Nokogiri_precompiled_aarch64_dedshit 1.14.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +44 -0
- data/LICENSE-DEPENDENCIES.md +2224 -0
- data/LICENSE.md +9 -0
- data/README.md +287 -0
- data/bin/nokogiri +131 -0
- data/dependencies.yml +41 -0
- data/ext/java/nokogiri/Html4Document.java +157 -0
- data/ext/java/nokogiri/Html4ElementDescription.java +133 -0
- data/ext/java/nokogiri/Html4EntityLookup.java +63 -0
- data/ext/java/nokogiri/Html4SaxParserContext.java +289 -0
- data/ext/java/nokogiri/Html4SaxPushParser.java +213 -0
- data/ext/java/nokogiri/NokogiriService.java +613 -0
- data/ext/java/nokogiri/XmlAttr.java +154 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +119 -0
- data/ext/java/nokogiri/XmlCdata.java +60 -0
- data/ext/java/nokogiri/XmlComment.java +77 -0
- data/ext/java/nokogiri/XmlDocument.java +705 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +163 -0
- data/ext/java/nokogiri/XmlDtd.java +516 -0
- data/ext/java/nokogiri/XmlElement.java +44 -0
- data/ext/java/nokogiri/XmlElementContent.java +412 -0
- data/ext/java/nokogiri/XmlElementDecl.java +148 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +151 -0
- data/ext/java/nokogiri/XmlEntityReference.java +79 -0
- data/ext/java/nokogiri/XmlNamespace.java +193 -0
- data/ext/java/nokogiri/XmlNode.java +1938 -0
- data/ext/java/nokogiri/XmlNodeSet.java +463 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +79 -0
- data/ext/java/nokogiri/XmlReader.java +615 -0
- data/ext/java/nokogiri/XmlRelaxng.java +133 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +329 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +288 -0
- data/ext/java/nokogiri/XmlSchema.java +423 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +137 -0
- data/ext/java/nokogiri/XmlText.java +90 -0
- data/ext/java/nokogiri/XmlXpathContext.java +305 -0
- data/ext/java/nokogiri/XsltStylesheet.java +368 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +13 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +27 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +178 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +99 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +140 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +65 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +339 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +817 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +228 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +110 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +86 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +107 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +62 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +165 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +50 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +37 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +70 -0
- data/ext/java/nokogiri/internals/ParserContext.java +262 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +564 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +865 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +50 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +174 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +11 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +265 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +40 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +122 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +178 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +106 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +278 