File size: 9,774 Bytes
9c6594c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
"""leveldb log datastore.

Format is described at:
    https://github.com/google/leveldb/blob/master/doc/log_format.md

block := record* trailer?
record :=
  checksum: uint32     // crc32c of type and data[] ; little-endian
  length: uint16       // little-endian
  type: uint8          // One of FULL, FIRST, MIDDLE, LAST
  data: uint8[length]

header :=
  ident: char[4]
  magic: uint16
  version: uint8
"""

# TODO: possibly restructure code by porting the C++ or go implementation

import logging
import os
import struct
import zlib
from typing import TYPE_CHECKING, Optional, Tuple

import wandb

if TYPE_CHECKING:
    from typing import IO, Any

    from wandb.proto.wandb_internal_pb2 import Record

logger = logging.getLogger(__name__)

LEVELDBLOG_HEADER_LEN = 7
LEVELDBLOG_BLOCK_LEN = 32768
LEVELDBLOG_DATA_LEN = LEVELDBLOG_BLOCK_LEN - LEVELDBLOG_HEADER_LEN

LEVELDBLOG_FULL = 1
LEVELDBLOG_FIRST = 2
LEVELDBLOG_MIDDLE = 3
LEVELDBLOG_LAST = 4

LEVELDBLOG_HEADER_IDENT = ":W&B"
LEVELDBLOG_HEADER_MAGIC = (
    0xBEE1  # zlib.crc32(bytes("Weights & Biases", 'iso8859-1')) & 0xffff
)
LEVELDBLOG_HEADER_VERSION = 0

try:
    bytes("", "ascii")

    def strtobytes(x):
        """Strtobytes."""
        return bytes(x, "iso8859-1")

    # def bytestostr(x):
    #     return str(x, 'iso8859-1')

except Exception:
    strtobytes = str
    # bytestostr = str


class DataStore:
    _index: int
    _flush_offset: int

    def __init__(self) -> None:
        self._opened_for_scan = False
        self._fp: Optional[IO[Any]] = None
        self._index = 0
        self._flush_offset = 0
        self._size_bytes = 0

        self._crc = [0] * (LEVELDBLOG_LAST + 1)
        for x in range(1, LEVELDBLOG_LAST + 1):
            self._crc[x] = zlib.crc32(strtobytes(chr(x))) & 0xFFFFFFFF

        assert (
            wandb._assert_is_internal_process  # type: ignore
        ), "DataStore can only be used in the internal process"

    def open_for_write(self, fname: str) -> None:
        self._fname = fname
        logger.info("open: %s", fname)
        open_flags = "xb"
        self._fp = open(fname, open_flags)
        self._write_header()

    def open_for_append(self, fname):
        # TODO: implement
        self._fname = fname
        logger.info("open: %s", fname)
        self._fp = open(fname, "wb")
        # do something with _index

    def open_for_scan(self, fname):
        self._fname = fname
        logger.info("open for scan: %s", fname)
        self._fp = open(fname, "r+b")
        self._index = 0
        self._size_bytes = os.stat(fname).st_size
        self._opened_for_scan = True
        self._read_header()

    def seek(self, offset: int) -> None:
        self._fp.seek(offset)  # type: ignore
        self._index = offset

    def get_offset(self) -> int:
        offset = self._fp.tell()  # type: ignore
        return offset

    def in_last_block(self):
        """Determine if we're in the last block to handle in-progress writes."""
        return self._index > self._size_bytes - LEVELDBLOG_DATA_LEN

    def scan_record(self):
        assert self._opened_for_scan, "file not open for scanning"
        # TODO(jhr): handle some assertions as file corruption issues
        # assume we have enough room to read header, checked by caller?
        header = self._fp.read(LEVELDBLOG_HEADER_LEN)
        if len(header) == 0:
            return None
        assert (
            len(header) == LEVELDBLOG_HEADER_LEN
        ), f"record header is {len(header)} bytes instead of the expected {LEVELDBLOG_HEADER_LEN}"
        fields = struct.unpack("<IHB", header)
        checksum, dlength, dtype = fields
        # check len, better fit in the block
        self._index += LEVELDBLOG_HEADER_LEN
        data = self._fp.read(dlength)
        checksum_computed = zlib.crc32(data, self._crc[dtype]) & 0xFFFFFFFF
        assert (
            checksum == checksum_computed
        ), "record checksum is invalid, data may be corrupt"
        self._index += dlength
        return dtype, data

    def scan_data(self):
        # TODO(jhr): handle some assertions as file corruption issues
        # how much left in the block.  if less than header len, read as pad,
        offset = self._index % LEVELDBLOG_BLOCK_LEN
        space_left = LEVELDBLOG_BLOCK_LEN - offset
        if space_left < LEVELDBLOG_HEADER_LEN:
            pad_check = strtobytes("\x00" * space_left)
            pad = self._fp.read(space_left)
            # verify they are zero
            assert pad == pad_check, "invalid padding"
            self._index += space_left

        record = self.scan_record()
        if record is None:  # eof
            return None
        dtype, data = record
        if dtype == LEVELDBLOG_FULL:
            return data

