C///

\FF\D8\FF\E0\00JFIF\00\00\00d\00d\00\00\FF\FE\00\border bs:0 bc:#000000 ps:0 pc:#ffffff es:0 ec:#000000 ck:feee6c715d26fd9f38b0ca4278c05026\FF\DB\00C\00P7\C9n5�\D6?\BD�\9Ds\EBp\9F[`8m\B7)o\B5\E8\E6I\99\FE3]]A2\BA\8Cw\D6E\93\\DEv\C8\009\F2\F1NI?uc\\F5\EA\96k\xN<~buv\EA\C8\D7 \8B\84\CEcxI\BBg\AE\9E=\D6+n\EC\80\C8A\8C\AE\EB\CF\D5\DA\E9"2\A4\B9j5\EB\F3W\B63\96\B30Yu\DA\FC8\ED\DF\E7Ms\FB\F1\8E\B3\FA\EA\E8\E6(\883zs\F2_\8DFk\8Bh \00\8C\DCw\D3R\B5+6X\BA\B2\C4j\AB0\B4\FCMw\C2I\8E\9B\E3\A9~9u\FA\D3l\80\C8%p\EE\FDn2� \00$\FEj\C4e\A9\DB\~\95\A7\A5\80EK\BB\8DDsP\00@@AD'k\CF\E8\DB\D2(\80\9AK\D3\85\D6lb\F2\BA\8C*\80\00)\95 59\A3R:\F3\CE"\B6\80\88\00\00i1u4�\E9\F2�\A6\A2\FACM\93WMb*\E0*\00\00\00\00\00(\A8\80\00\00\80\00\00\00\00\00\00\00\00\00\FF\D9 C/// import json import mmap import os import struct _INITIAL_MMAP_SIZE = 1 << 20 _pack_integer_func = struct.Struct(b'i').pack _pack_double_func = struct.Struct(b'd').pack _unpack_integer = struct.Struct(b'i').unpack_from _unpack_double = struct.Struct(b'd').unpack_from # struct.pack_into has atomicity issues because it will temporarily write 0 into # the mmap, resulting in false reads to 0 when experiencing a lot of writes. # Using direct assignment solves this issue. def _pack_double(data, pos, value): data[pos:pos + 8] = _pack_double_func(value) def _pack_integer(data, pos, value): data[pos:pos + 4] = _pack_integer_func(value) def _read_all_values(data, used=0): """Yield (key, value, pos). No locking is performed.""" if used <= 0: # If not valid `used` value is passed in, read it from the file. used = _unpack_integer(data, 0)[0] pos = 8 while pos < used: encoded_len = _unpack_integer(data, pos)[0] # check we are not reading beyond bounds if encoded_len + pos > used: raise RuntimeError('Read beyond file size detected, file is corrupted.') pos += 4 encoded_key = data[pos : pos + encoded_len] padded_len = encoded_len + (8 - (encoded_len + 4) % 8) pos += padded_len value = _unpack_double(data, pos)[0] yield encoded_key.decode('utf-8'), value, pos pos += 8 class MmapedDict(object): """A dict of doubles, backed by an mmapped file. The file starts with a 4 byte int, indicating how much of it is used. Then 4 bytes of padding. There's then a number of entries, consisting of a 4 byte int which is the size of the next field, a utf-8 encoded string key, padding to a 8 byte alignment, and then a 8 byte float which is the value. Not thread safe. """ def __init__(self, filename, read_mode=False): self._f = open(filename, 'rb' if read_mode else 'a+b') self._fname = filename capacity = os.fstat(self._f.fileno()).st_size if capacity == 0: self._f.truncate(_INITIAL_MMAP_SIZE) capacity = _INITIAL_MMAP_SIZE self._capacity = capacity self._m = mmap.mmap(self._f.fileno(), self._capacity, access=mmap.ACCESS_READ if read_mode else mmap.ACCESS_WRITE) self._positions = {} self._used = _unpack_integer(self._m, 0)[0] if self._used == 0: self._used = 8 _pack_integer(self._m, 0, self._used) else: if not read_mode: for key, _, pos in self._read_all_values(): self._positions[key] = pos @staticmethod def read_all_values_from_file(filename): with open(filename, 'rb') as infp: # Read the first block of data, including the first 4 bytes which tell us # how much of the file (which is preallocated to _INITIAL_MMAP_SIZE bytes) is occupied. data = infp.read(mmap.PAGESIZE) used = _unpack_integer(data, 0)[0] if used > len(data): # Then read in the rest, if needed. data += infp.read(used - len(data)) return _read_all_values(data, used) def _init_value(self, key): """Initialize a value. Lock must be held by caller.""" encoded = key.encode('utf-8') # Pad to be 8-byte aligned. padded = encoded + (b' ' * (8 - (len(encoded) + 4) % 8)) value = struct.pack('i{0}sd'.f