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
@@ -1,108 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'pp'
|
3
|
-
require 'htree/doc'
|
4
|
-
require 'htree/elem'
|
5
|
-
require 'htree/leaf'
|
6
|
-
require 'htree/tag'
|
7
|
-
require 'htree/output'
|
8
|
-
require 'htree/raw_string'
|
9
|
-
|
10
|
-
module HTree
|
11
|
-
class Doc
|
12
|
-
def pretty_print(q)
|
13
|
-
q.object_group(self) { @children.each {|elt| q.breakable; q.pp elt } }
|
14
|
-
end
|
15
|
-
alias inspect pretty_print_inspect
|
16
|
-
end
|
17
|
-
|
18
|
-
class Elem
|
19
|
-
def pretty_print(q)
|
20
|
-
if @empty
|
21
|
-
q.group(1, '{emptyelem', '}') {
|
22
|
-
q.breakable; q.pp @stag
|
23
|
-
}
|
24
|
-
else
|
25
|
-
q.group(1, "{elem", "}") {
|
26
|
-
q.breakable; q.pp @stag
|
27
|
-
@children.each {|elt| q.breakable; q.pp elt }
|
28
|
-
if @etag
|
29
|
-
q.breakable; q.pp @etag
|
30
|
-
end
|
31
|
-
}
|
32
|
-
end
|
33
|
-
end
|
34
|
-
alias inspect pretty_print_inspect
|
35
|
-
end
|
36
|
-
|
37
|
-
module Leaf
|
38
|
-
def pretty_print(q)
|
39
|
-
q.group(1, '{', '}') {
|
40
|
-
q.text self.class.name.sub(/.*::/,'').downcase
|
41
|
-
if rs = @raw_string
|
42
|
-
rs.scan(/[^\r\n]*(?:\r\n?|\n|[^\r\n]\z)/) {|line|
|
43
|
-
q.breakable
|
44
|
-
q.pp line
|
45
|
-
}
|
46
|
-
elsif self.respond_to? :display_xml
|
47
|
-
q.breakable
|
48
|
-
q.text self.display_xml('')
|
49
|
-
end
|
50
|
-
}
|
51
|
-
end
|
52
|
-
alias inspect pretty_print_inspect
|
53
|
-
end
|
54
|
-
|
55
|
-
class Name
|
56
|
-
def inspect
|
57
|
-
if xmlns?
|
58
|
-
@local_name ? "xmlns:#{@local_name}" : "xmlns"
|
59
|
-
elsif !@namespace_uri || @namespace_uri.empty?
|
60
|
-
@local_name
|
61
|
-
elsif @namespace_prefix
|
62
|
-
"#{@namespace_prefix}{#{@namespace_uri}}#{@local_name}"
|
63
|
-
elsif @namespace_prefix == false
|
64
|
-
"-{#{@namespace_uri}}#{@local_name}"
|
65
|
-
else
|
66
|
-
"{#{@namespace_uri}}#{@local_name}"
|
67
|
-
end
|
68
|
-
end
|
69
|
-
end
|
70
|
-
|
71
|
-
class STag
|
72
|
-
def pretty_print(q)
|
73
|
-
q.group(1, '<', '>') {
|
74
|
-
q.text @name.inspect
|
75
|
-
|
76
|
-
@attributes.each {|n, t|
|
77
|
-
q.breakable
|
78
|
-
q.text "#{n.inspect}=\"#{t.to_attvalue_content}\""
|
79
|
-
}
|
80
|
-
}
|
81
|
-
end
|
82
|
-
alias inspect pretty_print_inspect
|
83
|
-
end
|
84
|
-
|
85
|
-
class ETag
|
86
|
-
def pretty_print(q)
|
87
|
-
q.group(1, '</', '>') {
|
88
|
-
q.text @qualified_name
|
89
|
-
}
|
90
|
-
end
|
91
|
-
alias inspect pretty_print_inspect
|
92
|
-
end
|
93
|
-
|
94
|
-
class BogusETag
|
95
|
-
def pretty_print(q)
|
96
|
-
q.group(1, '{', '}') {
|
97
|
-
q.text self.class.name.sub(/.*::/,'').downcase
|
98
|
-
if rs = @raw_string
|
99
|
-
q.breakable
|
100
|
-
q.text rs
|
101
|
-
else
|
102
|
-
q.text "</#{@qualified_name}>"
|
103
|
-
end
|
104
|
-
}
|
105
|
-
end
|
106
|
-
end
|
107
|
-
end
|
108
|
-
# :startdoc:
|
@@ -1,94 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'htree/modules'
|
3
|
-
require 'htree/raw_string'
|
4
|
-
|
5
|
-
module HTree # :nodoc:
|
6
|
-
class XMLDecl # :nodoc:
|
7
|
-
def initialize(version, encoding=nil, standalone=nil)
|
8
|
-
init_raw_string
