rexle 0.9.5 → 0.9.6
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/rexle.rb +121 -330
- metadata +18 -5
data/lib/rexle.rb
CHANGED
@@ -1,4 +1,4 @@
|
|
1
|
-
#!/usr/bin/
|
1
|
+
#!/usr/bin/ruby
|
2
2
|
|
3
3
|
# file: rexle.rb
|
4
4
|
|
@@ -6,48 +6,21 @@ require 'rexml/document'
|
|
6
6
|
require 'rexleparser'
|
7
7
|
require 'dynarex-parser'
|
8
8
|
require 'polyrex-parser'
|
9
|
-
require 'cgi'
|
10
9
|
include REXML
|
11
10
|
|
12
|
-
# modifications:
|
13
|
-
# 20-Oct-2012: feature: added Rexle::Element#texts which is the equivalent
|
14
|
-
# of REXML::Element#texts
|
15
|
-
# 10-Sep-2012: bug fix: Removed code from method pretty_print in order to
|
16
|
-
# get the XML displayed properly
|
17
|
-
# 23-Aug-2012: feature: implemented xpath function contains()
|
18
|
-
# 17-Aug-2012: bug fix: pretty print now ignores text containing empty space
|
19
|
-
# 16-Aug-2012: the current element's text (if its not empty) is now returned
|
20
|
-
# from its children method
|
21
|
-
# 15-Aug-2012: feature: xpath containing child:: now supported
|
22
|
-
# 13-Aug-2012: bug fix: xpath can now handle the name() function
|
23
|
-
# 11-Aug-2012: bug fix: separated the max() method from 1 line into 3
|
24
|
-
# and that fixed it
|
25
|
-
# 08-Aug-2012: feature: added Element#insert_before and Element#insert_after
|
26
|
-
# 19-Jul-2012: Changed children to elements where appropriate
|
27
|
-
# 15-Jul-2012: bug fix: self.root.value is no longer appended
|
28
|
-
# to the body if there are no child elements
|
29
|
-
# 19-Jun-2012: a bug fix for .//*[@class]
|
30
|
-
# 17-Jun-2012: a couple of new xpath things are supported '.' and '|'
|
31
|
-
# 15-Apr-2012: bug fix: New element names are typecast as string
|
32
|
-
# 16-Mar-2012: bug fix: Element names which contain a colon can now be selected
|
33
|
-
# in the xpath.
|
34
|
-
# 22-Feb-2012: bug resolution: Deactivated the PolyrexParser; using RexleParser instead
|
35
|
-
# 14-Jan-2012: Implemented Rexle::Elements#each
|
36
|
-
# 21-Dec-2011: Bug fix: xpath modified to allow querying from the actual
|
37
|
-
# root rather than the 1st child element from the root
|
38
11
|
|
39
12
|
module XMLhelper
|
40
13
|
|
41
14
|
def doc_print(children)
|
42
|
-
|
43
|
-
body =
|
15
|
+
|
16
|
+
body = children.empty? ? self.root.value : scan_print(children).join
|
44
17
|
a = self.root.attributes.to_a.map{|k,v| "%s='%s'" % [k,v]}
|
45
18
|
"<%s%s>%s</%s>" % [self.root.name, a.empty? ? '' : ' ' + a.join(' '), body, self.root.name]
|
46
19
|
end
|
47
20
|
|
48
21
|
def doc_pretty_print(children)
|
49
22
|
|
50
|
-
body = pretty_print(children,2).join
|
23
|
+
body = children.empty? ? self.value : pretty_print(children,2).join
|
51
24
|
a = self.root.attributes.to_a.map{|k,v| "%s='%s'" % [k,v]}
|
52
25
|
ind = "\n "
|
53
26
|
"<%s%s>%s%s%s</%s>" % [self.root.name, a.empty? ? '' : ' ' + a.join(' '), ind, body, "\n", self.root.name]
|
@@ -56,24 +29,18 @@ module XMLhelper
|
|
56
29
|
def scan_print(nodes)
|
57
30
|
|
58
31
|
nodes.map do |x|
|
59
|
-
|
32
|
+
#puts 'x: ' + x.class.to_s #+ ' name: ' + x.name.to_s
|
60
33
|
if x.is_a? Rexle::Element then
|
61
|
-
|
34
|
+
unless x.name == '![' then
|
62
35
|
a = x.attributes.to_a.map{|k,v| "%s='%s'" % [k,v]}
|
63
36
|
tag = x.name + (a.empty? ? '' : ' ' + a.join(' '))
|
64
37
|
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
71
|
-
out = ["<%s/>" % tag]
|
72
|
-
end
|
73
|
-
elsif x.name == '!-' then
|
74
|
-
"<!--%s-->" % x.value
|
75
|
-
else
|
76
|
-
"<![CDATA[%s]]>" % x.value
|
38
|
+
out = ["<%s>" % tag]
|
39
|
+
out << x.value unless x.value.nil? || x.value.empty?
