feedtools 0.2.26 → 0.2.27
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +232 -216
- data/db/migration.rb +2 -0
- data/db/schema.mysql.sql +2 -0
- data/db/schema.postgresql.sql +3 -1
- data/db/schema.sqlite.sql +3 -1
- data/lib/feed_tools.rb +37 -14
- data/lib/feed_tools/database_feed_cache.rb +13 -2
- data/lib/feed_tools/feed.rb +430 -104
- data/lib/feed_tools/feed_item.rb +533 -268
- data/lib/feed_tools/helpers/generic_helper.rb +1 -1
- data/lib/feed_tools/helpers/html_helper.rb +78 -116
- data/lib/feed_tools/helpers/retrieval_helper.rb +33 -3
- data/lib/feed_tools/helpers/uri_helper.rb +46 -54
- data/lib/feed_tools/monkey_patch.rb +27 -1
- data/lib/feed_tools/vendor/html5/History.txt +10 -0
- data/lib/feed_tools/vendor/html5/Manifest.txt +117 -0
- data/lib/feed_tools/vendor/html5/README +45 -0
- data/lib/feed_tools/vendor/html5/Rakefile.rb +33 -0
- data/lib/feed_tools/vendor/html5/bin/html5 +217 -0
- data/lib/feed_tools/vendor/html5/lib/core_ext/string.rb +17 -0
- data/lib/feed_tools/vendor/html5/lib/html5.rb +13 -0
- data/lib/feed_tools/vendor/html5/lib/html5/constants.rb +1046 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/base.rb +10 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/inject_meta_charset.rb +82 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/iso639codes.rb +752 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/optionaltags.rb +198 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/rfc2046.rb +30 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/rfc3987.rb +89 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/sanitizer.rb +15 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/validator.rb +830 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/whitespace.rb +36 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser.rb +248 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/after_body_phase.rb +46 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/after_frameset_phase.rb +33 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/after_head_phase.rb +50 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/before_head_phase.rb +41 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_body_phase.rb +613 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_caption_phase.rb +69 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_cell_phase.rb +78 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_column_group_phase.rb +55 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_frameset_phase.rb +57 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_head_phase.rb +138 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_row_phase.rb +89 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_select_phase.rb +85 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_table_body_phase.rb +86 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_table_phase.rb +115 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/initial_phase.rb +133 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/phase.rb +154 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/root_element_phase.rb +41 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/trailing_end_phase.rb +35 -0
- data/lib/feed_tools/vendor/html5/lib/html5/inputstream.rb +648 -0
- data/lib/feed_tools/vendor/html5/lib/html5/liberalxmlparser.rb +158 -0
- data/lib/feed_tools/vendor/html5/lib/html5/sanitizer.rb +188 -0
- data/lib/feed_tools/vendor/html5/lib/html5/serializer.rb +2 -0
- data/lib/feed_tools/vendor/html5/lib/html5/serializer/htmlserializer.rb +179 -0
- data/lib/feed_tools/vendor/html5/lib/html5/serializer/xhtmlserializer.rb +20 -0
- data/lib/feed_tools/vendor/html5/lib/html5/sniffer.rb +45 -0
- data/lib/feed_tools/vendor/html5/lib/html5/tokenizer.rb +966 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders.rb +24 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/base.rb +334 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/hpricot.rb +231 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/rexml.rb +209 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/simpletree.rb +185 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers.rb +26 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/base.rb +162 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/hpricot.rb +48 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/rexml.rb +48 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/simpletree.rb +48 -0
- data/lib/feed_tools/vendor/html5/lib/html5/version.rb +3 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/chardet/test_big5.txt +51 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/test-yahoo-jp.dat +10 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/tests1.dat +394 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/tests2.dat +81 -0
- data/lib/feed_tools/vendor/html5/testdata/sanitizer/tests1.dat +416 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/core.test +104 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/injectmeta.test +65 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/optionaltags.test +900 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/options.test +60 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/whitespace.test +51 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/google-results.htm +1 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/python-ref-import.htm +1 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/web-apps-old.htm +1 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/web-apps.htm +34275 -0
