nokogiri 1.8.5 → 1.13.6
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 +4 -4
- data/Gemfile +3 -21
- data/LICENSE-DEPENDENCIES.md +1159 -868
- data/LICENSE.md +5 -28
- data/README.md +196 -90
- data/bin/nokogiri +63 -50
- data/dependencies.yml +13 -59
- data/ext/nokogiri/depend +38 -358
- data/ext/nokogiri/extconf.rb +750 -420
- 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 +119 -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 -91
- data/ext/nokogiri/nokogiri.h +191 -89
- data/ext/nokogiri/test_global_handlers.c +40 -0
- data/ext/nokogiri/xml_attr.c +41 -36
- data/ext/nokogiri/xml_attribute_decl.c +18 -18
- data/ext/nokogiri/xml_cdata.c +13 -18
- data/ext/nokogiri/xml_comment.c +19 -26
- data/ext/nokogiri/xml_document.c +291 -216
- 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 -52
- data/ext/nokogiri/xml_node.c +1044 -616
- data/ext/nokogiri/xml_node_set.c +174 -162
- 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 +112 -112
- data/ext/nokogiri/xml_sax_parser_context.c +112 -86
- data/ext/nokogiri/xml_sax_push_parser.c +36 -27
- 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 +397 -377
- data/lib/nokogiri/css/parser.y +250 -245
- 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 +3 -2
- data/lib/nokogiri/css/xpath_visitor.rb +218 -91
- 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 +103 -105
- 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 +17 -16
- 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 -108
- 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 +224 -86
- 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 +2 -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 +895 -377
- data/lib/nokogiri/xml/node_set.rb +92 -65
- data/lib/nokogiri/xml/notation.rb +13 -0
- data/lib/nokogiri/xml/parse_options.rb +22 -8
- 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 +38 -34
- 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 +112 -72
- data/lib/nokogiri/xml/syntax_error.rb +6 -4
- 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 +38 -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/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
- data/ports/archives/libxml2-2.9.14.tar.xz +0 -0
- data/ports/archives/libxslt-1.1.35.tar.xz +0 -0
- metadata +220 -266
- data/.autotest +0 -22
- data/.cross_rubies +0 -8
- data/.editorconfig +0 -17
- data/.gemtest +0 -0
- data/.travis.yml +0 -63
- data/CHANGELOG.md +0 -1368
- data/CONTRIBUTING.md +0 -42
- data/C_CODING_STYLE.rdoc +0 -33
- data/Gemfile-libxml-ruby +0 -3
- data/Manifest.txt +0 -370
- data/ROADMAP.md +0 -111
- data/Rakefile +0 -348
- data/SECURITY.md +0 -19
- data/STANDARD_RESPONSES.md +0 -47
- data/Y_U_NO_GEMSPEC.md +0 -155
- data/appveyor.yml +0 -29
- data/build_all +0 -44
- 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 -61
- 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 -15
- data/ext/nokogiri/xml_node.h +0 -13
- data/ext/nokogiri/xml_node_set.h +0 -12
- 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 -49
- data/lib/nokogiri/html/element_description_defaults.rb +0 -671
- data/lib/nokogiri/html/sax/parser_context.rb +0 -16
- data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
- data/patches/libxml2/0002-Fix-nullptr-deref-with-XPath-logic-ops.patch +0 -54
- data/patches/libxml2/0003-Fix-infinite-loop-in-LZMA-decompression.patch +0 -50
- data/patches/sort-patches-by-date +0 -25
- data/ports/archives/libxml2-2.9.8.tar.gz +0 -0
- data/ports/archives/libxslt-1.1.32.tar.gz +0 -0
- data/suppressions/README.txt +0 -1
- data/suppressions/nokogiri_ruby-2.supp +0 -10
- data/tasks/test.rb +0 -100
