dusty-noko_parser 0.0.9 → 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/examples/monkey.rb +51 -0
- data/examples/monkeys.xml +52 -0
- data/lib/noko_parser.rb +29 -0
- data/lib/noko_parser/parser.rb +86 -0
- data/lib/noko_parser/properties.rb +199 -0
- data/test/test_noko_parser.rb +0 -0
- metadata +11 -4
data/examples/monkey.rb
ADDED
@@ -0,0 +1,51 @@
|
|
1
|
+
module Fruits; end
|
2
|
+
module Animals; end
|
3
|
+
|
4
|
+
class Fruits::Banana
|
5
|
+
include NokoParser::Properties
|
6
|
+
nodes :xpath => '//banana'
|
7
|
+
|
8
|
+
property :id, :xpath => 'id'
|
9
|
+
property :size, :xpath => 'size'
|
10
|
+
|
11
|
+
end
|
12
|
+
|
13
|
+
|
14
|
+
class Animals::Monkey
|
15
|
+
include NokoParser::Properties
|
16
|
+
nodes :xpath => '//monkey'
|
17
|
+
|
18
|
+
property :id, :xpath => '.', :attribute => 'id'
|
19
|
+
|
20
|
+
property :first_name, :xpath => 'contact/name[@type="first"]'
|
21
|
+
|
22
|
+
property :last_name, :xpath => 'contact/name[@type="last"]'
|
23
|
+
|
24
|
+
property :age, :xpath => 'yrsold'
|
25
|
+
|
26
|
+
property :funny, :xpath => 'personality', :attribute => :hilarious
|
27
|
+
|
28
|
+
property :loud, :xpath => 'personality'
|
29
|
+
|
30
|
+
property :street, :xpath => 'street'
|
31
|
+
|
32
|
+
property :city, :xpath => 'city'
|
33
|
+
|
34
|
+
property :state, :xpath => 'state'
|
35
|
+
|
36
|
+
property :zip, :xpath => 'zip'
|
37
|
+
|
38
|
+
embed :bananas, :xpath => 'bananas', :class => 'Fruits::Banana'
|
39
|
+
|
40
|
+
def address
|
41
|
+
[street, city, state].compact.join(", ") + " #{zip}"
|
42
|
+
end
|
43
|
+
|
44
|
+
def age
|
45
|
+
@age.to_i
|
46
|
+
end
|
47
|
+
|
48
|
+
def funny=(value)
|
49
|
+
@funny = (value == "true" ? true : false)
|
50
|
+
end
|
51
|
+
end
|
@@ -0,0 +1,52 @@
|
|
1
|
+
<xml>
|
2
|
+
<animals>
|
3
|
+
<elephant id="3">
|
4
|
+
<name>elephant1</name>
|
5
|
+
<trunk>large</trunk>
|
6
|
+
</elephant>
|
7
|
+
<monkey id="1">
|
8
|
+
<contact>
|
9
|
+
<name type='first'>monkey1</name>
|
10
|
+
<name type='last'>monkeylastname</name>
|
11
|
+
</contact>
|
12
|
+
<bananas>
|
13
|
+
<banana>
|
14
|
+
<id>1</id>
|
15
|
+
<size>small</size>
|
16
|
+
</banana>
|
17
|
+
<banana>
|
18
|
+
<id>2</id>
|
19
|
+
<size>medium</size>
|
20
|
+
</banana>
|
21
|
+
</bananas>
|
22
|
+
<yrsold>23</yrsold>
|
23
|
+
<personality hilarious="true">quiet</personality>
|
24
|
+
<street>300 Monkey St</street>
|
25
|
+
<city>Cincinnati</city>
|
26
|
+
<state>Oh</state>
|
27
|
+
<zip>45219</zip>
|
28
|
+
</monkey>
|
29
|
+
<monkey id="2">
|
30
|
+
<contact>
|
31
|
+
<name type='first'>monkey2</name>
|
32
|
+
<name type='last'>monkeylastname</name>
|
33
|
+
</contact>
|
34
|
+
<bananas>
|
35
|
+
<banana>
|
36
|
+
<id>3</id>
|
37
|
+
<size>large</size>
|
38
|
+
</banana>
|
39
|
+
<banana>
|
40
|
+
<id>4</id>
|
41
|
+
<size>huge</size>
|