- data/lib/feed_tools/vendor/html5/testdata/sniffer/htmlOrFeed.json +43 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/contentModelFlags.test +48 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/entities.test +2339 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/escapeFlag.test +21 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test1.test +172 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test2.test +129 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test3.test +367 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test4.test +198 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests1.dat +1950 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests2.dat +773 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests3.dat +270 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests4.dat +60 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests5.dat +175 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests6.dat +196 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/attributes.test +1035 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/base-href-attribute.test +787 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/base-target-attribute.test +35 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/blockquote-cite-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/classattribute.test +152 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/contenteditableattribute.test +59 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/contextmenuattribute.test +115 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/dirattribute.test +59 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/draggableattribute.test +63 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/html-xmlns-attribute.test +23 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/idattribute.test +115 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/inputattributes.test +2795 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/irrelevantattribute.test +63 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/langattribute.test +5579 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/li-value-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/link-href-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/link-hreflang-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/link-rel-attribute.test +271 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/ol-start-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/starttags.test +375 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/style-scoped-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/tabindexattribute.test +79 -0
- data/lib/feed_tools/vendor/html5/tests/preamble.rb +72 -0
- data/lib/feed_tools/vendor/html5/tests/test_encoding.rb +35 -0
- data/lib/feed_tools/vendor/html5/tests/test_lxp.rb +279 -0
- data/lib/feed_tools/vendor/html5/tests/test_parser.rb +68 -0
- data/lib/feed_tools/vendor/html5/tests/test_sanitizer.rb +142 -0
- data/lib/feed_tools/vendor/html5/tests/test_serializer.rb +68 -0
- data/lib/feed_tools/vendor/html5/tests/test_sniffer.rb +27 -0
- data/lib/feed_tools/vendor/html5/tests/test_stream.rb +62 -0
- data/lib/feed_tools/vendor/html5/tests/test_tokenizer.rb +94 -0
- data/lib/feed_tools/vendor/html5/tests/test_treewalkers.rb +135 -0
- data/lib/feed_tools/vendor/html5/tests/test_validator.rb +31 -0
- data/lib/feed_tools/vendor/html5/tests/tokenizer_test_parser.rb +63 -0
- data/lib/feed_tools/vendor/uri.rb +781 -0
- data/lib/feed_tools/version.rb +1 -1
- data/rakefile +27 -6
- data/test/unit/atom_test.rb +298 -210
- data/test/unit/helper_test.rb +7 -12
- data/test/unit/rdf_test.rb +51 -1
- data/test/unit/rss_test.rb +13 -3
- metadata +239 -116
- data/lib/feed_tools/vendor/htree.rb +0 -97
- data/lib/feed_tools/vendor/htree/container.rb +0 -10
- data/lib/feed_tools/vendor/htree/context.rb +0 -67
- data/lib/feed_tools/vendor/htree/display.rb +0 -27
- data/lib/feed_tools/vendor/htree/doc.rb +0 -149
- data/lib/feed_tools/vendor/htree/elem.rb +0 -262
- data/lib/feed_tools/vendor/htree/encoder.rb +0 -163
- data/lib/feed_tools/vendor/htree/equality.rb +0 -218
- data/lib/feed_tools/vendor/htree/extract_text.rb +0 -37
- data/lib/feed_tools/vendor/htree/fstr.rb +0 -33
- data/lib/feed_tools/vendor/htree/gencode.rb +0 -97
- data/lib/feed_tools/vendor/htree/htmlinfo.rb +0 -672
- data/lib/feed_tools/vendor/htree/inspect.rb +0 -108
- data/lib/feed_tools/vendor/htree/leaf.rb +0 -94
- data/lib/feed_tools/vendor/htree/loc.rb +0 -367
- data/lib/feed_tools/vendor/htree/modules.rb +0 -48
- data/lib/feed_tools/vendor/htree/name.rb +0 -124
- data/lib/feed_tools/vendor/htree/output.rb +0 -207
- data/lib/feed_tools/vendor/htree/parse.rb +0 -409
- data/lib/feed_tools/vendor/htree/raw_string.rb +0 -124
- data/lib/feed_tools/vendor/htree/regexp-util.rb +0 -15
- data/lib/feed_tools/vendor/htree/rexml.rb +0 -130
- data/lib/feed_tools/vendor/htree/scan.rb +0 -166
- data/lib/feed_tools/vendor/htree/tag.rb +0 -111
- data/lib/feed_tools/vendor/htree/template.rb +0 -909
- data/lib/feed_tools/vendor/htree/text.rb +0 -115
- data/lib/feed_tools/vendor/htree/traverse.rb +0 -465
@@ -1,163 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'iconv'
