siren 0.2.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -0,0 +1,16 @@
1
+ === 0.2.0 / 2009-07-07
2
+
3
+ * Mostly complete implementation of JSONQuery as described
4
+ by SitePen. This would have been 0.1 but it supercedes
5
+ John's version.
6
+
7
+
8
+ === 0.1.1 / 2009-06-22
9
+
10
+ * Incomplete fork released through GitHub by John Nunemaker
11
+
12
+
13
+ === 0.1.0 / Never released
14
+
15
+ * Work begun January 2009, left incomplete.
16
+
@@ -0,0 +1,22 @@
1
+ History.txt
2
+ Manifest.txt
3
+ README.txt
4
+ Rakefile
5
+ lib/siren.rb
6
+ lib/siren/json.tt
7
+ lib/siren/json.rb
8
+ lib/siren/json_query.tt
9
+ lib/siren/json_query.rb
10
+ lib/siren/json_query_nodes.rb
11
+ lib/siren/node.rb
12
+ lib/siren/parser.rb
13
+ lib/siren/reference.rb
14
+ lib/siren/walker.rb
15
+ test/test_siren.rb
16
+ test/fixtures/beatles.json
17
+ test/fixtures/car.rb
18
+ test/fixtures/names.json
19
+ test/fixtures/people.json
20
+ test/fixtures/person.rb
21
+ test/fixtures/refs.json
22
+ test/fixtures/store.json
@@ -0,0 +1,318 @@
1
+ = Siren
2
+
3
+ * http://github.com/jcoglan/siren
4
+
5
+ Siren is a JSON and JSONQuery interpreter for Ruby. It extends the normal functionality
6
+ of JSON-to-Ruby processing with cross-references, automatic typecasting,
7
+ and a succinct query language for filtering JSON and Ruby object graphs.
8
+
9
+
10
+ == Installation
11
+
12
+ sudo gem install siren
13
+
14
+
15
+ == Usage
16
+
17
+ As expected, Siren can be used as a basic JSON parser, though if that's all you
18
+ want you'll be better off with more performant libraries like the +json+ gem.
19
+
20
+ Siren.parse '[{"name": "mike"}]'
21
+ #=> [{"name"=>"mike"}]
22
+
23
+ The features listed below go beyond standard JSON and are modelled on the feature
24
+ set listed in this SitePen article:
25
+
26
+ http://www.sitepen.com/blog/2008/07/16/jsonquery-data-querying-beyond-jsonpath/
27
+
28
+
29
+ === Cross-references
30
+
31
+ Siren allows JSON objects to be given IDs, and for other objects to make
32
+ references back to them from within the same document. This allows for cyclic
33
+ data structures and other objects inexpressible in standard JSON.
34
+
35
+ data = Siren.parse <<-JSON
36
+ [
37
+ {
38
+ "id": "obj1",
39
+ "name": "mike"
40
+ }, {
41
+ "name": "bob",
42
+ "friend": {"$ref": "obj1"}
43
+ }
44
+ ]
45
+ JSON
46
+
47
+ #=> [ {"name" => "mike", "id" => "obj1"},
48
+ {"name" => "bob", "friend" => {"name" => "mike", "id" => "obj1"} } ]
49
+
50
+ So +bob+ has a cross-reference to +mike+, which the parser resolves for us. Note
51
+ +mike+ is not copied but is referenced by +bob+:
52
+
53
+ data[0].__id__ #=> -607191558
54
+ data[1]['friend'].__id__ #=> -607191558
55
+
56
+ The general syntax for a cross-reference is <tt>{"$ref": ID_STRING}</tt>; the parser
57
+ will replace this with the referenced object in the Ruby output.
58
+
59
+
60
+ === Automatic typecasting
61
+
62
+ JSON parsers typically output hashes and arrays, these being Ruby's closest analogues
63
+ to the JavaScript types that JSON expresses. Siren allows objects to be imbued with
64
+ a +type+ attribute, which if present causes the parser to cast the object to an
65
+ instance of the named type instead of a hash. Instead of hash keys, the keys in the
66
+ JSON object become instance variables of the typed object. To allow the parser to use
67
+ a class for casting, it must be extended using <tt>Siren::Node</tt>:
68
+
69
+ class PersonModel
70
+ extend Siren::Node
71
+ end
72
+
73
+ data = Siren.parse <<-JSON
74
+ {
75
+ "type": "person_model",
76
+ "name": "Jimmy",
77
+ "age": 25
78
+ }
79
+ JSON
80
+
81
+ #=> #<PersonModel @type="person_model", @age=25, @name="Jimmy">
82
+
83
+
84
+ == JSONQuery
85
+
86
+ JSONQuery is a language designed for filtering and transforming JSON-like object graphs.