|
40
|
+
out << scan_print(x.children)
|
41
|
+
out << "</%s>" % x.name
|
42
|
+
else
|
43
|
+
"<![CDATA[%s]]>" % x.value
|
77
44
|
end
|
78
45
|
elsif x.is_a? String then
|
79
46
|
x
|
@@ -84,24 +51,19 @@ module XMLhelper
|
|
84
51
|
|
85
52
|
def pretty_print(nodes, indent='0')
|
86
53
|
indent = indent.to_i
|
87
|
-
|
88
|
-
nodes.select(){|x| x.is_a? Rexle::Element or x.strip.length > 0}
|
89
|
-
.map.with_index do |x, i|
|
90
|
-
|
54
|
+
nodes.map.with_index do |x, i|
|
91
55
|
if x.is_a? Rexle::Element then
|
92
56
|
unless x.name == '![' then
|
93
|
-
#return ["<%s/>" % x.name] if x.value = ''
|
94
57
|
a = x.attributes.to_a.map{|k,v| "%s='%s'" % [k,v]}
|
95
|
-
a ||= []
|
96
58
|
tag = x.name + (a.empty? ? '' : ' ' + a.join(' '))
|
97
59
|
|
98
|
-
|
99
|
-
|
100
|
-
("\n" + ' ' * indent) : ''
|
60
|
+
ind1 = x.children.length > 0 ? ("\n" + ' ' * indent) : ''
|
61
|
+
start = i > 0 ? ("\n" + ' ' * (indent - 1)) : ''
|
101
62
|
out = ["%s<%s>%s" % [start, tag, ind1]]
|
102
63
|
|
64
|
+
out << x.value.sub(/^[\n\s]+$/,'') unless x.value.nil? || x.value.empty?
|
103
65
|
out << pretty_print(x.children, (indent + 1).to_s.clone)
|
104
|
-
ind2 =
|
66
|
+
ind2 = x.children.length > 0 ? ("\n" + ' ' * (indent - 1)) : ''
|
105
67
|
out << "%s</%s>" % [ind2, x.name]
|
106
68
|
else
|
107
69
|
"<![CDATA[%s]]>" % x.value
|
@@ -119,10 +81,6 @@ class Rexle
|
|
119
81
|
include XMLhelper
|
120
82
|
|
121
83
|
attr_reader :prefixes
|
122
|
-
|
123
|
-
def self.version()
|
124
|
-
'0.9.xx'
|
125
|
-
end
|
126
84
|
|
127
85
|
def initialize(x=nil)
|
128
86
|
super()
|
@@ -134,144 +92,85 @@ class Rexle
|
|
134
92
|
Array: proc {|x| x},
|
135
93
|
:"REXML::Document" => proc {|x| scan_doc x.root}
|
136
94
|
}
|
137
|
-
|
138
|
-
doc_node = ['doc','',{}]
|
139
|
-
|
140
95
|
|
141
96
|
@a = procs[x.class.to_s.to_sym].call(x)
|
142
|
-
@doc = scan_element(
|
97
|
+
@doc = scan_element(*@a)
|
143
98
|
|
144
99
|
# fetch the namespaces
|
145
100
|
@prefixes = []
|
146
101
|
if @doc.root.attributes then
|
147
|
-
|
102
|
+
#puts 'attrs : ' + @doc.root.attributes.inspect
|
148
103
|
xmlns = @doc.root.attributes.select {|k,v| k[/^xmlns:/]}
|
149
104
|
@prefixes = xmlns.keys.map{|x| x[/\w+$/]}
|
150
105
|
end
|
106
|
+
end
|
107
|
+
|
108
|
+
end
|
109
|
+
|
110
|
+
def xpath(path, &blk)
|
111
|
+
|
112
|
+
# is it a function
|
113
|
+
fn_match = path.match(/^(\w+)\(([^\)]+)\)$/)
|
151
114
|
|
115
|
+
# Array: proc {|x| x.flatten.compact},
|
116
|
+
if (fn_match and fn_match.captures.first[/^(attribute|@)/]) or fn_match.nil? then
|
117
|
+
procs = {
|
118
|
+
Array: proc {|x| block_given? ? x : x.flatten },
|
119
|
+
String: proc {|x| x},
|
120
|
+
:"Rexle::Element" => proc {|x| [x]}
|
121
|
+
}
|
122
|
+
bucket = []
|
123
|
+
result = @doc.xpath(path, bucket, &blk)
|
124
|
+
|
125
|
+
procs[result.class.to_s.to_sym].call(result)
|
126
|
+
|
127
|
+
else
|
128
|
+
#puts 'fn ' + fn_match.captures.inspect
|
129
|
+
m, xpath_value = fn_match.captures
|
130
|
+
method(m.to_sym).call(xpath_value)
|
152
131
|
end
|
153
132
|
|
133
|
+
|
154
134
|
end
|
155
|
-
|
156
|
-
def xpath(path, &blk)
|
157
|
-
@doc.xpath(path, &blk)
|
158
|
-
end
|
159
135
|
|
160
136
|
class Element
|
161
137
|
include XMLhelper
|
162
138
|
|
163
139
|
attr_accessor :name, :value, :parent
|
164
|
-
attr_reader :child_lookup
|
165
|
-
|
166
|
-
alias original_clone clone
|
140
|
+
attr_reader :child_lookup
|
167
141
|
|
168
142
|
def initialize(name=nil, value='', attributes={}, rexle=nil)