- data/test/css/test_nthiness.rb +0 -226
- data/test/css/test_parser.rb +0 -386
- data/test/css/test_tokenizer.rb +0 -215
- data/test/css/test_xpath_visitor.rb +0 -96
- data/test/decorators/test_slop.rb +0 -23
- 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 -271
- data/test/html/sax/test_parser.rb +0 -168
- data/test/html/sax/test_parser_context.rb +0 -46
- data/test/html/sax/test_parser_text.rb +0 -163
- data/test/html/sax/test_push_parser.rb +0 -87
- data/test/html/test_attributes.rb +0 -85
- data/test/html/test_builder.rb +0 -164
- data/test/html/test_document.rb +0 -712
- data/test/html/test_document_encoding.rb +0 -143
- data/test/html/test_document_fragment.rb +0 -310
- 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 -91
- 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 -47
- 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 -402
- data/test/xml/sax/test_parser_context.rb +0 -115
- data/test/xml/sax/test_parser_text.rb +0 -202
- data/test/xml/sax/test_push_parser.rb +0 -265
- data/test/xml/test_attr.rb +0 -74
- 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 -54
- 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 -298
- 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 -262
- data/test/xml/test_namespace.rb +0 -96
- data/test/xml/test_node.rb +0 -1325
- data/test/xml/test_node_attributes.rb +0 -115
- data/test/xml/test_node_encoding.rb +0 -75
- data/test/xml/test_node_inheritance.rb +0 -32
- data/test/xml/test_node_reparenting.rb +0 -592
- data/test/xml/test_node_set.rb +0 -809
- 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 -620
- 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 -36
- data/test/xml/test_text.rb +0 -60
- data/test/xml/test_unparented_node.rb +0 -483
- data/test/xml/test_xinclude.rb +0 -83
- data/test/xml/test_xpath.rb +0 -470
- data/test/xslt/test_custom_functions.rb +0 -133
- data/test/xslt/test_exception_handling.rb +0 -37
@@ -1,151 +1,154 @@
|
|
1
|
+
# frozen_string_literal: true
|
1
2
|
#--
|
2
3
|
# DO NOT MODIFY!!!!
|
3
|
-
# This file is automatically generated by rex 1.0.
|
4
|
+
# This file is automatically generated by rex 1.0.7
|
4
5
|
# from lexical definition file "lib/nokogiri/css/tokenizer.rex".
|
5
6
|
#++
|
6
7
|
|
7
8
|
module Nokogiri
|
8
9
|
module CSS
|
9
|
-
|
10
|
-
|
10
|
+
# :nodoc: all
|
11
|
+
class Tokenizer
|
12
|
+
require 'strscan'
|
11
13
|
|
12
|
-
|
14
|
+
class ScanError < StandardError ; end
|
13
15
|
|
14
|
-
|
15
|
-
|
16
|
-
|
16
|
+
attr_reader :lineno
|
17
|
+
attr_reader :filename
|
18
|
+
attr_accessor :state
|
17
19
|
|
18
|
-
|
19
|
-
|
20
|
-
|
21
|
-
|
22
|
-
|
20
|
+
def scan_setup(str)
|
21
|
+
@ss = StringScanner.new(str)
|
22
|
+
@lineno = 1
|
23
|
+
@state = nil
|
24
|
+
end
|
23
25
|
|
24
|
-
|
25
|
-
|
26
|
-
|
26
|
+
def action
|
27
|
+
yield
|
28
|
+
end
|
27
29
|
|
28
|
-
|
29
|
-
|
30
|
-
|
31
|
-
|
32
|
-
|
30
|
+
def scan_str(str)
|
31
|
+
scan_setup(str)
|
32
|
+
do_parse
|
33
|
+
end
|
34
|
+
alias :scan :scan_str
|
33
35
|
|
34
|
-
|
35
|
-
|
36
|
-
|
37
|
-
|
38
|
-
|
39
|
-
|
36
|
+
def load_file( filename )
|
37
|
+
@filename = filename
|
38
|
+
File.open(filename, "r") do |f|
|
39
|
+
scan_setup(f.read)
|
40
|
+
end
|
41
|
+
end
|
40
42
|
|
41
|
-
|
42
|
-
|
43
|
-
|
44
|
-
|
43
|
+
def scan_file( filename )
|
44
|
+
load_file(filename)
|
45
|
+
do_parse
|
46
|
+
end
|
45
47
|
|
46
48
|
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
# skips empty actions
|
51
|
-
until token = _next_token or @ss.eos?; end
|
52
|
-
token
|
53
|
-
end
|
49
|
+
def next_token
|
50
|
+
return if @ss.eos?
