nokogiri 1.6.8.1 → 1.13.4
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 +5 -5
- data/Gemfile +3 -20
- data/LICENSE-DEPENDENCIES.md +1903 -0
- data/LICENSE.md +9 -0
- data/README.md +197 -83
- data/bin/nokogiri +63 -50
- data/dependencies.yml +16 -22
- data/ext/nokogiri/depend +38 -358
- data/ext/nokogiri/extconf.rb +751 -432
- data/ext/nokogiri/gumbo.c +584 -0
- data/ext/nokogiri/html4_document.c +166 -0
- data/ext/nokogiri/html4_element_description.c +294 -0
- data/ext/nokogiri/html4_entity_lookup.c +37 -0
- data/ext/nokogiri/html4_sax_parser_context.c +120 -0
- data/ext/nokogiri/html4_sax_push_parser.c +95 -0
- data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
- data/ext/nokogiri/nokogiri.c +228 -95
- data/ext/nokogiri/nokogiri.h +190 -98
- data/ext/nokogiri/test_global_handlers.c +40 -0
- data/ext/nokogiri/xml_attr.c +49 -40
- data/ext/nokogiri/xml_attribute_decl.c +18 -18
- data/ext/nokogiri/xml_cdata.c +24 -23
- data/ext/nokogiri/xml_comment.c +19 -26
- data/ext/nokogiri/xml_document.c +291 -217
- data/ext/nokogiri/xml_document_fragment.c +12 -16
- data/ext/nokogiri/xml_dtd.c +56 -50
- data/ext/nokogiri/xml_element_content.c +31 -26
- data/ext/nokogiri/xml_element_decl.c +22 -22
- data/ext/nokogiri/xml_encoding_handler.c +43 -18
- data/ext/nokogiri/xml_entity_decl.c +32 -30
- data/ext/nokogiri/xml_entity_reference.c +16 -18
- data/ext/nokogiri/xml_namespace.c +61 -58
- data/ext/nokogiri/xml_node.c +1194 -729
- data/ext/nokogiri/xml_node_set.c +178 -165
- data/ext/nokogiri/xml_processing_instruction.c +17 -19
- data/ext/nokogiri/xml_reader.c +226 -175
- data/ext/nokogiri/xml_relax_ng.c +52 -28
- data/ext/nokogiri/xml_sax_parser.c +115 -114
- data/ext/nokogiri/xml_sax_parser_context.c +105 -86
- data/ext/nokogiri/xml_sax_push_parser.c +88 -35
- data/ext/nokogiri/xml_schema.c +112 -33
- data/ext/nokogiri/xml_syntax_error.c +42 -21
- data/ext/nokogiri/xml_text.c +13 -17
- data/ext/nokogiri/xml_xpath_context.c +223 -115
- data/ext/nokogiri/xslt_stylesheet.c +265 -173
- data/gumbo-parser/CHANGES.md +63 -0
- data/gumbo-parser/Makefile +101 -0
- data/gumbo-parser/THANKS +27 -0
- data/gumbo-parser/src/Makefile +34 -0
- data/gumbo-parser/src/README.md +41 -0
- data/gumbo-parser/src/ascii.c +75 -0
- data/gumbo-parser/src/ascii.h +115 -0
- data/gumbo-parser/src/attribute.c +42 -0
- data/gumbo-parser/src/attribute.h +17 -0
- data/gumbo-parser/src/char_ref.c +22225 -0
- data/gumbo-parser/src/char_ref.h +29 -0
- data/gumbo-parser/src/char_ref.rl +2154 -0
- data/gumbo-parser/src/error.c +626 -0
- data/gumbo-parser/src/error.h +148 -0
- data/gumbo-parser/src/foreign_attrs.c +104 -0
- data/gumbo-parser/src/foreign_attrs.gperf +27 -0
- data/gumbo-parser/src/gumbo.h +943 -0
- data/gumbo-parser/src/insertion_mode.h +33 -0
- data/gumbo-parser/src/macros.h +91 -0
- data/gumbo-parser/src/parser.c +4875 -0
- data/gumbo-parser/src/parser.h +41 -0
- data/gumbo-parser/src/replacement.h +33 -0
- data/gumbo-parser/src/string_buffer.c +103 -0
- data/gumbo-parser/src/string_buffer.h +68 -0
- data/gumbo-parser/src/string_piece.c +48 -0
- data/gumbo-parser/src/svg_attrs.c +174 -0
- data/gumbo-parser/src/svg_attrs.gperf +77 -0
- data/gumbo-parser/src/svg_tags.c +137 -0
- data/gumbo-parser/src/svg_tags.gperf +55 -0
- data/gumbo-parser/src/tag.c +222 -0
- data/gumbo-parser/src/tag_lookup.c +382 -0
- data/gumbo-parser/src/tag_lookup.gperf +169 -0
- data/gumbo-parser/src/tag_lookup.h +13 -0
- data/gumbo-parser/src/token_buffer.c +79 -0
- data/gumbo-parser/src/token_buffer.h +71 -0
- data/gumbo-parser/src/token_type.h +17 -0
- data/gumbo-parser/src/tokenizer.c +3463 -0
- data/gumbo-parser/src/tokenizer.h +112 -0
- data/gumbo-parser/src/tokenizer_states.h +339 -0
- data/gumbo-parser/src/utf8.c +245 -0
- data/gumbo-parser/src/utf8.h +164 -0
- data/gumbo-parser/src/util.c +68 -0
- data/gumbo-parser/src/util.h +30 -0
- data/gumbo-parser/src/vector.c +111 -0
- data/gumbo-parser/src/vector.h +45 -0
- data/lib/nokogiri/class_resolver.rb +67 -0
- data/lib/nokogiri/css/node.rb +10 -8
- data/lib/nokogiri/css/parser.rb +410 -372
- data/lib/nokogiri/css/parser.y +260 -244
- data/lib/nokogiri/css/parser_extras.rb +54 -49
- data/lib/nokogiri/css/syntax_error.rb +3 -1
- data/lib/nokogiri/css/tokenizer.rb +107 -104
- data/lib/nokogiri/css/tokenizer.rex +7 -6
- data/lib/nokogiri/css/xpath_visitor.rb +226 -92
- data/lib/nokogiri/css.rb +50 -17
- data/lib/nokogiri/decorators/slop.rb +9 -7
- data/lib/nokogiri/extension.rb +31 -0
- data/lib/nokogiri/gumbo.rb +15 -0
- data/lib/nokogiri/html.rb +38 -27
- data/lib/nokogiri/{html → html4}/builder.rb +4 -2
- data/lib/nokogiri/{html → html4}/document.rb +104 -106
