Class: Ferret

Inherits:
Object
  • Object
show all
Includes:
Constants
Defined in:
lib/sql-ferret.rb

Defined Under Namespace

Modules: Constants Classes: Alias_Generator, Exemplar, Expression, Expression_Parser, Field, Lexical_Ruleset, Parameter_Collector, Scanner, Schema, Selectee, Stage, Table

Constant Summary collapse

LEXICAL_RULESET = 
Ferret::Lexical_Ruleset.new(
simple: ",:*()'\\<>",
multichar: %w{-> <-> <= >=},
intertoken: " \t\n\r\f")

Constants included from Constants

Constants::QSF_MULTICOL, Constants::QSF_MULTIROW

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(schema_source, sqlite = nil, use_mutex: true) ⇒ Ferret

If the caller can guarantee that this [[Ferret]] instance is never accessed from multiple threads at once, it can turn off using the internal mutex by passing [[use_mutex: false]] for a small performance increase.



20
21
22
23
24
25
26
27
28
29
30
31
32
 
# File 'lib/sql-ferret.rb', line 20

def initialize schema_source, sqlite = nil, use_mutex: true
  raise 'type mismatch' unless schema_source.is_a? String
  super()
  @schema = Ferret::Schema.new(schema_source)
  @sqlite = sqlite
  # Guards access to [[@sqlite]] and
  # [[@sqlite_locked]].
  @sync_mutex = use_mutex ? Mutex.new : nil
  # Are we currently in a transaction?  (This lets us
  # implement [[Ferret#transaction]] reëntrantly.)
  @sqlite_locked = false
  return
end

Instance Attribute Details

#schema ⇒ Object (readonly)

Returns the value of attribute schema.



14
15
16
 
# File 'lib/sql-ferret.rb', line 14

def schema
  @schema
end

Class Method Details

.deterpret(interpretation, object) ⇒ Object 



396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
 
# File 'lib/sql-ferret.rb', line 396

def self::deterpret interpretation, object
  # Note that we're not handling [[nil]] any specially.  If
  # this field permits [[null]] values, it's the caller's --
  # who lives somewhere in the query execution wrapper of
  # Ferret -- to handle [[nil]], and if it doesn't, passing
  # [[nil]] to [[deterpret]] is either an error or, in case of
  # YAML, requires special escaping.
  case interpretation
    when nil then
      return object
    when :unix_time then
      ugh 'deterpreted-value-type-error',
              input: object.inspect,
              expected: 'Time' \
          unless object.is_a? Time
      return object.to_i
    when :subsecond_unix_time then
      ugh 'deterpreted-value-type-error',
              input: object.inspect,
              expected: 'Time' \
          unless object.is_a? Time
      return object.to_f
    when :iso8601 then
      ugh 'deterpreted-value-type-error',
              input: object.inspect,
              expected: 'Time' \
          unless object.is_a? Time
      return object.xmlschema
    when :json then
      return JSON.generate(object)
    when :pretty_json then
      return JSON.pretty_generate(object)
    when :yaml then
      return YAML.dump(object)
    when :ruby_marshal then
      return Marshal.dump(value)
    when :packed_hex then
      ugh 'deterpreted-value-type-error',
              input: object.inspect,
              expected: 'String' \
          unless object.is_a? String
      return object.unpack('H*').first
    else
      raise 'assertion failed'
  end
end

.interpret(interpretation, value) ⇒ Object 



339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
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
 
# File 'lib/sql-ferret.rb', line 339

def self::interpret interpretation, value
  # If a [[null]] came from the database, we'll interpret it
  # as a [[nil]].
  return nil if value.nil?
  ugh? interpretation: interpretation.to_s,
      input: value.inspect do
    case interpretation
      when nil then
        return value
      when :unix_time, :subsecond_unix_time then
        ugh 'interpreted-value-type-error',
                input: value.inspect,
                expected: 'Numeric' \
            unless value.is_a? Numeric
        return Time.at(value)
      when :iso8601 then
        ugh 'interpreted-value-type-error',
                input: value.inspect,
                expected: 'String' \
            unless value.is_a? String
        return Time.xmlschema(value)
      when :json, :pretty_json then
        ugh 'interpreted-value-type-error',
                input: value.inspect,
                expected: 'String' \
            unless value.is_a? String
        return JSON.parse(value)
      when :yaml then
        ugh 'interpreted-value-type-error',
                input: value.inspect,
                expected: 'String' \
            unless value.is_a? String
        return YAML.load(value)
      when :ruby_marshal then
        ugh 'interpreted-value-type-error',
                input: value.inspect,
                expected: 'String' \
            unless value.is_a? String
        return Marshal.load(value)
      when :packed_hex then
        ugh 'interpreted-value-type-error',
                input: value.inspect,
                expected: 'String' \
            unless value.is_a? String
        ugh 'invalid-hex-data',
                input: value \
            unless value =~ /\A[\dabcdef]*\Z/
        ugh 'odd-length-hex-data',
                input: value \
            unless value.length % 2 == 0
        return [value].pack('H*')
      else
        raise 'assertion failed'
    end
  end
end