|
3
|
-
|
4
|
-
module HTree
|
5
|
-
class Encoder
|
6
|
-
# HTree::Encoder.internal_charset returns the MIME charset corresponding to $KCODE.
|
7
|
-
#
|
8
|
-
# - 'ISO-8859-1' when $KCODE=='NONE'
|
9
|
-
# - 'UTF-8' when $KCODE=='UTF8'
|
10
|
-
# - 'EUC-JP' when $KCODE=='EUC'
|
11
|
-
# - 'Shift_JIS' when $KCODE=='SJIS'
|
12
|
-
#
|
13
|
-
# This mapping ignores EUC-KR and various single byte charset other than ISO-8859-1 at least.
|
14
|
-
# This should be fixed when Ruby is m17nized.
|
15
|
-
def Encoder.internal_charset
|
16
|
-
KcodeCharset[$KCODE]
|
17
|
-
end
|
18
|
-
|
19
|
-
def initialize(output_encoding, internal_encoding=HTree::Encoder.internal_charset)
|
20
|
-
@buf = ''
|
21
|
-
@internal_encoding = internal_encoding
|
22
|
-
@output_encoding = output_encoding
|
23
|
-
@ic = Iconv.new(output_encoding, @internal_encoding)
|
24
|
-
@charpat = FirstCharPattern[internal_encoding]
|
25
|
-
|
26
|
-
@subcharset_list = SubCharset[output_encoding] || []
|
27
|
-
@subcharset_ic = {}
|
28
|
-
@subcharset_list.each {|subcharset|
|
29
|
-
@subcharset_ic[subcharset] = Iconv.new(subcharset, @internal_encoding)
|
30
|
-
}
|
31
|
-
end
|
32
|
-
|
33
|
-
def output_string(internal_str, external_str=@ic.iconv(internal_str))
|
34
|
-
@buf << external_str
|
35
|
-
@subcharset_ic.reject! {|subcharset, ic|
|
36
|
-
begin
|
37
|
-
ic.iconv(internal_str) != external_str
|
38
|
-
rescue Iconv::Failure
|
39
|
-
true
|
40
|
-
end
|
41
|
-
}
|
42
|
-
nil
|
43
|
-
end
|
44
|
-
|
45
|
-
def output_text(string)
|
46
|
-
begin
|
47
|
-
output_string string, @ic.iconv(string)
|
48
|
-
rescue Iconv::IllegalSequence, Iconv::InvalidCharacter => e
|
49
|
-
output_string string[0, string.length - e.failed.length], e.success
|
50
|
-
unless @charpat =~ e.failed
|
51
|
-
# xxx: shoule be configulable?
|
52
|
-
#raise ArgumentError, "cannot extract first character: #{e.failed.dump}"
|
53
|
-
string = e.failed[1, e.failed.length-1]
|
54
|
-
output_string '?'
|
55
|
-
retry
|
56
|
-
end
|
57
|
-
char = $&
|
58
|
-
rest = $'
|
59
|
-
begin
|
60
|
-
ucode = Iconv.conv("UTF-8", @internal_encoding, char).unpack("U")[0]
|
61
|
-
char = "&##{ucode};"
|
62
|
-
rescue Iconv::IllegalSequence, Iconv::InvalidCharacter
|
63
|
-
# xxx: shoule be configulable?
|
64
|
-
char = '?'