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +664 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +45 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +45 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +388 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +308 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +47 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +51 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +51 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +50 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +660 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +194 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +77 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +45 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +325 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +106 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +86 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +181 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +87 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +452 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +52 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +190 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +540 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1712 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +737 -0
- data/ext/nokogiri/depend +38 -0
- data/ext/nokogiri/extconf.rb +1086 -0
- data/ext/nokogiri/gumbo.c +594 -0
- data/ext/nokogiri/html4_document.c +167 -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 +116 -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 +265 -0
- data/ext/nokogiri/nokogiri.h +235 -0
- data/ext/nokogiri/test_global_handlers.c +42 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_cdata.c +57 -0
- data/ext/nokogiri/xml_comment.c +62 -0
- data/ext/nokogiri/xml_document.c +689 -0
- data/ext/nokogiri/xml_document_fragment.c +44 -0
- data/ext/nokogiri/xml_dtd.c +210 -0
- data/ext/nokogiri/xml_element_content.c +128 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_encoding_handler.c +104 -0
- data/ext/nokogiri/xml_entity_decl.c +112 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -0
- data/ext/nokogiri/xml_namespace.c +186 -0
- data/ext/nokogiri/xml_node.c +2426 -0
- data/ext/nokogiri/xml_node_set.c +496 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_reader.c +794 -0
- data/ext/nokogiri/xml_relax_ng.c +164 -0
- data/ext/nokogiri/xml_sax_parser.c +316 -0
- data/ext/nokogiri/xml_sax_parser_context.c +283 -0
- data/ext/nokogiri/xml_sax_push_parser.c +166 -0
- data/ext/nokogiri/xml_schema.c +260 -0
- data/ext/nokogiri/xml_syntax_error.c +85 -0
- data/ext/nokogiri/xml_text.c +48 -0
- data/ext/nokogiri/xml_xpath_context.c +415 -0
- data/ext/nokogiri/xslt_stylesheet.c +363 -0
- data/gumbo-parser/CHANGES.md +63 -0
- data/gumbo-parser/Makefile +111 -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/insertion_mode.h +33 -0
- data/gumbo-parser/src/macros.h +91 -0
- data/gumbo-parser/src/nokogiri_gumbo.h +944 -0
- data/gumbo-parser/src/parser.c +4878 -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 +223 -0
- data/gumbo-parser/src/tag_lookup.c +382 -0
- data/gumbo-parser/src/tag_lookup.gperf +170 -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 +66 -0
- data/gumbo-parser/src/util.h +34 -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 +54 -0
- data/lib/nokogiri/css/parser.rb +770 -0
- data/lib/nokogiri/css/parser.y +277 -0
- data/lib/nokogiri/css/parser_extras.rb +96 -0
- data/lib/nokogiri/css/syntax_error.rb +9 -0