42
|
+
</banana>
|
43
|
+
</bananas>
|
44
|
+
<yrsold>33</yrsold>
|
45
|
+
<personality hilarious="false">loud</personality>
|
46
|
+
<street>301 Monkey St</street>
|
47
|
+
<city>Cincinnati</city>
|
48
|
+
<state>Oh</state>
|
49
|
+
<zip>45219</zip>
|
50
|
+
</monkey>
|
51
|
+
</animals>
|
52
|
+
</xml>
|
data/lib/noko_parser.rb
ADDED
@@ -0,0 +1,29 @@
|
|
1
|
+
require 'nokogiri'
|
2
|
+
|
3
|
+
module NokoParser
|
4
|
+
|
5
|
+
class NokoParserError < StandardError; end
|
6
|
+
|
7
|
+
def self.version
|
8
|
+
"0.1.0"
|
9
|
+
end
|
10
|
+
|
11
|
+
end
|
12
|
+
|
13
|
+
## Stolen from merb-core
|
14
|
+
class Object
|
15
|
+
def full_const_get(name)
|
16
|
+
list = name.split("::")
|
17
|
+
list.shift if list.first.empty?
|
18
|
+
obj = self
|
19
|
+
list.each do |x|
|
20
|
+
# This is required because const_get tries to look for constants in the
|
21
|
+
# ancestor chain, but we only want constants that are HERE
|
22
|
+
obj = obj.const_defined?(x) ? obj.const_get(x) : obj.const_missing(x)
|
23
|
+
end
|
24
|
+
obj
|
25
|
+
end
|
26
|
+
end
|
27
|
+
|
28
|
+
require 'noko_parser/parser'
|
29
|
+
require 'noko_parser/properties'
|
@@ -0,0 +1,86 @@
|
|
1
|
+
module NokoParser
|
2
|
+
|
3
|
+
class Parser
|
4
|
+
|
5
|
+
##
|
6
|
+
# Initialize the parser by passing in a class that has included
|
7
|
+
# NokoParser::Properties
|
8
|
+
#
|
9
|
+
# @raises [NokoParserError] unless class included NokoParser::Properties
|
10
|
+
def initialize(klass, xml)
|
11
|
+
unless klass.respond_to?('_is_noko_parser?')
|
12
|
+
raise NokoParserError, 'Class must include NokoParser::Properties'
|
13
|
+
end
|
14
|
+
@object = klass
|
15
|
+
@xml = Nokogiri::XML(xml)
|
16
|
+
@nodes = []
|
17
|
+
@text_properties = klass._text_properties
|
18
|
+
@attribute_properties = klass._attribute_properties
|
19
|
+
@embedded_properties = klass._embedded_properties
|
20
|
+
end
|
21
|
+
|
22
|
+
##
|
23
|
+
# Runs the parser on the class. First finds the nodes that match
|
24
|
+
# and then parses the text and attribute properties on it
|
25
|
+
#
|
26
|
+
# return [Array<Object>] an array of objects
|
27
|
+
def run
|
28
|
+
@xml.xpath(@object._nodes_xpath).each do |node|
|
29
|
+
object = @object.new
|
30
|
+
parse_text_properties(node,object)
|
31
|
+
parse_attribute_properties(node,object)
|
32
|
+
parse_embedded_properties(node,object)
|
33
|
+
@nodes << object
|
34
|
+
end
|
35
|
+
@nodes
|
36
|
+
end
|
37
|
+
|
38
|
+
private
|
39
|
+
|
40
|
+
##
|
41
|
+
# Loop through the text properties of the class and use the defined
|
42
|
+
# xpath query to find the match. If found add to the object using
|
43
|
+
# the setter.
|
44
|
+
def parse_text_properties(node,object)
|
45
|
+
@text_properties.each do |property|
|
46
|
+
if match = node.xpath(property.xpath).first
|
47
|
+
unless match.inner_text.empty?