Instance Method Details

#change(table_name, **changes) ⇒ Object 



34
35
36
37
38
39
40
41
42
 
# File 'lib/sql-ferret.rb', line 34

def change table_name, **changes
  ugh? attempted: 'ferret-change' do
    table = @schema[table_name] or
        ugh 'unknown-table', table: table_name
    sql = table.sql_to_change changes.keys.map(&:to_s)
    _sync{@sqlite.execute sql, **changes}
  end
  return
end

#create_table(name) ⇒ Object 



330
331
332
333
334
335
336
337
 
# File 'lib/sql-ferret.rb', line 330

def create_table name
  ugh? attempted: 'ferret-create-table' do
    _sync do
      @sqlite.execute sql_to_create_table(name)
    end
  end
  return
end

#go(raw_expr, *inputs, **changes, &thunk) ⇒ Object 



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
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
 
# File 'lib/sql-ferret.rb', line 92

def go raw_expr, *inputs, **changes, &thunk
  expr = Ferret::Expression_Parser.new(raw_expr, @schema).expr

  ugh? expr: raw_expr do
    ugh? attempted: 'ferret-go' do
      if inputs.length > expr.exemplars.length then
        ugh 'too-many-exemplars-given',
            expected: expr.exemplars.length,
            given: inputs.length
      elsif inputs.length < expr.exemplars.length then
        ugh 'not-enough-exemplars-given',
            expected: expr.exemplars.length,
            given: inputs.length
      end
    end

    if thunk and ![:select, :select_distinct].
        include? expr.type then
      ugh 'superfluous-thunk-supplied',
          explanation: 'query-not-a-select'
    end

    case expr.type
    when :select, :select_distinct then
      ugh? attempted: 'ferret-select' do
        ugh 'superfluous-changes' \
            unless changes.empty?

        ast = expr.select

        # At least for now, all the parameters behave as
        # simple ANDed filter rules.
        inputs_imply_single_row = false
        coll = Ferret::Parameter_Collector.new
        expr.exemplars.zip(inputs).each_with_index do
            |(exemplar_spec, input), seq_no|
          test, selects_one_p = coll.feed input, exemplar_spec
          inputs_imply_single_row |= selects_one_p
          ast.sql.gsub! /\[test\s+#{seq_no}\]/, test
        end

        # Let's now compose the framework of executing the
        # query from [[proc]]:s.

        # [[tuple_preparer]] takes a tuple of raw values
        # fetched from SQL and prepares it into a deliverable
        # object.
        tuple_preparer = if ast.shape & QSF_MULTICOL == 0 then
          # A single column was requested.  The deliverable
          # object is the piece of data from this column.
          proc do |row|
            Ferret.interpret ast.outputs.values.first,
                row.first
          end
        else
          # Multiple columns were requested (or one column in
          # multicolumn mode).  The deliverable object is an
          # [[OpenStruct]] mapping field names to data from
          # these fields.
          proc do |row|
            output = OpenStruct.new
            raise 'assertion failed' \
                unless row.length == ast.outputs.size
            # Note that we're relying on modern Ruby's
            # [[Hash]]'s retention of key order here.
            ast.outputs.to_a.each_with_index do
                |(name, interpretation), i|
              output[name] =
                  Ferret.interpret interpretation, row[i]
            end
            output
          end
        end

        # [[query_executor]] takes a [[proc]], executes the
        # query, and calls [[proc]] with each tuple prepared
        # by [[tuple_preparer]].
        query_executor = proc do |&result_handler|
          @sqlite.execute ast.sql, **coll do |row|
            result_handler.call tuple_preparer.call(row)
          end
        end

        # [[processor]] executes the query and delivers
        # results either by yielding to [[thunk]] if it has
        # been given or by returning them if not, taking into
        # account the query's shape.
        if thunk then
          # A thunk was supplied -- we'll just pass prepared
          # rows to it.
          processor = proc do
            query_executor.call &thunk
          end
        else
          # Why [[and]] here?  Well, the shape flag tells us
          # whether the query can translate one input to more
          # than one, and [[inputs_imply_single_row]] tells us
          # whether there are more than one input values that
          # thus get translated.  We can only know that the
          # result is a single-row table if both of these
          # preconditions are satisfied.
          if (ast.shape & QSF_MULTIROW == 0) and
              inputs_imply_single_row then
            # A single row was requested (implicitly, by using
            # a unique field as an exemplar).  We'll return
            # this row, or [[nil]] if nothing was found.
            processor = lambda do
              query_executor.call do |output|
                return output
              end
              return nil
            end
          else
            # Many rows were requested.  We'll collect them to
            # a list and return it.
            processor = proc do
              results = []
              query_executor.call do |output|
                results.push output
              end
              return results
            end
          end
        end