- data/lib/nokogiri/html4/document_fragment.rb +54 -0
- data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
- data/lib/nokogiri/html4/element_description_defaults.rb +578 -0
- data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
- data/lib/nokogiri/{html → html4}/sax/parser.rb +24 -13
- data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
- data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
- data/lib/nokogiri/html4.rb +46 -0
- data/lib/nokogiri/html5/document.rb +91 -0
- data/lib/nokogiri/html5/document_fragment.rb +83 -0
- data/lib/nokogiri/html5/node.rb +100 -0
- data/lib/nokogiri/html5.rb +478 -0
- data/lib/nokogiri/jruby/dependencies.rb +21 -0
- data/lib/nokogiri/syntax_error.rb +2 -0
- data/lib/nokogiri/version/constant.rb +6 -0
- data/lib/nokogiri/version/info.rb +222 -0
- data/lib/nokogiri/version.rb +3 -107
- data/lib/nokogiri/xml/attr.rb +6 -3
- data/lib/nokogiri/xml/attribute_decl.rb +3 -1
- data/lib/nokogiri/xml/builder.rb +97 -53
- data/lib/nokogiri/xml/cdata.rb +3 -1
- data/lib/nokogiri/xml/character_data.rb +2 -0
- data/lib/nokogiri/xml/document.rb +231 -93
- data/lib/nokogiri/xml/document_fragment.rb +57 -44
- data/lib/nokogiri/xml/dtd.rb +4 -2
- data/lib/nokogiri/xml/element_content.rb +2 -0
- data/lib/nokogiri/xml/element_decl.rb +3 -1
- data/lib/nokogiri/xml/entity_decl.rb +4 -2
- data/lib/nokogiri/xml/entity_reference.rb +20 -0
- data/lib/nokogiri/xml/namespace.rb +3 -0
- data/lib/nokogiri/xml/node/save_options.rb +10 -5
- data/lib/nokogiri/xml/node.rb +911 -337
- data/lib/nokogiri/xml/node_set.rb +141 -84
- data/lib/nokogiri/xml/notation.rb +13 -0
- data/lib/nokogiri/xml/parse_options.rb +23 -9
- data/lib/nokogiri/xml/pp/character_data.rb +9 -6
- data/lib/nokogiri/xml/pp/node.rb +25 -26
- data/lib/nokogiri/xml/pp.rb +4 -2
- data/lib/nokogiri/xml/processing_instruction.rb +3 -1
- data/lib/nokogiri/xml/reader.rb +23 -28
- data/lib/nokogiri/xml/relax_ng.rb +8 -2
- data/lib/nokogiri/xml/sax/document.rb +45 -49
- data/lib/nokogiri/xml/sax/parser.rb +43 -40
- data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
- data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
- data/lib/nokogiri/xml/sax.rb +6 -4
- data/lib/nokogiri/xml/schema.rb +19 -9
- data/lib/nokogiri/xml/searchable.rb +138 -89
- data/lib/nokogiri/xml/syntax_error.rb +26 -1
- data/lib/nokogiri/xml/text.rb +2 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
- data/lib/nokogiri/xml/xpath.rb +15 -4
- data/lib/nokogiri/xml/xpath_context.rb +3 -3
- data/lib/nokogiri/xml.rb +39 -37
- data/lib/nokogiri/xslt/stylesheet.rb +3 -1
- data/lib/nokogiri/xslt.rb +29 -20
- data/lib/nokogiri.rb +49 -65
- data/lib/xsd/xmlparser/nokogiri.rb +26 -24
- data/patches/libxml2/0001-Remove-script-macro-support.patch +40 -0
- data/patches/libxml2/0002-Update-entities-to-remove-handling-of-ssi.patch +44 -0
- data/patches/libxml2/0003-libxml2.la-is-in-top_builddir.patch +25 -0
- data/patches/libxml2/0004-use-glibc-strlen.patch +53 -0
- data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
- data/patches/libxml2/0006-update-automake-files-for-arm64.patch +3040 -0
- data/patches/libxml2/0008-htmlParseComment-handle-abruptly-closed-comments.patch +61 -0
- data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
- data/patches/libxml2/0010-Revert-Different-approach-to-fix-quadratic-behavior.patch +45 -0
- data/patches/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
- data/ports/archives/libxml2-2.9.13.tar.xz +0 -0
- data/ports/archives/libxslt-1.1.35.tar.xz +0 -0
- metadata +233 -258
- data/.autotest +0 -26
- data/.cross_rubies +0 -9
- data/.editorconfig +0 -17
- data/.gemtest +0 -0
- data/.travis.yml +0 -51
- data/CHANGELOG.rdoc +0 -1160
- data/CONTRIBUTING.md +0 -42
- data/C_CODING_STYLE.rdoc +0 -33
- data/LICENSE.txt +0 -31
- data/Manifest.txt +0 -364
- data/ROADMAP.md +0 -111
- data/Rakefile +0 -375
- data/STANDARD_RESPONSES.md +0 -47
- data/Y_U_NO_GEMSPEC.md +0 -155
- data/appveyor.yml +0 -22
- data/build_all +0 -45
- data/ext/nokogiri/html_document.c +0 -170
- data/ext/nokogiri/html_document.h +0 -10
- data/ext/nokogiri/html_element_description.c +0 -279
- data/ext/nokogiri/html_element_description.h +0 -10
- data/ext/nokogiri/html_entity_lookup.c +0 -32
- data/ext/nokogiri/html_entity_lookup.h +0 -8
- data/ext/nokogiri/html_sax_parser_context.c +0 -116
- data/ext/nokogiri/html_sax_parser_context.h +0 -11
- data/ext/nokogiri/html_sax_push_parser.c +0 -87
- data/ext/nokogiri/html_sax_push_parser.h +0 -9
- data/ext/nokogiri/xml_attr.h +0 -9
- data/ext/nokogiri/xml_attribute_decl.h +0 -9
- data/ext/nokogiri/xml_cdata.h +0 -9
- data/ext/nokogiri/xml_comment.h +0 -9
- data/ext/nokogiri/xml_document.h +0 -23