87
+ Queries may be run against Ruby objects of any type, and may also be used as values in
88
+ JSON documents to generate data during parsing. For example:
89
+
90
+ # Run a filter against an array
91
+ adults = Siren.query "$[? @.age >= 18 ]", people
92
+
93
+ # Embed the query for expansion by the parser
94
+ people = Siren.parse <<-JSON
95
+ {
96
+ "people": [
97
+ {"name": "John", "age": 23},
98
+ {"name": "Paul", "age": 21},
99
+ {"name": "George", "age": 12},
100
+ {"name": "Ringo", "age": 17}
101
+ ],
102
+
103
+ "adults": {"$ref": "$.people[? @.age >= 18 ]"}
104
+ }
105
+ JSON
106
+
107
+ As for IDs, queries are embedded using the <tt>$ref</tt> syntax. Parsing the above
108
+ yields the following Ruby object:
109
+
110
+ { "adults" => [ {"name"=>"John", "age"=>23},
111
+ {"name"=>"Paul", "age"=>21}
112
+ ],
113
+ "people" => [ {"name"=>"John", "age"=>23},
114
+ {"name"=>"Paul", "age"=>21},
115
+ {"name"=>"George", "age"=>12},
116
+ {"name"=>"Ringo", "age"=>17}
117
+ ]
118
+ }
119
+
120
+ Siren supports the following sections of the JSONQuery language. A query is composed of
121
+ an object selector followed by zero or more filters.
122
+
123
+
124
+ === Object selectors
125
+
126
+ An object selector is either a valid object ID, or the special symbols <tt>$</tt>
127
+ or <tt>@</tt>.
128
+
129
+ * An object ID is a letter followed by zero or more letters, numbers or underscores.
130
+ Use of an ID produces a reference to the object with that +id+ property, as illustrated
131
+ above.
132
+ * <tt>$</tt> represents the root object the query is being run against, or the root of
133
+ the JSON document currently being parsed.
134
+ * <tt>@</tt> represents the current object in a looping operation, such as a map, sort
135
+ or filter.
136
+
137
+
138
+ === Expressions
139
+
140
+ Expressions are used within filters to generate data and perform comparisons. They work
141
+ in a similar fashion to JavaScript expressions, and may contain strings (enclosed in single
142
+ quotes), numbers, +true+, +false+, +null+ or other JSONQuery expressions. The following
143
+ operators are supported:
144
+
145
+ * Arithmetic: <tt>+</tt>, <tt>-</tt>, <tt>*</tt>, <tt>/</tt>, <tt>%</tt>
146
+ * Comparison: <tt><</tt>, <tt><=</tt>, <tt>></tt>, <tt>>=</tt>
147
+ * Equality: <tt>=</tt> for equality, <tt>!=</tt> for inequality
148
+ * String matching: <tt>=</tt> for case-sensitive matching, <tt>~</tt> for case-insensitive
149
+ * Logic: <tt>&</tt> for boolean AND, <tt>|</tt> boolean OR (not bitwise)
150
+ * Subexpressions are delimited using parentheses <tt>(</tt> and <tt>)</tt>
151
+
152
+
153
+ === String matching
154
+
155
+ The <tt>=</tt> and <tt>~</tt> operators, when used with strings, perform case-sensitive
156
+ and case-insensitive matching respectively. Within a string, <tt>*</tt> matches zero or
157
+ more characters of any type, and <tt>?</tt> matches a single character.
158
+
159
+ Siren.query "$[? @ = 'b*']", %w[foo bar baz]
160
+ #=> ["bar", "baz"]
161
+
162
+ Siren.query "$[? @ = 'b?']", %w[foo bar baz]
163
+ #=> []
164
+
165
+ Siren.query "$[? @ ~ 'BA?']", %w[foo bar baz]
166
+ #=> ["bar", "baz"]
167
+
168
+
169
+ === Field access filter
170
+
171
+ A field access filter selects a named field from an object. Fields are accessed using the
172
+ dot notation, or the bracket notation with an expression evaluating to a string or an integer.
173
+ This filter does one of the following based on the object's type:
174
+
175
+ * If the object is a +Hash+, the value for the named key is produced.
176
+ * If the object is an +Array+, the value at the given index is produced.
177
+ * If the object has the named method, the value of that method call is returned.