|
9
|
-
if /\A[a-zA-Z0-9_.:-]+\z/ !~ version
|
10
|
-
raise HTree::Error, "invalid version in XML declaration: #{version.inspect}"
|
11
|
-
end
|
12
|
-
if encoding && /\A[A-Za-z][A-Za-z0-9._-]*\z/ !~ encoding
|
13
|
-
raise HTree::Error, "invalid encoding in XML declaration: #{encoding.inspect}"
|
14
|
-
end
|
15
|
-
unless standalone == nil || standalone == true || standalone == false
|
16
|
-
raise HTree::Error, "invalid standalone document declaration in XML declaration: #{standalone.inspect}"
|
17
|
-
end
|
18
|
-
@version = version
|
19
|
-
@encoding = encoding
|
20
|
-
@standalone = standalone
|
21
|
-
end
|
22
|
-
attr_reader :version, :encoding, :standalone
|
23
|
-
end
|
24
|
-
|
25
|
-
class DocType # :nodoc:
|
26
|
-
def initialize(root_element_name, public_identifier=nil, system_identifier=nil)
|
27
|
-
init_raw_string
|
28
|
-
if public_identifier && /\A[ \x0d\x0aa-zA-Z0-9\-'()+,.\/:=?;!*\#@$_%]*\z/ !~ public_identifier
|
29
|
-
raise HTree::Error, "invalid public identifier in document type declaration: #{public_identifier.inspect}"
|
30
|
-
end
|
31
|
-
if system_identifier && /"/ =~ system_identifier && /'/ =~ system_identifier
|
32
|
-
raise HTree::Error, "invalid system identifier in document type declaration: #{system_identifier.inspect}"
|
33
|
-
end
|
34
|
-
|
35
|
-
@root_element_name = root_element_name
|
36
|
-
@public_identifier = public_identifier
|
37
|
-
@system_identifier = system_identifier
|
38
|
-
end
|
39
|
-
attr_reader :root_element_name, :public_identifier, :system_identifier
|
40
|
-
end
|
41
|
-
|
42
|
-
class ProcIns # :nodoc:
|
43
|
-
|
44
|
-
class << self
|
45
|
-
alias new! new
|
46
|
-
end
|
47
|
-
|
48
|
-
|
49
|
-
def ProcIns.new(target, content)
|
50
|
-
content = content.gsub(/\?>/, '? >') if content
|
51
|
-
new! target, content
|
52
|
-
end
|
53
|
-
|
54
|
-
def initialize(target, content) # :notnew:
|
55
|
-
init_raw_string
|
56
|
-
if content && /\?>/ =~ content
|
57
|
-
raise HTree::Error, "invalid processing instruction content: #{content.inspect}"
|
58
|
-
end
|
59
|
-
@target = target
|
60
|
-
@content = content
|
61
|
-
end
|
62
|
-
attr_reader :target, :content
|
63
|
-
end
|
64
|
-
|
65
|
-
class Comment # :nodoc:
|
66
|
-
|
67
|
-
class << self
|
68
|
-
alias new! new
|
69
|
-
end
|
70
|
-
|
71
|
-
|
72
|
-
def Comment.new(content)
|
73
|
-
content = content.gsub(/-(-+)/) { '-' + ' -' * $1.length }.sub(/-\z/, '- ')
|
74
|
-
new! content
|
75
|
-
end
|
76
|
-
|
77
|
-
def initialize(content) # :notnew:
|
78
|
-
init_raw_string
|
79
|
-
if /--/ =~ content || /-\z/ =~ content
|
80
|
-
raise HTree::Error, "invalid comment content: #{content.inspect}"
|
81
|
-
end
|
82
|
-
@content = content
|
83
|
-
end
|
84
|
-
attr_reader :content
|
85
|
-
end
|
86
|
-
|
87
|
-
class BogusETag # :nodoc:
|
88
|
-
def initialize(qualified_name)
|
89
|
-
init_raw_string
|
90
|
-
@etag = ETag.new(qualified_name)
|
91
|
-
end
|
92
|
-
end
|
93
|
-
end
|
94
|
-
# :startdoc:
|
@@ -1,367 +0,0 @@
|
|
1
|
-
# :stopdoc:
|
2
|
-
require 'htree/modules'
|
3
|
-
require 'htree/elem'
|
4
|
-
require 'htree/inspect'
|
5
|
-
|
6
|
-
module HTree # :nodoc:
|
7
|
-
module Node
|
8
|
-
# creates a location object which points to self.
|
9
|
-
def make_loc
|
10
|
-
self.class::Loc.new(nil, nil, self)
|
11
|
-
end
|
12
|
-
|
13
|
-
# return self.
|
14
|
-
def to_node
|
15
|
-
self
|
16
|
-
end
|
17
|
-
|
18
|
-
# +subst+ substitutes several subtrees at once.
|
19
|
-
#
|
20
|
-
# t = HTree('<r><x/><y/><z/></r>')
|
21
|
-
# l = t.make_loc
|
22
|
-
# t2 = t.subst({
|
23
|
-
# l.get_subnode(0, 'k') => 'v',
|
24
|
-
# l.get_subnode(0, -1) => HTree('<a/>'),
|
25
|
-
# l.get_subnode(0, 1) => nil,
|
26
|
-
# l.get_subnode(0, 2, 0) => HTree('<b/>'),
|
27
|
-
# })
|
28
|
-
# pp t2
|
29
|
-
# # =>
|
30
|
-
# #<HTree::Doc
|
31
|
-
# {elem <r k="v"> {emptyelem <a>} {emptyelem <x>} {elem <z> {emptyelem <b>}}}>
|
32
|
-
def subst(pairs)
|
33
|
-
pairs = pairs.map {|key, val|
|
34
|
-
key = key.index_list(self)
|
35
|
-
unless Array === val
|
36
|
-
val = [val]
|
37
|
-
end
|
38
|
-
[key, val]
|
39
|
-
}
|
40
|
-
|
41
|
-
pairs_empty_key, pairs_nonempty_key =
|
42
|
-
pairs.partition {|key, val| key.empty? }
|
43
|
-
if !pairs_empty_key.empty?
|
44
|
-
if !pairs_nonempty_key.empty?
|
45
|
-
raise ArgumentError, "cannot substitute a node under substituting tree."
|
46
|
-
end
|
47
|
-
result = []
|
48
|
-
pairs_empty_key.each {|key, val| result.concat val }
|
49
|
-
result.compact!
|
50
|
-
if result.length == 1
|
51
|
-
return result[0]
|
52
|
-
else
|
53
|
-
raise ArgumentError, "cannot substitute top node by multiple nodes: #{nodes.inspect}"
|
54
|
-
end
|
55
|
-
end
|
56
|
-
if pairs_nonempty_key.empty?
|
57
|
-
return self
|
58
|
-
end
|
59
|
-
|
60
|
-
subst_internal(pairs)
|
61
|
-
end
|
62
|
-
|
63
|
-
def subst_internal(pairs) # :nodoc:
|
64
|
-
subnode_pairs = {}
|
65
|
-
pairs.each {|key, val|
|
66
|
-
k = key.pop
|
67
|
-
(subnode_pairs[k] ||= []) << [key, val]
|
68
|
-
}
|
69
|
-
subnode_pairs = subnode_pairs.map {|k, subpairs|
|
70
|
-
s = get_subnode(k)
|
71
|
-
subpairs_empty_key, subpairs_nonempty_key =
|
72
|
-
subpairs.partition {|key, val| key.empty? }
|
73
|
-
if !subpairs_empty_key.empty?
|
74
|
-
if !subpairs_nonempty_key.empty?
|
75
|
-
raise ArgumentError, "cannot substitute a node under substituting tree."
|
76
|
-
end
|
77
|
-
r = []
|
78
|
-
subpairs_empty_key.each {|key, val| r.concat val }
|
79
|
-
[k, r.compact]
|
80
|
-
elsif subpairs_nonempty_key.empty?
|
81
|
-
[k, s]
|
82
|
-
else
|
83
|
-
[k, s.subst_internal(subpairs)]
|
84
|
-
end
|
85
|
-
}
|
86
|
-
subst_subnode(subnode_pairs)
|
87
|
-
end
|
88
|
-
end
|
89
|
-
|
90
|
-
class Doc; def node_test_string() 'doc()' end end
|
91
|
-
class Elem; def node_test_string() @stag.element_name.qualified_name end end
|
92
|
-
class Text; def node_test_string() 'text()' end end
|
93
|
-
class BogusETag; def node_test_string() 'bogus-etag()' end end
|
94
|
-
class XMLDecl; def node_test_string() 'xml-declaration()' end end
|
95
|
-
class DocType; def node_test_string() 'doctype()' end end
|
96
|
-
class ProcIns; def node_test_string() 'processing-instruction()' end end
|
97
|
-
class Comment; def node_test_string() 'comment()' end end
|
98
|
-
|
99
|
-
module Container
|
100
|
-
def find_loc_step(index)
|
101
|
-
if index < 0 || @children.length <= index
|
102
|
-
return "*[#{index}]"
|
103
|
-
end
|
104
|
-
|
105
|
-
return @loc_step_children[index].dup if defined? @loc_step_children
|
106
|
-
|
107
|
-
count = {}
|
108
|
-
count.default = 0
|
109
|
-
|
110
|
-
steps = []
|
111
|
-
|
112
|
-
@children.each {|c|
|
113
|
-
node_test = c.node_test_string
|
114
|
-
count[node_test] += 1
|
115
|
-
steps << [node_test, count[node_test]]
|
116
|
-
}
|
117
|
-
|
118
|
-
@loc_step_children = []
|
119
|
-
steps.each {|node_test, i|
|
120
|
-
if count[node_test] == 1
|
121
|
-
@loc_step_children << node_test
|
122
|
-
else
|
123
|
-
@loc_step_children << "#{node_test}[#{i}]"
|
124
|
-
end
|
125
|
-
}
|
126
|
-
|
127
|
-
return @loc_step_children[index].dup
|
128
|
-
end
|
129
|
-
end
|
130
|
-
|
131
|
-
class Elem
|
132
|
-
def find_loc_step(index)
|
133
|
-
return super if Integer === index
|
134
|
-
if String === index
|
135
|
-
index = Name.parse_attribute_name(index, DefaultContext)
|
136
|
-
end
|
137
|
-
unless Name === index
|
138
|
-
raise TypeError, "invalid index: #{index.inspect}"
|
139
|
-
end
|
140
|
-
"@#{index.qualified_name}"
|
141
|
-
end
|
142
|
-
end
|
143
|
-
end
|
144
|
-
|
145
|
-
class HTree::Location # :nodoc:
|
146
|
-
def initialize(parent, index, node) # :nodoc:
|
147
|
-
if parent
|
148
|
-
@parent = parent
|
149
|
-
@index = index
|
150
|
-
@node = parent.node.get_subnode(index)
|
151
|
-
if !@node.equal?(node)
|
152
|
-
raise ArgumentError, "unexpected node"
|
153
|
-
end
|
154
|
-
else
|
155
|
-
@parent = nil
|
156
|
-
@index = nil
|
157
|
-
@node = node
|
158
|
-
end
|
159
|
-
if @node && self.class != @node.class::Loc
|
160
|
-
raise ArgumentError, "invalid location class: #{self.class} should be #{node.class::Loc}"
|
161
|
-
end
|
162
|
-
@subloc = {}
|
163
|
-
end
|
164
|
-
attr_reader :parent, :index, :node
|
165
|
-
alias to_node node
|
166
|
-
|
167
|
-
# return self.
|
168
|
-
def make_loc
|
169
|
-
self
|
170
|
-
end
|
171
|
-
|
172
|
-
# +top+ returns the originator location.
|
173
|
-
#
|
174
|
-
# t = HTree('<a><b><c><d>')
|
175
|
-
# l = t.make_loc.get_subnode(0, 0, 0, 0)
|
176
|
-
# p l, l.top
|
177
|
-
# # =>
|
178
|
-
# #<HTree::Location: doc()/a/b/c/d>
|
179
|
-
# #<HTree::Location: doc()>
|
180
|
-
def top
|
181
|
-
result = self
|
182
|
-
while result.parent
|
183
|
-
result = result.parent
|
184
|
-
end
|
185
|
-
result
|
186
|
-
end
|
187
|
-
|
188
|
-
# +subst_itself+ substitutes the node pointed by the location.
|
189
|
-
# It returns the location of substituted node.
|
190
|
-
#
|
191
|
-
# t1 = HTree('<a><b><c><d>')
|
192
|
-
# p t1
|
193
|
-
# l1 = t1.make_loc.get_subnode(0, 0, 0, 0)
|
194
|
-
# p l1
|
195
|
-
# l2 = l1.subst_itself(HTree('<z/>'))
|
196
|
-
# p l2
|
197
|
-
# t2 = l2.top.to_node
|
198
|
-
# p t2
|
199
|
-
# # =>
|
200
|
-
# #<HTree::Doc {elem <a> {elem <b> {elem <c> {emptyelem <d>}}}}>
|
201
|
-
# #<HTree::Location: doc()/a/b/c/d>
|
202
|
-
# #<HTree::Location: doc()/a/b/c/z>
|
203
|
-
# #<HTree::Doc {elem <a> {elem <b> {elem <c> {emptyelem <z>}}}}>
|
204
|
-
#
|
205
|
-
def subst_itself(node)
|
206
|
-
if @parent
|
207
|
-
new_index = @index
|
208
|
-
if !@node
|
209
|
-
if Integer === @index
|
210
|
-
if @index < 0
|
211
|
-
new_index = 0
|
212
|
-
elsif @parent.to_node.children.length < @index
|
213
|
-
new_index = @parent.to_node.children.length
|
214
|
-
end
|
215
|
-
end
|
216
|
-
end
|
217
|
-
@parent.subst_itself(@parent.to_node.subst_subnode({@index=>node})).get_subnode(new_index)
|
218
|
-
else
|
219
|
-
node.make_loc
|
220
|
-
end
|
221
|
-
end
|
222
|
-
|
223
|
-
# +subst+ substitutes several subtrees at once.
|
224
|
-
#
|
225
|
-
# t = HTree('<r><x/><y/><z/></r>')
|
226
|
-
# l = t.make_loc
|
227
|
-
# l2 = l.subst({
|
228
|
-
# l.root.get_subnode('k') => 'v',
|
229
|
-
# l.root.get_subnode(-1) => HTree('<a/>'),
|
230
|
-
# l.find_element('y') => nil,
|
231
|
-
# l.find_element('z').get_subnode(0) => HTree('<b/>'),
|
232
|
-
# })
|
233
|
-
# pp l2, l2.to_node
|
234
|
-
# # =>
|
235
|
-
# #<HTree::Doc::Loc: doc()>
|
236
|
-
# #<HTree::Doc
|
237
|
-
# {elem <r k="v"> {emptyelem <a>} {emptyelem <x>} {elem <z> {emptyelem <b>}}}>
|
238
|
-
def subst(pairs)
|
239
|
-
subst_itself(@node.subst(pairs))
|
240
|
-
end
|
241
|
-
|
242
|
-
# +loc_list+ returns an array containing from location's root to itself.
|
243
|
-
#
|
244
|
-
# t = HTree('<a><b><c>')
|
245
|
-
# l = t.make_loc.get_subnode(0, 0, 0)
|
246
|
-
# pp l, l.loc_list
|
247
|
-
# # =>
|
248
|
-
# #<HTree::Location: doc()/a/b/c>
|
249
|
-
# [#<HTree::Location: doc()>,
|
250
|
-
# #<HTree::Location: doc()/a>,
|
251
|
-
# #<HTree::Location: doc()/a/b>,
|
252
|
-
# #<HTree::Location: doc()/a/b/c>]
|
253
|
-
#
|
254
|
-
def loc_list
|
255
|
-
loc = self
|
256
|
-
result = [self]
|
257
|
-
while loc = loc.parent
|
258
|
-
result << loc
|
259
|
-
end
|
260
|
-
result.reverse!
|
261
|
-
result
|
262
|
-
end
|
263
|
-
|
264
|
-
# +path+ returns the path of the location.
|
265
|
-
#
|
266
|
-
# l = HTree.parse("<a><b>x</b><b/><a/>").make_loc
|
267
|
-
# l = l.get_subnode(0, 0, 0)
|
268
|
-
# p l.path # => "doc()/a/b[1]/text()"
|
269
|
-
def path
|
270
|
-
result = ''
|
271
|
-
loc_list.each {|loc|
|
272
|
-
if parent = loc.parent
|
273
|
-
result << '/' << parent.node.find_loc_step(loc.index)
|
274
|
-
else
|
275
|
-
result << loc.node.node_test_string
|
276
|
-
end
|
277
|
-
}
|
278
|
-
result
|
279
|
-
end
|
280
|
-
|
281
|
-
def index_list(node) # :nodoc:
|
282
|
-
result = []
|
283
|
-
loc = self
|
284
|
-
while parent = loc.parent
|
285
|
-
return result if loc.to_node.equal? node
|
286
|
-
result << loc.index
|
287
|
-
loc = parent
|
288
|
-
end
|
289
|
-
return result if loc.to_node.equal? node
|
290
|
-
raise ArgumentError, "the location is not under the node: #{self.path}"
|
291
|
-
end
|
292
|
-
|
293
|
-
def pretty_print(q)
|
294
|
-
q.group(1, "#<#{self.class.name}", '>') {
|
295
|
-
q.text ':'
|
296
|
-
q.breakable
|
297
|
-
loc_list.each {|loc|
|
298
|
-
if parent = loc.parent
|
299
|
-
q.text '/'
|
300
|
-
q.group { q.breakable '' }
|
301
|
-
q.text parent.node.find_loc_step(loc.index)
|
302
|
-
else
|
303
|
-
q.text loc.node.node_test_string
|
304
|
-
end
|
305
|
-
}
|
306
|
-
}
|
307
|
-
end
|
308
|
-
alias inspect pretty_print_inspect
|
309
|
-
end
|
310
|
-
|
311
|
-
module HTree::Container::Loc # :nodoc:
|
312
|
-
# +get_subnode+ returns a location object which points to a subnode
|
313
|
-
# indexed by _index_.
|
314
|
-
def get_subnode_internal(index) # :nodoc:
|
315
|
-
return @subloc[index] if @subloc.include? index
|
316
|
-
node = @node.get_subnode(index)
|
317
|
-
if node
|
318
|
-
@subloc[index] = node.class::Loc.new(self, index, node)
|
319
|
-
else
|
320
|
-
@subloc[index] = HTree::Location.new(self, index, node)
|
321
|
-
end
|
322
|
-
end
|
323
|
-
|
324
|
-
# +subst_subnode+ returns the location which refers the substituted tree.
|
325
|
-
# loc.subst_subnode(pairs) -> loc
|
326
|
-
#
|
327
|
-
# t = HTree('<a><b><c>')
|
328
|
-
# l = t.make_loc.get_subnode(0, 0)
|
329
|
-
# l = l.subst_subnode({0=>HTree('<z/>')})
|
330
|
-
# pp t, l.top.to_node
|
331
|
-
# # =>
|
332
|
-
# #<HTree::Doc {elem <a> {elem <b> {emptyelem <c>}}}>
|
333
|
-
# #<HTree::Doc {elem <a> {elem <b> {emptyelem <z>}}}>
|
334
|
-
#
|
335
|
-
def subst_subnode(pairs)
|
336
|
-
self.subst_itself(@node.subst_subnode(pairs))
|
337
|
-
end
|
338
|
-
|
339
|
-
# +children+ returns an array of child locations.
|
340
|
-
def children
|
341
|
-
(0...@node.children.length).map {|i| get_subnode(i) }
|
342
|
-
end
|
343
|
-
end
|
344
|
-
|
345
|
-
class HTree::Elem::Loc # :nodoc:
|
346
|
-
def context() @node.context end
|
347
|
-
|
348
|
-
# +element_name+ returns the name of the element name as a Name object.
|
349
|
-
def element_name() @node.element_name end
|
350
|
-
|
351
|
-
def empty_element?() @node.empty_element? end
|
352
|
-
|
353
|
-
# +each_attribute+ iterates over each attributes.
|
354
|
-
def each_attribute
|
355
|
-
@node.each_attribute {|attr_name, attr_text|
|
356
|
-
attr_loc = get_subnode(attr_name)
|
357
|
-
yield attr_name, attr_loc
|
358
|
-
}
|
359
|
-
end
|
360
|
-
end
|
361
|
-
|
362
|
-
class HTree::Text::Loc # :nodoc:
|
363
|
-
def to_s() @node.to_s end
|
364
|
-
def strip() @node.strip end
|
365
|
-
def empty?() @node.empty? end
|
366
|
-
end
|
367
|
-
# :startdoc:
|