        _sync &processor
      end

    when :update then
      ugh? attempted: 'ferret-update' do
        ugh 'missing-changes' \
            if changes.empty?

        changed_table = expr.stages.last.table
        sql = "update #{changed_table.name} set "
        changes.keys.each_with_index do |fn, i|
          field = changed_table[fn.to_s]
          ugh 'unknown-field', field: fn,
                  table: changed_table.name,
                  role: 'changed-field' \
              unless field
          sql << ", " unless i.zero?
          sql << "#{field.name} = :#{fn}"
        end

        if expr.stages.length > 1 then
          ast = expr.select
          sql << " where " <<
              expr.stages.last.stalk.ref.name <<
              " in (#{ast.sql})"
        else
          # Special case: the criteria and the update live in
          # a single table, so we won't need to do any joining
          # or subquerying.
          unless expr.exemplars.empty? then
            sql << " " << expr.where_clause
          end
        end

        # We're going to pass the changes to
        # [[SQLite::Database#execute]] in a [[Hash]].
        # Unfortunately, the Ruby interface of SQLite does not
        # support mixing numbered and named arguments.  As a
        # workaround, we'll pass the etalon as a named
        # argument whose name is a number.  This is also
        # convenient because it avoids clashes with any other
        # named parameters -- those are necessarily column
        # names, and column names can not be numbers.
        coll = Ferret::Parameter_Collector.new
        expr.exemplars.zip(inputs).each_with_index do
            |(exemplar_spec, input), seq_no|
          test, selects_one_p = coll.feed input, exemplar_spec
          sql.gsub! /\[test\s+#{seq_no}\]/, test
        end

        _sync do
          @sqlite.execute sql, **coll, **changes
          return @sqlite.changes
        end
      end

    when :delete then
      ugh? attempted: 'ferret-delete' do
        ugh 'superfluous-changes' \
            unless changes.empty?

        affected_table = expr.stages.last.table
        sql = "delete from #{affected_table.name} "

        if expr.stages.length > 1 then
          ast = expr.select
          sql << " where " <<
              expr.stages.last.stalk.ref.name <<
              " in (#{ast.sql})"
        else
          # Special case: the criteria live in the affected
          # table, so we won't need to do any joining or
          # subquerying.
          unless expr.exemplars.empty? then
            sql << " " << expr.where_clause
          end
        end

        coll = Ferret::Parameter_Collector.new
        expr.exemplars.zip(inputs).each_with_index do
            |(exemplar_spec, input), seq_no|
          test, selects_one_p = coll.feed input, exemplar_spec
          sql.gsub! /\[test\s+#{seq_no}\]/, test
        end

        _sync do
          @sqlite.execute sql, **coll
          return @sqlite.changes
        end
      end

    else
      raise 'assertion failed'
    end
  end
end

#insert(table_name, **changes) ⇒ Object 



44
45
46
47
48
49
50
51
52
53
54
 
# File 'lib/sql-ferret.rb', line 44

def insert table_name, **changes
  ugh? attempted: 'ferret-insert' do
    table = @schema[table_name] or
        ugh 'unknown-table', table: table_name
    sql = table.sql_to_insert changes.keys.map(&:to_s)
    _sync do
      @sqlite.execute sql, **changes
      return @sqlite.last_insert_row_id
    end
  end
end

#pragma_user_version ⇒ Object 



316
317
318
319
320
 
# File 'lib/sql-ferret.rb', line 316

def pragma_user_version
  _sync do
    return @sqlite.get_first_value 'pragma user_version'
  end
end

#pragma_user_version=(new_version) ⇒ Object 



322
323
324
325
326
327
328
 
# File 'lib/sql-ferret.rb', line 322

def pragma_user_version= new_version
  raise 'type mismatch' unless new_version.is_a? Integer
  _sync do
    @sqlite.execute 'pragma user_version = ?', new_version
  end
  return new_version
end

#transaction(&thunk) ⇒ Object 



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
 
# File 'lib/sql-ferret.rb', line 56

def transaction &thunk
  # Note that [[_sync]] is reentrant, too.
  _sync do
    # If we get to this point, the only 'concurrent' access
    # might come from our very own thread -- that is, a
    # subroutine down the execution stack from present.  This
    # means that we can now access [[@sqlite_locked]] as
    # though we were in a single-threading environment, and
    # thus use it as a flag for 'this thread has already
    # acquired the SQLite-level lock so there's no need to
    # engage it again'.  (SQLite's transaction mechanism on
    # its own is not reentrant.)
    if @sqlite_locked then
      return yield
    else
      return @sqlite.transaction do
        begin
          @sqlite_locked = true
          return yield
        ensure
          @sqlite_locked = false
        end
      end
    end
  end
end