- data/lib/nokogiri/css/tokenizer.rb +155 -0
- data/lib/nokogiri/css/tokenizer.rex +56 -0
- data/lib/nokogiri/css/xpath_visitor.rb +359 -0
- data/lib/nokogiri/css.rb +66 -0
- data/lib/nokogiri/decorators/slop.rb +44 -0
- data/lib/nokogiri/encoding_handler.rb +57 -0
- data/lib/nokogiri/extension.rb +32 -0
- data/lib/nokogiri/gumbo.rb +15 -0
- data/lib/nokogiri/html.rb +48 -0
- data/lib/nokogiri/html4/builder.rb +37 -0
- data/lib/nokogiri/html4/document.rb +214 -0
- data/lib/nokogiri/html4/document_fragment.rb +54 -0
- data/lib/nokogiri/html4/element_description.rb +25 -0
- data/lib/nokogiri/html4/element_description_defaults.rb +572 -0
- data/lib/nokogiri/html4/encoding_reader.rb +121 -0
- data/lib/nokogiri/html4/entity_lookup.rb +15 -0
- data/lib/nokogiri/html4/sax/parser.rb +63 -0
- data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
- data/lib/nokogiri/html4/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html4.rb +47 -0
- data/lib/nokogiri/html5/document.rb +168 -0
- data/lib/nokogiri/html5/document_fragment.rb +90 -0
- data/lib/nokogiri/html5/node.rb +98 -0
- data/lib/nokogiri/html5.rb +389 -0
- data/lib/nokogiri/jruby/dependencies.rb +3 -0
- data/lib/nokogiri/jruby/isorelax/isorelax/20030108/isorelax-20030108.jar +0 -0
- data/lib/nokogiri/jruby/net/sf/saxon/Saxon-HE/9.6.0-4/Saxon-HE-9.6.0-4.jar +0 -0
- data/lib/nokogiri/jruby/net/sourceforge/htmlunit/neko-htmlunit/2.63.0/neko-htmlunit-2.63.0.jar +0 -0
- data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
- data/lib/nokogiri/jruby/nu/validator/jing/20200702VNU/jing-20200702VNU.jar +0 -0
- data/lib/nokogiri/jruby/org/nokogiri/nekodtd/0.1.11.noko2/nekodtd-0.1.11.noko2.jar +0 -0
- data/lib/nokogiri/jruby/xalan/serializer/2.7.3/serializer-2.7.3.jar +0 -0
- data/lib/nokogiri/jruby/xalan/xalan/2.7.3/xalan-2.7.3.jar +0 -0
- data/lib/nokogiri/jruby/xerces/xercesImpl/2.12.2/xercesImpl-2.12.2.jar +0 -0
- data/lib/nokogiri/jruby/xml-apis/xml-apis/1.4.01/xml-apis-1.4.01.jar +0 -0
- data/lib/nokogiri/syntax_error.rb +6 -0
- data/lib/nokogiri/version/constant.rb +6 -0
- data/lib/nokogiri/version/info.rb +223 -0
- data/lib/nokogiri/version.rb +4 -0
- data/lib/nokogiri/xml/attr.rb +66 -0
- data/lib/nokogiri/xml/attribute_decl.rb +20 -0
- data/lib/nokogiri/xml/builder.rb +487 -0
- data/lib/nokogiri/xml/cdata.rb +13 -0
- data/lib/nokogiri/xml/character_data.rb +9 -0
- data/lib/nokogiri/xml/document.rb +471 -0
- data/lib/nokogiri/xml/document_fragment.rb +205 -0
- data/lib/nokogiri/xml/dtd.rb +34 -0
- data/lib/nokogiri/xml/element_content.rb +38 -0
- data/lib/nokogiri/xml/element_decl.rb +15 -0
- data/lib/nokogiri/xml/entity_decl.rb +21 -0
- data/lib/nokogiri/xml/entity_reference.rb +20 -0
- data/lib/nokogiri/xml/namespace.rb +58 -0
- data/lib/nokogiri/xml/node/save_options.rb +68 -0
- data/lib/nokogiri/xml/node.rb +1563 -0
- data/lib/nokogiri/xml/node_set.rb +447 -0
- data/lib/nokogiri/xml/notation.rb +19 -0
- data/lib/nokogiri/xml/parse_options.rb +213 -0
- data/lib/nokogiri/xml/pp/character_data.rb +21 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +4 -0
- data/lib/nokogiri/xml/processing_instruction.rb +11 -0
- data/lib/nokogiri/xml/reader.rb +105 -0
- data/lib/nokogiri/xml/relax_ng.rb +38 -0
- data/lib/nokogiri/xml/sax/document.rb +167 -0
- data/lib/nokogiri/xml/sax/parser.rb +125 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +21 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +6 -0
- data/lib/nokogiri/xml/schema.rb +73 -0
- data/lib/nokogiri/xml/searchable.rb +270 -0
- data/lib/nokogiri/xml/syntax_error.rb +72 -0