- data/ext/nokogiri/xml_document_fragment.h +0 -10
- data/ext/nokogiri/xml_dtd.h +0 -10
- data/ext/nokogiri/xml_element_content.h +0 -10
- data/ext/nokogiri/xml_element_decl.h +0 -9
- data/ext/nokogiri/xml_encoding_handler.h +0 -8
- data/ext/nokogiri/xml_entity_decl.h +0 -10
- data/ext/nokogiri/xml_entity_reference.h +0 -9
- data/ext/nokogiri/xml_io.c +0 -60
- data/ext/nokogiri/xml_io.h +0 -11
- data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
- data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
- data/ext/nokogiri/xml_namespace.h +0 -13
- data/ext/nokogiri/xml_node.h +0 -13
- data/ext/nokogiri/xml_node_set.h +0 -13
- data/ext/nokogiri/xml_processing_instruction.h +0 -9
- data/ext/nokogiri/xml_reader.h +0 -10
- data/ext/nokogiri/xml_relax_ng.h +0 -9
- data/ext/nokogiri/xml_sax_parser.h +0 -39
- data/ext/nokogiri/xml_sax_parser_context.h +0 -10
- data/ext/nokogiri/xml_sax_push_parser.h +0 -9
- data/ext/nokogiri/xml_schema.h +0 -9
- data/ext/nokogiri/xml_syntax_error.h +0 -13
- data/ext/nokogiri/xml_text.h +0 -9
- data/ext/nokogiri/xml_xpath_context.h +0 -10
- data/ext/nokogiri/xslt_stylesheet.h +0 -14
- data/lib/nokogiri/html/document_fragment.rb +0 -39
- data/lib/nokogiri/html/element_description_defaults.rb +0 -671
- data/lib/nokogiri/html/sax/parser_context.rb +0 -16
- data/patches/sort-patches-by-date +0 -25
- 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 +0 -1
- data/suppressions/nokogiri_ree-1.8.7.358.supp +0 -61
- data/suppressions/nokogiri_ruby-1.8.7.370.supp +0 -0
- data/suppressions/nokogiri_ruby-1.9.2.320.supp +0 -28
- data/suppressions/nokogiri_ruby-1.9.3.327.supp +0 -28
- data/tasks/test.rb +0 -100
- data/test/css/test_nthiness.rb +0 -226
- data/test/css/test_parser.rb +0 -369
- data/test/css/test_tokenizer.rb +0 -215
- data/test/css/test_xpath_visitor.rb +0 -96
- data/test/decorators/test_slop.rb +0 -20
- data/test/files/2ch.html +0 -108
- data/test/files/GH_1042.html +0 -18
- data/test/files/address_book.rlx +0 -12
- data/test/files/address_book.xml +0 -10
- data/test/files/atom.xml +0 -344
- data/test/files/bar/bar.xsd +0 -4
- data/test/files/bogus.xml +0 -0
- data/test/files/dont_hurt_em_why.xml +0 -422
- data/test/files/encoding.html +0 -82
- data/test/files/encoding.xhtml +0 -84
- data/test/files/exslt.xml +0 -8
- data/test/files/exslt.xslt +0 -35
- data/test/files/foo/foo.xsd +0 -4
- data/test/files/metacharset.html +0 -10
- data/test/files/namespace_pressure_test.xml +0 -1684
- data/test/files/noencoding.html +0 -47
- data/test/files/po.xml +0 -32
- data/test/files/po.xsd +0 -66
- data/test/files/saml/saml20assertion_schema.xsd +0 -283
- data/test/files/saml/saml20protocol_schema.xsd +0 -302
- data/test/files/saml/xenc_schema.xsd +0 -146
- data/test/files/saml/xmldsig_schema.xsd +0 -318
- data/test/files/shift_jis.html +0 -10
- data/test/files/shift_jis.xml +0 -5
- data/test/files/shift_jis_no_charset.html +0 -9
- data/test/files/slow-xpath.xml +0 -25509
- data/test/files/snuggles.xml +0 -3
- data/test/files/staff.dtd +0 -10
- data/test/files/staff.xml +0 -59
- data/test/files/staff.xslt +0 -32
- data/test/files/test_document_url/bar.xml +0 -2
- data/test/files/test_document_url/document.dtd +0 -4
- data/test/files/test_document_url/document.xml +0 -6
- data/test/files/tlm.html +0 -851
- data/test/files/to_be_xincluded.xml +0 -2
- data/test/files/valid_bar.xml +0 -2
- data/test/files/xinclude.xml +0 -4
- data/test/helper.rb +0 -181
- data/test/html/sax/test_parser.rb +0 -141
- data/test/html/sax/test_parser_context.rb +0 -46
- data/test/html/sax/test_push_parser.rb +0 -87
- data/test/html/test_builder.rb +0 -164
- data/test/html/test_document.rb +0 -701
- data/test/html/test_document_encoding.rb +0 -145
- data/test/html/test_document_fragment.rb +0 -301
- data/test/html/test_element_description.rb +0 -105
- data/test/html/test_named_characters.rb +0 -14
- data/test/html/test_node.rb +0 -212
- data/test/html/test_node_encoding.rb +0 -85
- data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +0 -14
- data/test/namespaces/test_namespaces_aliased_default.rb +0 -24
- data/test/namespaces/test_namespaces_in_builder_doc.rb +0 -75
- data/test/namespaces/test_namespaces_in_cloned_doc.rb +0 -31
- data/test/namespaces/test_namespaces_in_created_doc.rb +0 -75
- data/test/namespaces/test_namespaces_in_parsed_doc.rb +0 -80
- data/test/namespaces/test_namespaces_preservation.rb +0 -31
- data/test/test_convert_xpath.rb +0 -135
- data/test/test_css_cache.rb +0 -45
- data/test/test_encoding_handler.rb +0 -48
- data/test/test_memory_leak.rb +0 -156
- data/test/test_nokogiri.rb +0 -138
- data/test/test_soap4r_sax.rb +0 -52
- data/test/test_xslt_transforms.rb +0 -314
