Class: Tapsoob::DataStream::Base

Inherits:
Object
  • Object
show all
Defined in:
lib/tapsoob/data_stream/base.rb

Direct Known Subclasses

FilePartition, Interleaved, Keyed, KeyedPartition

Constant Summary collapse

DEFAULT_CHUNKSIZE = 
1000

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(db, state, opts = {}) ⇒ Base

Returns a new instance of Base.



12
13
14
15
16
17
18
19
20
21
22
23
 
# File 'lib/tapsoob/data_stream/base.rb', line 12

def initialize(db, state, opts = {})
  @db = db
  @state = {
    :offset          => 0,
    :avg_chunksize   => 0,
    :num_chunksize   => 0,
    :total_chunksize => 0
  }.merge(state)
  @state[:chunksize] ||= DEFAULT_CHUNKSIZE
  @options = opts
  @complete = false
end

Instance Attribute Details

#db ⇒ Object (readonly)

Returns the value of attribute db.



10
11
12
 
# File 'lib/tapsoob/data_stream/base.rb', line 10

def db
  @db
end

#options ⇒ Object (readonly)

Returns the value of attribute options.



10
11
12
 
# File 'lib/tapsoob/data_stream/base.rb', line 10

def options
  @options
end

#state ⇒ Object (readonly)

Returns the value of attribute state.



10
11
12
 
# File 'lib/tapsoob/data_stream/base.rb', line 10

def state
  @state
end

Class Method Details

.factory(db, state, opts) ⇒ Object 



298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
 
# File 'lib/tapsoob/data_stream/base.rb', line 298

def self.factory(db, state, opts)
  if defined?(Sequel::MySQL) && Sequel::MySQL.respond_to?(:convert_invalid_date_time=)
    Sequel::MySQL.convert_invalid_date_time = :nil
  end

  if state.has_key?(:klass)
    return eval(state[:klass]).new(db, state, opts)
  end

  if Tapsoob::Utils.single_integer_primary_key(db, state[:table_name].to_sym)
    Tapsoob::DataStream::Keyed.new(db, state, opts)
  else
    Tapsoob::DataStream::Base.new(db, state, opts)
  end
end

.parse_json(json) ⇒ Object 



215
216
217
218
219
 
# File 'lib/tapsoob/data_stream/base.rb', line 215

def self.parse_json(json)
  hash = JSON.parse(json).symbolize_keys
  hash[:state].symbolize_keys! if hash.has_key?(:state)
  hash
end

Instance Method Details

#complete? ⇒ Boolean

Returns:

  • (Boolean) 


179
180
181
182
183
184
185
186
187
188
189
190
191
 
# File 'lib/tapsoob/data_stream/base.rb', line 179

def complete?
  # For file-based loading, check if we've read all lines
  if state[:file_initialized]
    result = state[:lines_read] >= state[:total_lines]
    log.debug "DataStream::Base#complete? (file) lines_read=#{state[:lines_read]} total_lines=#{state[:total_lines]} result=#{result} table=#{table_name}"
    result
  else
    # For database fetching, check offset vs size
    result = state[:offset] >= state[:size]
    log.debug "DataStream::Base#complete? (db) offset=#{state[:offset]} size=#{state[:size]} result=#{result} table=#{table_name}"
    result
  end
end

#encode_rows(rows) ⇒ Object 



155
156
157
 
# File 'lib/tapsoob/data_stream/base.rb', line 155

def encode_rows(rows)
  Tapsoob::Utils.base64encode(Marshal.dump(rows))
end

#error ⇒ Object 



34
35
36
 
# File 'lib/tapsoob/data_stream/base.rb', line 34

def error
  state[:error] || false
end

#error=(val) ⇒ Object 



30
31
32
 
# File 'lib/tapsoob/data_stream/base.rb', line 30

def error=(val)
  state[:error] = val
end

#fetch(opts = {}) ⇒ Object 



159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
 
# File 'lib/tapsoob/data_stream/base.rb', line 159

def fetch(opts = {})
  opts = (opts.empty? ? { :type => "database", :source => db.uri } : opts)

  log.debug "DataStream::Base#fetch state -> #{state.inspect}"

  t1 = Time.now
  rows = (opts[:type] == "file" ? fetch_file(opts[:source]) : fetch_rows)
  encoded_data = encode_rows(rows)
  t2 = Time.now
  elapsed_time = t2 - t1

  # Only increment offset for database fetches
  # For file fetches, offset is managed by fetch_file (tracks lines read, not rows)
  if opts[:type] != "file"
    state[:offset] += (rows == {} ? 0 : rows[:data].size)
  end

  [encoded_data, (rows == {} ? 0 : rows[:data].size), elapsed_time]
end

#fetch_data_from_database(params) {|rows| ... } ⇒ Object

Yields:

  • (rows) 


193
194
195
196
197
198
199
200
201
202
203
 
# File 'lib/tapsoob/data_stream/base.rb', line 193

def fetch_data_from_database(params)
  encoded_data = params[:encoded_data]

  rows = parse_encoded_data(encoded_data, params[:checksum])

  # update local state
  state.merge!(params[:state].merge(:chunksize => state[:chunksize]))

  yield rows if block_given?
  (rows == {} ? 0 : rows[:data].size)
end

#fetch_data_to_database(params) ⇒ Object 



205
206
207
208
209
210
211
212
213
 
# File 'lib/tapsoob/data_stream/base.rb', line 205

def fetch_data_to_database(params)
  encoded_data = params[:encoded_data]

  rows = parse_encoded_data(encoded_data, params[:checksum])

  log.debug "DataStream::Base#fetch_data_to_database: importing #{rows[:data] ? rows[:data].size : 0} rows for table #{table_name rescue 'unknown'}"
  import_rows(rows)
  (rows == {} ? 0 : rows[:data].size)
end

#fetch_file(dump_path) ⇒ Object 



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
 
# File 'lib/tapsoob/data_stream/base.rb', line 90

def fetch_file(dump_path)
  # Stream NDJSON format - read line by line without loading entire file
  file_path = File.join(dump_path, "data", "#{table_name}.json")

  # Initialize state on first call
  unless state[:file_initialized]
    state[:file_initialized] = true
    state[:lines_read] = 0
    state[:total_lines] = File.foreach(file_path).count
  end

  table_name_val = nil
  header_val = nil
  types_val = nil
  data_batch = []

  # Read from current offset
  File.open(file_path, 'r') do |file|
    # Skip to current offset
    state[:lines_read].times { file.gets }

    # Read chunksize worth of lines
    state[:chunksize].times do
      break if file.eof?
      line = file.gets
      next unless line

      chunk = JSON.parse(line.strip)
      table_name_val ||= chunk["table_name"]
      header_val ||= chunk["header"]
      types_val ||= chunk["types"]
      data_batch.concat(chunk["data"]) if chunk["data"]

      state[:lines_read] += 1
    end
  end

  # Apply skip-duplicates if needed
  data_batch = data_batch.uniq if @options[:"skip-duplicates"]

  # Don't set state[:size] or state[:offset] here - they're managed separately
  # for completion tracking based on actual data rows imported
  log.debug "DataStream::Base#fetch_file: read #{data_batch.size} rows from #{state[:lines_read]} lines (total #{state[:total_lines]} lines in file)"

  rows = {
    :table_name => table_name_val,
    :header     => header_val,
    :data       => data_batch,
    :types      => types_val
  }
  update_chunksize_stats
  rows
end

#fetch_rows ⇒ Object

keep a record of the average chunksize within the first few hundred thousand records, after chunksize goes below 100 or maybe if offset is > 1000



75
76
77
78
79
80
81
82
83
84
85
86
87
88
 
# File 'lib/tapsoob/data_stream/base.rb', line 75

def fetch_rows
  # Only count once on first fetch
  state[:size] ||= table.count

  ds = table.order(*order_by).limit(state[:chunksize], state[:offset])
  log.debug "DataStream::Base#fetch_rows SQL -> #{ds.sql}"
  rows = Tapsoob::Utils.format_data(db, ds.all,
    :string_columns => string_columns,
    :schema => db.schema(table_name),
    :table => table_name
  )
  update_chunksize_stats
  rows
end

#import_rows(rows) ⇒ Object 



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
 
# File 'lib/tapsoob/data_stream/base.rb', line 235

def import_rows(rows)
  columns = rows[:header]
  data    = rows[:data]

  # Only import existing columns
  if table.columns.size != columns.size
    existing_columns        = table.columns.map(&:to_s)
    additional_columns      = columns - existing_columns
    additional_columns_idxs = additional_columns.map { |c| columns.index(c) }
    additional_columns_idxs.reverse.each do |idx|
      columns.delete_at(idx)
      rows[:types].delete_at(idx)
    end
    data.each_index { |didx| additional_columns_idxs.reverse.each { |idx| data[didx].delete_at(idx) } }
  end

  # Decode blobs
  if rows.has_key?(:types) && rows[:types].include?("blob")
    blob_indices = rows[:types].each_index.select { |idx| rows[:types][idx] == "blob" }
    data.each_index do |idx|
      blob_indices.each do |bi|
        data[idx][bi] = Sequel::SQL::Blob.new(Tapsoob::Utils.base64decode(data[idx][bi])) unless data[idx][bi].nil?
      end
    end
  end

  # Parse date/datetime/time columns
  if rows.has_key?(:types)
    %w(date datetime time).each do |type|
      if rows[:types].include?(type)
        type_indices = rows[:types].each_index.select { |idx| rows[:types][idx] == type }
        data.each_index do |idx|
          type_indices.each do |ti|
            data[idx][ti] = Sequel.send("string_to_#{type}".to_sym, data[idx][ti]) unless data[idx][ti].nil?
          end
        end
      end
    end
  end

  # Remove id column
  if @options[:"discard-identity"] && rows[:header].include?("id")
    columns = rows[:header] - ["id"]
    data    = data.map { |d| d[1..-1] }
  end

  table.import(columns, data, :commit_every => 100)
rescue Exception => ex
  case ex.message
  when /integer out of range/ then
    raise Tapsoob::InvalidData, "  \\nDetected integer data that exceeds the maximum allowable size for an integer type.\n  This generally occurs when importing from SQLite due to the fact that SQLite does\n  not enforce maximum values on integer types.\n    ERROR\n  else raise ex\n  end\nend\n", []

#increment(row_count) ⇒ Object 



69
70
71
 
# File 'lib/tapsoob/data_stream/base.rb', line 69

def increment(row_count)
  state[:offset] += row_count
end

#log ⇒ Object 



25
26
27
28
 
# File 'lib/tapsoob/data_stream/base.rb', line 25

def log
  Tapsoob.log.level = Logger::DEBUG if state[:debug]
  Tapsoob.log
end

#max_chunksize_training ⇒ Object 



144
145
146
 
# File 'lib/tapsoob/data_stream/base.rb', line 144

def max_chunksize_training
  20
end

#order_by(name = nil) ⇒ Object 



62
63
64
65
66
67
 
# File 'lib/tapsoob/data_stream/base.rb', line 62

def order_by(name=nil)
  @order_by ||= begin
    name ||= table_name
    Tapsoob::Utils.order_by(db, name)
  end
end

#parse_encoded_data(encoded_data, checksum) ⇒ Object

Raises:



221
222
223
224
225
226
227
228
229
230
231
232
233
 
# File 'lib/tapsoob/data_stream/base.rb', line 221

def parse_encoded_data(encoded_data, checksum)
  raise Tapsoob::CorruptedData.new("Checksum Failed") unless Tapsoob::Utils.valid_data?(encoded_data, checksum)

  begin
    return Marshal.load(Tapsoob::Utils.base64decode(encoded_data))
  rescue Object => e
    unless ENV['NO_DUMP_MARSHAL_ERRORS']
      puts "Error encountered loading data, wrote the data chunk to dump.#{Process.pid}.dat"
      File.open("dump.#{Process.pid}.dat", "w") { |f| f.write(encoded_data) }
    end
    raise e
  end
end

#string_columns ⇒ Object 



54
55
56
 
# File 'lib/tapsoob/data_stream/base.rb', line 54

def string_columns
  @string_columns ||= Tapsoob::Utils.incorrect_blobs(db, table_name)
end

#table ⇒ Object 



58
59
60
 
# File 'lib/tapsoob/data_stream/base.rb', line 58

def table
  @table ||= db[table_name_sql]
end

#table_name ⇒ Object 



38
39
40
 
# File 'lib/tapsoob/data_stream/base.rb', line 38

def table_name
  state[:table_name].to_sym
end

#table_name_sql ⇒ Object 



42
43
44
 
# File 'lib/tapsoob/data_stream/base.rb', line 42

def table_name_sql
  table_name
end

#to_hash ⇒ Object 



46
47
48
 
# File 'lib/tapsoob/data_stream/base.rb', line 46

def to_hash
  state.merge(:klass => self.class.to_s)
end

#to_json ⇒ Object 



50
51
52
 
# File 'lib/tapsoob/data_stream/base.rb', line 50

def to_json
  JSON.generate(to_hash)
end

#update_chunksize_stats ⇒ Object 



148
149
150
151
152
153
 
# File 'lib/tapsoob/data_stream/base.rb', line 148

def update_chunksize_stats
  return if state[:num_chunksize] >= max_chunksize_training
  state[:total_chunksize] += state[:chunksize]
  state[:num_chunksize] += 1
  state[:avg_chunksize] = state[:total_chunksize] / state[:num_chunksize] rescue state[:chunksize]
end

#verify_stream ⇒ Object 



294
295
296
 
# File 'lib/tapsoob/data_stream/base.rb', line 294

def verify_stream
  state[:offset] = table.count
end