- data/lib/nokogiri/xml/text.rb +11 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +13 -0
- data/lib/nokogiri/xml/xpath.rb +21 -0
- data/lib/nokogiri/xml/xpath_context.rb +16 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +27 -0
- data/lib/nokogiri/xslt.rb +65 -0
- data/lib/nokogiri.rb +120 -0
- data/lib/xsd/xmlparser/nokogiri.rb +106 -0
- metadata +391 -0
@@ -0,0 +1,167 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtml4Document ;
|
4
|
+
|
5
|
+
static ID id_encoding_found;
|
6
|
+
static ID id_to_s;
|
7
|
+
|
8
|
+
void xmlSetStructuredErrorFunc(void *rb_error_list, void *Nokogiri_error_array_pusher);
|
9
|
+
/*
|
10
|
+
* call-seq:
|
11
|
+
* new
|
12
|
+
*
|
13
|
+
* Create a new document
|
14
|
+
*/
|
15
|
+
static VALUE
|
16
|
+
rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
|
17
|
+
{
|
18
|
+
VALUE uri, external_id, rest, rb_doc;
|
19
|
+
htmlDocPtr doc;
|
20
|
+
|
21
|
+
rb_scan_args(argc, argv, "0*", &rest);
|
22
|
+
uri = rb_ary_entry(rest, (long)0);
|
23
|
+
external_id = rb_ary_entry(rest, (long)1);
|
24
|
+
|
25
|
+
doc = htmlNewDoc(
|
26
|
+
RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
|
27
|
+
RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
|
28
|
+
);
|
29
|
+
rb_doc = noko_xml_document_wrap_with_init_args(klass, doc, argc, argv);
|
30
|
+
return rb_doc ;
|
31
|
+
}
|
32
|
+
|
33
|
+
/*
|
34
|
+
* call-seq:
|
35
|
+
* read_io(io, url, encoding, options)
|
36
|
+
*
|
37
|
+
* Read the HTML document from +io+ with given +url+, +encoding+,
|
38
|
+
* and +options+. See Nokogiri::HTML4.parse
|
39
|
+
*/
|
40
|
+
static VALUE
|
41
|
+
rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
|
42
|
+
{
|
43
|
+
VALUE rb_doc;
|
44
|
+
VALUE rb_error_list = rb_ary_new();
|
45
|
+
htmlDocPtr c_doc;
|
46
|
+
const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
|
47
|
+
const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
|
48
|
+
int options = NUM2INT(rb_options);
|
49
|
+
|
50
|
+
xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
|
51
|
+
|
52
|
+
c_doc = htmlReadIO(noko_io_read, noko_io_close, (void *)rb_io, c_url, c_encoding, options);
|
53
|
+
|
54
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
55
|
+
|
56
|
+
/*
|
57
|
+
* If EncodingFound has occurred in EncodingReader, make sure to do
|
58
|
+
* a cleanup and propagate the error.
|
59
|
+
*/
|
60
|
+
if (rb_respond_to(rb_io, id_encoding_found)) {
|
61
|
+
VALUE encoding_found = rb_funcall(rb_io, id_encoding_found, 0);
|
62
|
+
if (!NIL_P(encoding_found)) {
|
63
|
+
xmlFreeDoc(c_doc);
|
64
|
+
rb_exc_raise(encoding_found);
|
65
|
+
}
|
66
|
+
}
|
67
|
+
|
68
|
+
if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
|
69
|
+
VALUE rb_error ;
|
70
|
+
|
71
|
+
xmlFreeDoc(c_doc);
|
72
|
+
|
73
|
+
rb_error = rb_ary_entry(rb_error_list, 0);
|
74
|
+
if (rb_error == Qnil) {
|
75
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
76
|
+
} else {
|
77
|
+
VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
|
78
|
+
exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
|
79
|
+
exception_message);
|
80
|
+
rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
|
81
|
+
}
|
82
|
+
|
83
|
+
return Qnil;
|
84
|
+
}
|
85
|
+
|
86
|
+
rb_doc = noko_xml_document_wrap(klass, c_doc);
|
87
|
+
rb_iv_set(rb_doc, "@errors", rb_error_list);
|
88
|
+
return rb_doc;
|
89
|
+
}
|
90
|
+
|
91
|
+
/*
|
92
|
+
* call-seq:
|
93
|
+
* read_memory(string, url, encoding, options)
|
94
|
+
*
|
95
|
+
* Read the HTML document contained in +string+ with given +url+, +encoding+,
|
96
|
+