|
169
143
|
@rexle = rexle
|
170
144
|
super()
|
171
|
-
@name, @value, @attributes = name
|
145
|
+
@name, @value, @attributes = name, value, attributes
|
172
146
|
raise "Element name must not be blank" unless name
|
173
147
|
@child_elements = []
|
174
148
|
@child_lookup = []
|
175
149
|
end
|
176
150
|
|
177
|
-
def contains(raw_args)
|
178
|
-
path, raw_val = raw_args.split(',',2)
|
179
|
-
val = raw_val.strip[/^["']?.*["']?$/]
|
180
|
-
|
181
|
-
anode = query_xpath(path)
|
182
|
-
return unless anode
|
183
|
-
a = scan_contents(anode.first)
|
184
|
-
|
185
|
-
[a.grep(/#{val}/).length > 0]
|
186
|
-
end
|
187
|
-
|
188
|
-
def count(path)
|
189
|
-
length = query_xpath(path).flatten.compact.length
|
190
|
-
length
|
191
|
-
end
|
192
|
-
|
193
|
-
def max(path)
|
194
|
-
a = query_xpath(path).flatten.compact.map(&:to_i)
|
195
|
-
a.max
|
196
|
-
end
|
197
|
-
|
198
151
|
def name()
|
152
|
+
#puts 'zzz : ' + @rexle.prefixes.inspect
|
153
|
+
#puts '@rexle : ' + @rexle.inspect
|
199
154
|
if @rexle then
|
200
|
-
|
201
|
-
|
155
|
+
prefix = @rexle.prefixes.find {|x| x == @name[/^(\w+):/,1] } if @rexle.prefixes.is_a? Array
|
156
|
+
#puts 'ppp: ' + prefix.inspect
|
157
|
+
prefix ? @name.sub(prefix + ':', '') : @name
|
202
158
|
else
|
203
|
-
|
159
|
+
@name
|
204
160
|
end
|
205
161
|
end
|
206
162
|
|
207
|
-
def xpath(
|
208
|
-
r = filter_xpath(path, rlist=[], &blk)
|
209
|
-
r.is_a?(Array) ? r.compact : r
|
210
|
-
end
|
211
|
-
|
212
|
-
def filter_xpath(path, rlist=[], &blk)
|
213
|
-
|
214
|
-
# is it a function
|
215
|
-
fn_match = path.match(/^(\w+)\(["']?([^\)]*)["']?\)$/)
|
216
|
-
|
217
|
-
# Array: proc {|x| x.flatten.compact},
|
218
|
-
if (fn_match and fn_match.captures.first[/^(attribute|@)/]) or fn_match.nil? then
|
219
|
-
procs = {
|
220
|
-
Array: proc {|x| block_given? ? x : x.flatten.uniq },
|
221
|
-
String: proc {|x| x},
|
222
|
-
Hash: proc {|x| x},
|
223
|
-
TrueClass: proc{|x| x},
|
224
|
-
FalseClass: proc{|x| x},
|
225
|
-
:"Rexle::Element" => proc {|x| [x]}
|
226
|
-
}
|
227
|
-
bucket = []
|
228
|
-
raw_results = path.split('|').map do |xp|
|
229
|
-
query_xpath(xp, bucket, &blk)
|
230
|
-
end
|
231
|
-
|
232
|
-
#results = raw_results.inject(&:+)
|
233
|
-
results = raw_results.last
|
234
|
-
procs[results.class.to_s.to_sym].call(results) if results
|
235
|
-
|
236
|
-
else
|
237
|
-
m, xpath_value = fn_match.captures
|
238
|
-
xpath_value.empty? ? method(m.to_sym).call : method(m.to_sym).call(xpath_value)
|
239
|
-
end
|
240
|
-
|
241
|
-
end
|
242
|
-
|
243
|
-
def query_xpath(raw_xpath_value, rlist=[], &blk)
|
163
|
+
def xpath(xpath_value, rlist=[], &blk)
|
244
164
|
|
245
|
-
#remove any
|
165
|
+
#remove any prefixes
|
246
166
|
#@rexle.prefixes.each {|x| xpath_value.sub!(x + ':','') }
|
247
|
-
|
248
|
-
|
249
|
-
xpath_value = raw_xpath_value.sub('child::','./')
|
250
|
-
#xpath_value.sub!(/\.\/(?=[\/])/,'')
|
251
|
-
|
252
|
-
if xpath_value[/^[\w\/]+\s*=.*/] then
|
253
|
-
flag_func = true
|
254
|
-
|
255
|
-
xpath_value.sub!(/^\w+\s*=.*/,'.[\0]')
|
256
|
-
xpath_value.sub!(/\/([\w]+\s*=.*)/,'[\1]')
|
257
|
-
|
258
|
-
#result = self.element xpath_value
|
259
|
-
#return [(result.is_a?(Rexle::Element) ? true : false)]
|
260
|
-
end
|
261
|
-
|
167
|
+
|
168
|
+
xpath_value.sub!(/^\[/,'*[')
|
262
169
|
#xpath_value.sub!(/^attribute::/,'*/attribute::')
|
263
|
-
raw_path, raw_condition = xpath_value.sub(
|
264
|
-
.match(/([^\[]+)(\[[^\]]+\])?/).captures
|
170
|
+
