nokogiri-backport 1.11.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +3 -0
- data/LICENSE-DEPENDENCIES.md +1682 -0
- data/LICENSE.md +9 -0
- data/README.md +272 -0
- data/bin/nokogiri +118 -0
- data/dependencies.yml +74 -0
- data/ext/java/nokogiri/EncodingHandler.java +124 -0
- data/ext/java/nokogiri/HtmlDocument.java +178 -0
- data/ext/java/nokogiri/HtmlElementDescription.java +148 -0
- data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
- data/ext/java/nokogiri/HtmlSaxParserContext.java +282 -0
- data/ext/java/nokogiri/HtmlSaxPushParser.java +222 -0
- data/ext/java/nokogiri/NokogiriService.java +597 -0
- data/ext/java/nokogiri/XmlAttr.java +162 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +129 -0
- data/ext/java/nokogiri/XmlCdata.java +82 -0
- data/ext/java/nokogiri/XmlComment.java +97 -0
- data/ext/java/nokogiri/XmlDocument.java +633 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +185 -0
- data/ext/java/nokogiri/XmlDtd.java +481 -0
- data/ext/java/nokogiri/XmlElement.java +68 -0
- data/ext/java/nokogiri/XmlElementContent.java +382 -0
- data/ext/java/nokogiri/XmlElementDecl.java +147 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +157 -0
- data/ext/java/nokogiri/XmlEntityReference.java +101 -0
- data/ext/java/nokogiri/XmlNamespace.java +199 -0
- data/ext/java/nokogiri/XmlNode.java +1684 -0
- data/ext/java/nokogiri/XmlNodeSet.java +434 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +100 -0
- data/ext/java/nokogiri/XmlReader.java +531 -0
- data/ext/java/nokogiri/XmlRelaxng.java +151 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +374 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +286 -0
- data/ext/java/nokogiri/XmlSchema.java +388 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +138 -0
- data/ext/java/nokogiri/XmlText.java +110 -0
- data/ext/java/nokogiri/XmlXpathContext.java +301 -0
- data/ext/java/nokogiri/XsltStylesheet.java +347 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +10 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +20 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +151 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +116 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +69 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +327 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +734 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +217 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +127 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +100 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +180 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +72 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +60 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
- data/ext/java/nokogiri/internals/ParserContext.java +259 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +488 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +778 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +73 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +168 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +274 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +119 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +159 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +37 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +93 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +252 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +639 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +367 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +295 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +40 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +630 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +173 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +76 