* and +options+. See Nokogiri::HTML4.parse
|
97
|
+
*/
|
98
|
+
static VALUE
|
99
|
+
rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
|
100
|
+
{
|
101
|
+
VALUE rb_doc;
|
102
|
+
VALUE rb_error_list = rb_ary_new();
|
103
|
+
htmlDocPtr c_doc;
|
104
|
+
const char *c_buffer = StringValuePtr(rb_html);
|
105
|
+
const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
|
106
|
+
const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
|
107
|
+
int html_len = (int)RSTRING_LEN(rb_html);
|
108
|
+
int options = NUM2INT(rb_options);
|
109
|
+
|
110
|
+
xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
|
111
|
+
|
112
|
+
c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
|
113
|
+
|
114
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
115
|
+
|
116
|
+
if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
|
117
|
+
VALUE rb_error ;
|
118
|
+
|
119
|
+
xmlFreeDoc(c_doc);
|
120
|
+
|
121
|
+
rb_error = rb_ary_entry(rb_error_list, 0);
|
122
|
+
if (rb_error == Qnil) {
|
123
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
124
|
+
} else {
|
125
|
+
VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
|
126
|
+
exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
|
127
|
+
exception_message);
|
128
|
+
rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
|
129
|
+
}
|
130
|
+
|
131
|
+
return Qnil;
|
132
|
+
}
|
133
|
+
|
134
|
+
rb_doc = noko_xml_document_wrap(klass, c_doc);
|
135
|
+
rb_iv_set(rb_doc, "@errors", rb_error_list);
|
136
|
+
return rb_doc;
|
137
|
+
}
|
138
|
+
|
139
|
+
/*
|
140
|
+
* call-seq:
|
141
|
+
* type
|
142
|
+
*
|
143
|
+
* The type for this document
|
144
|
+
*/
|
145
|
+
static VALUE
|
146
|
+
rb_html_document_type(VALUE self)
|
147
|
+
{
|
148
|
+
htmlDocPtr doc;
|
149
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
150
|
+
return INT2NUM(doc->type);
|
151
|
+
}
|
152
|
+
|
153
|
+
void
|
154
|
+
noko_init_html_document(void)
|
155
|
+
{
|
156
|
+
assert(cNokogiriXmlDocument);
|
157
|
+
cNokogiriHtml4Document = rb_define_class_under(mNokogiriHtml4, "Document", cNokogiriXmlDocument);
|
158
|
+
|
159
|
+
rb_define_singleton_method(cNokogiriHtml4Document, "read_memory", rb_html_document_s_read_memory, 4);
|
160
|
+
rb_define_singleton_method(cNokogiriHtml4Document, "read_io", rb_html_document_s_read_io, 4);
|
161
|
+
rb_define_singleton_method(cNokogiriHtml4Document, "new", rb_html_document_s_new, -1);
|
162
|
+
|
163
|
+
rb_define_method(cNokogiriHtml4Document, "type", rb_html_document_type, 0);
|
164
|
+
|
165
|
+
id_encoding_found = rb_intern("encoding_found");
|
166
|
+
id_to_s = rb_intern("to_s");
|
167
|
+
}
|
@@ -0,0 +1,294 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtml4ElementDescription ;
|
4
|
+
|
5
|
+
/*
|
6
|
+
* call-seq:
|
7
|
+
* required_attributes
|
8
|
+
*
|
9
|
+
* A list of required attributes for this element
|
10
|
+
*/
|
11
|
+
static VALUE
|
12
|
+
required_attributes(VALUE self)
|
13
|
+
{
|
14
|
+
const htmlElemDesc *description;
|
15
|
+
VALUE list;
|
16
|
+
int i;
|
17
|
+
|
18
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
19
|
+
|
20
|
+
list = rb_ary_new();
|
21
|
+
|
22
|
+
if (NULL == description->attrs_req) { return list; }
|
23
|
+
|
24
|
+
for (i = 0; description->attrs_depr[i]; i++) {
|
25
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
|
26
|
+
}
|
27
|
+
|
28
|
+
return list;
|
29
|
+
}
|
30
|
+
|
31
|
+
/*
|
32
|
+
* call-seq:
|
33
|
+
* deprecated_attributes
|
34
|
+
*
|
35
|
+
* A list of deprecated attributes for this element
|
36
|
+
*/
|
37
|
+
static VALUE
|
38
|
+
deprecated_attributes(VALUE self)
|
39
|
+
{
|
40
|
+
const htmlElemDesc *description;
|
41
|
+
VALUE list;
|
42
|
+
int i;
|
43
|
+
|
44
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
45
|
+
|
46
|
+
list = rb_ary_new();
|
47
|
+
|
48
|
+
if (NULL == description->attrs_depr) { return list; }
|
49
|
+
|
50
|
+
for (i = 0; description->attrs_depr[i]; i++) {
|
51
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
|
52
|
+
}
|
53
|
+
|
54
|
+
return list;
|
55
|
+
}
|
56
|
+
|
57
|
+
/*
|
58
|
+
* call-seq:
|
59
|
+
* optional_attributes
|
60
|
+
*
|
61
|
+
* A list of optional attributes for this element
|
62
|
+
*/
|
63
|
+
static VALUE
|
64
|
+
optional_attributes(VALUE self)
|
65
|
+
{
|
66
|
+
const htmlElemDesc *description;
|
67
|
+
VALUE list;
|
68
|
+
int i;
|
69
|
+
|
70
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
71
|
+
|
72
|
+
list = rb_ary_new();
|
73
|
+
|
74
|
+
if (NULL == description->attrs_opt) { return list; }
|
75
|
+
|
76
|
+
for (i = 0; description->attrs_opt[i]; i++) {
|
77
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
|
78
|
+
}
|
79
|
+
|
80
|
+
return list;
|
81
|
+
}
|
82
|
+
|
83
|
+
/*
|
84
|
+
* call-seq:
|
85
|
+
* default_sub_element
|
86
|
+
*
|
87
|
+
* The default sub element for this element
|
88
|
+
*/
|
89
|
+
static VALUE
|
90
|
+
default_sub_element(VALUE self)
|
91
|
+
{
|
92
|
+
const htmlElemDesc *description;
|
93
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
94
|
+
|
95
|
+
if (description->defaultsubelt) {
|
96
|
+
return NOKOGIRI_STR_NEW2(description->defaultsubelt);
|
97
|
+
}
|
98
|
+
|
99
|
+
return Qnil;
|
100
|
+
}
|
101
|
+
|
102
|
+
/*
|
103
|
+
* call-seq:
|
104
|
+
* sub_elements
|
105
|
+
*
|
106
|
+
* A list of allowed sub elements for this element.
|
107
|
+
*/
|
108
|
+
static VALUE
|
109
|
+
sub_elements(VALUE self)
|
110
|
+
{
|
111
|
+
const htmlElemDesc *description;
|
112
|
+
VALUE list;
|
113
|
+
int i;
|
114
|
+
|
115
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
116
|
+
|
117
|
+
list = rb_ary_new();
|
118
|
+
|
119
|
+
if (NULL == description->subelts) { return list; }
|
120
|
+
|
121
|
+
for (i = 0; description->subelts[i]; i++) {
|
122
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
|
123
|
+
}
|
124
|
+
|
125
|
+
return list;
|
126
|
+
}
|
127
|
+
|
128
|
+
/*
|
129
|
+
* call-seq:
|
130
|
+
* description
|
131
|
+
*
|
132
|
+
* The description for this element
|
133
|
+
*/
|
134
|
+
static VALUE
|
135
|
+
description(VALUE self)
|
136
|
+
{
|
137
|
+
const htmlElemDesc *description;
|
138
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
139
|
+
|
140
|
+
return NOKOGIRI_STR_NEW2(description->desc);
|
141
|
+
}
|
142
|
+
|
143
|
+
/*
|
144
|
+
* call-seq:
|
145
|
+
* inline?
|
146
|
+
*
|
147
|
+
* Is this element an inline element?
|
148
|
+
*/
|
149
|
+
static VALUE
|
150
|
+
inline_eh(VALUE self)
|
151
|
+
{
|
152
|
+
const htmlElemDesc *description;
|
153
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
154
|
+
|
155
|
+
if (description->isinline) { return Qtrue; }
|
156
|
+
return Qfalse;
|
157
|
+
}
|
158
|
+
|
159
|
+
/*
|
160
|
+
* call-seq:
|
161
|
+
* deprecated?
|
162
|
+
*
|
163
|
+
* Is this element deprecated?
|
164
|
+
*/
|
165
|
+
static VALUE
|
166
|
+
deprecated_eh(VALUE self)
|
167
|
+
{
|
168
|
+
const htmlElemDesc *description;
|
169
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
170
|
+
|
171
|
+
if (description->depr) { return Qtrue; }
|
172
|
+
return Qfalse;
|
173
|
+
}
|
174
|
+
|
175
|
+
/*
|
176
|
+
* call-seq:
|
177
|
+
* empty?
|
178
|
+
*
|
179
|
+
* