Class: MysqlBinlog::BinlogFieldParser

Inherits:
Object
  • Object
show all
Defined in:
lib/mysql_binlog/binlog_field_parser.rb

Overview

Parse various types of standard and non-standard data types from a provided binary log using its reader to read data.

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(binlog_instance) ⇒ BinlogFieldParser 



51
52
53
54
55
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 51

def initialize(binlog_instance)
  @format_cache = {}
  @binlog = binlog_instance
  @reader = binlog_instance.reader
end

Instance Attribute Details

#binlog ⇒ Object

Returns the value of attribute binlog.



48
49
50
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 48

def binlog
  @binlog
end

#reader ⇒ Object

Returns the value of attribute reader.



49
50
51
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 49

def reader
  @reader
end

Instance Method Details

#convert_mysql_type_date(value) ⇒ Object

Convert a packed DATE from a uint24 into a string representing the date.



361
362
363
364
365
366
367
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 361

def convert_mysql_type_date(value)
  "%04i-%02i-%02i" % [
    extract_bits(value, 15, 9),
    extract_bits(value,  4, 5),
    extract_bits(value,  5, 0),
  ]
end

#convert_mysql_type_datetime(value) ⇒ Object

Convert a packed DATETIME from a uint64 into a string representing the date and time.



381
382
383
384
385
386
387
388
389
390
391
392
393
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 381

def convert_mysql_type_datetime(value)
  date = value / 1000000
  time = value % 1000000

  "%04i-%02i-%02i %02i:%02i:%02i" % [
    date / 10000,
    (date % 10000) / 100,
    date % 100,
    time / 10000,
    (time % 10000) / 100,
    time % 100,
  ]
end

#convert_mysql_type_datetimef(int_part, frac_part) ⇒ Object 



395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 395

def convert_mysql_type_datetimef(int_part, frac_part)
  year_month = extract_bits(int_part, 17, 22)
  year = year_month / 13
  month = year_month % 13
  day = extract_bits(int_part, 5, 17)
  hour = extract_bits(int_part, 5, 12)
  minute = extract_bits(int_part, 6, 6)
  second = extract_bits(int_part, 6, 0)

  "%04i-%02i-%02i %02i:%02i:%02i.%06i" % [
    year,
    month,
    day,
    hour,
    minute,
    second,
    frac_part,
  ]
end

#convert_mysql_type_time(value) ⇒ Object

Convert a packed TIME from a uint24 into a string representing the time.



371
372
373
374
375
376
377
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 371

def convert_mysql_type_time(value)
  "%02i:%02i:%02i" % [
    value / 10000,
    (value % 10000) / 100,
    value % 100,
  ]
end

#extract_bits(value, bits, offset) ⇒ Object

Extract a number of sequential bits at a given offset within an integer. This is used to unpack bit-packed fields.



355
356
357
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 355

def extract_bits(value, bits, offset)
  (value & ((1 << bits) - 1) << offset) >> offset
end

#read_bit_array(length) ⇒ Object

Read an arbitrary-length bitmap, provided its length. Returns an array of true/false values. This is used both for internal usage in RBR events that need bitmaps, as well as for the BIT type.



319
320
321
322
323
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 319

def read_bit_array(length)
  data = reader.read((length+7)/8)
  data.unpack("b*").first.  # Unpack into a string of "10101"
    split("").map { |c| c == "1" }.shift(length) # Return true/false array
end

#read_datetimef(decimals) ⇒ Object 



428
429
430
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 428

def read_datetimef(decimals)
  convert_mysql_type_datetimef(read_uint40_be, read_frac_part(decimals))
end

#read_double ⇒ Object

Read a double-precision (8-byte) floating point number.



195
196
197
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 195

def read_double
  reader.read(8).unpack("E").first
end

#read_float ⇒ Object

Read a single-precision (4-byte) floating point number.



190
191
192
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 190

def read_float
  reader.read(4).unpack("e").first
end

#read_frac_part(decimals) ⇒ Object 



415
416
417
418
419
420
421
422
423
424
425
426
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 415

def read_frac_part(decimals)
  case decimals
  when 0
    0
  when 1, 2
    read_uint8 * 10000
  when 3, 4
    read_uint16_be * 100
  when 5, 6
    read_uint24_be
  end
end

#read_int16_be ⇒ Object

Read a signed 16-bit (2-byte) big-endian integer.



134
135
136
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 134

def read_int16_be
  reader.read(2).unpack('n').first
end

#read_int24_be ⇒ Object

Read a signed 24-bit (3-byte) big-endian integer.



139
140
141
142
143
144
145
146
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 139

def read_int24_be
  a, b, c = reader.read(3).unpack('CCC')
  if (a & 128) == 0
    (a << 16) | (b << 8) | c
  else
    (-1 << 24) | (a << 16) | (b << 8) | c
  end
end

#read_int32_be ⇒ Object

Read a signed 32-bit (4-byte) big-endian integer.



149
150
151
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 149

def read_int32_be
  reader.read(4).unpack('N').first
end

#read_int8 ⇒ Object

Read a signed 8-bit (1-byte) integer.