raw_path, raw_condition = xpath_value.sub(/^\/(?!\/)/,'').match(/([^\[]+)(\[[^\]]+\])?/).captures
|
265
171
|
|
266
172
|
remaining_path = ($').to_s
|
267
|
-
|
268
|
-
r = raw_path[/([^\/]+)(?=\/\/)/,1]
|
269
|
-
if r then
|
270
|
-
a_path = raw_path.split(/(?=\/\/)/,2)
|
271
|
-
else
|
272
|
-
a_path = raw_path.split('/',2)
|
273
|
-
end
|
274
|
-
|
173
|
+
a_path = raw_path.split('/')
|
275
174
|
condition = raw_condition if a_path.length <= 1
|
276
175
|
|
277
176
|
if raw_path[0,2] == '//' then
|
@@ -282,23 +181,20 @@ class Rexle
|
|
282
181
|
else
|
283
182
|
|
284
183
|
attribute = xpath_value[/^(attribute::|@)(.*)/,2]
|
285
|
-
|
286
|
-
|
287
|
-
return [@attributes[attribute.to_sym]] if attribute and @attributes and @attributes.has_key?(attribute.to_sym)
|
184
|
+
return [@attributes[attribute]] if attribute and @attributes and @attributes.has_key?(attribute)
|
185
|
+
|
288
186
|
s = a_path.shift
|
289
187
|
end
|
290
188
|
|
291
189
|
# isolate the xpath to return just the path to the current element
|
292
|
-
|
293
|
-
elmnt_path = s[/^([\w:\*]+\[[^\]]+\])|[\/]+{,2}[^\/]+/]
|
190
|
+
elmnt_path = s[/^([\w\*]+\[[^\]]+\])|[\/]+{,2}[^\/]+/]
|
294
191
|
element_part = elmnt_path[/(^@?[^\[]+)?/,1] if elmnt_path
|
295
|
-
|
192
|
+
|
296
193
|
if element_part then
|
297
194
|
unless element_part[/^@/] then
|
298
|
-
element_name = element_part
|
195
|
+
element_name = element_part
|
299
196
|
else
|
300
|
-
|
301
|
-
condition = xpath_value[/^\[/] ? xpath_value : element_part
|
197
|
+
condition = element_part
|
302
198
|
element_name = nil
|
303
199
|
end
|
304
200
|
|
@@ -306,32 +202,15 @@ class Rexle
|
|
306
202
|
|
307
203
|
#element_name ||= '*'
|
308
204
|
raw_condition = '' if condition
|
309
|
-
attr_search = format_condition(condition) if condition and condition.length > 0
|
310
|
-
|
311
|
-
attr_search2 = xpath_value[/^\[(.*)\]$/,1]
|
312
|
-
if attr_search2 then
|
313
|
-
r4 = attribute_search(attr_search, self, self.attributes)
|
314
|
-
return r4
|
315
|
-
end
|
316
|
-
|
317
|
-
return_elements = []
|
318
|
-
|
319
|
-
if raw_path[0,2] == '//' then
|
320
205
|
|
321
|
-
|
322
|
-
n = xpath_value[regex,1]
|
323
|
-
xpath_value.slice!(regex)
|
324
|
-
|
325
|
-
rs = scan_match(self, xpath_value).flatten.compact
|
326
|
-
return n ? rs[n.to_i-1] : rs
|
206
|
+
attr_search = format_condition(condition) if condition and condition.length > 0
|
327
207
|
|
328
|
-
|
329
|
-
return
|
208
|
+
if raw_path[0,2] == '//'
|
209
|
+
return scan_match(self, xpath_value)
|
330
210
|
else
|
331
211
|
|
332
|
-
return_elements = @child_lookup.map.with_index.select do |x|
|
333
|
-
|
334
|
-
(element_name == '*' && x[0].is_a?(Array))
|
212
|
+
return_elements = @child_lookup.map.with_index.select do |x|
|
213
|
+
x[0][0] == element_name or element_name == '*'
|
335
214
|
end
|
336
215
|
|
337
216
|
end
|
@@ -339,6 +218,7 @@ class Rexle
|
|
339
218
|
if return_elements.length > 0 then
|
340
219
|
|
341
220
|
if (a_path + [remaining_path]).join.empty? then
|
221
|
+
|
342
222
|
rlist = return_elements.map.with_index {|x,i| filter(x, i+1, attr_search, &blk)}.compact
|
343
223
|
rlist = rlist[0] if rlist.length == 1
|
344
224
|
else
|
@@ -348,9 +228,7 @@ class Rexle
|
|
348
228
|
rtn_element = filter(x, i+1, attr_search){|e| r = e.xpath(a_path.join('/') + raw_condition.to_s + remaining_path, &blk); (r || e) }
|
349
229
|
next if rtn_element.nil? or (rtn_element.is_a? Array and rtn_element.empty?)
|
350
230
|
|
351
|
-
if rtn_element.is_a?
