nokogiri 1.0.0 → 1.6.8.1
Sign up to get free protection for your applications and to get access to all the features.
Potentially problematic release.
This version of nokogiri might be problematic. Click here for more details.
- checksums.yaml +7 -0
- data/.autotest +26 -0
- data/.cross_rubies +9 -0
- data/.editorconfig +17 -0
- data/.gemtest +0 -0
- data/.travis.yml +51 -0
- data/CHANGELOG.rdoc +1160 -0
- data/CONTRIBUTING.md +42 -0
- data/C_CODING_STYLE.rdoc +33 -0
- data/Gemfile +22 -0
- data/LICENSE.txt +31 -0
- data/Manifest.txt +284 -40
- data/README.md +166 -0
- data/ROADMAP.md +111 -0
- data/Rakefile +310 -199
- data/STANDARD_RESPONSES.md +47 -0
- data/Y_U_NO_GEMSPEC.md +155 -0
- data/appveyor.yml +22 -0
- data/bin/nokogiri +118 -0
- data/build_all +45 -0
- data/dependencies.yml +29 -0
- data/ext/nokogiri/depend +358 -0
- data/ext/nokogiri/extconf.rb +664 -34
- data/ext/nokogiri/html_document.c +120 -33
- data/ext/nokogiri/html_document.h +1 -1
- data/ext/nokogiri/html_element_description.c +279 -0
- data/ext/nokogiri/html_element_description.h +10 -0
- data/ext/nokogiri/html_entity_lookup.c +32 -0
- data/ext/nokogiri/html_entity_lookup.h +8 -0
- data/ext/nokogiri/html_sax_parser_context.c +116 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/html_sax_push_parser.c +87 -0
- data/ext/nokogiri/html_sax_push_parser.h +9 -0
- data/ext/nokogiri/nokogiri.c +145 -0
- data/ext/nokogiri/nokogiri.h +131 -0
- data/ext/nokogiri/xml_attr.c +94 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +23 -19
- data/ext/nokogiri/xml_cdata.h +1 -1
- data/ext/nokogiri/xml_comment.c +69 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +501 -54
- data/ext/nokogiri/xml_document.h +14 -1
- data/ext/nokogiri/xml_document_fragment.c +48 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +109 -24
- data/ext/nokogiri/xml_dtd.h +3 -1
- data/ext/nokogiri/xml_element_content.c +123 -0
- data/ext/nokogiri/xml_element_content.h +10 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_element_decl.h +9 -0
- data/ext/nokogiri/xml_encoding_handler.c +79 -0
- data/ext/nokogiri/xml_encoding_handler.h +8 -0
- data/ext/nokogiri/xml_entity_decl.c +110 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +52 -0
- data/ext/nokogiri/xml_entity_reference.h +9 -0
- data/ext/nokogiri/xml_io.c +60 -0
- data/ext/nokogiri/xml_io.h +11 -0
- data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
- data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
- data/ext/nokogiri/xml_namespace.c +117 -0
- data/ext/nokogiri/xml_namespace.h +13 -0
- data/ext/nokogiri/xml_node.c +1285 -315
- data/ext/nokogiri/xml_node.h +4 -6
- data/ext/nokogiri/xml_node_set.c +415 -54
- data/ext/nokogiri/xml_node_set.h +6 -2
- data/ext/nokogiri/xml_processing_instruction.c +56 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +316 -77
- data/ext/nokogiri/xml_reader.h +1 -1
- data/ext/nokogiri/xml_relax_ng.c +161 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +215 -80
- data/ext/nokogiri/xml_sax_parser.h +30 -1
- data/ext/nokogiri/xml_sax_parser_context.c +262 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +115 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +205 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +45 -175
- data/ext/nokogiri/xml_syntax_error.h +4 -2
- data/ext/nokogiri/xml_text.c +37 -14
- data/ext/nokogiri/xml_text.h +1 -1
- data/ext/nokogiri/xml_xpath_context.c +230 -13
- data/ext/nokogiri/xml_xpath_context.h +2 -1
- data/ext/nokogiri/xslt_stylesheet.c +196 -34
- data/ext/nokogiri/xslt_stylesheet.h +6 -1
- data/lib/nokogiri/css/node.rb +18 -61
- data/lib/nokogiri/css/parser.rb +725 -17
- data/lib/nokogiri/css/parser.y +126 -63
- data/lib/nokogiri/css/parser_extras.rb +91 -0
- data/lib/nokogiri/css/syntax_error.rb +7 -0
- data/lib/nokogiri/css/tokenizer.rb +148 -5
- data/lib/nokogiri/css/tokenizer.rex +31 -39
- data/lib/nokogiri/css/xpath_visitor.rb +109 -51
- data/lib/nokogiri/css.rb +24 -3
- data/lib/nokogiri/decorators/slop.rb +42 -0
- data/lib/nokogiri/html/builder.rb +27 -1
- data/lib/nokogiri/html/document.rb +329 -3
- data/lib/nokogiri/html/document_fragment.rb +39 -0
- data/lib/nokogiri/html/element_description.rb +23 -0
- data/lib/nokogiri/html/element_description_defaults.rb +671 -0
- data/lib/nokogiri/html/entity_lookup.rb +13 -0
- data/lib/nokogiri/html/sax/parser.rb +35 -4
- data/lib/nokogiri/html/sax/parser_context.rb +16 -0
- data/lib/nokogiri/html/sax/push_parser.rb +36 -0
- data/lib/nokogiri/html.rb +18 -76
- data/lib/nokogiri/syntax_error.rb +4 -0
- data/lib/nokogiri/version.rb +106 -1
- data/lib/nokogiri/xml/attr.rb +14 -0
- data/lib/nokogiri/xml/attribute_decl.rb +18 -0
- data/lib/nokogiri/xml/builder.rb +395 -31
- data/lib/nokogiri/xml/cdata.rb +4 -2
- data/lib/nokogiri/xml/character_data.rb +7 -0
- data/lib/nokogiri/xml/document.rb +267 -12
- data/lib/nokogiri/xml/document_fragment.rb +149 -0
- data/lib/nokogiri/xml/dtd.rb +27 -1
- data/lib/nokogiri/xml/element_content.rb +36 -0
- data/lib/nokogiri/xml/element_decl.rb +13 -0
- data/lib/nokogiri/xml/entity_decl.rb +19 -0
- data/lib/nokogiri/xml/namespace.rb +13 -0
- data/lib/nokogiri/xml/node/save_options.rb +61 -0
- data/lib/nokogiri/xml/node.rb +748 -109
- data/lib/nokogiri/xml/node_set.rb +200 -72
- data/lib/nokogiri/xml/parse_options.rb +120 -0
- data/lib/nokogiri/xml/pp/character_data.rb +18 -0
- data/lib/nokogiri/xml/pp/node.rb +56 -0
- data/lib/nokogiri/xml/pp.rb +2 -0
- data/lib/nokogiri/xml/processing_instruction.rb +8 -0
- data/lib/nokogiri/xml/reader.rb +102 -4
- data/lib/nokogiri/xml/relax_ng.rb +32 -0
- data/lib/nokogiri/xml/sax/document.rb +114 -2
- data/lib/nokogiri/xml/sax/parser.rb +97 -7
- data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
- data/lib/nokogiri/xml/sax.rb +2 -7