129
130
131
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 129

def read_int8
  reader.read(1).unpack("c").first
end

#read_int_be_by_size(size) ⇒ Object 



174
175
176
177
178
179
180
181
182
183
184
185
186
187
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 174

def read_int_be_by_size(size)
  case size
  when 1
    read_int8
  when 2
    read_int16_be
  when 3
    read_int24_be
  when 4
    read_int32_be
  else
    raise "read_int#{size*8}_be not implemented"
  end
end

#read_lpstring(size = 1) ⇒ Object

Read a (Pascal-style) length-prefixed string. The length is stored as a 8-bit (1-byte) to 32-bit (4-byte) unsigned integer, depending on the optional size parameter (default 1), followed by the string itself with no termination character.



240
241
242
243
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 240

def read_lpstring(size=1)
  length = read_uint_by_size(size)
  read_nstring(length)
end

#read_lpstringz(size = 1) ⇒ Object

Read an lpstring (as above) which is also terminated with a null byte.



246
247
248
249
250
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 246

def read_lpstringz(size=1)
  string = read_lpstring(size)
  reader.read(1) # null
  string
end

#read_mysql_type(type, metadata = nil) ⇒ Object

Read a single field, provided the MySQL column type as a symbol. Not all types are currently supported.



438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 438

def read_mysql_type(type, metadata=nil)
  case type
  when :tiny
    read_uint8
  when :short
    read_uint16
  when :int24
    read_uint24
  when :long
    read_uint32
  when :longlong
    read_uint64
  when :float
    read_float
  when :double
    read_double
  when :var_string
    read_varstring
  when :varchar, :string
    prefix_size = (metadata[:max_length] > 255) ? 2 : 1
    read_lpstring(prefix_size)
  when :blob, :geometry, :json
    read_lpstring(metadata[:length_size])
  when :timestamp
    read_uint32
  when :timestamp2
    read_timestamp2(metadata[:decimals])
  when :year
    read_uint8 + 1900
  when :date
    convert_mysql_type_date(read_uint24)
  when :time
    convert_mysql_type_time(read_uint24)
  when :datetime
    convert_mysql_type_datetime(read_uint64)
  when :datetime2
    read_datetimef(metadata[:decimals])
  when :enum, :set
    read_uint_by_size(metadata[:size])
  when :bit
    byte_length = (metadata[:bits]+7)/8
    read_uint_by_size(byte_length)
  when :newdecimal
    precision = metadata[:precision]
    scale = metadata[:decimals]
    read_newdecimal(precision, scale)
  else
    raise UnsupportedTypeException.new("Type #{type} is not supported.")
  end
end

#read_newdecimal(precision, scale) ⇒ Object

Read a (new) decimal value. The value is stored as a sequence of signed big-endian integers, each representing up to 9 digits of the integral and fractional parts. The first integer of the integral part and/or the last integer of the fractional part might be compressed (or packed) and are of variable length. The remaining integers (if any) are uncompressed and 32 bits wide.



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
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 266

def read_newdecimal(precision, scale)
  digits_per_integer = 9
  compressed_bytes = [0, 1, 1, 2, 2, 3, 3, 4, 4, 4]
  integral = (precision - scale)
  uncomp_integral = integral / digits_per_integer
  uncomp_fractional = scale / digits_per_integer
  comp_integral = integral - (uncomp_integral * digits_per_integer)
  comp_fractional = scale - (uncomp_fractional * digits_per_integer)

  # The sign is encoded in the high bit of the first byte/digit. The byte
  # might be part of a larger integer, so apply the optional bit-flipper
  # and push back the byte into the input stream.
  value = read_uint8
  str, mask = (value & 0x80 != 0) ? ["", 0] : ["-", -1]
  reader.unget(value ^ 0x80)

  size = compressed_bytes[comp_integral]

  if size > 0
    value = read_int_be_by_size(size) ^ mask
    str << value.to_s
  end

  (1..uncomp_integral).each do
    value = read_int32_be ^ mask
    str << value.to_s
  end

  str << "."

  (1..uncomp_fractional).each do
    value = read_int32_be ^ mask
    str << value.to_s
  end

  size = compressed_bytes[comp_fractional]

  if size > 0
    value = read_int_be_by_size(size) ^ mask
    str << value.to_s
  end

  BigDecimal(str)
end

#read_nstring(length) ⇒ Object

Read a non-terminated string, provided its length.



227
228
229
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 227

def read_nstring(length)
  reader.read(length)
end

#read_nstringz(length) ⇒ Object

Read a null-terminated string, provided its length (with the null).



232
233
234
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 232

def read_nstringz(length)
  reader.read(length).unpack("A*").first
end

#read_timestamp2(decimals) ⇒ Object 



432
433
434
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 432

def read_timestamp2(decimals)
  read_uint32_be + (read_frac_part(decimals) / 1000000)
end

#read_uint16 ⇒ Object

Read an unsigned 16-bit (2-byte) integer.