|
65
|
-
end
|
66
|
-
output_string char
|
67
|
-
string = rest
|
68
|
-
retry
|
69
|
-
end
|
70
|
-
end
|
71
|
-
|
72
|
-
ChRef = {
|
73
|
-
'&' => '&',
|
74
|
-
'<' => '<',
|
75
|
-
'>' => '>',
|
76
|
-
'"' => '"',
|
77
|
-
}
|
78
|
-
|
79
|
-
def output_dynamic_text(string)
|
80
|
-
if string.respond_to? :rcdata
|
81
|
-
output_text(string.rcdata.gsub(/[<>]/) { ChRef[$&] })
|
82
|
-
else
|
83
|
-
output_text(string.to_s.gsub(/[&<>]/) { ChRef[$&] })
|
84
|
-
end
|
85
|
-
end
|
86
|
-
|
87
|
-
def output_dynamic_attvalue(string)
|
88
|
-
if string.respond_to? :rcdata
|
89
|
-
output_text(string.rcdata.gsub(/[<>"]/) { ChRef[$&] })
|
90
|
-
else
|
91
|
-
output_text(string.to_s.gsub(/[&<>"]/) { ChRef[$&] })
|
92
|
-
end
|
93
|
-
end
|
94
|
-
|
95
|
-
def finish
|
96
|
-
external_str = @ic.close
|
97
|
-
@buf << external_str
|
98
|
-
@subcharset_ic.reject! {|subcharset, ic|
|
99
|
-
begin
|
100
|
-
ic.close != external_str
|
101
|
-
rescue Iconv::Failure
|
102
|
-
true
|
103
|
-
end
|
104
|
-
}
|
105
|
-
@buf
|
106
|
-
end
|
107
|
-
|
108
|
-
def finish_with_xmldecl
|
109
|
-
content = finish
|
110
|
-
xmldecl = Iconv.conv(@output_encoding, 'US-ASCII',
|
111
|
-
"<?xml version=\"1.0\" encoding=\"#{minimal_charset}\"?>")
|
112
|
-
xmldecl + content
|
113
|
-
end
|
114
|
-
|
115
|
-
def minimal_charset
|
116
|
-
@subcharset_list.each {|subcharset|
|
117
|
-
if @subcharset_ic.include? subcharset
|
118
|
-
return subcharset
|
119
|
-
end
|
120
|
-
}
|
121
|
-
@output_encoding
|
122
|
-
end
|
123
|
-
|
124
|
-
KcodeCharset = {
|
125
|
-
'EUC' => 'EUC-JP',
|
126
|
-
'SJIS' => 'Shift_JIS',
|
127
|
-
'UTF8' => 'UTF-8',
|
128
|
-
'NONE' => 'ISO-8859-1',
|
129
|
-
}
|
130
|
-
|
131
|
-
FirstCharPattern = {
|
132
|
-
'EUC-JP' => /\A(?:
|
133
|
-
[\x00-\x7f]
|
134
|
-
|[\xa1-\xfe][\xa1-\xfe]
|
135
|
-
|\x8e[\xa1-\xfe]
|
136
|
-
|\x8f[\xa1-\xfe][\xa1-\xfe])/nx,
|
137
|
-
'Shift_JIS' => /\A(?:
|
138
|
-
[\x00-\x7f]
|
139
|
-
|[\x81-\x9f][\x40-\x7e\x80-\xfc]
|
140
|
-
|[\xa1-\xdf]
|
141
|
-
|[\xe0-\xfc][\x40-\x7e\x80-\xfc])/nx,
|
142
|
-
'UTF-8' => /\A(?:
|
143
|
-
[\x00-\x7f]
|
144
|
-
|[\xc0-\xdf][\x80-\xbf]
|
145
|
-
|[\xe0-\xef][\x80-\xbf][\x80-\xbf]
|
146
|
-
|[\xf0-\xf7][\x80-\xbf][\x80-\xbf][\x80-\xbf]
|
147
|
-
|[\xf8-\xfb][\x80-\xbf][\x80-\xbf][\x80-\xbf][\x80-\xbf]
|
148
|
-
|[\xfc-\xfd][\x80-\xbf][\x80-\xbf][\x80-\xbf][\x80-\xbf][\x80-\xbf])/nx,
|
149
|
-
'ISO-8859-1' => /\A[\x00-\xff]/n
|
150
|
-
}
|
151
|
-
|
152
|
-
SubCharset = {
|
153
|
-
'ISO-2022-JP-2' => ['US-ASCII', 'ISO-2022-JP'],
|
154
|
-
'ISO-2022-JP-3' => ['US-ASCII', 'ISO-2022-JP'],
|
155
|
-
'UTF-16BE' => [],
