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

Returns a new instance of BinlogFieldParser.



45
46
47
48
49
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 45

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.



42
43
44
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 42

def binlog
  @binlog
end

#reader ⇒ Object

Returns the value of attribute reader.



43
44
45
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 43

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.



328
329
330
331
332
333
334
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 328

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.



348
349
350
351
352
353
354
355
356
357
358
359
360
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 348

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_time(value) ⇒ Object

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



338
339
340
341
342
343
344
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 338

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.



322
323
324
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 322

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.



286
287
288
289
290
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 286

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_double ⇒ Object

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



162
163
164
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 162

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

#read_float ⇒ Object

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



157
158
159
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 157

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

#read_int16_be ⇒ Object

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



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

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

#read_int24_be ⇒ Object

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



106
107
108
109
110
111
112
113
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 106

def read_int24_be
  a, b, c = reader.read(3).unpack('CCC')
  if a & 128
    (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.



116
117
118
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 116

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

#read_int8 ⇒ Object

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



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

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

#read_int_be_by_size(size) ⇒ Object 



141
142
143
144
145
146
147
148
149
150
151
152
153
154
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 141

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.



207
208
209
210
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 207

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.



213
214
215
216
217
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 213

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.



364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 364

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
    read_lpstring(metadata[:length_size])
  when :timestamp
    read_uint32
  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 :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.



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

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.new(str)
end

#read_nstring(length) ⇒ Object

Read a non-terminated string, provided its length.



194
195
196
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 194

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

#read_nstringz(length) ⇒ Object

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



199
200
201
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 199

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

#read_uint16 ⇒ Object

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



57
58
59
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 57

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

#read_uint24 ⇒ Object

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



62
63
64
65
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 62

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

#read_uint32 ⇒ Object

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



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

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

#read_uint40 ⇒ Object

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



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

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

#read_uint48 ⇒ Object

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



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

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.



85
86
87
88
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 85

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.



91
92
93
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 91

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

#read_uint8 ⇒ Object

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



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

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.



279
280
281
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 279

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.



294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 294

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 



120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 120

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



174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 174

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.



222
223
224
225
 
# File 'lib/mysql_binlog/binlog_field_parser.rb', line 222

def read_varstring
  length = read_varint
  read_nstring(length)
end