- data/test/xml/node/test_save_options.rb +0 -28
- data/test/xml/node/test_subclass.rb +0 -44
- data/test/xml/sax/test_parser.rb +0 -394
- data/test/xml/sax/test_parser_context.rb +0 -115
- data/test/xml/sax/test_push_parser.rb +0 -157
- data/test/xml/test_attr.rb +0 -67
- data/test/xml/test_attribute_decl.rb +0 -86
- data/test/xml/test_builder.rb +0 -341
- data/test/xml/test_c14n.rb +0 -180
- data/test/xml/test_cdata.rb +0 -48
- data/test/xml/test_comment.rb +0 -40
- data/test/xml/test_document.rb +0 -982
- data/test/xml/test_document_encoding.rb +0 -31
- data/test/xml/test_document_fragment.rb +0 -271
- data/test/xml/test_dtd.rb +0 -187
- data/test/xml/test_dtd_encoding.rb +0 -31
- data/test/xml/test_element_content.rb +0 -56
- data/test/xml/test_element_decl.rb +0 -73
- data/test/xml/test_entity_decl.rb +0 -122
- data/test/xml/test_entity_reference.rb +0 -251
- data/test/xml/test_namespace.rb +0 -96
- data/test/xml/test_node.rb +0 -1244
- data/test/xml/test_node_attributes.rb +0 -115
- data/test/xml/test_node_encoding.rb +0 -69
- data/test/xml/test_node_inheritance.rb +0 -32
- data/test/xml/test_node_reparenting.rb +0 -549
- data/test/xml/test_node_set.rb +0 -775
- data/test/xml/test_parse_options.rb +0 -64
- data/test/xml/test_processing_instruction.rb +0 -30
- data/test/xml/test_reader.rb +0 -589
- data/test/xml/test_reader_encoding.rb +0 -134
- data/test/xml/test_relax_ng.rb +0 -60
- data/test/xml/test_schema.rb +0 -142
- data/test/xml/test_syntax_error.rb +0 -30
- data/test/xml/test_text.rb +0 -60
- data/test/xml/test_unparented_node.rb +0 -440
- data/test/xml/test_xinclude.rb +0 -83
- data/test/xml/test_xpath.rb +0 -445
- data/test/xslt/test_custom_functions.rb +0 -133
- data/test/xslt/test_exception_handling.rb +0 -37
- data/test_all +0 -107
@@ -1,64 +1,143 @@
|
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
1
4
|
module Nokogiri
|
2
5
|
module CSS
|
3
|
-
|
4
|
-
|
6
|
+
# When translating CSS selectors to XPath queries with Nokogiri::CSS.xpath_for, the XPathVisitor
|
7
|
+
# class allows for changing some of the behaviors related to builtin xpath functions and quirks
|
8
|
+
# of HTML5.
|
9
|
+
class XPathVisitor
|
10
|
+
WILDCARD_NAMESPACES = Nokogiri.libxml2_patches.include?("0009-allow-wildcard-namespaces.patch") # :nodoc:
|
11
|
+
|
12
|
+
# Enum to direct XPathVisitor when to use Nokogiri builtin XPath functions.
|
13
|
+
module BuiltinsConfig
|
14
|
+
# Never use Nokogiri builtin functions, always generate vanilla XPath 1.0 queries. This is
|
15
|
+
# the default when calling Nokogiri::CSS.xpath_for directly.
|
16
|
+
NEVER = :never
|
17
|
+
|
18
|
+
# Always use Nokogiri builtin functions whenever possible. This is probably only useful for testing.
|
19
|
+
ALWAYS = :always
|
20
|
+
|
21
|
+
# Only use Nokogiri builtin functions when they will be faster than vanilla XPath. This is
|
22
|
+
# the behavior chosen when searching for CSS selectors on a Nokogiri document, fragment, or
|
23
|
+
# node.
|
24
|
+
OPTIMAL = :optimal
|
25
|
+
|
26
|
+
# :nodoc: array of values for validation
|
27
|
+
VALUES = [NEVER, ALWAYS, OPTIMAL]
|
28
|
+
end
|
5
29
|
|
6
|
-
|
7
|
-
|
30
|
+
# Enum to direct XPathVisitor when to tweak the XPath query to suit the nature of the document
|
31
|
+
# being searched. Note that searches for CSS selectors from a Nokogiri document, fragment, or
|
32
|
+
# node will choose the correct option automatically.
|
33
|
+
module DoctypeConfig
|
34
|
+
# The document being searched is an XML document. This is the default.
|
35
|
+
XML = :xml
|
36
|
+
|
37
|
+
# The document being searched is an HTML4 document.
|
38
|
+
HTML4 = :html4
|
39
|
+
|
40
|
+
# The document being searched is an HTML5 document.
|
41
|
+
HTML5 = :html5
|
42
|
+
|
43
|
+
# :nodoc: array of values for validation
|
44
|
+
VALUES = [XML, HTML4, HTML5]
|
45
|
+
end
|
46
|
+
|
47
|
+
# :call-seq:
|
48
|
+
# new() → XPathVisitor
|
49
|
+
# new(builtins:, doctype:) → XPathVisitor
|
50
|
+
#
|
51
|
+
# [Parameters]
|
52
|
+
# - +builtins:+ (BuiltinsConfig) Determine when to use Nokogiri's built-in xpath functions for performance improvements.
|
53
|
+
# - +doctype:+ (DoctypeConfig) Make document-type-specific accommodations for CSS queries.
|
54
|
+
#
|
55
|
+
# [Returns] XPathVisitor
|
56
|
+
#
|
57
|
+
def initialize(builtins: BuiltinsConfig::NEVER, doctype: DoctypeConfig::XML)
|
58
|
+
unless BuiltinsConfig::VALUES.include?(builtins)
|
59
|
+
raise(ArgumentError, "Invalid values #{builtins.inspect} for builtins: keyword parameter")
|
60
|
+
end
|
61
|
+
unless DoctypeConfig::VALUES.include?(doctype)
|
62
|
+
raise(ArgumentError, "Invalid values #{doctype.inspect} for doctype: keyword parameter")