|
48
|
+
object.send("#{property.name}=",match.inner_text)
|
49
|
+
end
|
50
|
+
end
|
51
|
+
end
|
52
|
+
end
|
53
|
+
|
54
|
+
##
|
55
|
+
# Loop through the attribute properties of the class and use the defined
|
56
|
+
# xpath query to find the match. If found add to the object using
|
57
|
+
# the setter.
|
58
|
+
def parse_attribute_properties(node,object)
|
59
|
+
@attribute_properties.each do |property|
|
60
|
+
if match = node.xpath(property.xpath).first
|
61
|
+
object.send("#{property.name}=",match.attribute(property.attribute))
|
62
|
+
end
|
63
|
+
end
|
64
|
+
end
|
65
|
+
|
66
|
+
##
|
67
|
+
# Loop through the embedded properties of the class and use the defined
|
68
|
+
# xpath query to find the match. If found add to the object using the
|
69
|
+
# setter.
|
70
|
+
def parse_embedded_properties(node,object)
|
71
|
+
@embedded_properties.each do |property|
|
72
|
+
unless parser = Object.full_const_get(property.class_name) rescue nil
|
73
|
+
raise NokoParserError,
|
74
|
+
"Class #{property.class_name} not found for embed :#{property.name}"
|
75
|
+
end
|
76
|
+
if match = node.xpath(property.xpath).first
|
77
|
+
result = parser.parse(match.to_xml)
|
78
|
+
result = result.first if property.single
|
79
|
+
object.send("#{property.name}=", result)
|
80
|
+
end
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
end
|
85
|
+
|
86
|
+
end
|
@@ -0,0 +1,199 @@
|
|
1
|
+
module NokoParser
|
2
|
+
|
3
|
+
module Properties
|
4
|
+
|
5
|
+
##
|
6
|
+
# Text property is used to find the innerhtml of an element
|
7
|
+
class TextProperty
|
8
|
+
attr_reader :name, :xpath
|
9
|
+
|
10
|
+
def initialize(name,xpath)
|
11
|
+
@name = name.to_s
|
12
|
+
@xpath = xpath.to_s
|
13
|
+
end
|
14
|
+
|
15
|
+
end
|
16
|
+
|
17
|
+
##
|
18
|
+
# Attribute property is used to find the value of an attribute of an
|
19
|
+
# element
|
20
|
+
class AttributeProperty
|
21
|
+
attr_reader :name, :xpath, :attribute
|
22
|
+
|
23
|
+
def initialize(name,xpath,attribute)
|
24
|
+
@name = name.to_s.downcase
|
25
|
+
@xpath = xpath.to_s
|
26
|
+
@attribute = attribute.to_s
|
27
|
+
end
|
28
|
+
|
29
|
+
end
|
30
|
+
|
31
|
+
##
|
32
|
+
# Embedded property is used to find the value of sub-elements inside the
|
33
|
+
# element
|
34
|
+
#
|
35
|
+
# If single is present, only the first element will be returned
|
36
|
+
# otherwise an array is returned
|
37
|
+
class EmbeddedProperty
|
38
|
+
