):
    if buf[0:2] == EMPTY_HEADER[0:2]:
        bser_version = 1
        bser_capabilities = 0
        expected_len, pos2 = Bunser.unser_int(buf, 2)
    elif buf[0:2] == EMPTY_HEADER_V2[0:2]:
        if len(buf) < 8:
            raise ValueError("Invalid BSER header")
        bser_version = 2
        bser_capabilities = struct.unpack_from("I", buf, 2)[0]
        expected_len, pos2 = Bunser.unser_int(buf, 6)
    else:
        raise ValueError("Invalid BSER header")

    return bser_version, bser_capabilities, expected_len, pos2


def pdu_info(buf):
    info = _pdu_info_helper(buf)
    return info[0], info[1], info[2] + info[3]


def pdu_len(buf):
    info = _pdu_info_helper(buf)
    return info[2] + info[3]


def loads(buf, mutable=True, value_encoding=None, value_errors=None):
    """Deserialize a BSER-encoded blob.

    @param buf: The buffer to deserialize.
    @type buf: bytes

    @param mutable: Whether to return mutable results.
    @type mutable: bool

    @param value_encoding: Optional codec to use to decode values. If
                           unspecified or None, return values as bytestrings.
    @type value_encoding: str

    @param value_errors: Optional error handler for codec. 'strict' by default.
                         The other most common argument is 'surrogateescape' on
                         Python 3. If value_encoding is None, this is ignored.
    @type value_errors: str
    """

    info = _pdu_info_helper(buf)
    expected_len = info[2]
    pos = info[3]

    if len(buf) != expected_len + pos:
        raise ValueError(
            "bser data len %d != header len %d" % (expected_len + pos, len(buf))
        )

    bunser = Bunser(
        mutable=mutable,
        value_encoding=value_encoding,
        value_errors=value_errors,
    )

    return bunser.loads_recursive(buf, pos)[0]


def load(fp, mutable=True, value_encoding=None, value_errors=None):
    from . import load

    return load.load(fp, mutable, value_encoding, value_errors)