|
156
|
-
'UTF-16LE' => [],
|
157
|
-
'UTF-16' => [],
|
158
|
-
}
|
159
|
-
SubCharset.default = ['US-ASCII']
|
160
|
-
|
161
|
-
end
|
162
|
-
end
|
163
|
-
# :startdoc:
|
@@ -1,218 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'htree/doc'
|
3
|
-
require 'htree/elem'
|
4
|
-
require 'htree/leaf'
|
5
|
-
require 'htree/tag'
|
6
|
-
require 'htree/raw_string'
|
7
|
-
require 'htree/context'
|
8
|
-
|
9
|
-
module HTree
|
10
|
-
# compare tree structures.
|
11
|
-
def ==(other)
|
12
|
-
check_equality(self, other, :usual_equal_object)
|
13
|
-
end
|
14
|
-
alias eql? ==
|
15
|
-
|
16
|
-
# hash value for the tree structure.
|
17
|
-
def hash
|
18
|
-
return @hash_code if defined? @hash_code
|
19
|
-
@hash_code = usual_equal_object.hash
|
20
|
-
end
|
21
|
-
|
22
|
-
def usual_equal_object
|
23
|
-
return @usual_equal_object if defined? @usual_equal_object
|
24
|
-
@usual_equal_object = make_usual_equal_object
|
25
|
-
end
|
26
|
-
|
27
|
-
def make_usual_equal_object
|
28
|
-
raise NotImplementedError
|
29
|
-
end
|
30
|
-
|
31
|
-
def exact_equal_object
|
32
|
-
return @exact_equal_object if defined? @exact_equal_object
|
33
|
-
@exact_equal_object = make_exact_equal_object
|
34
|
-
end
|
35
|
-
|
36
|
-
def make_exact_equal_object
|
37
|
-
raise NotImplementedError
|
38
|
-
end
|
39
|
-
|
40
|
-
def exact_equal?(other)
|
41
|
-
check_equality(self, other, :exact_equal_object)
|
42
|
-
end
|
43
|
-
|
44
|
-
def check_equality(obj1, obj2, equal_object_method)
|
45
|
-
return false unless obj1.class == obj2.class
|
46
|
-
if obj1.class == Array
|
47
|
-
return false unless obj1.length == obj2.length
|
48
|
-
obj1.each_with_index {|c1, i|
|
49
|
-
return false unless c1.class == obj2[i].class
|
50
|
-
}
|
51
|
-
obj1.each_with_index {|c1, i|
|
52
|
-
return false unless check_equality(c1, obj2[i], equal_object_method)
|
53
|
-
}
|
54
|
-
true
|
55
|
-
elsif obj1.respond_to? equal_object_method
|
56
|
-
o1 = obj1.send(equal_object_method)
|
57
|
-
o2 = obj2.send(equal_object_method)
|
58
|
-
check_equality(o1, o2, equal_object_method)
|
59
|
-
else
|
60
|
-
obj1 == obj2
|
61
|
-
end
|
62
|
-
end
|
63
|
-
|
64
|
-
class Doc
|
65
|
-
alias exact_equal_object children
|
66
|
-
alias usual_equal_object children
|
67
|
-
end
|
68
|
-
|
69
|
-
class Elem
|
70
|
-
def make_exact_equal_object
|
71
|
-
[@stag, @children, @empty, @etag]
|
72
|
-
end
|
73
|
-
|
74
|
-
def make_usual_equal_object
|
75
|
-
[@stag, @children]
|
76
|
-
end
|
77
|
-
end
|
78
|
-
|
79
|
-
class Name
|
80
|
-
def make_exact_equal_object
|
81
|
-
[@namespace_prefix, @namespace_uri, @local_name]
|
82