        assert (
            dtype == LEVELDBLOG_FIRST
        ), f"expected record to be type {LEVELDBLOG_FIRST} but found {dtype}"
        while True:
            offset = self._index % LEVELDBLOG_BLOCK_LEN
            record = self.scan_record()
            if record is None:  # eof
                return None
            dtype, new_data = record
            if dtype == LEVELDBLOG_LAST:
                data += new_data
                break
            assert (
                dtype == LEVELDBLOG_MIDDLE
            ), f"expected record to be type {LEVELDBLOG_MIDDLE} but found {dtype}"
            data += new_data
        return data

    def _write_header(self):
        data = struct.pack(
            "<4sHB",
            strtobytes(LEVELDBLOG_HEADER_IDENT),
            LEVELDBLOG_HEADER_MAGIC,
            LEVELDBLOG_HEADER_VERSION,
        )
        assert (
            len(data) == LEVELDBLOG_HEADER_LEN
        ), f"header size is {len(data)} bytes, expected {LEVELDBLOG_HEADER_LEN}"
        self._fp.write(data)
        self._index += len(data)

    def _read_header(self):
        header = self._fp.read(LEVELDBLOG_HEADER_LEN)
        assert (
            len(header) == LEVELDBLOG_HEADER_LEN
        ), f"header is {len(header)} bytes instead of the expected {LEVELDBLOG_HEADER_LEN}"
        ident, magic, version = struct.unpack("<4sHB", header)
        if ident != strtobytes(LEVELDBLOG_HEADER_IDENT):
            raise Exception("Invalid header")
        if magic != LEVELDBLOG_HEADER_MAGIC:
            raise Exception("Invalid header")
        if version != LEVELDBLOG_HEADER_VERSION:
            raise Exception("Invalid header")
        self._index += len(header)

    def _write_record(self, s, dtype=None):
        """Write record that must fit into a block."""
        # double check that there is enough space
        # (this is a precondition to calling this method)
        assert len(s) + LEVELDBLOG_HEADER_LEN <= (
            LEVELDBLOG_BLOCK_LEN - self._index % LEVELDBLOG_BLOCK_LEN
        ), "not enough space to write new records"

        dlength = len(s)
        dtype = dtype or LEVELDBLOG_FULL
        # print("record: length={} type={}".format(dlength, dtype))
        checksum = zlib.crc32(s, self._crc[dtype]) & 0xFFFFFFFF
        # logger.info("write_record: index=%d len=%d dtype=%d",
        #     self._index, dlength, dtype)
        self._fp.write(struct.pack("<IHB", checksum, dlength, dtype))
        if dlength:
            self._fp.write(s)
        self._index += LEVELDBLOG_HEADER_LEN + len(s)

    def _write_data(self, s):
        start_offset = self._index

        offset = self._index % LEVELDBLOG_BLOCK_LEN
        space_left = LEVELDBLOG_BLOCK_LEN - offset
        data_used = 0
        data_left = len(s)
        # logger.info("write_data: index=%d offset=%d len=%d",
        #     self._index, offset, data_left)
        if space_left < LEVELDBLOG_HEADER_LEN:
            pad = "\x00" * space_left
            self._fp.write(strtobytes(pad))
            self._index += space_left
            offset = 0
            space_left = LEVELDBLOG_BLOCK_LEN

        # does it fit in first (possibly partial) block?
        if data_left + LEVELDBLOG_HEADER_LEN <= space_left:
            self._write_record(s)
        else:
            # write first record (we could still be in the middle of a block,
            # but this write will end on a block boundary)
            data_room = space_left - LEVELDBLOG_HEADER_LEN
            self._write_record(s[:data_room], LEVELDBLOG_FIRST)
            data_used += data_room
            data_left -= data_room
            assert data_left, "data_left should be non-zero"

            # write middles (if any)
            while data_left > LEVELDBLOG_DATA_LEN:
                self._write_record(
                    s[data_used : data_used + LEVELDBLOG_DATA_LEN],
                    LEVELDBLOG_MIDDLE,
                )
                data_used += LEVELDBLOG_DATA_LEN
                data_left -= LEVELDBLOG_DATA_LEN

            # write last and flush the entire block to disk
            self._write_record(s[data_used:], LEVELDBLOG_LAST)
            self._fp.flush()
            os.fsync(self._fp.fileno())
            self._flush_offset = self._index

        return start_offset, self._index, self._flush_offset

    def ensure_flushed(self, off: int) -> None:
        self._fp.flush()  # type: ignore

    def write(self, obj: "Record") -> Tuple[int, int, int]:
        """Write a protocol buffer.

        Args:
            obj: Protocol buffer to write.

        Returns:
            (start_offset, end_offset, flush_offset) if successful,
            None otherwise

        """
        raw_size = obj.ByteSize()
        s = obj.SerializeToString()
        assert len(s) == raw_size, "invalid serialization"
        ret = self._write_data(s)
        return ret

    def close(self) -> None:
        if self._fp is not None:
            logger.info("close: %s", self._fname)
            self._fp.close()