178
+ * Otherwise, the value of the named instance variable is produced.
179
+
180
+ data = {
181
+ "key" => "foo",
182
+ "name" => { "foo" => "bar" }
183
+ }
184
+
185
+ Siren.query "$.key", data
186
+ #=> "foo"
187
+
188
+ Siren.query "$[1]", %w[foo bar whizz]
189
+ #=> "bar"
190
+
191
+ Siren.query "$.name[ $.key ]", data
192
+ #=> "bar"
193
+
194
+ Siren.query "$.size", [3,4,5]
195
+ #=> 3
196
+
197
+ Fields can also be accessed recursively; the syntax <tt>..symbol</tt> returns an array
198
+ of all the values in the object with the property name +symbol+.
199
+
200
+ data = {"rec" => 6, "key" => {"rec" => 7}}
201
+
202
+ Siren.query "$..rec", data
203
+ #=> [6, 7]
204
+
205
+
206
+ === Array slice filter
207
+
208
+ The array slice filter selects a subset of values from an array. The syntax is <tt>[a:b:s]</tt>,
209
+ where +a+ is the start index, +b+ the end index and +s+ the step amount. The step may be
210
+ omitted (as in <tt>[a:b]</tt>), in which case the step defaults to <tt>1</tt>.
211
+
212
+ Siren.query "$[1:5:2]", %w[a b c d e f g h]
213
+ #=> ["b", "d", "f"]
214
+
215
+
216
+ === Selection filter
217
+
218
+ The selection filter extracts a subset of an array using a predicate condition.
219
+ Predicates are enclosed in <tt>[? ... ]</tt>, and within this block the symbol <tt>@</tt>
220
+ refers to each member of the array in turn. For example:
221
+
222
+ Siren.query "$.strings[? @.length > 3]", {'strings' => %w[the quick brown fox]}
223
+ #=> ["quick", "brown"]
224
+
225
+ Siren.query "$[? @ > 2 & @ <= 5]", [9,2,7,3,8,5]
226
+ #=> [3, 5]
227
+
228
+ Selections can be made recursively by prefixing the filter with two dots (<tt>..</tt>).
229
+ This will search the filtered object recursively for any properties that match the
230
+ predicate expression, and return them as a flat array.
231
+
232
+ data = [
233
+ 8,2,5,7,3,
234
+ {
235
+ 'foo' => 12,
236
+ 'bar' => 7,
237
+ 'whizz' => [9,2,6,4]
238
+ }
239
+ ]
240
+
241
+ Siren.query "$..[? @ % 4 = 0]", data
242
+ #=> [8, 4, 12]
243
+
244
+
245
+ === Mapping filter
246
+
247
+ Mappings work like Ruby's <tt>Enumerable#map</tt> method, producing a new array by
248
+ applying some expression to the members of another. Mappings are expressed by enclosing
249
+ the mapping expression in <tt>[= ... ]</tt>
250
+
251
+ Siren.query "$[= @[ 'upcase' ] ]", %w[the quick brown fox]
252
+ #=> ["THE", "QUICK", "BROWN", "FOX"]
253
+
254
+ Siren.query "$[= @.length + 2]", %w[the quick brown fox]
255
+ #=> [5, 7, 7, 5]
256
+
257
+ Just like other types of filters, mapping filters may use subqueries inside the
258
+ main query:
259
+
260
+ Siren.query "$[= @[? @ > 4] ]", [ [2,9,4,3], [5,8,2], [3,6] ]
261
+ #=> [[9], [5, 8], [6]]
262
+
263
+ Siren.query "$[= @[? @ > 4][0] + @.size ]", [ [2,9,4,3], [5,8,2], [3,6] ]
264
+ #=> [13, 8, 8]
265
+
266
+
267
+ === Sorting filter
268
+
269
+ The sorting filter returns a sorted version of the array it is operating on. The sort
270
+ must contain one or more comma-separated expressions to use to compare the items in the
271
+ array, and each expression must specify ascending (<tt>/</tt>) or descending (<tt>\\</tt>)
272
+ order. Expressions are prioritized from left to right. For example, the following sorts
273
+ the people in the list in ascending order by last name, then in descending order by first
274
+ name for people with the same last name:
275
+
276
+ people = Siren.parse <<-JSON
277
+ [
278
+ {"first": "Thom", "last": "Yorke"},
279
+ {"first": "Jonny", "last": "Greenwood"},
280
+ {"first": "Ed", "last": "O'Brien"},
281
+ {"first": "Colin", "last": "Greenwood"},
282
+ {"first": "Phil", "last": "Selway"}
283
+ ]
284
+ JSON
285
+
286
+ Siren.query "$[ /@.last, \\@.first ]", people
287
+
288
+ # => [ { "last"=>"Greenwood", "first"=>"Jonny"},
289
+ # { "last"=>"Greenwood", "first"=>"Colin"},
290
+ # { "last"=>"O'Brien", "first"=>"Ed"},
291
+ # { "last"=>"Selway", "first"=>"Phil"},
292
+ # { "last"=>"Yorke", "first"=>"Thom"} ]
293
+
294
+
295
+ == License
296
+
297
+ (The MIT License)
298
+
299
+ Copyright (c) 2009 James Coglan
300
+
301
+ Permission is hereby granted, free of charge, to any person obtaining
302
+ a copy of this software and associated documentation files (the
303
+ 'Software'), to deal in the Software without restriction, including
304
+ without limitation the rights to use, copy, modify, merge, publish,
305
+ distribute, sublicense, and/or sell copies of the Software, and to
306
+ permit persons to whom the Software is furnished to do so, subject to
307
+ the following conditions:
308
+
309
+ The above copyright notice and this permission notice shall be
310
+ included in all copies or substantial portions of the Software.