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +42 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +293 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +93 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +79 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +166 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +76 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +402 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +51 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +179 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +507 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1745 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +685 -0
- data/ext/nokogiri/depend +477 -0
- data/ext/nokogiri/extconf.rb +836 -0
- data/ext/nokogiri/html_document.c +171 -0
- data/ext/nokogiri/html_document.h +10 -0
- 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 +135 -0
- data/ext/nokogiri/nokogiri.h +130 -0
- data/ext/nokogiri/xml_attr.c +103 -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 +62 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +69 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +622 -0
- data/ext/nokogiri/xml_document.h +23 -0
- data/ext/nokogiri/xml_document_fragment.c +48 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +202 -0
- data/ext/nokogiri/xml_dtd.h +10 -0
- 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 +63 -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 +111 -0
- data/ext/nokogiri/xml_namespace.h +14 -0
- data/ext/nokogiri/xml_node.c +1773 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +486 -0
- data/ext/nokogiri/xml_node_set.h +12 -0
- data/ext/nokogiri/xml_processing_instruction.c +56 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +657 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +179 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +305 -0
- data/ext/nokogiri/xml_sax_parser.h +39 -0
- 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 +159 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +276 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +64 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +52 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath_context.c +374 -0
- data/ext/nokogiri/xml_xpath_context.h +10 -0
- data/ext/nokogiri/xslt_stylesheet.c +263 -0
- data/ext/nokogiri/xslt_stylesheet.h +14 -0
- data/lib/isorelax.jar +0 -0
- data/lib/jing.jar +0 -0
- data/lib/nekodtd.jar +0 -0
- data/lib/nekohtml.jar +0 -0
- data/lib/nokogiri/css/node.rb +53 -0
- data/lib/nokogiri/css/parser.rb +751 -0
- data/lib/nokogiri/css/parser.y +272 -0
- data/lib/nokogiri/css/parser_extras.rb +94 -0
- data/lib/nokogiri/css/syntax_error.rb +8 -0
- data/lib/nokogiri/css/tokenizer.rb +154 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +260 -0
- data/lib/nokogiri/css.rb +28 -0
- data/lib/nokogiri/decorators/slop.rb +43 -0
- data/lib/nokogiri/html/builder.rb +36 -0
- data/lib/nokogiri/html/document.rb +322 -0
- data/lib/nokogiri/html/document_fragment.rb +50 -0
- data/lib/nokogiri/html/element_description.rb +24 -0
- data/lib/nokogiri/html/element_description_defaults.rb +672 -0
- data/lib/nokogiri/html/entity_lookup.rb +14 -0
- data/lib/nokogiri/html/sax/parser.rb +63 -0
- data/lib/nokogiri/html/sax/parser_context.rb +17 -0
- data/lib/nokogiri/html/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html.rb +38 -0
- data/lib/nokogiri/jruby/dependencies.rb +20 -0
- data/lib/nokogiri/syntax_error.rb +5 -0
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +182 -0
- data/lib/nokogiri/version.rb +3 -0
- data/lib/nokogiri/xml/attr.rb +15 -0
- data/lib/nokogiri/xml/attribute_decl.rb +19 -0
- data/lib/nokogiri/xml/builder.rb +447 -0
- data/lib/nokogiri/xml/cdata.rb +12 -0
- data/lib/nokogiri/xml/character_data.rb +8 -0