attr_reader :name, :xpath, :class_name, :single
|
39
|
+
|
40
|
+
def initialize(name,xpath,class_name,single)
|
41
|
+
@name = name.to_s.downcase
|
42
|
+
@xpath = xpath.to_s
|
43
|
+
@class_name = class_name.to_s
|
44
|
+
@single = single || false
|
45
|
+
end
|
46
|
+
|
47
|
+
end
|
48
|
+
|
49
|
+
def self.included(base)
|
50
|
+
base.instance_variable_set(:@_text_properties,[])
|
51
|
+
base.instance_variable_set(:@_attribute_properties,[])
|
52
|
+
base.instance_variable_set(:@_embedded_properties,[])
|
53
|
+
base.send(:extend, ClassMethods)
|
54
|
+
base.send(:include, InstanceMethods)
|
55
|
+
end
|
56
|
+
|
57
|
+
module ClassMethods
|
58
|
+
|
59
|
+
##
|
60
|
+
# Parse an xml string
|
61
|
+
#
|
62
|
+
# @return [Array<Object>] an array of objects
|
63
|
+
# @see NokoParser::Parser
|
64
|
+
def parse(xml)
|
65
|
+
parser = NokoParser::Parser.new(self,xml)
|
66
|
+
parser.run
|
67
|
+
end
|
68
|
+
|
69
|
+
##
|
70
|
+
# Parse an xml file
|
71
|
+
#
|
72
|
+
# @return [Array<Object>] an array of objects
|
73
|
+
# @see NokoParser::Parser
|
74
|
+
def parse_file(file)
|
75
|
+
parse(open(file))
|
76
|
+
end
|
77
|
+
|
78
|
+
##
|
79
|
+
# Make sure this class is a noko_parser compatible class
|
80
|
+
def _is_noko_parser?
|
81
|
+
true
|
82
|
+
end
|
83
|
+
|
84
|
+
##
|
85
|
+
# Returns the TextProperties defined on the class
|
86
|
+
#
|
87
|
+
# @return [Array<TextProperty>]
|
88
|
+
def _text_properties
|
89
|
+
@_text_properties
|
90
|
+
end
|
91
|
+
|
92
|
+
##
|
93
|
+
# Returns the AttributeProperties defined in the class
|
94
|
+
#
|
95
|
+
# @return [Array<AttributeProperty]
|
96
|
+
def _attribute_properties
|
97
|
+
@_attribute_properties
|
98
|
+
end
|
99
|
+
|
100
|
+
##
|
101
|
+
# Returns the EmbeddedProperties defined on the class
|
102
|
+
#
|
103
|
+
# @return [Array<EmbedProperty>]
|
104
|
+
def _embedded_properties
|
105
|
+
@_embedded_properties
|
106
|
+
end
|
107
|
+
|
108
|
+
##
|
109
|
+
# Returns the xpath required to find the nodes
|
110
|
+
def _nodes_xpath
|
111
|
+
@_nodes_xpath || (raise NokoParserError, 'nodes :xpath => "" is required')
|
112
|
+
end
|
113
|
+
|
114
|
+
##
|
115
|
+
# Set the xpath for finding the nodes.
|
116
|
+
#
|
117
|
+
# examples:
|
118
|
+
# nodes '//monkeys'
|
119
|
+
# nodes '//animals/monkeys/friendly'
|
120
|
+
def nodes(opts={})
|
121
|
+
unless opts[:xpath]
|
122
|
+
raise NokoParserError, ":xpath required for nodes definition"