311
+
312
+ THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND,
313
+ EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
314
+ MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
315
+ IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
316
+ CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
317
+ TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
318
+ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
@@ -0,0 +1,19 @@
1
+ # -*- ruby -*-
2
+
3
+ require 'rubygems'
4
+ require 'hoe'
5
+ require './lib/siren.rb'
6
+
7
+ Hoe.spec('siren') do |p|
8
+ # p.rubyforge_name = 'sirenx' # if different than lowercase project name
9
+ p.developer('James Coglan', 'jcoglan@googlemail.com')
10
+ p.extra_deps = %w[treetop eventful]
11
+ end
12
+
13
+ task :tt do
14
+ %w(json json_query).each do |grammar|
15
+ `tt lib/siren/#{grammar}.tt`
16
+ end
17
+ end
18
+
19
+ # vim: syntax=Ruby
@@ -0,0 +1,76 @@
1
+ require 'set'
2
+ require 'rubygems'
3
+ require 'treetop'
4
+ require 'eventful'
5
+
6
+ %w[ json json_query json_query_nodes
7
+ walker parser node reference
8
+ ].each do |path|
9
+ require File.dirname(__FILE__) + '/siren/' + path
10
+ end
11
+
12
+ module Siren
13
+ VERSION = '0.2.0'
14
+
15
+ TYPE_FIELD = "type"
16
+ ID_FIELD = "id"
17
+ REF_FIELD = "$ref"
18
+
19
+ class JsonParser
20
+ include Siren::Walker
21
+ end
22
+
23
+ def self.each(object, &block)
24
+ case object
25
+ when Array then object.each_with_index { |x,i| yield(i,x) }
26
+ when Hash then object.each { |k,v| yield(k,v) }
27
+ when Enumerable then each(object.to_a, &block)
28
+ else nil
29
+ end
30
+ end
31
+
32
+ def self.parse(string, &block)
33
+ @json_parser ||= JsonParser.new
34
+ Reference.flush!
35
+ @symbols = {}
36
+
37
+ result = @json_parser.parse(string).value rescue nil
38
+
39
+ @json_parser.walk(result) do |holder, key, value|
40
+ if Hash === value && value[REF_FIELD]
41
+ value = Reference.new(value)
42
+ value.on(:resolve) do |ref, root, symbols|
43
+ holder[key] = ref.find(root, symbols, holder)
44
+ end
45
+ end
46
+ value
47
+ end
48
+
49
+ @json_parser.walk(result) do |holder, key, value|
50
+ if Hash === value
51
+ id = value[ID_FIELD]
52
+ value = Node.from_json(value)
53
+ @symbols[id] = value
54
+ end
55
+ value
56
+ end
57
+
58
+ result = Node.from_json(result)
59
+
60
+ Reference.resolve!(result, @symbols)
61
+ @json_parser.walk(result, &block) if block_given?
62
+
63
+ result
64
+ end
65
+
66
+ def self.query(expression, root)
67
+ compile_query(expression).value(root, @symbols || {})
68
+ end
69
+
70
+ def self.compile_query(expression)
71
+ @query_parser ||= JsonQueryParser.new
72
+ @query_cache ||= {}
73
+ @query_cache[expression] ||= @query_parser.parse(expression)
74
+ end
75
+ end
76
+