- data/lib/nokogiri/xml/document.rb +290 -0
- data/lib/nokogiri/xml/document_fragment.rb +159 -0
- data/lib/nokogiri/xml/dtd.rb +33 -0
- data/lib/nokogiri/xml/element_content.rb +37 -0
- data/lib/nokogiri/xml/element_decl.rb +14 -0
- data/lib/nokogiri/xml/entity_decl.rb +20 -0
- data/lib/nokogiri/xml/entity_reference.rb +19 -0
- data/lib/nokogiri/xml/namespace.rb +14 -0
- data/lib/nokogiri/xml/node/save_options.rb +62 -0
- data/lib/nokogiri/xml/node.rb +1240 -0
- data/lib/nokogiri/xml/node_set.rb +372 -0
- data/lib/nokogiri/xml/notation.rb +7 -0
- data/lib/nokogiri/xml/parse_options.rb +127 -0
- data/lib/nokogiri/xml/pp/character_data.rb +19 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +3 -0
- data/lib/nokogiri/xml/processing_instruction.rb +9 -0
- data/lib/nokogiri/xml/reader.rb +116 -0
- data/lib/nokogiri/xml/relax_ng.rb +37 -0
- data/lib/nokogiri/xml/sax/document.rb +172 -0
- data/lib/nokogiri/xml/sax/parser.rb +123 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +17 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +5 -0
- data/lib/nokogiri/xml/schema.rb +72 -0
- data/lib/nokogiri/xml/searchable.rb +239 -0
- data/lib/nokogiri/xml/syntax_error.rb +71 -0
- data/lib/nokogiri/xml/text.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +12 -0
- data/lib/nokogiri/xml/xpath.rb +11 -0
- data/lib/nokogiri/xml/xpath_context.rb +17 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +26 -0
- data/lib/nokogiri/xslt.rb +57 -0
- data/lib/nokogiri.rb +144 -0
- data/lib/serializer.jar +0 -0
- data/lib/xalan.jar +0 -0
- data/lib/xercesImpl.jar +0 -0
- data/lib/xml-apis.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +103 -0
- metadata +531 -0
@@ -0,0 +1,171 @@
|
|
1
|
+
#include <html_document.h>
|
2
|
+
|
3
|
+
static ID id_encoding_found;
|
4
|
+
static ID id_to_s;
|
5
|
+
|
6
|
+
/*
|
7
|
+
* call-seq:
|
8
|
+
* new
|
9
|
+
*
|
10
|
+
* Create a new document
|
11
|
+
*/
|
12
|
+
static VALUE
|
13
|
+
rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
|
14
|
+
{
|
15
|
+
VALUE uri, external_id, rest, rb_doc;
|
16
|
+
htmlDocPtr doc;
|
17
|
+
|
18
|
+
rb_scan_args(argc, argv, "0*", &rest);
|
19
|
+
uri = rb_ary_entry(rest, (long)0);
|
20
|
+
external_id = rb_ary_entry(rest, (long)1);
|
21
|
+
|
22
|
+
doc = htmlNewDoc(
|
23
|
+
RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
|
24
|
+
RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
|
25
|
+
);
|
26
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, doc);
|
27
|
+
rb_obj_call_init(rb_doc, argc, argv);
|
28
|
+
return rb_doc ;
|
29
|
+
}
|
30
|
+
|
31
|
+
/*
|
32
|
+
* call-seq:
|
33
|
+
* read_io(io, url, encoding, options)
|
34
|
+
*
|
35
|
+
* Read the HTML document from +io+ with given +url+, +encoding+,
|
36
|
+
* and +options+. See Nokogiri::HTML.parse
|
37
|
+
*/
|
38
|
+
static VALUE
|
39
|
+
rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
|
40
|
+
{
|
41
|
+
VALUE rb_doc;
|
42
|
+
VALUE rb_error_list = rb_ary_new();
|
43
|
+
htmlDocPtr c_doc;
|
44
|
+
const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
|
45
|
+
const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
|
46
|
+
int options = NUM2INT(rb_options);
|
47
|
+
|
48
|
+
xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
|
49
|
+
|
50
|
+
c_doc = htmlReadIO(io_read_callback, io_close_callback, (void *)rb_io, c_url, c_encoding, options);
|
51
|
+
|
52
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
53
|
+
|
54
|
+
/*
|
55
|
+
* If EncodingFound has occurred in EncodingReader, make sure to do
|
56
|
+
* a cleanup and propagate the error.
|
57
|
+
*/
|
58
|
+
if (rb_respond_to(rb_io, id_encoding_found)) {
|
59
|
+
VALUE encoding_found = rb_funcall(rb_io, id_encoding_found, 0);
|
60
|
+
if (!NIL_P(encoding_found)) {
|
61
|
+
xmlFreeDoc(c_doc);
|
62
|
+