|
-
end
|
83
|
-
|
84
|
-
def make_usual_equal_object
|
85
|
-
xmlns? ? @local_name : [@namespace_uri, @local_name]
|
86
|
-
end
|
87
|
-
end
|
88
|
-
|
89
|
-
module Util
|
90
|
-
module_function
|
91
|
-
def cmp_with_nil(a, b)
|
92
|
-
if a == nil
|
93
|
-
if b == nil
|
94
|
-
0
|
95
|
-
else
|
96
|
-
-1
|
97
|
-
end
|
98
|
-
else
|
99
|
-
if b == nil
|
100
|
-
1
|
101
|
-
else
|
102
|
-
a <=> b
|
103
|
-
end
|
104
|
-
end
|
105
|
-
end
|
106
|
-
end
|
107
|
-
|
108
|
-
class Context
|
109
|
-
def make_exact_equal_object
|
110
|
-
@namespaces.keys.sort {|prefix1, prefix2|
|
111
|
-
Util.cmp_with_nil(prefix1, prefix2)
|
112
|
-
}.map {|prefix| [prefix, @namespaces[prefix]] }
|
113
|
-
end
|
114
|
-
|
115
|
-
# make_usual_equal_object is not used through STag#make_usual_equal_object
|
116
|
-
# NotImplementedError is suitable?
|
117
|
-
alias make_usual_equal_object make_exact_equal_object
|
118
|
-
end
|
119
|
-
|
120
|
-
class STag
|
121
|
-
def make_exact_equal_object
|
122
|
-
[@raw_string,
|
123
|
-
@name,
|
124
|
-
@attributes.sort {|(n1,t1), (n2, t2)|
|
125
|
-
Util.cmp_with_nil(n1.namespace_prefix, n2.namespace_prefix).nonzero? ||
|
126
|
-
Util.cmp_with_nil(n1.namespace_uri, n2.namespace_uri).nonzero? ||
|
127
|
-
Util.cmp_with_nil(n1.local_name, n2.local_name)
|
128
|
-
},
|
129
|
-
@inherited_context
|
130
|
-
]
|
131
|
-
end
|
132
|
-
|
133
|
-
def make_usual_equal_object
|
134
|
-
[@name,
|
135
|
-
@attributes.find_all {|n,t| !n.xmlns? }.sort {|(n1,t1), (n2, t2)|
|
136
|
-
Util.cmp_with_nil(n1.namespace_prefix, n2.namespace_prefix).nonzero? ||
|
137
|
-
Util.cmp_with_nil(n1.namespace_uri, n2.namespace_uri).nonzero? ||
|
138
|
-
Util.cmp_with_nil(n1.local_name, n2.local_name)
|
139
|
-
}
|
140
|
-
]
|
141
|
-
end
|
142
|
-
|
143
|
-
end
|
144
|
-
|
145
|
-
class ETag
|
146
|
-
def make_exact_equal_object
|
147
|
-
[@raw_string, @qualified_name]
|
148
|
-
end
|
149
|
-
|
150
|
-
alias usual_equal_object qualified_name
|
151
|
-
end
|
152
|
-
|
153
|
-
class Text
|
154
|
-
def make_exact_equal_object
|
155
|
-
[@raw_string, @rcdata]
|
156
|
-
end
|
157
|
-
|
158
|
-
def make_usual_equal_object
|
159
|
-
@normalized_rcdata
|
160
|
-
end
|
161
|
-
end
|
162
|
-
|
163
|
-
class XMLDecl
|
164
|
-
def make_exact_equal_object
|
165
|
-
[@raw_string, @version, @encoding, @standalone]
|
166
|
-
end
|
167
|
-
|
168
|
-
def make_usual_equal_object
|
169
|
-
[@version, @encoding, @standalone]
|
170
|
-
end
|
171
|
-
end
|
172
|
-
|
173
|
-
class DocType
|
174
|
-
def make_exact_equal_object
|
175
|
-
[@raw_string, @root_element_name, @system_identifier, @public_identifier]