|
352
|
-
rtn_element
|
353
|
-
elsif rtn_element.is_a? Array then
|
231
|
+
if rtn_element.is_a? Array then
|
354
232
|
rtn_element
|
355
233
|
elsif (rtn_element.is_a? String) || (rtn_element.is_a?(Array) and not(rtn_element[0].is_a? String))
|
356
234
|
rtn_element
|
@@ -359,26 +237,23 @@ class Rexle
|
|
359
237
|
end
|
360
238
|
end
|
361
239
|
#
|
362
|
-
|
363
240
|
rlist = rlist.flatten(1) unless rlist.length > 1 and rlist[0].is_a? Array
|
364
241
|
|
365
242
|
end
|
366
|
-
|
367
243
|
rlist.compact! if rlist.is_a? Array
|
368
244
|
|
369
245
|
else
|
370
246
|
|
371
247
|
# strip off the 1st element from the XPath
|
372
|
-
new_xpath = xpath_value[
|
373
|
-
|
248
|
+
new_xpath = xpath_value[/^\/\/\w+\/(.*)/,1]
|
249
|
+
|
374
250
|
if new_xpath then
|
375
251
|
self.xpath(new_xpath + raw_condition.to_s + remaining_path, rlist,&blk)
|
376
252
|
end
|
377
253
|
end
|
378
|
-
|
254
|
+
|
379
255
|
rlist = rlist.flatten(1) unless not(rlist.is_a? Array) or (rlist.length > 1 and rlist[0].is_a? Array)
|
380
256
|
rlist = [rlist] if rlist.is_a? Rexle::Element
|
381
|
-
rlist = (rlist.length > 0 ? true : false) if flag_func == true
|
382
257
|
rlist
|
383
258
|
end
|
384
259
|
|
@@ -398,11 +273,7 @@ class Rexle
|
|
398
273
|
end
|
399
274
|
|
400
275
|
def inspect()
|
401
|
-
if self.xml.length > 30 then
|
402
276
|
"%s ... </>" % self.xml[/<[^>]+>/]
|
403
|
-
else
|
404
|
-
self.xml
|
405
|
-
end
|
406
277
|
end
|
407
278
|
|
408
279
|
alias add add_element
|
@@ -422,28 +293,11 @@ class Rexle
|
|
422
293
|
|
423
294
|
def add_text(s) @value = s; self end
|
424
295
|
|
425
|
-
def attribute(key)
|
426
|
-
key = key.to_sym if key.is_a? String
|
427
|
-
@attributes[key].gsub('<','<').gsub('>','>')
|
428
|
-
end
|
429
|
-
|
296
|
+
def attribute(key) @attributes[key] end
|
430
297
|
def attributes() @attributes end
|
431
|
-
|
432
|
-
def children()
|
433
|
-
|
434
|
-
r = (@value.empty? ? [] : [@value]) + @child_elements
|
435
|
-
def r.is_an_empty_string?()
|
436
|
-
self.length == 1 and self.first == ''
|
437
|
-
end
|
438
|
-
|
439
|
-
return r
|
440
|
-
end
|
441
|
-
|
442
|
-
def children=(a) @child_elements = a end
|
443
|
-
|
444
|
-
def deep_clone() Rexle.new(self.xml).root end
|
445
|
-
def clone() Element.new(@name, @value, @attributes) end
|
446
|
-
|
298
|
+
def children() @child_elements end
|
299
|
+
def children=(a) @child_elements = a end
|
300
|
+
|
447
301
|
def delete(obj=nil)
|
448
302
|
if obj then
|
449
303
|
i = @child_elements.index(obj)
|
@@ -453,14 +307,11 @@ class Rexle
|
|
453
307
|
end
|
454
308
|
end
|
455
309
|
|
456
|
-
def element(s)
|
457
|
-
r = self.xpath(s)
|
458
|
-
r.is_a?(Array) ? r.first : r
|
459
|
-
end
|
310
|
+
def element(s) self.xpath(s).first end
|
460
311
|
|
461
312
|
def elements(s=nil)
|
462
313
|
procs = {
|
463
|
-
NilClass: proc {Elements.new(@child_elements
|
314
|
+
NilClass: proc {Elements.new(@child_elements)},
|
464
315
|
String: proc {|x| @child_elements[x]}
|
465
316
|
}
|
466
317
|
|
@@ -468,14 +319,7 @@ class Rexle
|
|
468
319
|
end
|
469
320
|
|
470
321
|
def doc_root() @rexle.root end
|
471
|
-
def each(&blk)
|
472
|
-
@child_elements.each(&blk) #unless @child_elements.empty?
|
473
|
-
end
|
474
|
-
def has_elements?() !self.elements.empty? end
|
475
|
-
|
476
|
-
def insert_after(node) insert(node, 1) end
|
477
|
-
def insert_before(node) insert(node) end
|
478
|
-
|
322
|
+
def each(&blk) @child_elements.each(&blk) end
|
479
323
|
def root() self end #@rexle.root end
|
480
324
|
|
481
325
|
def text(s='')
|
@@ -486,7 +330,6 @@ class Rexle
|
|
486
330
|
e = self.element(s)
|
487
331
|
result = e.value if e
|
488
332
|
end
|
489
|
-
result = CGI.unescape_html result.to_s
|
490
333
|
|
491
334
|
def result.unescape()
|
492
335
|
s = self.clone
|
@@ -496,14 +339,10 @@ class Rexle
|
|
496
339
|
|
497
340
|
result
|
498
341
|
end
|
499
|
-
|
500
|
-
def texts()
|
501
|
-
[@value] + @child_elements.select {|x| x.is_a? String}
|
502
|
-
end
|
503
342
|
|
504
343
|
def value=(raw_s)
|
505
344
|
|
506
|
-
@value =
|
345
|
+
@value = raw_s.to_s.clone
|
507
346
|
escape_chars = %w(& & < < > >).each_slice(2).to_a
|
508
347
|
escape_chars.each{|x| @value.gsub!(*x)}
|
509
348
|
|
@@ -523,27 +362,17 @@ class Rexle
|
|
523
362
|
method(msg).call(self.children)
|
524
363
|
end
|
525
364
|
|
526
|
-
alias to_s xml
|
527
365
|
|
528
366
|
private
|
529
|
-
|
530
|
-
def insert(node,offset=0)
|
531
|
-
i = parent.child_elements.index(self)
|
532
|
-
return unless i
|
533
|
-
parent.child_elements.insert(i+offset,node)
|
534
|
-
parent.child_lookup.insert(i+offset, [node.name, node.attributes, node.value])
|
535
|
-
self
|
536
|
-
end
|
537
367
|
|
538
368
|
def format_condition(condition)
|
539
|
-
|
540
|
-
raw_items = condition[1..-1].scan(/\'[^\']*\'
|
369
|
+
#raw_items = condition[1..-1].scan(/\'[^\']*\'|and|or|\d+|[!=]+|[@\w\.\/]+/)
|
370
|
+
raw_items = condition[1..-1].scan(/\'[^\']*\'|and|or|\d+|[!=<>]+|position\(\)|[@\w\.\/]+/)
|
541
371
|
|
542
372
|
if raw_items[0][/^\d+$/] then
|
543
373
|
return raw_items[0].to_i
|
544
374
|
elsif raw_items[0] == 'position()' then
|
545
|
-
|
546
|
-
return rrr
|
375
|
+
return "i %s %s" % raw_items[1..-1]
|
547
376
|
else
|
548
377
|
|
549
378
|
andor_items = raw_items.map.with_index.select{|x,i| x[/\band\b|\bor\b/]}.map{|x| [x.last, x.last + 1]}.flatten
|
@@ -558,10 +387,9 @@ class Rexle
|
|
558
387
|
|
559
388
|
if x.length >= 3 then
|
560
389
|
x[1] = '==' if x[1] == '='
|
561
|
-
"h[
|
390
|
+
"h['%s'] %s %s" % x
|
562
391
|
else
|
563
|
-
|
564
|
-
x.join[/^(and|or)$/] ? x : ("h[:'%s']" % x)
|
392
|
+
x
|
565
393
|
end
|
566
394
|
end
|
567
395
|
|
@@ -572,7 +400,6 @@ class Rexle
|
|
572
400
|
items = cons_items.map do |x|
|
573
401
|
|
574
402
|
if x.length >= 3 then
|
575
|
-
|
576
403
|
x[1] = '==' if x[1] == '='
|
577
404
|
if x[0] != '.' then
|
578
405
|
if x[0][/\//] then
|
@@ -580,7 +407,7 @@ class Rexle
|
|
580
407
|
|
581
408
|
"e.xpath('#{path}').first.value == #{value}"
|
582
409
|
else
|
583
|
-
"(name == '%s' and value %s
|
410
|
+
"(name == '%s' and value %s %s)" % [x[0], x[1], x[2]]
|
584
411
|
end
|
585
412
|
else
|
586
413
|
"e.value %s %s" % [x[1], x[2]]
|
@@ -589,7 +416,7 @@ class Rexle
|
|
589
416
|
x
|
590
417
|
end
|
591
418
|
end
|
592
|
-
|
419
|
+
|
593
420
|
return items.join(' ')
|
594
421
|
end
|
595
422
|
end
|
@@ -598,46 +425,35 @@ class Rexle
|
|
598
425
|
end
|
599
426
|
|
600
427
|
|
601
|
-
def scan_match(node,
|
602
|
-
|
603
|
-
r = []
|
604
|
-
xpath2 = path[2..-1]
|
605
|
-
xpath2.sub!(/^\*\//,'')
|
606
|
-
xpath2.sub!(/^\*/,self.name)
|
607
|
-
xpath2.sub!(/^\w+/,'').sub!(/^\//,'') if xpath2[/^\w+/] == self.name
|
608
|
-
|
428
|
+
def scan_match(node, xpath)
|
609
429
|
|
610
|
-
r
|
611
|
-
r << node.
