| Class | Sequel::Dataset::PlaceholderLiteralizer |
| In: |
lib/sequel/dataset/placeholder_literalizer.rb
|
| Parent: | Object |
PlaceholderLiteralizer allows you to record the application of arbitrary changes to a dataset with placeholder arguments, recording where those placeholder arguments are used in the query. When running the query, the literalization process is much faster as Sequel can skip most of the work it normally has to do when literalizing a dataset.
Basically, this enables optimizations that allow Sequel to cache the SQL produced for a given dataset, so that it doesn‘t need to recompute that information every time.
Example:
loader = Sequel::Dataset::PlaceholderLiteralizer.loader(DB[:items]) do |pl, ds|
ds.where(id: pl.arg).exclude(name: pl.arg).limit(1)
end
loader.first(1, "foo")
# SELECT * FROM items WHERE ((id = 1) AND (name != 'foo')) LIMIT 1
loader.first(2, "bar")
# SELECT * FROM items WHERE ((id = 2) AND (name != 'bar')) LIMIT 1
Caveats:
Note that this method does not handle all possible cases. For example:
loader = Sequel::Dataset::PlaceholderLiteralizer.loader(DB[:items]) do |pl, ds|
ds.join(pl.arg, item_id: :id)
end
loader.all(:cart_items)
Will not qualify the item_id column with cart_items. In this type of situation it‘s best to add a table alias when joining:
loader = Sequel::Dataset::PlaceholderLiteralizer.loader(DB[:items]) do |pl, ds|
ds.join(Sequel.as(pl.arg, :t), item_id: :id)
end
loader.all(:cart_items)
There are other similar cases that are not handled, mainly when Sequel changes the SQL produced depending on the types of the arguments.
Create a PlaceholderLiteralizer by yielding a Recorder and dataset to the given block, recording the offsets at which the recorders arguments are used in the query.
# File lib/sequel/dataset/placeholder_literalizer.rb, line 119
119: def self.loader(dataset, &block)
120: Recorder.new.loader(dataset, &block)
121: end
Save the dataset, array of SQL fragments, and ending SQL string.
# File lib/sequel/dataset/placeholder_literalizer.rb, line 124
124: def initialize(dataset, fragments, final_sql, arity)
125: @dataset = dataset
126: @fragments = fragments
127: @final_sql = final_sql
128: @arity = arity
129: freeze
130: end
Return an array of all objects by running the SQL query for the given arguments. If a block is given, yields all objects to the block after loading them.
# File lib/sequel/dataset/placeholder_literalizer.rb, line 148
148: def all(*args, &block)
149: @dataset.with_sql_all(sql(*args), &block)
150: end
Run the SQL query for the given arguments, returning the first value. For this to make sense, the dataset should return a single row with a single value (or no rows).
# File lib/sequel/dataset/placeholder_literalizer.rb, line 164
164: def get(*args)
165: @dataset.with_sql_single_value(sql(*args))
166: end
Return the SQL query to use for the given arguments.
# File lib/sequel/dataset/placeholder_literalizer.rb, line 169
169: def sql(*args)
170: raise Error, "wrong number of arguments (#{args.length} for #{@arity})" unless args.length == @arity
171: s = String.new
172: ds = @dataset
173: @fragments.each do |sql, i, transformer|
174: s << sql
175: if i.is_a?(Integer)
176: v = args.fetch(i)
177: v = transformer.call(v) if transformer
178: else
179: v = i.call
180: end
181: ds.literal_append(s, v)
182: end
183: if sql = @final_sql
184: s << sql
185: end
186: s
187: end
Return a new PlaceholderLiteralizer with a modified dataset. This yields the receiver‘s dataset to the block, and the block should return the new dataset to use.
# File lib/sequel/dataset/placeholder_literalizer.rb, line 142
142: def with_dataset
143: dup.instance_exec{@dataset = yield @dataset; self}.freeze
144: end