Class: SimpleOpencBot

Inherits:
Object
  • Object
show all
Includes:
OpencBot
Defined in:
lib/simple_openc_bot.rb

Direct Known Subclasses

MyLicence

Defined Under Namespace

Classes: BaseLicenceRecord

Constant Summary

Constants included from OpencBot

OpencBot::VERSION

Class Method Summary collapse

Instance Method Summary collapse

Methods included from OpencBot

#db_location, #db_name, #export, extended, #insert_or_update, #root_directory, #save_data, #save_run_report, #spotcheck, #sqlite_busy_timeout, #table_summary, #unlock_database, #verbose?

Methods included from OpencBot::Helpers::Text

#normalise_utf8_spaces, #strip_all_spaces

Class Method Details

.inherited(obj) ⇒ Object 



17
18
19
20
21
 
# File 'lib/simple_openc_bot.rb', line 17

def self.inherited(obj)
  path, = caller[0].partition(":")
  path = File.expand_path(File.join(File.dirname(path), ".."))
  @@simple_app_directory = path
end

.yields(*fields) ⇒ Object 



12
13
14
15
 
# File 'lib/simple_openc_bot.rb', line 12

def self.yields(*fields)
  raise "We currently only support one Record type per bot" if fields.count > 1
  self._yields = fields
end

Instance Method Details

#all_stored_records(opts = {}) ⇒ Object 



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
 
# File 'lib/simple_openc_bot.rb', line 99

def all_stored_records(opts={})
  if opts[:only_unexported]
    opts[:limit] ||= opts[:batch]
  end

  select = opts[:select] || "ocdata.*"
  table = opts[:table] || "ocdata"
  where = (opts[:where] ?  "\nWHERE #{opts[:where]}\n" : "\nWHERE 1 \n")
  order = (opts[:order] ?  "\nORDER BY #{opts[:order]}\n" : "")
  limit = (opts[:limit] ? "\nLIMIT #{opts[:limit]}\n" : "")

  if opts[:only_unexported]
    where += " AND (_last_exported_at IS NULL "\
      "OR _last_exported_at < _last_updated_at)"

    if !opts[:specific_ids].blank?
      ids = opts[:specific_ids].map{|id| "'#{id}'"}.join(",")
      where += " AND #{_yields[0].unique_field} IN (#{ids})"
    end
  end

  if opts[:count]
    sql = "COUNT(*) AS count from #{table} #{where}"
    puts sql if opts[:debug]
    select(sql)
  else
    sql = "#{select} from #{table} #{where} #{order} #{limit}"
    puts sql if opts[:debug]
    select_records(sql)
  end
end

#check_unique_index(record_class) ⇒ Object

should this be a method in sqlite_magic gem?



73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
 
# File 'lib/simple_openc_bot.rb', line 73

def check_unique_index(record_class)
  indexes = sqlite_magic_connection.execute("PRAGMA INDEX_LIST('ocdata')")
  db_unique_fields = indexes.map do |i|
    next if i["unique"] != 1
    next unless i["name"] =~ /autoindex/
    info = sqlite_magic_connection.execute("PRAGMA INDEX_INFO('#{i["name"]}')")
    info.map{|x| x["name"]}
  end.compact.flatten
  record_unique_fields = record_class.unique_fields.map(&:to_s)
  if db_unique_fields != record_unique_fields
    sqlite_magic_connection.execute("ROLLBACK")
    error = "Unique fields #{record_unique_fields} do not match the unique index(es) in `ocdata` table!"
    error += "\nThis is usually because the value of unique_field has changed since the table was automatically created."
    error += "\nUnique fields in `ocdata`: #{db_unique_fields.flatten}; in record #{record_class.name}: #{record_unique_fields}"
    raise error
  end
end

#count_stored_records ⇒ Object 



91
92
93
94
95
96
97
 
# File 'lib/simple_openc_bot.rb', line 91

def count_stored_records
  begin
    all_stored_records(:count => true).first["count"]
  rescue SqliteMagic::NoSuchTable
    0
  end
end