|
430
|
+
r = node.xpath(xpath[2..-1])
|
431
|
+
r << node.children.map {|n| scan_match(n, xpath)}
|
432
|
+
#puts 'r: ' + r.inspect
|
612
433
|
r
|
613
434
|
end
|
614
|
-
|
615
|
-
# used by xpath function contains()
|
616
|
-
#
|
617
|
-
def scan_contents(node)
|
618
|
-
|
619
|
-
a = []
|
620
|
-
a << node.text
|
621
|
-
|
622
|
-
node.elements.each do |child|
|
623
|
-
a.concat scan_contents(child)
|
624
|
-
end
|
625
|
-
a
|
626
|
-
end
|
627
|
-
|
628
435
|
|
629
436
|
def filter(raw_element, i, attr_search, &blk)
|
630
437
|
|
631
438
|
x = raw_element
|
632
439
|
e = @child_elements[x.last]
|
633
440
|
|
634
|
-
|
635
|
-
name, value = e.name, e.value if e.is_a? Rexle::Element
|
441
|
+
h = x[0][1] # <-- fetch the attributes
|
636
442
|
|
637
|
-
h = x[0][1] # <-- fetch the attributes
|
638
|
-
|
639
443
|
if attr_search then
|
640
|
-
|
444
|
+
if attr_search.is_a? Fixnum then
|
445
|
+
block_given? ? blk.call(e) : e if i == attr_search
|
446
|
+
elsif attr_search[/i\s[<>\=]\s\d+/] and eval(attr_search) then
|
447
|
+
block_given? ? blk.call(e) : e
|
448
|
+
elsif h and attr_search[/^h\[/] and eval(attr_search)
|
449
|
+
block_given? ? blk.call(e) : e
|
450
|
+
elsif attr_search[/^\(name ==/] and e.child_lookup.select{|name, attributes, value| eval(attr_search) }.length > 0
|
451
|
+
block_given? ? blk.call(e) : e
|
452
|
+
elsif attr_search[/^e\.value/] and eval(attr_search)
|
453
|
+
block_given? ? blk.call(e) : e
|
454
|
+
elsif attr_search[/^e\.xpath/] and eval(attr_search)
|
455
|
+
block_given? ? blk.call(e) : e
|
456
|
+
end
|
641
457
|
else
|
642
458
|
|
643
459
|
block_given? ? blk.call(e) : e
|
@@ -645,28 +461,9 @@ class Rexle
|
|
645
461
|
|
646
462
|
end
|
647
463
|
|
648
|
-
def attribute_search(attr_search, e, h, i=nil, &blk)
|
649
|
-
if attr_search.is_a? Fixnum then
|
650
|
-
block_given? ? blk.call(e) : e if i == attr_search
|
651
|
-
elsif attr_search[/i\s[<>\=]\s\d+/] and eval(attr_search) then
|
652
|
-
block_given? ? blk.call(e) : e
|
653
|
-
elsif h and attr_search[/^h\[/] and eval(attr_search)
|
654
|
-
block_given? ? blk.call(e) : e
|
655
|
-
elsif attr_search[/^\(name ==/] and e.child_lookup.select{|name, attributes, value| eval(attr_search) }.length > 0
|
656
|
-
block_given? ? blk.call(e) : e
|
657
|
-
elsif attr_search[/^\(name ==/] and eval(attr_search)
|
658
|
-
block_given? ? blk.call(e) : e
|
659
|
-
elsif attr_search[/^e\.value/] and eval(attr_search)
|
660
|
-
block_given? ? blk.call(e) : e
|
661
|
-
elsif attr_search[/^e\.xpath/] and eval(attr_search)
|
662
|
-
block_given? ? blk.call(e) : e
|
663
|
-
end
|
664
|
-
end
|
665
464
|
end # -- end of element --
|
666
465
|
|
667
466
|
class Elements
|
668
|
-
include Enumerable
|
669
|
-
|
670
467
|
def initialize(elements=[])
|
671
468
|
super()
|
672
469
|
@elements = elements
|
@@ -675,10 +472,6 @@ class Rexle
|
|
675
472
|
def [](i)
|
676
473
|
@elements[i-1]
|
677
474
|
end
|
678
|
-
|
679
|
-
def each(&blk) @elements.each(&blk) end
|
680
|
-
def to_a() @elements end
|
681
|
-
|
682
475
|
end # -- end of elements --
|
683
476
|
|
684
477
|
|
@@ -696,9 +489,8 @@ class Rexle
|
|
696
489
|
else
|
697
490
|
a = yield
|
698
491
|
end
|
699
|
-
|
700
|
-
@
|
701
|
-
@doc = scan_element(*(doc_node << @a))
|
492
|
+
|
493
|
+
@doc = scan_element(*a)
|
702
494
|
self
|
703
495
|
end
|
704
496
|
|
@@ -710,18 +502,14 @@ class Rexle
|
|
710
502
|
|
711
503
|
alias add add_element
|
712
504
|
|
713
|
-
def delete(xpath)
|
714
|
-
e = @doc.element(xpath)
|
715
|
-
e.delete if e
|
716
|
-
end
|
717
|
-
|
505
|
+
def delete(xpath) @doc.element(xpath).delete end
|
718
506
|
def element(xpath) self.xpath(xpath).first end
|
719
507
|
def elements(s=nil) @doc.elements(s) end
|
720
508
|
def name() @doc.root.name end
|
721
509
|
def to_a() @a end
|
722
510
|
def to_s(options={}) self.xml options end
|
723
511
|
def text(xpath) @doc.text(xpath) end
|
724
|
-
def root() @doc
|
512
|
+
def root() @doc end
|
725
513
|
|
726
514
|
def write(f)
|
727
515
|
f.write xml
|
@@ -730,11 +518,9 @@ class Rexle
|
|
730
518
|
def xml(options={})
|
731
519
|
o = {pretty: false, declaration: true}.merge(options)
|
732
520
|
msg = o[:pretty] == false ? :doc_print : :doc_pretty_print
|
733
|
-
|
734
521
|
r = ''
|
735
522
|
r = "<?xml version='1.0' encoding='UTF-8'?>\n" if o[:declaration] == true
|
736
523
|
r << method(msg).call(self.root.children)
|
737
|
-
|
738
524
|
r
|
739
525
|
end
|
740
526
|
|
@@ -751,23 +537,20 @@ class Rexle
|
|
751
537
|
if recordx_type then
|
752
538
|
procs = {
|
753
539
|
'dynarex' => proc {|x| DynarexParser.new(x).to_a},
|
754
|
-
'polyrex' => proc {|x| PolyrexParser.new(x).to_a}
|
755
|
-
'polyrex' => proc {|x| RexleParser.new(x).to_a}
|
540
|
+
'polyrex' => proc {|x| PolyrexParser.new(x).to_a}
|
756
541
|
}
|
757
542
|
procs[recordx_type].call(x)
|
758
543
|
else
|
759
|
-
|
760
544
|
RexleParser.new(x).to_a
|
761
545
|
end
|
762
546
|
else
|
763
|
-
|
764
547
|
RexleParser.new(x).to_a
|
765
548
|
end
|
766
549
|
|
767
550
|
end
|
768
551
|
|
769
552
|
def scan_element(name, value=nil, attributes=nil, *children)
|
770
|
-
|
553
|
+
#puts 'name : ' + name.inspect
|
771
554
|
element = Element.new(name, value, attributes, self)
|
772
555
|
|
773
556
|
if children then
|
@@ -782,6 +565,14 @@ class Rexle
|
|
782
565
|
return element
|
783
566
|
end
|
784
567
|
|
568
|
+
def count(path)
|
569
|
+
#puts 'path : ' + path
|
570
|
+
length = @doc.xpath(path).flatten.compact.length
|
571
|
+
#puts 'length = ' + length
|
572
|
+
length
|
573
|
+
end
|
574
|
+
|
575
|
+
def max(path) @doc.xpath(path).flatten.compact.map(&:to_i).max end
|
785
576
|
|
786
577
|
# scan a rexml doc
|
787
578
|
#
|
metadata
CHANGED
@@ -2,7 +2,7 @@
|
|
2
2
|
name: rexle
|
3
3
|
version: !ruby/object:Gem::Version
|
4
4
|
prerelease:
|
5
|
-
version: 0.9.
|
5
|
+
version: 0.9.6
|
6
6
|
platform: ruby
|
7
7
|
authors:
|
8
8
|
- James Robertson
|
@@ -10,7 +10,8 @@ autorequire:
|
|
10
10
|
bindir: bin
|
11
11
|
cert_chain: []
|
12
12
|
|
13
|
-
date:
|
13
|
+
date: 2011-07-19 00:00:00 +01:00
|
14
|
+
default_executable:
|
14
15
|
dependencies:
|
15
16
|
- !ruby/object:Gem::Dependency
|
16
17
|
name: rexleparser
|
@@ -46,7 +47,7 @@ dependencies:
|
|
46
47
|
type: :runtime
|
47
48
|
version_requirements: *id003
|
48
49
|
- !ruby/object:Gem::Dependency
|
49
|
-
name:
|
50
|
+
name: nokogiri
|
50
51
|
prerelease: false
|
51
52
|
requirement: &id004 !ruby/object:Gem::Requirement
|
52
53
|
none: false
|
@@ -56,6 +57,17 @@ dependencies:
|
|
56
57
|
version: "0"
|
57
58
|
type: :runtime
|
58
59
|
version_requirements: *id004
|
60
|
+
- !ruby/object:Gem::Dependency
|
61
|
+
name: rexle-builder
|
62
|
+
prerelease: false
|
63
|
+
requirement: &id005 !ruby/object:Gem::Requirement
|
64
|
+
none: false
|
65
|
+
requirements:
|
66
|
+
- - ">="
|
67
|
+
- !ruby/object:Gem::Version
|
68
|
+
version: "0"
|
69
|
+
type: :runtime
|
70
|
+
version_requirements: *id005
|
59
71
|
description:
|
60
72
|
email:
|
61
73
|
executables: []
|
@@ -66,6 +78,7 @@ extra_rdoc_files: []
|
|
66
78
|
|
67
79
|
files:
|
68
80
|
- lib/rexle.rb
|
81
|
+
has_rdoc: true
|
69
82
|
homepage:
|
70
83
|
licenses: []
|
71
84
|
|
@@ -89,9 +102,9 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
89
102
|
requirements: []
|
90
103
|
|
91
104
|
rubyforge_project:
|
92
|
-
rubygems_version: 1.
|
105
|
+
rubygems_version: 1.5.2
|
93
106
|
signing_key:
|
94
107
|
specification_version: 3
|
95
|
-
summary:
|
108
|
+
summary: rexle
|
96
109
|
test_files: []
|
97
110
|
|