|
54
51
|
|
55
|
-
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
when nil
|
60
|
-
case
|
61
|
-
when (text = @ss.scan(/has\([\s]*/))
|
62
|
-
action { [:HAS, text] }
|
52
|
+
# skips empty actions
|
53
|
+
until token = _next_token or @ss.eos?; end
|
54
|
+
token
|
55
|
+
end
|
63
56
|
|
64
|
-
|
65
|
-
|
57
|
+
def _next_token
|
58
|
+
text = @ss.peek(1)
|
59
|
+
@lineno += 1 if text == "\n"
|
60
|
+
token = case @state
|
61
|
+
when nil
|
62
|
+
case
|
63
|
+
when (text = @ss.scan(/has\([\s]*/))
|
64
|
+
action { [:HAS, text] }
|
66
65
|
|
67
|
-
|
68
|
-
|
66
|
+
when (text = @ss.scan(/-?([_A-Za-z]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*\([\s]*/))
|
67
|
+
action { [:FUNCTION, text] }
|
69
68
|
|
70
|
-
|
71
|
-
|
69
|
+
when (text = @ss.scan(/-?([_A-Za-z]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*/))
|
70
|
+
action { [:IDENT, text] }
|
72
71
|
|
73
|
-
|
74
|
-
|
72
|
+
when (text = @ss.scan(/\#([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])+/))
|
73
|
+
action { [:HASH, text] }
|
75
74
|
|
76
|
-
|
77
|
-
|
75
|
+
when (text = @ss.scan(/[\s]*~=[\s]*/))
|
76
|
+
action { [:INCLUDES, text] }
|
78
77
|
|
79
|
-
|
80
|
-
|
78
|
+
when (text = @ss.scan(/[\s]*\|=[\s]*/))
|
79
|
+
action { [:DASHMATCH, text] }
|
81
80
|
|
82
|
-
|
83
|
-
|
81
|
+
when (text = @ss.scan(/[\s]*\^=[\s]*/))
|
82
|
+
action { [:PREFIXMATCH, text] }
|
84
83
|
|
85
|
-
|
86
|
-
|
84
|
+
when (text = @ss.scan(/[\s]*\$=[\s]*/))
|
85
|
+
action { [:SUFFIXMATCH, text] }
|
87
86
|
|
88
|
-
|
89
|
-
|
87
|
+
when (text = @ss.scan(/[\s]*\*=[\s]*/))
|
88
|
+
action { [:SUBSTRINGMATCH, text] }
|
90
89
|
|
91
|
-
|
92
|
-
|
90
|
+
when (text = @ss.scan(/[\s]*!=[\s]*/))
|
91
|
+
action { [:NOT_EQUAL, text] }
|
93
92
|
|
94
|
-
|
95
|
-
|
93
|
+
when (text = @ss.scan(/[\s]*=[\s]*/))
|
94
|
+
action { [:EQUAL, text] }
|
96
95
|
|
97
|
-
|
98
|
-
|
96
|
+
when (text = @ss.scan(/[\s]*\)/))
|
97
|
+
action { [:RPAREN, text] }
|
99
98
|
|
100
|
-
|
101
|
-
|
99
|
+
when (text = @ss.scan(/\[[\s]*/))
|
100
|
+
action { [:LSQUARE, text] }
|
102
101
|
|
103
|
-
|
104
|
-
|
102
|
+
when (text = @ss.scan(/[\s]*\]/))
|
103
|
+
action { [:RSQUARE, text] }
|
105
104
|
|
106
|
-
|
107
|
-
|
105
|
+
when (text = @ss.scan(/[\s]*\+[\s]*/))
|
106
|
+
action { [:PLUS, text] }
|
108
107
|
|
109
|
-
|
110
|
-
|
108
|
+
when (text = @ss.scan(/[\s]*>[\s]*/))
|
109
|
+
action { [:GREATER, text] }
|
111
110
|
|
112
|