|
176
|
-
end
|
177
|
-
|
178
|
-
def make_usual_equal_object
|
179
|
-
[@root_element_name, @system_identifier, @public_identifier]
|
180
|
-
end
|
181
|
-
end
|
182
|
-
|
183
|
-
class ProcIns
|
184
|
-
def make_exact_equal_object
|
185
|
-
[@raw_string, @target, @content]
|
186
|
-
end
|
187
|
-
|
188
|
-
def make_usual_equal_object
|
189
|
-
[@target, @content]
|
190
|
-
end
|
191
|
-
end
|
192
|
-
|
193
|
-
class Comment
|
194
|
-
def make_exact_equal_object
|
195
|
-
[@raw_string, @content]
|
196
|
-
end
|
197
|
-
|
198
|
-
alias usual_equal_object content
|
199
|
-
end
|
200
|
-
|
201
|
-
class BogusETag
|
202
|
-
def make_exact_equal_object
|
203
|
-
[@etag]
|
204
|
-
end
|
205
|
-
|
206
|
-
alias usual_equal_object make_exact_equal_object
|
207
|
-
end
|
208
|
-
|
209
|
-
class Location
|
210
|
-
def make_exact_equal_object
|
211
|
-
[@parent, @index, @node]
|
212
|
-
end
|
213
|
-
|
214
|
-
alias usual_equal_object make_exact_equal_object
|
215
|
-
end
|
216
|
-
|
217
|
-
end
|
218
|
-
# :startdoc:
|
@@ -1,37 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'htree/text'
|
3
|
-
require 'htree/doc'
|
4
|
-
require 'htree/elem'
|
5
|
-
|
6
|
-
module HTree
|
7
|
-
module Node
|
8
|
-
def extract_text
|
9
|
-
raise NotImplementedError
|
10
|
-
end
|
11
|
-
end
|
12
|
-
|
13
|
-
class Location
|
14
|
-
def extract_text
|
15
|
-
to_node.extract_text
|
16
|
-
end
|
17
|
-
end
|
18
|
-
|
19
|
-
module Container
|
20
|
-
def extract_text
|
21
|
-
Text.concat(*@children.map {|n| n.extract_text })
|
22
|
-
end
|
23
|
-
end
|
24
|
-
|
25
|
-
module Leaf
|
26
|
-
def extract_text
|
27
|
-
Text.new('')
|
28
|
-
end
|
29
|
-
end
|
30
|
-
|
31
|
-
class Text
|
32
|
-
def extract_text
|
33
|
-
self
|
34
|
-
end
|
35
|
-
end
|
36
|
-
end
|
37
|
-
# :startdoc:
|
@@ -1,33 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'htree/modules'
|
3
|
-
|
4
|
-
module HTree
|
5
|
-
def HTree.with_frozen_string_hash
|
6
|
-
if Thread.current[:htree_frozen_string_hash]
|
7
|
-
yield
|
8
|
-
else
|
9
|
-
begin
|
10
|
-
Thread.current[:htree_frozen_string_hash] = {}
|
11
|
-
yield
|
12
|
-
ensure
|
13
|
-
Thread.current[:htree_frozen_string_hash] = nil
|
14
|
-
end
|
15
|
-
end
|
16
|
-
end
|
17
|
-
|
18
|
-
def HTree.frozen_string(str)
|
19
|
-
if h = Thread.current[:htree_frozen_string_hash]
|
20
|
-
if s = h[str]
|
21
|
-
s
|
22
|
-
else
|
23
|
-
h[str] = str unless str.frozen?
|
24
|
-
str = str.dup.freeze
|
25
|
-
h[str] = str
|
26
|
-
end
|
27
|
-
else
|
28
|
-
str = str.dup.freeze unless str.frozen?
|
29
|
-
str
|
30
|
-
end
|
31
|
-
end
|
32
|
-
end
|
33
|
-
# :startdoc:
|