|
123
|
+
end
|
124
|
+
@_nodes_xpath = opts[:xpath].to_s
|
125
|
+
end
|
126
|
+
|
127
|
+
##
|
128
|
+
# Sets a property on the object. If :attribute is defined it will
|
129
|
+
# set an AttributeProperty. Otherwise, it will set a TextProperty
|
130
|
+
# :xpath => 'somexpathstring' is required. The xpath should be the
|
131
|
+
# query required to find this element inside the node.
|
132
|
+
#
|
133
|
+
# examples
|
134
|
+
# property :name, :xpath => 'MyName[@type="real_name"]'
|
135
|
+
# property :eye_color, :xpath => 'eyes', :attribute => 'color'
|
136
|
+
def property(name,opts={})
|
137
|
+
unless opts[:xpath]
|
138
|
+
raise NokoParserError, ":xpath required for #{name}"
|
139
|
+
end
|
140
|
+
if opts[:attribute]
|
141
|
+
_add_attribute_property(name,opts[:xpath],opts[:attribute])
|
142
|
+
else
|
143
|
+
_add_text_property(name,opts[:xpath])
|
144
|
+
end
|
145
|
+
end
|
146
|
+
|
147
|
+
##
|
148
|
+
# Sets an embedded property on the option.
|
149
|
+
#
|
150
|
+
# examples:
|
151
|
+
# embed :names, :xpath => 'names', :class => Name
|
152
|
+
def embed(name,opts={})
|
153
|
+
unless opts[:class]
|
154
|
+
raise NokoParserError, ":xpath and :class required for #{name}"
|
155
|
+
end
|
156
|
+
_add_embedded_property(name,opts[:xpath],opts[:class],opts[:single])
|
157
|
+
end
|
158
|
+
|
159
|
+
##
|
160
|
+
# Adds to the class variable @text_properties in addition to
|
161
|
+
# creating an accessor for this property
|
162
|
+
def _add_text_property(name,xpath)
|
163
|
+
@_text_properties << TextProperty.new(name,xpath)
|
164
|
+
send(:attr_accessor, name)
|
165
|
+
end
|
166
|
+
|
167
|
+
##
|
168
|
+
# Adds to the class variable @attribute_properties in addition to
|
169
|
+
# creating an accessor for this property
|
170
|
+
def _add_attribute_property(name,xpath,attribute)
|
171
|
+
@_attribute_properties << AttributeProperty.new(
|
172
|
+
name,xpath,attribute
|
173
|
+
)
|
174
|
+
send(:attr_accessor, name)
|
175
|
+
end
|
176
|
+
|
177
|
+
##
|
178
|
+
# Adds to the class variable @embedded_properties in addition to
|
179
|
+
# creating a reader for this property
|
180
|
+
def _add_embedded_property(name,xpath,class_name,single)
|
181
|
+
@_embedded_properties << EmbeddedProperty.new(
|
182
|
+
name,xpath,class_name,single
|
183
|
+
)
|
184
|
+
send(:attr_accessor, name)
|
185
|
+
end
|
186
|
+
|
187
|
+
end # ClassMethods
|
188
|
+
|
189
|
+
module InstanceMethods
|
190
|
+
|
191
|
+
# def _something
|
192
|
+
# true
|
193
|
+
# end
|
194
|
+
|
195
|
+
end
|
196
|
+
|
197
|
+
end
|
198
|
+
|
199
|
+
end # NokoParser
|
File without changes
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: dusty-noko_parser
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.0
|
4
|
+
version: 0.1.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Dusty Doris
|
@@ -30,10 +30,17 @@ extensions: []
|
|
30
30
|
|
31
31
|
extra_rdoc_files:
|
32
32
|
- README.txt
|
33
|
-
files:
|
34
|
-
|
33
|
+
files:
|
34
|
+
- README.txt
|
35
|
+
- examples/monkey.rb
|
36
|
+
- examples/monkeys.xml
|
37
|
+
- lib/noko_parser.rb
|
38
|
+
- lib/noko_parser/parser.rb
|
39
|
+
- lib/noko_parser/properties.rb
|
40
|
+
- test/test_noko_parser.rb
|
35
41
|
has_rdoc: true
|
36
42
|
homepage: http://code.dusty.name
|
43
|
+
licenses:
|
37
44
|
post_install_message:
|
38
45
|
rdoc_options: []
|
39
46
|
|
@@ -54,7 +61,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
54
61
|
requirements: []
|
55
62
|
|
56
63
|
rubyforge_project: none
|
57
|
-
rubygems_version: 1.
|
64
|
+
rubygems_version: 1.3.5
|
58
65
|
signing_key:
|
59
66
|
specification_version: 2
|
60
67
|
summary: Wrapper around Nokogiri to easily parse xml files with xpath
|