rb_exc_raise(encoding_found);
|
63
|
+
}
|
64
|
+
}
|
65
|
+
|
66
|
+
if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
|
67
|
+
VALUE rb_error ;
|
68
|
+
|
69
|
+
xmlFreeDoc(c_doc);
|
70
|
+
|
71
|
+
rb_error = rb_ary_entry(rb_error_list, 0);
|
72
|
+
if (rb_error == Qnil) {
|
73
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
74
|
+
} else {
|
75
|
+
VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
|
76
|
+
exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
|
77
|
+
exception_message);
|
78
|
+
rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
|
79
|
+
}
|
80
|
+
|
81
|
+
return Qnil;
|
82
|
+
}
|
83
|
+
|
84
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
|
85
|
+
rb_iv_set(rb_doc, "@errors", rb_error_list);
|
86
|
+
return rb_doc;
|
87
|
+
}
|
88
|
+
|
89
|
+
/*
|
90
|
+
* call-seq:
|
91
|
+
* read_memory(string, url, encoding, options)
|
92
|
+
*
|
93
|
+
* Read the HTML document contained in +string+ with given +url+, +encoding+,
|
94
|
+
* and +options+. See Nokogiri::HTML.parse
|
95
|
+
*/
|
96
|
+
static VALUE
|
97
|
+
rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
|
98
|
+
{
|
99
|
+
VALUE rb_doc;
|
100
|
+
VALUE rb_error_list = rb_ary_new();
|
101
|
+
htmlDocPtr c_doc;
|
102
|
+
const char *c_buffer = StringValuePtr(rb_html);
|
103
|
+
const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
|
104
|
+
const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
|
105
|
+
int html_len = (int)RSTRING_LEN(rb_html);
|
106
|
+
int options = NUM2INT(rb_options);
|
107
|
+
|
108
|
+
xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
|
109
|
+
|
110
|
+
c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
|
111
|
+
|
112
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
113
|
+
|
114
|
+
if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
|
115
|
+
VALUE rb_error ;
|
116
|
+
|
117
|
+
xmlFreeDoc(c_doc);
|
118
|
+
|
119
|
+
rb_error = rb_ary_entry(rb_error_list, 0);
|
120
|
+
if (rb_error == Qnil) {
|
121
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
122
|
+
} else {
|
123
|
+
VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
|
124
|
+
exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
|
125
|
+
exception_message);
|
126
|
+
rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
|
127
|
+
}
|
128
|
+
|
129
|
+
return Qnil;
|
130
|
+
}
|
131
|
+
|
132
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
|
133
|
+
rb_iv_set(rb_doc, "@errors", rb_error_list);
|
134
|
+
return rb_doc;
|
135
|
+
}
|
136
|
+
|
137
|
+
/*
|
138
|
+
* call-seq:
|
139
|
+
* type
|
140
|
+
*
|
141
|
+
* The type for this document
|
142
|
+
*/
|
143
|
+
static VALUE
|
144
|
+
rb_html_document_type(VALUE self)
|
145
|
+
{
|
146
|
+
htmlDocPtr doc;
|
147
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
148
|
+
return INT2NUM((long)doc->type);
|
149
|
+
}
|
150
|
+
|
151
|
+
VALUE cNokogiriHtmlDocument ;
|
152
|
+
|
153
|
+
void
|
154
|
+
init_html_document()
|
155
|
+
{
|
156
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
157
|
+
VALUE nokogiri_xml = rb_define_module_under(nokogiri, "XML");
|
158
|
+
VALUE nokogiri_xml_node = rb_define_class_under(nokogiri_xml, "Node", rb_cObject);
|
159
|
+
VALUE nokogiri_xml_document = rb_define_class_under(nokogiri_xml, "Document", nokogiri_xml_node);
|
160
|
+
VALUE nokogiri_html = rb_define_module_under(nokogiri, "HTML");
|
161
|
+
cNokogiriHtmlDocument = rb_define_class_under(nokogiri_html, "Document", nokogiri_xml_document);
|
162
|
+
|
163
|
+
rb_define_singleton_method(cNokogiriHtmlDocument, "read_memory", rb_html_document_s_read_memory, 4);
|
164
|
+
rb_define_singleton_method(cNokogiriHtmlDocument, "read_io", rb_html_document_s_read_io, 4);
|
165
|
+
rb_define_singleton_method(cNokogiriHtmlDocument, "new", rb_html_document_s_new, -1);
|
166
|
+
|
167
|
+
rb_define_method(cNokogiriHtmlDocument, "type", rb_html_document_type, 0);
|
168
|
+
|
169
|
+
id_encoding_found = rb_intern("encoding_found");
|
170
|
+
id_to_s = rb_intern("to_s");
|
171
|
+
}
|
@@ -0,0 +1,279 @@
|
|
1
|
+
#include <html_element_description.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* required_attributes
|
6
|
+
*
|
7
|
+
* A list of required attributes for this element
|
8
|
+
*/
|
9
|
+
static VALUE required_attributes(VALUE self)
|
10
|
+
{
|
11
|
+
const htmlElemDesc * description;
|
12
|
+
VALUE list;
|
13
|
+
int i;
|
14
|
+
|
15
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
16
|
+
|
17
|
+
list = rb_ary_new();
|
18
|
+
|
19
|
+
if(NULL == description->attrs_req) return list;
|
20
|
+
|
21
|
+
for(i = 0; description->attrs_depr[i]; i++) {
|
22
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
|
23
|
+
}
|
24
|
+
|
25
|
+
return list;
|
26
|
+
}
|
27
|
+
|
28
|
+
/*
|
29
|
+
* call-seq:
|
30
|
+
* deprecated_attributes
|
31
|
+
*
|
32
|
+
* A list of deprecated attributes for this element
|
33
|
+
*/
|
34
|
+
static VALUE deprecated_attributes(VALUE self)
|
35
|
+
{
|
36
|
+
const htmlElemDesc * description;
|
37
|
+
VALUE list;
|
38
|
+
int i;
|
39
|
+
|
40
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
41
|
+
|
42
|
+
list = rb_ary_new();
|
43
|
+
|
44
|
+
if(NULL == description->attrs_depr) return list;
|
45
|
+
|
46
|
+
for(i = 0; description->attrs_depr[i]; i++) {
|
47
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
|
48
|
+
}
|
49
|
+
|
50
|
+
return list;
|
51
|
+
}
|
52
|
+
|
53
|
+
/*
|
54
|
+
* call-seq:
|
55
|
+
* optional_attributes
|
56
|
+
*
|
57
|
+
* A list of optional attributes for this element
|
58
|
+
*/
|
59
|
+
static VALUE optional_attributes(VALUE self)
|
60
|
+
{
|
61
|
+
const htmlElemDesc * description;
|
62
|
+
VALUE list;
|
63
|
+
int i;
|
64
|
+
|
65
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
66
|
+
|
67
|
+
list = rb_ary_new();
|
68
|
+
|
69
|
+
if(NULL == description->attrs_opt) return list;
|
70
|
+
|
71
|
+
for(i = 0; description->attrs_opt[i]; i++) {
|
72
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
|
73
|
+
}
|
74
|
+
|
75
|
+
return list;
|
76
|
+
}
|
77
|
+
|
78
|
+
/*
|
79
|
+
* call-seq:
|
80
|
+
* default_sub_element
|
81
|
+
*
|
82
|
+
* The default sub element for this element
|
83
|
+
*/
|
84
|
+
static VALUE default_sub_element(VALUE self)
|
85
|
+
{
|
86
|
+
const htmlElemDesc * description;
|
87
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
88
|
+
|
89
|
+
if (description->defaultsubelt)
|
90
|
+
return NOKOGIRI_STR_NEW2(description->defaultsubelt);
|
91
|
+
|
92
|
+
return Qnil;
|
93
|
+
}
|
94
|
+
|
95
|
+
/*
|
96
|
+
* call-seq:
|
97
|
+
* sub_elements
|
98
|
+
*
|
99
|
+
* A list of allowed sub elements for this element.
|
100
|
+
*/
|
101
|
+
static VALUE sub_elements(VALUE self)
|
102
|
+
{
|
103
|
+
const htmlElemDesc * description;
|
104
|
+
VALUE list;
|
105
|
+
int i;
|
106
|
+
|
107
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
108
|
+
|
109
|
+
list = rb_ary_new();
|
110
|
+
|
111
|
+
if(NULL == description->subelts) return list;
|
112
|
+
|
113
|
+
for(i = 0; description->subelts[i]; i++) {
|
114
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
|
115
|
+
}
|
116
|
+
|
117
|
+
return list;
|
118
|
+
}
|
119
|
+
|
120
|
+
/*
|
121
|
+
* call-seq:
|
122
|
+
* description
|
123
|
+
*
|
124
|
+
* The description for this element
|
125
|
+
*/
|
126
|
+
static VALUE description(VALUE self)
|
127
|
+
{
|
128
|
+
const htmlElemDesc * description;
|
129
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
130
|
+
|
131
|
+
return NOKOGIRI_STR_NEW2(description->desc);
|
132
|
+
}
|
133
|
+
|
134
|
+
/*
|
135
|
+
* call-seq:
|
136
|
+
* inline?
|
137
|
+
*
|
138
|
+
* Is this element an inline element?
|
139
|
+
*/
|
140
|
+
static VALUE inline_eh(VALUE self)
|
141
|
+
{
|
142
|
+
const htmlElemDesc * description;
|
143
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
144
|
+
|
145
|
+
if(description->isinline) return Qtrue;
|
146
|
+
return Qfalse;
|
147
|
+
}
|
148
|
+
|
149
|
+
/*
|
150
|
+
* call-seq:
|
151
|
+
* deprecated?
|
152
|
+
*
|
153
|
+
* Is this element deprecated?
|
154
|
+
*/
|
155
|
+
static VALUE deprecated_eh(VALUE self)
|
156
|
+
{
|
157
|
+
const htmlElemDesc * description;
|
158
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
159
|
+
|
160
|
+
if(description->depr) return Qtrue;
|
161
|
+
return Qfalse;
|
162
|
+
}
|
163
|
+
|
164
|
+
/*
|
165
|
+
* call-seq:
|
166
|
+
* empty?
|
167
|
+
*
|
168
|
+
* Is this an empty element?
|
169
|
+
*/
|
170
|
+
static VALUE empty_eh(VALUE self)
|
171
|
+
{
|
172
|
+
const htmlElemDesc * description;
|
173
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
174
|
+
|
175
|
+
if(description->empty) return Qtrue;
|
176
|
+
return Qfalse;
|
177
|
+
}
|
178
|
+
|
179
|
+
/*
|
180
|
+
* call-seq:
|
181
|
+
* save_end_tag?
|
182
|
+
*
|
183
|
+
* Should the end tag be saved?
|
184
|
+
*/
|
185
|
+
static VALUE save_end_tag_eh(VALUE self)
|
186
|
+
{
|
187
|
+
const htmlElemDesc * description;
|
188
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
189
|
+
|
190
|
+
if(description->saveEndTag) return Qtrue;
|
191
|
+
return Qfalse;
|
192
|
+
}
|
193
|
+
|
194
|
+
/*
|
195
|
+
* call-seq:
|
196
|
+
* implied_end_tag?
|
197
|
+
*
|
198
|
+
* Can the end tag be implied for this tag?
|
199
|
+
*/
|
200
|
+
static VALUE implied_end_tag_eh(VALUE self)
|
201
|
+
{
|
202
|
+
const htmlElemDesc * description;
|
203
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
204
|
+
|
205
|
+
if(description->endTag) return Qtrue;
|
206
|
+
return Qfalse;
|
207
|
+
}
|
208
|
+
|
209
|
+
/*
|
210
|
+
* call-seq:
|
211
|
+
* implied_start_tag?
|
212
|
+
*
|
213
|
+
* Can the start tag be implied for this tag?
|
214
|
+
*/
|
215
|
+
static VALUE implied_start_tag_eh(VALUE self)
|
216
|
+
{
|
217
|
+
const htmlElemDesc * description;
|
218
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
219
|
+
|
220
|
+
if(description->startTag) return Qtrue;
|
221
|
+
return Qfalse;
|
222
|
+
}
|
223
|
+
|
224
|
+
/*
|
225
|
+
* call-seq:
|
226
|
+
* name
|
227
|
+
*
|
228
|
+
* Get the tag name for this ElemementDescription
|
229
|
+
*/
|
230
|
+
static VALUE name(VALUE self)
|
231
|
+
{
|
232
|
+
const htmlElemDesc * description;
|
233
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
234
|
+
|
235
|
+
if(NULL == description->name) return Qnil;
|
236
|
+
return NOKOGIRI_STR_NEW2(description->name);
|
237
|
+
}
|
238
|
+
|
239
|
+
/*
|
240
|
+
* call-seq:
|
241
|
+
* [](tag_name)
|
242
|
+
*
|
243
|
+
* Get ElemementDescription for +tag_name+
|
244
|
+
*/
|
245
|
+
static VALUE get_description(VALUE klass, VALUE tag_name)
|
246
|
+
{
|
247
|
+
const htmlElemDesc * description = htmlTagLookup(
|
248
|
+
(const xmlChar *)StringValueCStr(tag_name)
|
249
|
+
);
|
250
|
+
|
251
|
+
if(NULL == description) return Qnil;
|
252
|
+
return Data_Wrap_Struct(klass, 0, 0, (void *)(uintptr_t)description);
|
253
|
+
}
|
254
|
+
|
255
|
+
VALUE cNokogiriHtmlElementDescription ;
|
256
|
+
void init_html_element_description()
|
257
|
+
{
|
258
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
259
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
260
|
+
VALUE klass = rb_define_class_under(html, "ElementDescription",rb_cObject);
|
261
|
+
|
262
|
+
cNokogiriHtmlElementDescription = klass;
|
263
|
+
|
264
|
+
rb_define_singleton_method(klass, "[]", get_description, 1);
|
265
|
+
|
266
|
+
rb_define_method(klass, "name", name, 0);
|
267
|
+
rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
|
268
|
+
rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
|
269
|
+
rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
|
270
|
+
rb_define_method(klass, "empty?", empty_eh, 0);
|
271
|
+
rb_define_method(klass, "deprecated?", deprecated_eh, 0);
|
272
|
+
rb_define_method(klass, "inline?", inline_eh, 0);
|
273
|
+
rb_define_method(klass, "description", description, 0);
|
274
|
+
rb_define_method(klass, "sub_elements", sub_elements, 0);
|
275
|
+
rb_define_method(klass, "default_sub_element", default_sub_element, 0);
|
276
|
+
rb_define_method(klass, "optional_attributes", optional_attributes, 0);
|
277
|
+
rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
|
278
|
+
rb_define_method(klass, "required_attributes", required_attributes, 0);
|
279
|
+
}
|
@@ -0,0 +1,32 @@
|
|
1
|
+
#include <html_entity_lookup.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* get(key)
|
6
|
+
*
|
7
|
+
* Get the HTML::EntityDescription for +key+
|
8
|
+
*/
|
9
|
+
static VALUE get(VALUE self, VALUE key)
|
10
|
+
{
|
11
|
+
const htmlEntityDesc * desc =
|
12
|
+
htmlEntityLookup((const xmlChar *)StringValueCStr(key));
|
13
|
+
VALUE klass, args[3];
|
14
|
+
|
15
|
+
if(NULL == desc) return Qnil;
|
16
|
+
klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
|
17
|
+
|
18
|
+
args[0] = INT2NUM((long)desc->value);
|
19
|
+
args[1] = NOKOGIRI_STR_NEW2(desc->name);
|
20
|
+
args[2] = NOKOGIRI_STR_NEW2(desc->desc);
|
21
|
+
|
22
|
+
return rb_class_new_instance(3, args, klass);
|
23
|
+
}
|
24
|
+
|
25
|
+
void init_html_entity_lookup()
|
26
|
+
{
|
27
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
28
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
29
|
+
VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
|
30
|
+
|
31
|
+
rb_define_method(klass, "get", get, 1);
|
32
|
+
}
|
@@ -0,0 +1,116 @@
|
|
1
|
+
#include <html_sax_parser_context.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtmlSaxParserContext ;
|
4
|
+
|
5
|
+
static void deallocate(xmlParserCtxtPtr ctxt)
|
6
|
+
{
|
7
|
+
NOKOGIRI_DEBUG_START(ctxt);
|
8
|
+
|
9
|
+
ctxt->sax = NULL;
|
10
|
+
|
11
|
+
htmlFreeParserCtxt(ctxt);
|
12
|
+
|
13
|
+
NOKOGIRI_DEBUG_END(ctxt);
|
14
|
+
}
|
15
|
+
|
16
|
+
static VALUE
|
17
|
+
parse_memory(VALUE klass, VALUE data, VALUE encoding)
|
18
|
+
{
|
19
|
+
htmlParserCtxtPtr ctxt;
|
20
|
+
|
21
|
+
if (NIL_P(data))
|
22
|
+
rb_raise(rb_eArgError, "data cannot be nil");
|
23
|
+
if (!(int)RSTRING_LEN(data))
|
24
|
+
rb_raise(rb_eRuntimeError, "data cannot be empty");
|
25
|
+
|
26
|
+
ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
|
27
|
+
(int)RSTRING_LEN(data));
|
28
|
+
if (ctxt->sax) {
|
29
|
+
xmlFree(ctxt->sax);
|
30
|
+
ctxt->sax = NULL;
|
31
|
+
}
|
32
|
+
|
33
|
+
if (RTEST(encoding)) {
|
34
|
+
xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
|
35
|
+
if (enc != NULL) {
|
36
|
+
xmlSwitchToEncoding(ctxt, enc);
|
37
|
+
if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
|
38
|
+
rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
|
39
|
+
StringValueCStr(encoding));
|
40
|
+
}
|
41
|
+
}
|
42
|
+
}
|
43
|
+
|
44
|
+
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
45
|
+
}
|
46
|
+
|
47
|
+
static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
|
48
|
+
{
|
49
|
+
htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
|
50
|
+
StringValueCStr(filename),
|
51
|
+
StringValueCStr(encoding)
|
52
|
+
);
|
53
|
+
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
54
|
+
}
|
55
|
+
|
56
|
+
static VALUE
|
57
|
+
parse_doc(VALUE ctxt_val)
|
58
|
+
{
|
59
|
+
htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
|
60
|
+
htmlParseDocument(ctxt);
|
61
|
+
return Qnil;
|
62
|
+
}
|
63
|
+
|
64
|
+
static VALUE
|
65
|
+
parse_doc_finalize(VALUE ctxt_val)
|
66
|
+
{
|
67
|
+
htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
|
68
|
+
|
69
|
+
if (ctxt->myDoc)
|
70
|
+
xmlFreeDoc(ctxt->myDoc);
|
71
|
+
|
72
|
+
NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
|
73
|
+
return Qnil;
|
74
|
+
}
|
75
|
+
|
76
|
+
static VALUE
|
77
|
+
parse_with(VALUE self, VALUE sax_handler)
|
78
|
+
{
|
79
|
+
htmlParserCtxtPtr ctxt;
|
80
|
+
htmlSAXHandlerPtr sax;
|
81
|
+
|
82
|
+
if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
|
83
|
+
rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
|
84
|
+
|
85
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
86
|
+
Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
|
87
|
+
|
88
|
+
/* Free the sax handler since we'll assign our own */
|
89
|
+
if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
|
90
|
+
xmlFree(ctxt->sax);
|
91
|
+
|
92
|
+
ctxt->sax = sax;
|
93
|
+
ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
|
94
|
+
|
95
|
+
rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
|
96
|
+
|
97
|
+
return self;
|
98
|
+
}
|
99
|
+
|
100
|
+
void init_html_sax_parser_context()
|
101
|
+
{
|
102
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
103
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
104
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
105
|
+
VALUE sax = rb_define_module_under(xml, "SAX");
|
106
|
+
VALUE hsax = rb_define_module_under(html, "SAX");
|
107
|
+
VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
|
108
|
+
VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
|
109
|
+
|
110
|
+
cNokogiriHtmlSaxParserContext = klass;
|
111
|
+
|
112
|
+
rb_define_singleton_method(klass, "memory", parse_memory, 2);
|
113
|
+
rb_define_singleton_method(klass, "file", parse_file, 2);
|
114
|
+
|
115
|
+
rb_define_method(klass, "parse_with", parse_with, 1);
|
116
|
+
}
|