-
|
113
|
-
|
111
|
+
when (text = @ss.scan(/[\s]*,[\s]*/))
|
112
|
+
action { [:COMMA, text] }
|
114
113
|
|
115
|
-
|
116
|
-
|
114
|
+
when (text = @ss.scan(/[\s]*~[\s]*/))
|
115
|
+
action { [:TILDE, text] }
|
117
116
|
|
118
|
-
|
119
|
-
|
117
|
+
when (text = @ss.scan(/\:not\([\s]*/))
|
118
|
+
action { [:NOT, text] }
|
120
119
|
|
121
|
-
|
122
|
-
|
120
|
+
when (text = @ss.scan(/-?([0-9]+|[0-9]*\.[0-9]+)/))
|
121
|
+
action { [:NUMBER, text] }
|
123
122
|
|
124
|
-
|
125
|
-
|
123
|
+
when (text = @ss.scan(/[\s]*\/\/[\s]*/))
|
124
|
+
action { [:DOUBLESLASH, text] }
|
126
125
|
|
127
|
-
|
128
|
-
|
126
|
+
when (text = @ss.scan(/[\s]*\/[\s]*/))
|
127
|
+
action { [:SLASH, text] }
|
129
128
|
|
130
|
-
|
131
|
-
|
129
|
+
when (text = @ss.scan(/U\+[0-9a-f?]{1,6}(-[0-9a-f]{1,6})?/))
|
130
|
+
action {[:UNICODE_RANGE, text] }
|
132
131
|
|
133
|
-
|
134
|
-
|
132
|
+
when (text = @ss.scan(/[\s]+/))
|
133
|
+
action { [:S, text] }
|
135
134
|
|
136
|
-
|
137
|
-
|
135
|
+
when (text = @ss.scan(/"([^\n\r\f"]|\n|\r\n|\r|\f|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*(?<!\\)(?:\\{2})*"|'([^\n\r\f']|\n|\r\n|\r|\f|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*(?<!\\)(?:\\{2})*'/))
|
136
|
+
action { [:STRING, text] }
|
138
137
|
|
139
|
-
|
140
|
-
|
141
|
-
raise ScanError, "can not match: '" + text + "'"
|
142
|
-
end # if
|
138
|
+
when (text = @ss.scan(/./))
|
139
|
+
action { [text, text] }
|
143
140
|
|
144
|
-
|
145
|
-
|
146
|
-
|
147
|
-
|
148
|
-
|
141
|
+
|
142
|
+
else
|
143
|
+
text = @ss.string[@ss.pos .. -1]
|
144
|
+
raise ScanError, "can not match: '" + text + "'"
|
145
|
+
end # if
|
146
|
+
|
147
|
+
else
|
148
|
+
raise ScanError, "undefined state: '" + state.to_s + "'"
|
149
|
+
end # case state
|
150
|
+
token
|
151
|
+
end # def _next_token
|
149
152
|
|
150
153
|
end # class
|
151
154
|
end
|
@@ -1,6 +1,7 @@
|
|
1
1
|
module Nokogiri
|
2
2
|
module CSS
|
3
|
-
|
3
|
+
# :nodoc: all
|
4
|
+
class Tokenizer
|
4
5
|
|
5
6
|
macro
|
6
7
|
nl \n|\r\n|\r|\f
|
@@ -12,7 +13,7 @@ macro
|
|
12
13
|
escape {unicode}|\\[^\n\r\f0-9A-Fa-f]
|
13
14
|
nmchar [_A-Za-z0-9-]|{nonascii}|{escape}
|
14
15
|
nmstart [_A-Za-z]|{nonascii}|{escape}
|
15
|
-
ident
|
16
|
+
ident -?({nmstart})({nmchar})*
|
16
17
|
name ({nmchar})+
|
17
18
|
string1 "([^\n\r\f"]|{nl}|{nonascii}|{escape})*(?<!\\)(?:\\{2})*"
|
18
19
|
string2 '([^\n\r\f']|{nl}|{nonascii}|{escape})*(?<!\\)(?:\\{2})*'
|