#export_data(opts = {}) ⇒ Object 



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
 
# File 'lib/simple_openc_bot.rb', line 143

def export_data(opts={})
  begin
    sqlite_magic_connection.add_columns(
      'ocdata', [:_last_exported_at, :_last_updated_at])
  rescue SQLite3::SQLException
  end
  Enumerator.new do |yielder|
    b = 1
    loop do
      if opts[:all]
        break if b > 1
        batch = all_stored_records(opts)
      else
        batch = unexported_stored_records(:batch => 100, :specific_ids => opts[:specific_ids])
      end
      break if batch.empty?
      updates = {}
      batch.map do |record|
        pipeline_data = record.to_pipeline
        next if pipeline_data.nil?
        updates[record.class.name] ||= []
        # opts[:all] is currently called in the bot:test rake task
        # This has the unfortunate side effect of updating the _last_exported_at
        # time when running the validation task, so I've added the following conditional
        if !opts[:all]
          updates[record.class.name] << record.to_hash.merge(
            :_last_exported_at => Time.now.iso8601(2))
        else
          updates[record.class.name] << record.to_hash
        end
        yielder << pipeline_data
      end
      sqlite_magic_connection.execute("BEGIN TRANSACTION")
      if b == 1
        check_unique_index(_yields[0])
      end
      updates.each do |k, v|
        save_data(k.constantize.unique_fields, v)
      end
      sqlite_magic_connection.execute("COMMIT")
      b += 1
    end
  end
end

#select_records(sql) ⇒ Object 



139
140
141
 
# File 'lib/simple_openc_bot.rb', line 139

def select_records(sql)
  select(sql).map { |record| record['_type'].constantize.new(record) }
end

#spotcheck_data ⇒ Object 



188
189
190
191
192
193
 
# File 'lib/simple_openc_bot.rb', line 188

def spotcheck_data
  batch = spotcheck_records
  batch.collect do |record|
    record.to_pipeline
  end
end

#spotcheck_records(limit = 5) ⇒ Object 



135
136
137
 
# File 'lib/simple_openc_bot.rb', line 135

def spotcheck_records(limit = 5)
  all_stored_records(:order => "RANDOM()", :limit => limit)
end

#sqlite_magic_connection ⇒ Object

Override default in ScraperWiki gem



24
25
26
27
 
# File 'lib/simple_openc_bot.rb', line 24

def sqlite_magic_connection
  db = @config ? @config[:db] : File.expand_path(File.join(@@simple_app_directory, 'db', db_name))
  @sqlite_magic_connection ||= SqliteMagic::Connection.new(db)
end

#unexported_stored_records(opts = {}) ⇒ Object 



131
132
133
 
# File 'lib/simple_openc_bot.rb', line 131

def unexported_stored_records(opts={})
  all_stored_records(opts.merge!(:only_unexported => true))
end

#update_data(opts = {}) ⇒ Object 



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
 
# File 'lib/simple_openc_bot.rb', line 29

def update_data(opts={})
  if opts[:specific_ids].nil? || opts[:specific_ids].empty?
    # fetch everything
    record_enumerator = Enumerator.new do |yielder|
      fetch_all_records(opts) do |result|
        yielder.yield(result)
      end
    end
  else
    # fetch records with specified ids
    record_enumerator = Enumerator.new do |yielder|
      fetch_specific_records(opts) do |result|
        yielder.yield(result)
      end
    end
  end
  saves_count = 0
  batch_size = opts[:test_mode] ? 1 : 500
  record_enumerator.each_slice(batch_size) do |records|
    begin
      sqlite_magic_connection.execute("BEGIN TRANSACTION")
      records.each do |record|
        insert_or_update(record.class.unique_fields,
          record.to_hash)
        saves_count += 1
        if saves_count == 1
          # TODO: move this validation to somewhere more explicit
          raise "Bot must specify what record type it will yield" if _yields.nil?
          check_unique_index(_yields[0])
        end
        STDOUT.print(".")
        STDOUT.flush
      end
    ensure
      sqlite_magic_connection.execute("COMMIT") if sqlite_magic_connection.database.transaction_active?
    end
  end
  # NB #save_run_report saves a timestamp called :run_at, so :completed_at is redundant.
  # Probably should either remove or replace with another datapoint, e.g. started_at
  save_run_report(:status => 'success', :completed_at => Time.now)
  saves_count
end

#validate_data(opts = {}) ⇒ Object 



195
196
197
198
199
200
201
202
203
204
 
# File 'lib/simple_openc_bot.rb', line 195

def validate_data(opts={})
  opts = {:limit => 1000}.merge(opts)
  errors = all_stored_records(opts).map do |record|
    record.errors
  end.compact
  total = count_stored_records
  selected = [opts[:limit], total].min
  puts "NOTICE: only validated first #{selected} of #{total} records"
  errors
end