|
63
|
+
end
|
64
|
+
|
65
|
+
@builtins = builtins
|
66
|
+
@doctype = doctype
|
67
|
+
end
|
68
|
+
|
69
|
+
# :call-seq: config() → Hash
|
70
|
+
#
|
71
|
+
# [Returns]
|
72
|
+
# a Hash representing the configuration of the XPathVisitor, suitable for use as
|
73
|
+
# part of the CSS cache key.
|
74
|
+
def config
|
75
|
+
{ builtins: @builtins, doctype: @doctype }
|
76
|
+
end
|
77
|
+
|
78
|
+
# :stopdoc:
|
79
|
+
def visit_function(node)
|
80
|
+
msg = :"visit_function_#{node.value.first.gsub(/[(]/, "")}"
|
81
|
+
return send(msg, node) if respond_to?(msg)
|
8
82
|
|
9
83
|
case node.value.first
|
10
84
|
when /^text\(/
|
11
|
-
|
85
|
+
"child::text()"
|
12
86
|
when /^self\(/
|
13
87
|
"self::#{node.value[1]}"
|
14
88
|
when /^eq\(/
|
15
|
-
"position()
|
89
|
+
"position()=#{node.value[1]}"
|
16
90
|
when /^(nth|nth-of-type)\(/
|
17
|
-
if node.value[1].is_a?(Nokogiri::CSS::Node)
|
91
|
+
if node.value[1].is_a?(Nokogiri::CSS::Node) && (node.value[1].type == :NTH)
|
18
92
|
nth(node.value[1])
|
19
93
|
else
|
20
|
-
"position()
|
94
|
+
"position()=#{node.value[1]}"
|
21
95
|
end
|
22
96
|
when /^nth-child\(/
|
23
|
-
if node.value[1].is_a?(Nokogiri::CSS::Node)
|
24
|
-
nth(node.value[1], :
|
97
|
+
if node.value[1].is_a?(Nokogiri::CSS::Node) && (node.value[1].type == :NTH)
|
98
|
+
nth(node.value[1], child: true)
|
25
99
|
else
|
26
|
-
"count(preceding-sibling::*)
|
100
|
+
"count(preceding-sibling::*)=#{node.value[1].to_i - 1}"
|
27
101
|
end
|
28
102
|
when /^nth-last-of-type\(/
|
29
|
-
if node.value[1].is_a?(Nokogiri::CSS::Node)
|
30
|
-
nth(node.value[1], :
|
103
|
+
if node.value[1].is_a?(Nokogiri::CSS::Node) && (node.value[1].type == :NTH)
|
104
|
+
nth(node.value[1], last: true)
|
31
105
|
else
|
32
106
|
index = node.value[1].to_i - 1
|
33
|
-
index == 0 ? "position()
|
107
|
+
index == 0 ? "position()=last()" : "position()=last()-#{index}"
|
34
108
|
end
|
35
109
|
when /^nth-last-child\(/
|
36
|
-
if node.value[1].is_a?(Nokogiri::CSS::Node)
|
37
|
-
nth(node.value[1], :
|
110
|
+
if node.value[1].is_a?(Nokogiri::CSS::Node) && (node.value[1].type == :NTH)
|
111
|
+
nth(node.value[1], last: true, child: true)
|
38
112
|
else
|
39
|
-
"count(following-sibling::*)
|
113
|
+
"count(following-sibling::*)=#{node.value[1].to_i - 1}"
|
40
114
|
end
|
41
115
|
when /^(first|first-of-type)\(/
|
42
|
-
"position()
|
116
|
+
"position()=1"
|
43
117
|
when /^(last|last-of-type)\(/
|
44
|
-
"position()
|
118
|
+
"position()=last()"
|
45
119
|
when /^contains\(/
|
46
|
-
"contains(
|
120
|
+
"contains(.,#{node.value[1]})"
|
47
121
|
when /^gt\(/
|
48
|
-
"position()
|
122
|
+
"position()>#{node.value[1]}"
|
49
123
|
when /^only-child\(/
|
50
|
-
"last()
|
124
|
+
"last()=1"
|
51
125
|
when /^comment\(/
|
52
126
|
"comment()"
|
53
127
|
when /^has\(/
|
54
|
-
node.value[1].
|
128
|
+
is_direct = node.value[1].value[0].nil? # e.g. "has(> a)", "has(~ a)", "has(+ a)"
|
129
|
+
".#{"//" unless is_direct}#{node.value[1].accept(self)}"
|
55
130
|
else
|
56
|
-
|
57
|
-
|
131
|
+
# xpath function call, let's marshal those arguments
|
132
|
+
args = ["."]
|
133
|
+
args += node.value[1..-1].map do |n|
|
134
|
+
n.is_a?(Nokogiri::CSS::Node) ? n.accept(self) : n
|
135
|
+
end
|
136
|
+
"#{node.value.first}#{args.join(",")})"
|
58
137
|
end
|
59
138
|
end
|
60
139
|
|
61
|
-
def visit_not
|
140
|
+
def visit_not(node)
|
62
141
|
child = node.value.first
|
63
142
|
if :ELEMENT_NAME == child.type
|
64
143
|
"not(self::#{child.accept(self)})"
|
@@ -67,64 +146,63 @@ module Nokogiri
|
|
67
146
|
end
|
68
147
|
end
|
69
148
|
|
70
|
-
def visit_id
|
149
|
+
def visit_id(node)
|
71
150
|
node.value.first =~ /^#(.*)$/
|
72
|
-
"@id
|
151
|
+
"@id='#{Regexp.last_match(1)}'"
|
73
152
|
end
|
74
153
|
|
75
|
-
def visit_attribute_condition
|
76
|
-
|
77
|
-
|
78
|
-
else
|
79
|
-
'@'
|
80
|
-
end
|
81
|
-
attribute += node.value.first.accept(self)
|
82
|
-
|
83
|
-
# Support non-standard css
|
84
|
-
attribute.gsub!(/^@@/, '@')
|
85
|
-
|
86
|
-
return attribute unless node.value.length == 3
|
154
|
+
def visit_attribute_condition(node)
|
155
|
+
attribute = node.value.first.accept(self)
|
156
|
+
return attribute if node.value.length == 1
|
87
157
|
|
88
158
|
value = node.value.last
|
89
|
-
value = "'#{value}'"
|
159
|
+
value = "'#{value}'" unless /^['"]/.match?(value)
|
160
|
+
|
161
|
+
# quoted values - see test_attribute_value_with_quotes in test/css/test_parser.rb
|
162
|
+
if (value[0] == value[-1]) && %q{"'}.include?(value[0])
|
163
|
+
str_value = value[1..-2]
|
164
|
+
if str_value.include?(value[0])
|
165
|
+
value = 'concat("' + str_value.split('"', -1).join(%q{",'"',"}) + '","")'
|
166
|
+
end
|
167
|
+
end
|
90
168
|
|
91
169
|
case node.value[1]
|
92
170
|
when :equal
|
93
|
-
attribute + "
|
171
|
+
attribute + "=" + value.to_s
|
94
172
|
when :not_equal
|
95
|
-
attribute + "
|
173
|
+
attribute + "!=" + value.to_s
|
96
174
|
when :substring_match
|
97
|
-
"contains(#{attribute}
|
175
|
+
"contains(#{attribute},#{value})"
|
98
176
|
when :prefix_match
|
99
|
-
"starts-with(#{attribute}
|
177
|
+
"starts-with(#{attribute},#{value})"
|
100
178
|
when :dash_match
|
101
|
-
"#{attribute}
|
179
|
+
"#{attribute}=#{value} or starts-with(#{attribute},concat(#{value},'-'))"
|
102
180
|
when :includes
|
103
|
-
|
181
|
+
value = value[1..-2] # strip quotes
|
182
|
+
css_class(attribute, value)
|
104
183
|
when :suffix_match
|
105
|
-
"substring(#{attribute},
|
106
|
-
"string-length(#{value}) + 1, string-length(#{value})) = #{value}"
|
184
|
+
"substring(#{attribute},string-length(#{attribute})-string-length(#{value})+1,string-length(#{value}))=#{value}"
|
107
185
|
else
|
108
|
-
attribute + " #{node.value[1]} " +
|
186
|
+
attribute + " #{node.value[1]} " + value.to_s
|
109
187
|
end
|
110
188
|
end
|
111
189
|
|
112
|
-
def visit_pseudo_class
|
113
|
-
if node.value.first.is_a?(Nokogiri::CSS::Node)
|
190
|
+
def visit_pseudo_class(node)
|
191
|
+
if node.value.first.is_a?(Nokogiri::CSS::Node) && (node.value.first.type == :FUNCTION)
|
114
192
|
node.value.first.accept(self)
|
115
193
|
else
|
116
|
-
msg = :"visit_pseudo_class_#{node.value.first.gsub(/[(]/,
|
117
|
-
return
|
194
|
+
msg = :"visit_pseudo_class_#{node.value.first.gsub(/[(]/, "")}"
|
195
|
+
return send(msg, node) if respond_to?(msg)
|
118
196
|
|
119
197
|
case node.value.first
|
120
|
-
when "first" then "position()
|
121
|
-
when "first-child" then "count(preceding-sibling::*)
|
122
|
-
when "last" then "position()
|
123
|
-
when "last-child" then "count(following-sibling::*)
|
124
|
-
when "first-of-type" then "position()
|
125
|
-
when "last-of-type" then "position()
|
126
|
-
when "only-child" then "count(preceding-sibling::*)
|
127
|
-
when "only-of-type" then "last()
|
198
|
+
when "first" then "position()=1"
|
199
|
+
when "first-child" then "count(preceding-sibling::*)=0"
|
200
|
+
when "last" then "position()=last()"
|
201
|
+
when "last-child" then "count(following-sibling::*)=0"
|
202
|
+
when "first-of-type" then "position()=1"
|
203
|
+
when "last-of-type" then "position()=last()"
|
204
|
+
when "only-child" then "count(preceding-sibling::*)=0 and count(following-sibling::*)=0"
|
205
|
+
when "only-of-type" then "last()=1"
|
128
206
|
when "empty" then "not(node())"
|
129
207
|
when "parent" then "node()"
|
130
208
|
when "root" then "not(parent::*)"
|
@@ -134,24 +212,24 @@ module Nokogiri
|
|
134
212
|
end
|
135
213
|
end
|
136
214
|
|
137
|
-
def visit_class_condition
|
138
|
-
"
|
215
|
+
def visit_class_condition(node)
|
216
|
+
css_class("@class", node.value.first)
|
139
217
|
end
|
140
218
|
|
141
|
-
def visit_combinator
|
219
|
+
def visit_combinator(node)
|
142
220
|
if is_of_type_pseudo_class?(node.value.last)
|
143
|
-
"#{node.value.first
|
221
|
+
"#{node.value.first&.accept(self)}][#{node.value.last.accept(self)}"
|
144
222
|
else
|
145
|
-
"#{node.value.first
|
223
|
+
"#{node.value.first&.accept(self)} and #{node.value.last.accept(self)}"
|
146
224
|
end
|
147
225
|
end
|
148
|
-
|
226
|
+
|
149
227
|
{
|
150
|
-
|
151
|
-
|
152
|
-
|
153
|
-
|
154
|
-
}.each do |k,v|
|
228
|
+
"direct_adjacent_selector" => "/following-sibling::*[1]/self::",
|
229
|
+
"following_selector" => "/following-sibling::",
|
230
|
+
"descendant_selector" => "//",
|
231
|
+
"child_selector" => "/",
|
232
|
+
}.each do |k, v|
|
155
233
|
class_eval %{
|
156
234
|
def visit_#{k} node
|
157
235
|
"\#{node.value.first.accept(self) if node.value.first}#{v}\#{node.value.last.accept(self)}"
|
@@ -159,43 +237,69 @@ module Nokogiri
|
|
159
237
|
}
|
160
238
|
end
|
161
239
|
|
162
|
-
def visit_conditional_selector
|
163
|
-
node.value.first.accept(self) +
|
164
|
-
|
240
|
+
def visit_conditional_selector(node)
|
241
|
+
node.value.first.accept(self) + "[" +
|
242
|
+
node.value.last.accept(self) + "]"
|
243
|
+
end
|
244
|
+
|
245
|
+
def visit_element_name(node)
|
246
|
+
if @doctype == DoctypeConfig::HTML5 && html5_element_name_needs_namespace_handling(node)
|
247
|
+
# HTML5 has namespaces that should be ignored in CSS queries
|
248
|
+
# https://github.com/sparklemotion/nokogiri/issues/2376
|
249
|
+
if @builtins == BuiltinsConfig::ALWAYS || (@builtins == BuiltinsConfig::OPTIMAL && Nokogiri.uses_libxml?)
|
250
|
+
if WILDCARD_NAMESPACES
|
251
|
+
"*:#{node.value.first}"
|
252
|
+
else
|
253
|
+
"*[nokogiri-builtin:local-name-is('#{node.value.first}')]"
|
254
|
+
end
|
255
|
+
else
|
256
|
+
"*[local-name()='#{node.value.first}']"
|
257
|
+
end
|
258
|
+
else
|
259
|
+
node.value.first
|
260
|
+
end
|
165
261
|
end
|
166
262
|
|
167
|
-
def
|
168
|
-
node.value.first
|
263
|
+
def visit_attrib_name(node)
|
264
|
+
"@#{node.value.first}"
|
169
265
|
end
|
170
266
|
|
171
|
-
def accept
|
267
|
+
def accept(node)
|
172
268
|
node.accept(self)
|
173
269
|
end
|
174
270
|
|
175
|
-
|
176
|
-
|
271
|
+
private
|
272
|
+
|
273
|
+
def html5_element_name_needs_namespace_handling(node)
|
274
|
+
# if this is the wildcard selector "*", use it as normal
|
275
|
+
node.value.first != "*" &&
|
276
|
+
# if there is already a namespace (i.e., it is a prefixed QName), use it as normal
|
277
|
+
!node.value.first.include?(":")
|
278
|
+
end
|
279
|
+
|
280
|
+
def nth(node, options = {})
|
177
281
|
raise ArgumentError, "expected an+b node to contain 4 tokens, but is #{node.value.inspect}" unless node.value.size == 4
|
178
282
|
|
179
|
-
a, b = read_a_and_positive_b
|
283
|
+
a, b = read_a_and_positive_b(node.value)
|
180
284
|
position = if options[:child]
|
181
|
-
options[:last] ? "(count(following-sibling::*)
|
285
|
+
options[:last] ? "(count(following-sibling::*)+1)" : "(count(preceding-sibling::*)+1)"
|
182
286
|
else
|
183
287
|
options[:last] ? "(last()-position()+1)" : "position()"
|
184
288
|
end
|
185
289
|
|
186
290
|
if b.zero?
|
187
|
-
"(#{position} mod #{a})
|
291
|
+
"(#{position} mod #{a})=0"
|
188
292
|
else
|
189
293
|
compare = a < 0 ? "<=" : ">="
|
190
294
|
if a.abs == 1
|
191
|
-
"#{position}
|
295
|
+
"#{position}#{compare}#{b}"
|
192
296
|
else
|
193
|
-
"(#{position}
|
297
|
+
"(#{position}#{compare}#{b}) and (((#{position}-#{b}) mod #{a.abs})=0)"
|
194
298
|
end
|
195
299
|
end
|
196
300
|
end
|
197
301
|
|
198
|
-
def read_a_and_positive_b
|
302
|
+
def read_a_and_positive_b(values)
|
199
303
|
op = values[2]
|
200
304
|
if op == "+"
|
201
305
|
a = values[0].to_i
|
@@ -208,15 +312,45 @@ module Nokogiri
|
|
208
312
|
end
|
209
313
|
[a, b]
|
210
314
|
end
|
211
|
-
|
212
|
-
def is_of_type_pseudo_class?
|
213
|
-
if node.type
|
214
|
-
if node.value[0].is_a?(Nokogiri::CSS::Node)
|
315
|
+
|
316
|
+
def is_of_type_pseudo_class?(node) # rubocop:disable Naming/PredicateName
|
317
|
+
if node.type == :PSEUDO_CLASS
|
318
|
+
if node.value[0].is_a?(Nokogiri::CSS::Node) && (node.value[0].type == :FUNCTION)
|
215
319
|
node.value[0].value[0]
|
216
320
|
else
|
217
321
|
node.value[0]
|
218
322
|
end =~ /(nth|first|last|only)-of-type(\()?/
|
219
|
-
end
|
323
|
+
end
|
324
|
+
end
|
325
|
+
|
326
|
+
def css_class(hay, needle)
|
327
|
+
if @builtins == BuiltinsConfig::ALWAYS || (@builtins == BuiltinsConfig::OPTIMAL && Nokogiri.uses_libxml?)
|
328
|
+
# use the builtin implementation
|
329
|
+
"nokogiri-builtin:css-class(#{hay},'#{needle}')"
|
330
|
+
else
|
331
|
+
# use only ordinary xpath functions
|
332
|
+
"contains(concat(' ',normalize-space(#{hay}),' '),' #{needle} ')"
|
333
|
+
end
|
334
|
+
end
|
335
|
+
end
|
336
|
+
|
337
|
+
module XPathVisitorAlwaysUseBuiltins # :nodoc:
|
338
|
+
def self.new
|
339
|
+
warn(
|
340
|
+
"Nokogiri::CSS::XPathVisitorAlwaysUseBuiltins is deprecated and will be removed in a future version of Nokogiri",
|
341
|
+
{ uplevel: 1 },
|
342
|
+
)
|
343
|
+
XPathVisitor.new(builtins: :always)
|
344
|
+
end
|
345
|
+
end
|
346
|
+
|
347
|
+
module XPathVisitorOptimallyUseBuiltins # :nodoc:
|
348
|
+
def self.new
|
349
|
+
warn(
|
350
|
+
"Nokogiri::CSS::XPathVisitorOptimallyUseBuiltins is deprecated and will be removed in a future version of Nokogiri",
|
351
|
+
{ uplevel: 1 },
|
352
|
+
)
|
353
|
+
XPathVisitor.new(builtins: :optimal)
|
220
354
|
end
|
221
355
|
end
|
222
356
|
end
|
data/lib/nokogiri/css.rb
CHANGED
@@ -1,27 +1,60 @@
|
|
1
|
-
|
2
|
-
|
3
|
-
x = $-w
|
4
|
-
$-w = false
|
5
|
-
require 'nokogiri/css/parser'
|
6
|
-
$-w = x
|
7
|
-
|
8
|
-
require 'nokogiri/css/tokenizer'
|
9
|
-
require 'nokogiri/css/syntax_error'
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
10
3
|
|
11
4
|
module Nokogiri
|
5
|
+
# Translate a CSS selector into an XPath 1.0 query
|
12
6
|
module CSS
|
13
7
|
class << self
|
14
|
-
|
15
|
-
#
|
16
|
-
def parse
|
17
|
-
Parser.new.parse
|
8
|
+
# TODO: Deprecate this method ahead of 2.0 and delete it in 2.0.
|
9
|
+
# It is not used by Nokogiri and shouldn't be part of the public API.
|
10
|
+
def parse(selector) # :nodoc:
|
11
|
+
Parser.new.parse(selector)
|
18
12
|
end
|
19
13
|
|
20
|
-
|
21
|
-
#
|
22
|
-
|
23
|
-
|
14
|
+
# :call-seq:
|
15
|
+
# xpath_for(selector) → String
|
16
|
+
# xpath_for(selector [, prefix:] [, visitor:] [, ns:]) → String
|
17
|
+
#
|
18
|
+
# Translate a CSS selector to the equivalent XPath query.
|
19
|
+
#
|
20
|
+
# [Parameters]
|
21
|
+
# - +selector+ (String) The CSS selector to be translated into XPath
|
22
|
+
#
|
23
|
+
# - +prefix:+ (String)
|
24
|
+
#
|
25
|
+
# The XPath prefix for the query, see Nokogiri::XML::XPath for some options. Default is
|
26
|
+
# +XML::XPath::GLOBAL_SEARCH_PREFIX+.
|
27
|
+
#
|
28
|
+
# - +visitor:+ (Nokogiri::CSS::XPathVisitor)
|
29
|
+
#
|
30
|
+
# The visitor class to use to transform the AST into XPath. Default is
|
31
|
+
# +Nokogiri::CSS::XPathVisitor.new+.
|
32
|
+
#
|
33
|
+
# - +ns:+ (Hash<String ⇒ String>)
|
34
|
+
#
|
35
|
+
# The namespaces that are referenced in the query, if any. This is a hash where the keys are
|
36
|
+
# the namespace prefix and the values are the namespace URIs. Default is an empty Hash.
|
37
|
+
#
|
38
|
+
# [Returns] (String) The equivalent XPath query for +selector+
|
39
|
+
#
|
40
|
+
# 💡 Note that translated queries are cached for performance concerns.
|
41
|
+
#
|
42
|
+
def xpath_for(selector, options = {})
|
43
|
+
prefix = options.fetch(:prefix, Nokogiri::XML::XPath::GLOBAL_SEARCH_PREFIX)
|
44
|
+
visitor = options.fetch(:visitor) { Nokogiri::CSS::XPathVisitor.new }
|
45
|
+
ns = options.fetch(:ns, {})
|
46
|
+
Parser.new(ns).xpath_for(selector, prefix, visitor)
|
24
47
|
end
|
25
48
|
end
|
26
49
|
end
|
27
50
|
end
|
51
|
+
|
52
|
+
require_relative "css/node"
|
53
|
+
require_relative "css/xpath_visitor"
|
54
|
+
x = $-w
|
55
|
+
$-w = false
|
56
|
+
require_relative "css/parser"
|
57
|
+
$-w = x
|
58
|
+
|
59
|
+
require_relative "css/tokenizer"
|
60
|
+
require_relative "css/syntax_error"
|
@@ -1,3 +1,5 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
module Nokogiri
|
2
4
|
module Decorators
|
3
5
|
###
|
@@ -9,21 +11,21 @@ module Nokogiri
|
|
9
11
|
|
10
12
|
###
|
11
13
|
# look for node with +name+. See Nokogiri.Slop
|
12
|
-
def method_missing
|
14
|
+
def method_missing(name, *args, &block)
|
13
15
|
if args.empty?
|
14
|
-
list = xpath("#{XPATH_PREFIX}#{name.to_s.sub(/^_/,
|
15
|
-
elsif args.first.is_a?
|
16
|
+
list = xpath("#{XPATH_PREFIX}#{name.to_s.sub(/^_/, "")}")
|
17
|
+
elsif args.first.is_a?(Hash)
|
16
18
|
hash = args.first
|
17
19
|
if hash[:css]
|
18
20
|
list = css("#{name}#{hash[:css]}")
|
19
21
|
elsif hash[:xpath]
|
20
|
-
conds = Array(hash[:xpath]).join(
|
22
|
+
conds = Array(hash[:xpath]).join(" and ")
|
21
23
|
list = xpath("#{XPATH_PREFIX}#{name}[#{conds}]")
|
22
24
|
end
|
23
25
|
else
|
24
26
|
CSS::Parser.without_cache do
|
25
27
|
list = xpath(
|
26
|
-
*CSS.xpath_for("#{name}#{args.first}", :
|
28
|
+
*CSS.xpath_for("#{name}#{args.first}", prefix: XPATH_PREFIX)
|
27
29
|
)
|
28
30
|
end
|
29
31
|
end
|
@@ -32,8 +34,8 @@ module Nokogiri
|
|
32
34
|
list.length == 1 ? list.first : list
|
33
35
|
end
|
34
36
|
|
35
|
-
def respond_to_missing?
|
36
|
-
list = xpath("#{XPATH_PREFIX}#{name.to_s.sub(/^_/,
|
37
|
+
def respond_to_missing?(name, include_private = false)
|
38
|
+
list = xpath("#{XPATH_PREFIX}#{name.to_s.sub(/^_/, "")}")
|
37
39
|
|
38
40
|
!list.empty?
|
39
41
|
end
|
@@ -0,0 +1,31 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
# load the C or Java extension
|
4
|
+
begin
|
5
|
+
# native precompiled gems package shared libraries in <gem_dir>/lib/nokogiri/<ruby_version>
|
6
|
+
::RUBY_VERSION =~ /(\d+\.\d+)/
|
7
|
+
require_relative "#{Regexp.last_match(1)}/nokogiri"
|
8
|
+
rescue LoadError => e
|
9
|
+
if /GLIBC/.match?(e.message)
|
10
|
+
warn(<<~EOM)
|
11
|
+
|
12
|
+
ERROR: It looks like you're trying to use Nokogiri as a precompiled native gem on a system with glibc < 2.17:
|
13
|
+
|
14
|
+
#{e.message}
|
15
|
+
|
16
|
+
If that's the case, then please install Nokogiri via the `ruby` platform gem:
|
17
|
+
gem install nokogiri --platform=ruby
|
18
|
+
or:
|
19
|
+
bundle config set force_ruby_platform true
|
20
|
+
|
21
|
+
Please visit https://nokogiri.org/tutorials/installing_nokogiri.html for more help.
|
22
|
+
|
23
|
+
EOM
|
24
|
+
raise e
|
25
|
+
end
|
26
|
+
|
27
|
+
# use "require" instead of "require_relative" because non-native gems will place C extension files
|
28
|
+
# in Gem::BasicSpecification#extension_dir after compilation (during normal installation), which
|
29
|
+
# is in $LOAD_PATH but not necessarily relative to this file (see #2300)
|
30
|
+
require "nokogiri/nokogiri"
|
31
|
+
end
|
@@ -0,0 +1,15 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Nokogiri
|
4
|
+
module Gumbo
|
5
|
+
# The default maximum number of attributes per element.
|
6
|
+
DEFAULT_MAX_ATTRIBUTES = 400
|
7
|
+
|
8
|
+
# The default maximum number of errors for parsing a document or a fragment.
|
9
|
+
DEFAULT_MAX_ERRORS = 0
|
10
|
+
|
11
|
+
# The default maximum depth of the DOM tree produced by parsing a document
|
12
|
+
# or fragment.
|
13
|
+
DEFAULT_MAX_TREE_DEPTH = 400
|
14
|
+
end
|
15
|
+
end
|