- data/lib/nokogiri/xml/schema.rb +63 -0
- data/lib/nokogiri/xml/searchable.rb +221 -0
- data/lib/nokogiri/xml/syntax_error.rb +27 -1
- data/lib/nokogiri/xml/text.rb +4 -1
- data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
- data/lib/nokogiri/xml/xpath.rb +4 -0
- data/lib/nokogiri/xml/xpath_context.rb +3 -1
- data/lib/nokogiri/xml.rb +45 -38
- data/lib/nokogiri/xslt/stylesheet.rb +19 -0
- data/lib/nokogiri/xslt.rb +47 -2
- data/lib/nokogiri.rb +117 -24
- data/lib/xsd/xmlparser/nokogiri.rb +102 -0
- data/patches/sort-patches-by-date +25 -0
- data/ports/archives/libxml2-2.9.4.tar.gz +0 -0
- data/ports/archives/libxslt-1.1.29.tar.gz +0 -0
- data/suppressions/README.txt +1 -0
- data/suppressions/nokogiri_ree-1.8.7.358.supp +61 -0
- data/suppressions/nokogiri_ruby-1.8.7.370.supp +0 -0
- data/suppressions/nokogiri_ruby-1.9.2.320.supp +28 -0
- data/suppressions/nokogiri_ruby-1.9.3.327.supp +28 -0
- data/tasks/test.rb +100 -0
- data/test/css/test_nthiness.rb +73 -6
- data/test/css/test_parser.rb +184 -39
- data/test/css/test_tokenizer.rb +72 -19
- data/test/css/test_xpath_visitor.rb +44 -2
- data/test/decorators/test_slop.rb +20 -0
- data/test/files/2ch.html +108 -0
- data/test/files/GH_1042.html +18 -0
- data/test/files/address_book.rlx +12 -0
- data/test/files/address_book.xml +10 -0
- data/test/files/atom.xml +344 -0
- data/test/files/bar/bar.xsd +4 -0
- data/test/files/bogus.xml +0 -0
- data/test/files/dont_hurt_em_why.xml +422 -0
- data/test/files/encoding.html +82 -0
- data/test/files/encoding.xhtml +84 -0
- data/test/files/exslt.xml +8 -0
- data/test/files/exslt.xslt +35 -0
- data/test/files/foo/foo.xsd +4 -0
- data/test/files/metacharset.html +10 -0
- data/test/files/namespace_pressure_test.xml +1684 -0
- data/test/files/noencoding.html +47 -0
- data/test/files/po.xml +32 -0
- data/test/files/po.xsd +66 -0
- data/test/files/saml/saml20assertion_schema.xsd +283 -0
- data/test/files/saml/saml20protocol_schema.xsd +302 -0
- data/test/files/saml/xenc_schema.xsd +146 -0
- data/test/files/saml/xmldsig_schema.xsd +318 -0
- data/test/files/shift_jis.html +10 -0
- data/test/files/shift_jis.xml +5 -0
- data/test/files/shift_jis_no_charset.html +9 -0
- data/test/files/slow-xpath.xml +25509 -0
- data/test/files/snuggles.xml +3 -0
- data/test/files/staff.dtd +10 -0
- data/test/files/test_document_url/bar.xml +2 -0
- data/test/files/test_document_url/document.dtd +4 -0
- data/test/files/test_document_url/document.xml +6 -0
- data/test/files/tlm.html +2 -1
- data/test/files/to_be_xincluded.xml +2 -0
- data/test/files/valid_bar.xml +2 -0
- data/test/files/xinclude.xml +4 -0
- data/test/helper.rb +124 -13
- data/test/html/sax/test_parser.rb +118 -4
- data/test/html/sax/test_parser_context.rb +46 -0
- data/test/html/sax/test_push_parser.rb +87 -0
- data/test/html/test_builder.rb +94 -8
- data/test/html/test_document.rb +626 -11
- data/test/html/test_document_encoding.rb +145 -0
- data/test/html/test_document_fragment.rb +301 -0
- data/test/html/test_element_description.rb +105 -0
- data/test/html/test_named_characters.rb +14 -0
- data/test/html/test_node.rb +212 -0
- data/test/html/test_node_encoding.rb +85 -0
- data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +14 -0
- data/test/namespaces/test_namespaces_aliased_default.rb +24 -0
- data/test/namespaces/test_namespaces_in_builder_doc.rb +75 -0
- data/test/namespaces/test_namespaces_in_cloned_doc.rb +31 -0
- data/test/namespaces/test_namespaces_in_created_doc.rb +75 -0
- data/test/namespaces/test_namespaces_in_parsed_doc.rb +80 -0
- data/test/namespaces/test_namespaces_preservation.rb +31 -0
- data/test/test_convert_xpath.rb +2 -47
- data/test/test_css_cache.rb +45 -0
- data/test/test_encoding_handler.rb +48 -0
- data/test/test_memory_leak.rb +156 -0
- data/test/test_nokogiri.rb +103 -1
- data/test/test_soap4r_sax.rb +52 -0
- data/test/test_xslt_transforms.rb +293 -8
- data/test/xml/node/test_save_options.rb +28 -0
- data/test/xml/node/test_subclass.rb +44 -0
- data/test/xml/sax/test_parser.rb +309 -8
- data/test/xml/sax/test_parser_context.rb +115 -0
- data/test/xml/sax/test_push_parser.rb +157 -0
- data/test/xml/test_attr.rb +67 -0
- data/test/xml/test_attribute_decl.rb +86 -0
- data/test/xml/test_builder.rb +327 -2
- data/test/xml/test_c14n.rb +180 -0
- data/test/xml/test_cdata.rb +32 -2
- data/test/xml/test_comment.rb +40 -0
- data/test/xml/test_document.rb +846 -35
- data/test/xml/test_document_encoding.rb +31 -0
- data/test/xml/test_document_fragment.rb +271 -0
- data/test/xml/test_dtd.rb +153 -9
- data/test/xml/test_dtd_encoding.rb +31 -0
- data/test/xml/test_element_content.rb +56 -0
- data/test/xml/test_element_decl.rb +73 -0
- data/test/xml/test_entity_decl.rb +122 -0
- data/test/xml/test_entity_reference.rb +251 -0
- data/test/xml/test_namespace.rb +96 -0
- data/test/xml/test_node.rb +1126 -105
- data/test/xml/test_node_attributes.rb +115 -0
- data/test/xml/test_node_encoding.rb +69 -0
- data/test/xml/test_node_inheritance.rb +32 -0
- data/test/xml/test_node_reparenting.rb +549 -0
- data/test/xml/test_node_set.rb +668 -9
- data/test/xml/test_parse_options.rb +64 -0
- data/test/xml/test_processing_instruction.rb +30 -0
- data/test/xml/test_reader.rb +589 -0
- data/test/xml/test_reader_encoding.rb +134 -0
- data/test/xml/test_relax_ng.rb +60 -0
- data/test/xml/test_schema.rb +142 -0
- data/test/xml/test_syntax_error.rb +30 -0
- data/test/xml/test_text.rb +49 -2
- data/test/xml/test_unparented_node.rb +440 -0
- data/test/xml/test_xinclude.rb +83 -0
- data/test/xml/test_xpath.rb +445 -0
- data/test/xslt/test_custom_functions.rb +133 -0
- data/test/xslt/test_exception_handling.rb +37 -0
- data/test_all +107 -0
- metadata +459 -115
- data/History.txt +0 -6
- data/README.ja.txt +0 -86
- data/README.txt +0 -87
- data/ext/nokogiri/html_sax_parser.c +0 -32
- data/ext/nokogiri/html_sax_parser.h +0 -11
- data/ext/nokogiri/native.c +0 -40
- data/ext/nokogiri/native.h +0 -51
- data/ext/nokogiri/xml_xpath.c +0 -46
- data/ext/nokogiri/xml_xpath.h +0 -11
- data/lib/nokogiri/css/generated_parser.rb +0 -653
- data/lib/nokogiri/css/generated_tokenizer.rb +0 -159
- data/lib/nokogiri/decorators/hpricot/node.rb +0 -58
- data/lib/nokogiri/decorators/hpricot/node_set.rb +0 -14
- data/lib/nokogiri/decorators/hpricot/xpath_visitor.rb +0 -17
- data/lib/nokogiri/decorators/hpricot.rb +0 -3
- data/lib/nokogiri/decorators.rb +0 -1
- data/lib/nokogiri/hpricot.rb +0 -47
- data/lib/nokogiri/xml/after_handler.rb +0 -18
- data/lib/nokogiri/xml/before_handler.rb +0 -32
- data/lib/nokogiri/xml/element.rb +0 -6
- data/lib/nokogiri/xml/entity_declaration.rb +0 -9
- data/nokogiri.gemspec +0 -34
- data/test/hpricot/files/basic.xhtml +0 -17
- data/test/hpricot/files/boingboing.html +0 -2266
- data/test/hpricot/files/cy0.html +0 -3653
- data/test/hpricot/files/immob.html +0 -400
- data/test/hpricot/files/pace_application.html +0 -1320
- data/test/hpricot/files/tenderlove.html +0 -16
- data/test/hpricot/files/uswebgen.html +0 -220
- data/test/hpricot/files/utf8.html +0 -1054
- data/test/hpricot/files/week9.html +0 -1723
- data/test/hpricot/files/why.xml +0 -19
- data/test/hpricot/load_files.rb +0 -7
- data/test/hpricot/test_alter.rb +0 -67
- data/test/hpricot/test_builder.rb +0 -27
- data/test/hpricot/test_parser.rb +0 -423
- data/test/hpricot/test_paths.rb +0 -15
- data/test/hpricot/test_preserved.rb +0 -78
- data/test/hpricot/test_xml.rb +0 -30
- data/test/test_reader.rb +0 -222
data/lib/nokogiri/xml/node.rb
CHANGED
@@ -1,195 +1,834 @@
|
|
1
|
+
# encoding: UTF-8
|
2
|
+
require 'stringio'
|
3
|
+
require 'nokogiri/xml/node/save_options'
|
4
|
+
|
1
5
|
module Nokogiri
|
2
6
|
module XML
|
7
|
+
####
|
8
|
+
# Nokogiri::XML::Node is your window to the fun filled world of dealing
|
9
|
+
# with XML and HTML tags. A Nokogiri::XML::Node may be treated similarly
|
10
|
+
# to a hash with regard to attributes. For example (from irb):
|
11
|
+
#
|
12
|
+
# irb(main):004:0> node
|
13
|
+
# => <a href="#foo" id="link">link</a>
|
14
|
+
# irb(main):005:0> node['href']
|
15
|
+
# => "#foo"
|
16
|
+
# irb(main):006:0> node.keys
|
17
|
+
# => ["href", "id"]
|
18
|
+
# irb(main):007:0> node.values
|
19
|
+
# => ["#foo", "link"]
|
20
|
+
# irb(main):008:0> node['class'] = 'green'
|
21
|
+
# => "green"
|
22
|
+
# irb(main):009:0> node
|
23
|
+
# => <a href="#foo" id="link" class="green">link</a>
|
24
|
+
# irb(main):010:0>
|
25
|
+
#
|
26
|
+
# See Nokogiri::XML::Node#[] and Nokogiri::XML#[]= for more information.
|
27
|
+
#
|
28
|
+
# Nokogiri::XML::Node also has methods that let you move around your
|
29
|
+
# tree. For navigating your tree, see:
|
30
|
+
#
|
31
|
+
# * Nokogiri::XML::Node#parent
|
32
|
+
# * Nokogiri::XML::Node#children
|
33
|
+
# * Nokogiri::XML::Node#next
|
34
|
+
# * Nokogiri::XML::Node#previous
|
35
|
+
#
|
36
|
+
#
|
37
|
+
# When printing or otherwise emitting a document or a node (and
|
38
|
+
# its subtree), there are a few methods you might want to use:
|
39
|
+
#
|
40
|
+
# * content, text, inner_text, to_str: emit plaintext
|
41
|
+
#
|
42
|
+
# These methods will all emit the plaintext version of your
|
43
|
+
# document, meaning that entities will be replaced (e.g., "<"
|
44
|
+
# will be replaced with "<"), meaning that any sanitizing will
|
45
|
+
# likely be un-done in the output.
|
46
|
+
#
|
47
|
+
# * to_s, to_xml, to_html, inner_html: emit well-formed markup
|
48
|
+
#
|
49
|
+
# These methods will all emit properly-escaped markup, meaning
|
50
|
+
# that it's suitable for consumption by browsers, parsers, etc.
|
51
|
+
#
|
52
|
+
# You may search this node's subtree using Searchable#xpath and Searchable#css
|
3
53
|
class Node
|
54
|
+
include Nokogiri::XML::PP::Node
|
55
|
+
include Nokogiri::XML::Searchable
|
56
|
+
include Enumerable
|
57
|
+
|
58
|
+
# Element node type, see Nokogiri::XML::Node#element?
|
59
|
+
ELEMENT_NODE = 1
|
60
|
+
# Attribute node type
|
61
|
+
ATTRIBUTE_NODE = 2
|
62
|
+
# Text node type, see Nokogiri::XML::Node#text?
|
63
|
+
TEXT_NODE = 3
|
64
|
+
# CDATA node type, see Nokogiri::XML::Node#cdata?
|
4
65
|
CDATA_SECTION_NODE = 4
|
5
|
-
|
6
|
-
|
66
|
+
# Entity reference node type
|
67
|
+
ENTITY_REF_NODE = 5
|
68
|
+
# Entity node type
|
69
|
+
ENTITY_NODE = 6
|
70
|
+
# PI node type
|
71
|
+
PI_NODE = 7
|
72
|
+
# Comment node type, see Nokogiri::XML::Node#comment?
|
73
|
+
COMMENT_NODE = 8
|
74
|
+
# Document node type, see Nokogiri::XML::Node#xml?
|
75
|
+
DOCUMENT_NODE = 9
|
76
|
+
# Document type node type
|
77
|
+
DOCUMENT_TYPE_NODE = 10
|
78
|
+
# Document fragment node type
|
79
|
+
DOCUMENT_FRAG_NODE = 11
|
80
|
+
# Notation node type
|
81
|
+
NOTATION_NODE = 12
|
82
|
+
# HTML document node type, see Nokogiri::XML::Node#html?
|
7
83
|
HTML_DOCUMENT_NODE = 13
|
8
|
-
|
9
|
-
|
10
|
-
|
11
|
-
|
12
|
-
|
13
|
-
|
14
|
-
|
84
|
+
# DTD node type
|
85
|
+
DTD_NODE = 14
|
86
|
+
# Element declaration type
|
87
|
+
ELEMENT_DECL = 15
|
88
|
+
# Attribute declaration type
|
89
|
+
ATTRIBUTE_DECL = 16
|
90
|
+
# Entity declaration type
|
91
|
+
ENTITY_DECL = 17
|
92
|
+
# Namespace declaration type
|
93
|
+
NAMESPACE_DECL = 18
|
94
|
+
# XInclude start type
|
95
|
+
XINCLUDE_START = 19
|
96
|
+
# XInclude end type
|
97
|
+
XINCLUDE_END = 20
|
98
|
+
# DOCB document node type
|
15
99
|
DOCB_DOCUMENT_NODE = 21
|
16
100
|
|
17
|
-
|
101
|
+
def initialize name, document # :nodoc:
|
102
|
+
# ... Ya. This is empty on purpose.
|
103
|
+
end
|
18
104
|
|
19
105
|
###
|
20
106
|
# Decorate this node with the decorators set up in this node's Document
|
21
107
|
def decorate!
|
22
|
-
document.decorate(self)
|
108
|
+
document.decorate(self)
|
23
109
|
end
|
24
110
|
|
25
111
|
###
|
26
|
-
#
|
27
|
-
def
|
28
|
-
|
29
|
-
|
30
|
-
document.decorate(list)
|
31
|
-
|
32
|
-
first = self.child
|
33
|
-
return list unless first # Empty list
|
34
|
-
|
35
|
-
list << first unless first.blank?
|
36
|
-
while first = first.next
|
37
|
-
list << first unless first.blank?
|
38
|
-
end
|
39
|
-
list
|
112
|
+
# Search this node's immediate children using CSS selector +selector+
|
113
|
+
def > selector
|
114
|
+
ns = document.root.namespaces
|
115
|
+
xpath CSS.xpath_for(selector, :prefix => "./", :ns => ns).first
|
40
116
|
end
|
41
117
|
|
42
118
|
###
|
43
|
-
#
|
44
|
-
|
45
|
-
|
46
|
-
def search *paths
|
47
|
-
ns = paths.last.is_a?(Hash) ? paths.pop : {}
|
48
|
-
xpath(*(paths.map { |path|
|
49
|
-
path =~ /^(\.\/|\/)/ ? path : CSS::Parser.parse(path).map { |ast|
|
50
|
-
ast.to_xpath
|
51
|
-
}
|
52
|
-
}.flatten.uniq) + [ns])
|
119
|
+
# Get the attribute value for the attribute +name+
|
120
|
+
def [] name
|
121
|
+
get(name.to_s)
|
53
122
|
end
|
54
|
-
alias :/ :search
|
55
123
|
|
56
|
-
|
57
|
-
|
124
|
+
###
|
125
|
+
# Set the attribute value for the attribute +name+ to +value+
|
126
|
+
def []= name, value
|
127
|
+
set name.to_s, value.to_s
|
128
|
+
end
|
58
129
|
|
59
|
-
|
130
|
+
###
|
131
|
+
# Add +node_or_tags+ as a child of this Node.
|
132
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
133
|
+
#
|
134
|
+
# Returns the reparented node (if +node_or_tags+ is a Node), or NodeSet (if +node_or_tags+ is a DocumentFragment, NodeSet, or string).
|
135
|
+
#
|
136
|
+
# Also see related method +<<+.
|
137
|
+
def add_child node_or_tags
|
138
|
+
node_or_tags = coerce(node_or_tags)
|
139
|
+
if node_or_tags.is_a?(XML::NodeSet)
|
140
|
+
node_or_tags.each { |n| add_child_node_and_reparent_attrs n }
|
141
|
+
else
|
142
|
+
add_child_node_and_reparent_attrs node_or_tags
|
143
|
+
end
|
144
|
+
node_or_tags
|
145
|
+
end
|
60
146
|
|
61
|
-
|
62
|
-
|
63
|
-
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
71
|
-
|
72
|
-
|
73
|
-
|
74
|
-
|
75
|
-
combined << node
|
76
|
-
end
|
77
|
-
end
|
147
|
+
###
|
148
|
+
# Add +node_or_tags+ as the first child of this Node.
|
149
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
150
|
+
#
|
151
|
+
# Returns the reparented node (if +node_or_tags+ is a Node), or NodeSet (if +node_or_tags+ is a DocumentFragment, NodeSet, or string).
|
152
|
+
#
|
153
|
+
# Also see related method +add_child+.
|
154
|
+
def prepend_child node_or_tags
|
155
|
+
if first = children.first
|
156
|
+
# Mimic the error add_child would raise.
|
157
|
+
raise RuntimeError, "Document already has a root node" if document? && !node_or_tags.processing_instruction?
|
158
|
+
first.__send__(:add_sibling, :previous, node_or_tags)
|
159
|
+
else
|
160
|
+
add_child(node_or_tags)
|
78
161
|
end
|
79
162
|
end
|
80
163
|
|
81
|
-
|
82
|
-
|
83
|
-
|
84
|
-
|
164
|
+
###
|
165
|
+
# Add +node_or_tags+ as a child of this Node.
|
166
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
167
|
+
#
|
168
|
+
# Returns self, to support chaining of calls (e.g., root << child1 << child2)
|
169
|
+
#
|
170
|
+
# Also see related method +add_child+.
|
171
|
+
def << node_or_tags
|
172
|
+
add_child node_or_tags
|
173
|
+
self
|
85
174
|
end
|
175
|
+
###
|
176
|
+
# Insert +node_or_tags+ before this Node (as a sibling).
|
177
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
178
|
+
#
|
179
|
+
# Returns the reparented node (if +node_or_tags+ is a Node), or NodeSet (if +node_or_tags+ is a DocumentFragment, NodeSet, or string).
|
180
|
+
#
|
181
|
+
# Also see related method +before+.
|
182
|
+
def add_previous_sibling node_or_tags
|
183
|
+
raise ArgumentError.new("A document may not have multiple root nodes.") if (parent && parent.document?) && !node_or_tags.processing_instruction?
|
86
184
|
|
87
|
-
|
88
|
-
search("#{path}", ns).first
|
185
|
+
add_sibling :previous, node_or_tags
|
89
186
|
end
|
90
187
|
|
91
|
-
|
92
|
-
|
93
|
-
|
188
|
+
###
|
189
|
+
# Insert +node_or_tags+ after this Node (as a sibling).
|
190
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
191
|
+
#
|
192
|
+
# Returns the reparented node (if +node_or_tags+ is a Node), or NodeSet (if +node_or_tags+ is a DocumentFragment, NodeSet, or string).
|
193
|
+
#
|
194
|
+
# Also see related method +after+.
|
195
|
+
def add_next_sibling node_or_tags
|
196
|
+
raise ArgumentError.new("A document may not have multiple root nodes.") if (parent && parent.document?) && !node_or_tags.processing_instruction?
|
197
|
+
|
198
|
+
add_sibling :next, node_or_tags
|
94
199
|
end
|
95
200
|
|
96
|
-
|
97
|
-
|
201
|
+
####
|
202
|
+
# Insert +node_or_tags+ before this node (as a sibling).
|
203
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
204
|
+
#
|
205
|
+
# Returns self, to support chaining of calls.
|
206
|
+
#
|
207
|
+
# Also see related method +add_previous_sibling+.
|
208
|
+
def before node_or_tags
|
209
|
+
add_previous_sibling node_or_tags
|
210
|
+
self
|
98
211
|
end
|
99
212
|
|
100
|
-
|
101
|
-
|
213
|
+
####
|
214
|
+
# Insert +node_or_tags+ after this node (as a sibling).
|
215
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a Nokogiri::XML::DocumentFragment, or a string containing markup.
|
216
|
+
#
|
217
|
+
# Returns self, to support chaining of calls.
|
218
|
+
#
|
219
|
+
# Also see related method +add_next_sibling+.
|
220
|
+
def after node_or_tags
|
221
|
+
add_next_sibling node_or_tags
|
222
|
+
self
|
102
223
|
end
|
103
224
|
|
104
225
|
####
|
105
|
-
#
|
106
|
-
#
|
107
|
-
|
108
|
-
|
109
|
-
|
226
|
+
# Set the inner html for this Node to +node_or_tags+
|
227
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a Nokogiri::XML::DocumentFragment, or a string containing markup.
|
228
|
+
#
|
229
|
+
# Returns self.
|
230
|
+
#
|
231
|
+
# Also see related method +children=+
|
232
|
+
def inner_html= node_or_tags
|
233
|
+
self.children = node_or_tags
|
234
|
+
self
|
235
|
+
end
|
110
236
|
|
111
|
-
|
112
|
-
|
237
|
+
####
|
238
|
+
# Set the inner html for this Node +node_or_tags+
|
239
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a Nokogiri::XML::DocumentFragment, or a string containing markup.
|
240
|
+
#
|
241
|
+
# Returns the reparented node (if +node_or_tags+ is a Node), or NodeSet (if +node_or_tags+ is a DocumentFragment, NodeSet, or string).
|
242
|
+
#
|
243
|
+
# Also see related method +inner_html=+
|
244
|
+
def children= node_or_tags
|
245
|
+
node_or_tags = coerce(node_or_tags)
|
246
|
+
children.unlink
|
247
|
+
if node_or_tags.is_a?(XML::NodeSet)
|
248
|
+
node_or_tags.each { |n| add_child_node_and_reparent_attrs n }
|
249
|
+
else
|
250
|
+
add_child_node_and_reparent_attrs node_or_tags
|
251
|
+
end
|
252
|
+
node_or_tags
|
113
253
|
end
|
114
254
|
|
115
255
|
####
|
116
|
-
#
|
117
|
-
#
|
118
|
-
|
119
|
-
|
120
|
-
|
256
|
+
# Replace this Node with +node_or_tags+.
|
257
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
258
|
+
#
|
259
|
+
# Returns the reparented node (if +node_or_tags+ is a Node), or NodeSet (if +node_or_tags+ is a DocumentFragment, NodeSet, or string).
|
260
|
+
#
|
261
|
+
# Also see related method +swap+.
|
262
|
+
def replace node_or_tags
|
263
|
+
# We cannot replace a text node directly, otherwise libxml will return
|
264
|
+
# an internal error at parser.c:13031, I don't know exactly why
|
265
|
+
# libxml is trying to find a parent node that is an element or document
|
266
|
+
# so I can't tell if this is bug in libxml or not. issue #775.
|
267
|
+
if text?
|
268
|
+
replacee = Nokogiri::XML::Node.new 'dummy', document
|
269
|
+
add_previous_sibling_node replacee
|
270
|
+
unlink
|
271
|
+
return replacee.replace node_or_tags
|
272
|
+
end
|
273
|
+
|
274
|
+
node_or_tags = coerce(node_or_tags)
|
121
275
|
|
122
|
-
|
123
|
-
|
124
|
-
|
125
|
-
|
126
|
-
|
276
|
+
if node_or_tags.is_a?(XML::NodeSet)
|
277
|
+
node_or_tags.each { |n| add_previous_sibling n }
|
278
|
+
unlink
|
279
|
+
else
|
280
|
+
replace_node node_or_tags
|
127
281
|
end
|
282
|
+
node_or_tags
|
128
283
|
end
|
129
284
|
|
130
|
-
|
131
|
-
|
285
|
+
####
|
286
|
+
# Swap this Node for +node_or_tags+
|
287
|
+
# +node_or_tags+ can be a Nokogiri::XML::Node, a ::DocumentFragment, a ::NodeSet, or a string containing markup.
|
288
|
+
#
|
289
|
+
# Returns self, to support chaining of calls.
|
290
|
+
#
|
291
|
+
# Also see related method +replace+.
|
292
|
+
def swap node_or_tags
|
293
|
+
replace node_or_tags
|
294
|
+
self
|
132
295
|
end
|
133
296
|
|
134
|
-
alias :
|
135
|
-
|
136
|
-
|
297
|
+
alias :next :next_sibling
|
298
|
+
alias :previous :previous_sibling
|
299
|
+
|
300
|
+
# :stopdoc:
|
301
|
+
# HACK: This is to work around an RDoc bug
|
302
|
+
alias :next= :add_next_sibling
|
303
|
+
# :startdoc:
|
304
|
+
|
305
|
+
alias :previous= :add_previous_sibling
|
306
|
+
alias :remove :unlink
|
307
|
+
alias :get_attribute :[]
|
308
|
+
alias :attr :[]
|
309
|
+
alias :set_attribute :[]=
|
310
|
+
alias :text :content
|
311
|
+
alias :inner_text :content
|
312
|
+
alias :has_attribute? :key?
|
313
|
+
alias :name :node_name
|
314
|
+
alias :name= :node_name=
|
315
|
+
alias :type :node_type
|
316
|
+
alias :to_str :text
|
317
|
+
alias :clone :dup
|
318
|
+
alias :elements :element_children
|
319
|
+
|
320
|
+
####
|
321
|
+
# Returns a hash containing the node's attributes. The key is
|
322
|
+
# the attribute name without any namespace, the value is a Nokogiri::XML::Attr
|
323
|
+
# representing the attribute.
|
324
|
+
# If you need to distinguish attributes with the same name, with different namespaces
|
325
|
+
# use #attribute_nodes instead.
|
326
|
+
def attributes
|
327
|
+
Hash[attribute_nodes.map { |node|
|
328
|
+
[node.node_name, node]
|
329
|
+
}]
|
137
330
|
end
|
138
331
|
|
139
|
-
|
140
|
-
|
332
|
+
###
|
333
|
+
# Get the attribute values for this Node.
|
334
|
+
def values
|
335
|
+
attribute_nodes.map(&:value)
|
336
|
+
end
|
337
|
+
|
338
|
+
###
|
339
|
+
# Get the attribute names for this Node.
|
340
|
+
def keys
|
341
|
+
attribute_nodes.map(&:node_name)
|
342
|
+
end
|
343
|
+
|
344
|
+
###
|
345
|
+
# Iterate over each attribute name and value pair for this Node.
|
346
|
+
def each
|
347
|
+
attribute_nodes.each { |node|
|
348
|
+
yield [node.node_name, node.value]
|
349
|
+
}
|
350
|
+
end
|
351
|
+
|
352
|
+
###
|
353
|
+
# Remove the attribute named +name+
|
354
|
+
def remove_attribute name
|
355
|
+
attr = attributes[name].remove if key? name
|
356
|
+
clear_xpath_context if Nokogiri.jruby?
|
357
|
+
attr
|
358
|
+
end
|
359
|
+
alias :delete :remove_attribute
|
360
|
+
|
361
|
+
###
|
362
|
+
# Returns true if this Node matches +selector+
|
363
|
+
def matches? selector
|
364
|
+
ancestors.last.search(selector).include?(self)
|
365
|
+
end
|
366
|
+
|
367
|
+
###
|
368
|
+
# Create a DocumentFragment containing +tags+ that is relative to _this_
|
369
|
+
# context node.
|
370
|
+
def fragment tags
|
371
|
+
type = document.html? ? Nokogiri::HTML : Nokogiri::XML
|
372
|
+
type::DocumentFragment.new(document, tags, self)
|
373
|
+
end
|
374
|
+
|
375
|
+
###
|
376
|
+
# Parse +string_or_io+ as a document fragment within the context of
|
377
|
+
# *this* node. Returns a XML::NodeSet containing the nodes parsed from
|
378
|
+
# +string_or_io+.
|
379
|
+
def parse string_or_io, options = nil
|
380
|
+
##
|
381
|
+
# When the current node is unparented and not an element node, use the
|
382
|
+
# document as the parsing context instead. Otherwise, the in-context
|
383
|
+
# parser cannot find an element or a document node.
|
384
|
+
# Document Fragments are also not usable by the in-context parser.
|
385
|
+
if !element? && !document? && (!parent || parent.fragment?)
|
386
|
+
return document.parse(string_or_io, options)
|
387
|
+
end
|
388
|
+
|
389
|
+
options ||= (document.html? ? ParseOptions::DEFAULT_HTML : ParseOptions::DEFAULT_XML)
|
390
|
+
if Fixnum === options
|
391
|
+
options = Nokogiri::XML::ParseOptions.new(options)
|
392
|
+
end
|
393
|
+
# Give the options to the user
|
394
|
+
yield options if block_given?
|
395
|
+
|
396
|
+
contents = string_or_io.respond_to?(:read) ?
|
397
|
+
string_or_io.read :
|
398
|
+
string_or_io
|
399
|
+
|
400
|
+
return Nokogiri::XML::NodeSet.new(document) if contents.empty?
|
401
|
+
|
402
|
+
##
|
403
|
+
# This is a horrible hack, but I don't care. See #313 for background.
|
404
|
+
error_count = document.errors.length
|
405
|
+
node_set = in_context(contents, options.to_i)
|
406
|
+
if node_set.empty? and document.errors.length > error_count and options.recover?
|
407
|
+
fragment = Nokogiri::HTML::DocumentFragment.parse contents
|
408
|
+
node_set = fragment.children
|
409
|
+
end
|
410
|
+
node_set
|
141
411
|
end
|
142
|
-
alias :inner_text :text
|
143
412
|
|
144
413
|
####
|
145
|
-
# Set the content to +string+.
|
146
|
-
|
147
|
-
|
148
|
-
|
414
|
+
# Set the Node's content to a Text node containing +string+. The string gets XML escaped, not interpreted as markup.
|
415
|
+
def content= string
|
416
|
+
self.native_content = encode_special_chars(string.to_s)
|
417
|
+
end
|
418
|
+
|
419
|
+
###
|
420
|
+
# Set the parent Node for this Node
|
421
|
+
def parent= parent_node
|
422
|
+
parent_node.add_child(self)
|
423
|
+
parent_node
|
149
424
|
end
|
150
425
|
|
426
|
+
###
|
427
|
+
# Returns a Hash of {prefix => value} for all namespaces on this
|
428
|
+
# node and its ancestors.
|
429
|
+
#
|
430
|
+
# This method returns the same namespaces as #namespace_scopes.
|
431
|
+
#
|
432
|
+
# Returns namespaces in scope for self -- those defined on self
|
433
|
+
# element directly or any ancestor node -- as a Hash of
|
434
|
+
# attribute-name/value pairs. Note that the keys in this hash
|
435
|
+
# XML attributes that would be used to define this namespace,
|
436
|
+
# such as "xmlns:prefix", not just the prefix. Default namespace
|
437
|
+
# set on self will be included with key "xmlns". However,
|
438
|
+
# default namespaces set on ancestor will NOT be, even if self
|
439
|
+
# has no explicit default namespace.
|
440
|
+
def namespaces
|
441
|
+
Hash[namespace_scopes.map { |nd|
|
442
|
+
key = ['xmlns', nd.prefix].compact.join(':')
|
443
|
+
[key, nd.href]
|
444
|
+
}]
|
445
|
+
end
|
446
|
+
|
447
|
+
# Returns true if this is a Comment
|
151
448
|
def comment?
|
152
449
|
type == COMMENT_NODE
|
153
450
|
end
|
154
451
|
|
452
|
+
# Returns true if this is a CDATA
|
155
453
|
def cdata?
|
156
454
|
type == CDATA_SECTION_NODE
|
157
455
|
end
|
158
456
|
|
457
|
+
# Returns true if this is an XML::Document node
|
159
458
|
def xml?
|
160
459
|
type == DOCUMENT_NODE
|
161
460
|
end
|
162
461
|
|
462
|
+
# Returns true if this is an HTML::Document node
|
163
463
|
def html?
|
164
464
|
type == HTML_DOCUMENT_NODE
|
165
465
|
end
|
166
466
|
|
167
|
-
|
168
|
-
|
467
|
+
# Returns true if this is a Document
|
468
|
+
def document?
|
469
|
+
is_a? XML::Document
|
470
|
+
end
|
471
|
+
|
472
|
+
# Returns true if this is a ProcessingInstruction node
|
473
|
+
def processing_instruction?
|
474
|
+
type == PI_NODE
|
475
|
+
end
|
476
|
+
|
477
|
+
# Returns true if this is a Text node
|
478
|
+
def text?
|
479
|
+
type == TEXT_NODE
|
480
|
+
end
|
481
|
+
|
482
|
+
# Returns true if this is a DocumentFragment
|
483
|
+
def fragment?
|
484
|
+
type == DOCUMENT_FRAG_NODE
|
485
|
+
end
|
486
|
+
|
487
|
+
###
|
488
|
+
# Fetch the Nokogiri::HTML::ElementDescription for this node. Returns
|
489
|
+
# nil on XML documents and on unknown tags.
|
490
|
+
def description
|
491
|
+
return nil if document.xml?
|
492
|
+
Nokogiri::HTML::ElementDescription[name]
|
493
|
+
end
|
494
|
+
|
495
|
+
###
|
496
|
+
# Is this a read only node?
|
497
|
+
def read_only?
|
498
|
+
# According to gdome2, these are read-only node types
|
499
|
+
[NOTATION_NODE, ENTITY_NODE, ENTITY_DECL].include?(type)
|
169
500
|
end
|
170
|
-
alias :to_s :to_html
|
171
|
-
alias :inner_html :to_html
|
172
501
|
|
502
|
+
# Returns true if this is an Element node
|
503
|
+
def element?
|
504
|
+
type == ELEMENT_NODE
|
505
|
+
end
|
506
|
+
alias :elem? :element?
|
507
|
+
|
508
|
+
###
|
509
|
+
# Turn this node in to a string. If the document is HTML, this method
|
510
|
+
# returns html. If the document is XML, this method returns XML.
|
511
|
+
def to_s
|
512
|
+
document.xml? ? to_xml : to_html
|
513
|
+
end
|
514
|
+
|
515
|
+
# Get the inner_html for this node's Node#children
|
516
|
+
def inner_html *args
|
517
|
+
children.map { |x| x.to_html(*args) }.join
|
518
|
+
end
|
519
|
+
|
520
|
+
# Get the path to this node as a CSS expression
|
173
521
|
def css_path
|
174
522
|
path.split(/\//).map { |part|
|
175
523
|
part.length == 0 ? nil : part.gsub(/\[(\d+)\]/, ':nth-of-type(\1)')
|
176
524
|
}.compact.join(' > ')
|
177
525
|
end
|
178
526
|
|
179
|
-
|
180
|
-
|
181
|
-
|
182
|
-
|
183
|
-
|
184
|
-
|
527
|
+
###
|
528
|
+
# Get a list of ancestor Node for this Node. If +selector+ is given,
|
529
|
+
# the ancestors must match +selector+
|
530
|
+
def ancestors selector = nil
|
531
|
+
return NodeSet.new(document) unless respond_to?(:parent)
|
532
|
+
return NodeSet.new(document) unless parent
|
533
|
+
|
534
|
+
parents = [parent]
|
535
|
+
|
536
|
+
while parents.last.respond_to?(:parent)
|
537
|
+
break unless ctx_parent = parents.last.parent
|
538
|
+
parents << ctx_parent
|
539
|
+
end
|
540
|
+
|
541
|
+
return NodeSet.new(document, parents) unless selector
|
542
|
+
|
543
|
+
root = parents.last
|
544
|
+
search_results = root.search(selector)
|
545
|
+
|
546
|
+
NodeSet.new(document, parents.find_all { |parent|
|
547
|
+
search_results.include?(parent)
|
548
|
+
})
|
549
|
+
end
|
550
|
+
|
551
|
+
###
|
552
|
+
# Adds a default namespace supplied as a string +url+ href, to self.
|
553
|
+
# The consequence is as an xmlns attribute with supplied argument were
|
554
|
+
# present in parsed XML. A default namespace set with this method will
|
555
|
+
# now show up in #attributes, but when this node is serialized to XML an
|
556
|
+
# "xmlns" attribute will appear. See also #namespace and #namespace=
|
557
|
+
def default_namespace= url
|
558
|
+
add_namespace_definition(nil, url)
|
559
|
+
end
|
560
|
+
alias :add_namespace :add_namespace_definition
|
561
|
+
|
562
|
+
###
|
563
|
+
# Set the default namespace on this node (as would be defined with an
|
564
|
+
# "xmlns=" attribute in XML source), as a Namespace object +ns+. Note that
|
565
|
+
# a Namespace added this way will NOT be serialized as an xmlns attribute
|
566
|
+
# for this node. You probably want #default_namespace= instead, or perhaps
|
567
|
+
# #add_namespace_definition with a nil prefix argument.
|
568
|
+
def namespace= ns
|
569
|
+
return set_namespace(ns) unless ns
|
570
|
+
|
571
|
+
unless Nokogiri::XML::Namespace === ns
|
572
|
+
raise TypeError, "#{ns.class} can't be coerced into Nokogiri::XML::Namespace"
|
573
|
+
end
|
574
|
+
if ns.document != document
|
575
|
+
raise ArgumentError, 'namespace must be declared on the same document'
|
576
|
+
end
|
577
|
+
|
578
|
+
set_namespace ns
|
185
579
|
end
|
186
580
|
|
187
581
|
####
|
188
582
|
# Yields self and all children to +block+ recursively.
|
189
|
-
def traverse
|
583
|
+
def traverse &block
|
190
584
|
children.each{|j| j.traverse(&block) }
|
191
585
|
block.call(self)
|
192
586
|
end
|
587
|
+
|
588
|
+
###
|
589
|
+
# Accept a visitor. This method calls "visit" on +visitor+ with self.
|
590
|
+
def accept visitor
|
591
|
+
visitor.visit(self)
|
592
|
+
end
|
593
|
+
|
594
|
+
###
|
595
|
+
# Test to see if this Node is equal to +other+
|
596
|
+
def == other
|
597
|
+
return false unless other
|
598
|
+
return false unless other.respond_to?(:pointer_id)
|
599
|
+
pointer_id == other.pointer_id
|
600
|
+
end
|
601
|
+
|
602
|
+
###
|
603
|
+
# Serialize Node using +options+. Save options can also be set using a
|
604
|
+
# block. See SaveOptions.
|
605
|
+
#
|
606
|
+
# These two statements are equivalent:
|
607
|
+
#
|
608
|
+
# node.serialize(:encoding => 'UTF-8', :save_with => FORMAT | AS_XML)
|
609
|
+
#
|
610
|
+
# or
|
611
|
+
#
|
612
|
+
# node.serialize(:encoding => 'UTF-8') do |config|
|
613
|
+
# config.format.as_xml
|
614
|
+
# end
|
615
|
+
#
|
616
|
+
def serialize *args, &block
|
617
|
+
options = args.first.is_a?(Hash) ? args.shift : {
|
618
|
+
:encoding => args[0],
|
619
|
+
:save_with => args[1]
|
620
|
+
}
|
621
|
+
|
622
|
+
encoding = options[:encoding] || document.encoding
|
623
|
+
options[:encoding] = encoding
|
624
|
+
|
625
|
+
outstring = ""
|
626
|
+
if encoding && outstring.respond_to?(:force_encoding)
|
627
|
+
outstring.force_encoding(Encoding.find(encoding))
|
628
|
+
end
|
629
|
+
io = StringIO.new(outstring)
|
630
|
+
write_to io, options, &block
|
631
|
+
io.string
|
632
|
+
end
|
633
|
+
|
634
|
+
###
|
635
|
+
# Serialize this Node to HTML
|
636
|
+
#
|
637
|
+
# doc.to_html
|
638
|
+
#
|
639
|
+
# See Node#write_to for a list of +options+. For formatted output,
|
640
|
+
# use Node#to_xhtml instead.
|
641
|
+
def to_html options = {}
|
642
|
+
to_format SaveOptions::DEFAULT_HTML, options
|
643
|
+
end
|
644
|
+
|
645
|
+
###
|
646
|
+
# Serialize this Node to XML using +options+
|
647
|
+
#
|
648
|
+
# doc.to_xml(:indent => 5, :encoding => 'UTF-8')
|
649
|
+
#
|
650
|
+
# See Node#write_to for a list of +options+
|
651
|
+
def to_xml options = {}
|
652
|
+
options[:save_with] ||= SaveOptions::DEFAULT_XML
|
653
|
+
serialize(options)
|
654
|
+
end
|
655
|
+
|
656
|
+
###
|
657
|
+
# Serialize this Node to XHTML using +options+
|
658
|
+
#
|
659
|
+
# doc.to_xhtml(:indent => 5, :encoding => 'UTF-8')
|
660
|
+
#
|
661
|
+
# See Node#write_to for a list of +options+
|
662
|
+
def to_xhtml options = {}
|
663
|
+
to_format SaveOptions::DEFAULT_XHTML, options
|
664
|
+
end
|
665
|
+
|
666
|
+
###
|
667
|
+
# Write Node to +io+ with +options+. +options+ modify the output of
|
668
|
+
# this method. Valid options are:
|
669
|
+
#
|
670
|
+
# * +:encoding+ for changing the encoding
|
671
|
+
# * +:indent_text+ the indentation text, defaults to one space
|
672
|
+
# * +:indent+ the number of +:indent_text+ to use, defaults to 2
|
673
|
+
# * +:save_with+ a combination of SaveOptions constants.
|
674
|
+
#
|
675
|
+
# To save with UTF-8 indented twice:
|
676
|
+
#
|
677
|
+
# node.write_to(io, :encoding => 'UTF-8', :indent => 2)
|
678
|
+
#
|
679
|
+
# To save indented with two dashes:
|
680
|
+
#
|
681
|
+
# node.write_to(io, :indent_text => '-', :indent => 2
|
682
|
+
#
|
683
|
+
def write_to io, *options
|
684
|
+
options = options.first.is_a?(Hash) ? options.shift : {}
|
685
|
+
encoding = options[:encoding] || options[0]
|
686
|
+
if Nokogiri.jruby?
|
687
|
+
save_options = options[:save_with] || options[1]
|
688
|
+
indent_times = options[:indent] || 0
|
689
|
+
else
|
690
|
+
save_options = options[:save_with] || options[1] || SaveOptions::FORMAT
|
691
|
+
indent_times = options[:indent] || 2
|
692
|
+
end
|
693
|
+
indent_text = options[:indent_text] || ' '
|
694
|
+
|
695
|
+
config = SaveOptions.new(save_options.to_i)
|
696
|
+
yield config if block_given?
|
697
|
+
|
698
|
+
native_write_to(io, encoding, indent_text * indent_times, config.options)
|
699
|
+
end
|
700
|
+
|
701
|
+
###
|
702
|
+
# Write Node as HTML to +io+ with +options+
|
703
|
+
#
|
704
|
+
# See Node#write_to for a list of +options+
|
705
|
+
def write_html_to io, options = {}
|
706
|
+
write_format_to SaveOptions::DEFAULT_HTML, io, options
|
707
|
+
end
|
708
|
+
|
709
|
+
###
|
710
|
+
# Write Node as XHTML to +io+ with +options+
|
711
|
+
#
|
712
|
+
# See Node#write_to for a list of +options+
|
713
|
+
def write_xhtml_to io, options = {}
|
714
|
+
write_format_to SaveOptions::DEFAULT_XHTML, io, options
|
715
|
+
end
|
716
|
+
|
717
|
+
###
|
718
|
+
# Write Node as XML to +io+ with +options+
|
719
|
+
#
|
720
|
+
# doc.write_xml_to io, :encoding => 'UTF-8'
|
721
|
+
#
|
722
|
+
# See Node#write_to for a list of options
|
723
|
+
def write_xml_to io, options = {}
|
724
|
+
options[:save_with] ||= SaveOptions::DEFAULT_XML
|
725
|
+
write_to io, options
|
726
|
+
end
|
727
|
+
|
728
|
+
###
|
729
|
+
# Compare two Node objects with respect to their Document. Nodes from
|
730
|
+
# different documents cannot be compared.
|
731
|
+
def <=> other
|
732
|
+
return nil unless other.is_a?(Nokogiri::XML::Node)
|
733
|
+
return nil unless document == other.document
|
734
|
+
compare other
|
735
|
+
end
|
736
|
+
|
737
|
+
###
|
738
|
+
# Do xinclude substitution on the subtree below node. If given a block, a
|
739
|
+
# Nokogiri::XML::ParseOptions object initialized from +options+, will be
|
740
|
+
# passed to it, allowing more convenient modification of the parser options.
|
741
|
+
def do_xinclude options = XML::ParseOptions::DEFAULT_XML, &block
|
742
|
+
options = Nokogiri::XML::ParseOptions.new(options) if Fixnum === options
|
743
|
+
|
744
|
+
# give options to user
|
745
|
+
yield options if block_given?
|
746
|
+
|
747
|
+
# call c extension
|
748
|
+
process_xincludes(options.to_i)
|
749
|
+
end
|
750
|
+
|
751
|
+
def canonicalize(mode=XML::XML_C14N_1_0,inclusive_namespaces=nil,with_comments=false)
|
752
|
+
c14n_root = self
|
753
|
+
document.canonicalize(mode, inclusive_namespaces, with_comments) do |node, parent|
|
754
|
+
tn = node.is_a?(XML::Node) ? node : parent
|
755
|
+
tn == c14n_root || tn.ancestors.include?(c14n_root)
|
756
|
+
end
|
757
|
+
end
|
758
|
+
|
759
|
+
private
|
760
|
+
|
761
|
+
def add_sibling next_or_previous, node_or_tags
|
762
|
+
impl = (next_or_previous == :next) ? :add_next_sibling_node : :add_previous_sibling_node
|
763
|
+
iter = (next_or_previous == :next) ? :reverse_each : :each
|
764
|
+
|
765
|
+
node_or_tags = coerce node_or_tags
|
766
|
+
if node_or_tags.is_a?(XML::NodeSet)
|
767
|
+
if text?
|
768
|
+
pivot = Nokogiri::XML::Node.new 'dummy', document
|
769
|
+
send impl, pivot
|
770
|
+
else
|
771
|
+
pivot = self
|
772
|
+
end
|
773
|
+
node_or_tags.send(iter) { |n| pivot.send impl, n }
|
774
|
+
pivot.unlink if text?
|
775
|
+
else
|
776
|
+
send impl, node_or_tags
|
777
|
+
end
|
778
|
+
node_or_tags
|
779
|
+
end
|
780
|
+
|
781
|
+
def to_format save_option, options
|
782
|
+
# FIXME: this is a hack around broken libxml versions
|
783
|
+
return dump_html if Nokogiri.uses_libxml? && %w[2 6] === LIBXML_VERSION.split('.')[0..1]
|
784
|
+
|
785
|
+
options[:save_with] = save_option unless options[:save_with]
|
786
|
+
serialize(options)
|
787
|
+
end
|
788
|
+
|
789
|
+
def write_format_to save_option, io, options
|
790
|
+
# FIXME: this is a hack around broken libxml versions
|
791
|
+
return (io << dump_html) if Nokogiri.uses_libxml? && %w[2 6] === LIBXML_VERSION.split('.')[0..1]
|
792
|
+
|
793
|
+
options[:save_with] ||= save_option
|
794
|
+
write_to io, options
|
795
|
+
end
|
796
|
+
|
797
|
+
def inspect_attributes
|
798
|
+
[:name, :namespace, :attribute_nodes, :children]
|
799
|
+
end
|
800
|
+
|
801
|
+
def coerce data # :nodoc:
|
802
|
+
case data
|
803
|
+
when XML::NodeSet
|
804
|
+
return data
|
805
|
+
when XML::DocumentFragment
|
806
|
+
return data.children
|
807
|
+
when String
|
808
|
+
return fragment(data).children
|
809
|
+
when Document, XML::Attr
|
810
|
+
# unacceptable
|
811
|
+
when XML::Node
|
812
|
+
return data
|
813
|
+
end
|
814
|
+
|
815
|
+
raise ArgumentError, <<-EOERR
|
816
|
+
Requires a Node, NodeSet or String argument, and cannot accept a #{data.class}.
|
817
|
+
(You probably want to select a node from the Document with at() or search(), or create a new Node via Node.new().)
|
818
|
+
EOERR
|
819
|
+
end
|
820
|
+
|
821
|
+
def implied_xpath_contexts # :nodoc:
|
822
|
+
[".//"]
|
823
|
+
end
|
824
|
+
|
825
|
+
def add_child_node_and_reparent_attrs node # :nodoc:
|
826
|
+
add_child_node node
|
827
|
+
node.attribute_nodes.find_all { |a| a.name =~ /:/ }.each do |attr_node|
|
828
|
+
attr_node.remove
|
829
|
+
node[attr_node.name] = attr_node.value
|
830
|
+
end
|
831
|
+
end
|
193
832
|
end
|
194
833
|
end
|
195
834
|
end
|