63
64
65
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 63

def read_uint16
  reader.read(2).unpack("v").first
end

#read_uint16_be ⇒ Object

Read an unsigned 16-bit (2-byte) big-endian integer.



68
69
70
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 68

def read_uint16_be
  reader.read(2).unpack("n").first
end

#read_uint24 ⇒ Object

Read an unsigned 24-bit (3-byte) integer.



73
74
75
76
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 73

def read_uint24
  a, b, c = reader.read(3).unpack("CCC")
  a + (b << 8) + (c << 16)
end

#read_uint24_be ⇒ Object

Read an unsigned 24-bit (3-byte) big-endian integer.



79
80
81
82
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 79

def read_uint24_be
  a, b = reader.read(3).unpack("nC")
  (a << 8) + b
end

#read_uint32 ⇒ Object

Read an unsigned 32-bit (4-byte) integer.



90
91
92
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 90

def read_uint32
  reader.read(4).unpack("V").first
end

#read_uint32_be ⇒ Object

Read an unsigned 32-bit (4-byte) integer.



85
86
87
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 85

def read_uint32_be
  reader.read(4).unpack("N").first
end

#read_uint40 ⇒ Object

Read an unsigned 40-bit (5-byte) integer.



95
96
97
98
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 95

def read_uint40
  a, b = reader.read(5).unpack("CV")
  a + (b << 8)
end

#read_uint40_be ⇒ Object

Read an unsigned 40-bit (5-byte) big-endian integer.



101
102
103
104
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 101

def read_uint40_be
  a, b = reader.read(5).unpack("NC")
  (a << 8) + b
end

#read_uint48 ⇒ Object

Read an unsigned 48-bit (6-byte) integer.



107
108
109
110
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 107

def read_uint48
  a, b, c = reader.read(6).unpack("vvv")
  a + (b << 16) + (c << 32)
end

#read_uint56 ⇒ Object

Read an unsigned 56-bit (7-byte) integer.



113
114
115
116
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 113

def read_uint56
  a, b, c = reader.read(7).unpack("CvV")
  a + (b << 8) + (c << 24)
end

#read_uint64 ⇒ Object

Read an unsigned 64-bit (8-byte) integer.



119
120
121
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 119

def read_uint64
  reader.read(8).unpack("Q<").first
end

#read_uint64_be ⇒ Object

Read an unsigned 64-bit (8-byte) integer.



124
125
126
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 124

def read_uint64_be
  reader.read(8).unpack("Q>").first
end

#read_uint8 ⇒ Object

Read an unsigned 8-bit (1-byte) integer.



58
59
60
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 58

def read_uint8
  reader.read(1).unpack("C").first
end

#read_uint8_array(length) ⇒ Object

Read an array of unsigned 8-bit (1-byte) integers.



312
313
314
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 312

def read_uint8_array(length)
  reader.read(length).bytes.to_a
end

#read_uint_bitmap_by_size_and_name(size, bit_names) ⇒ Object

Read a uint value using the provided size, and convert it to an array of symbols derived from a mapping table provided.



327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 327

def read_uint_bitmap_by_size_and_name(size, bit_names)
  value = read_uint_by_size(size)
  named_bits = []

  # Do an efficient scan for the named bits we know about using the hash
  # provided.
  bit_names.each do |(name, bit_value)|
    if (value & bit_value) != 0
      value -= bit_value
      named_bits << name
    end
  end

  # If anything is left over in +value+, add "unknown" names to the result
  # so that they can be identified and corrected.
  if value > 0
    0.upto(size * 8).map { |n| 1 << n }.each do |bit_value|
      if (value & bit_value) != 0
        named_bits << "unknown_#{bit_value}".to_sym
      end
    end
  end

  named_bits
end

#read_uint_by_size(size) ⇒ Object 



153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 153

def read_uint_by_size(size)
  case size
  when 1
    read_uint8
  when 2
    read_uint16
  when 3
    read_uint24
  when 4
    read_uint32
  when 5
    read_uint40
  when 6
    read_uint48
  when 7
    read_uint56
  when 8
    read_uint64
  end
end

#read_varint ⇒ Object

Read a variable-length “Length Coded Binary” integer. This is derived from the MySQL protocol, and re-used in the binary log format. This format uses the first byte to alternately store the actual value for integer values <= 250, or to encode the number of following bytes used to store the actual value, which can be 2, 3, or 8. It also includes support for SQL NULL as a special case.

See: forge.mysql.com/wiki/MySQL_Internals_ClientServer_Protocol#Elements



207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 207

def read_varint
  first_byte = read_uint8

  case
  when first_byte <= 250
    first_byte
  when first_byte == 251
    nil
  when first_byte == 252
    read_uint16
  when first_byte == 253
    read_uint24
  when first_byte == 254
    read_uint64
  when first_byte == 255
    raise "Invalid variable-length integer"
  end
end

#read_varstring ⇒ Object

Read a MySQL-style varint length-prefixed string. The length is stored as a variable-length “Length Coded Binary” value (see read_varint) which is followed by the string content itself. No termination is included.



255
256
257
258
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 255

def read_varstring
  length = read_varint
  read_nstring(length)
end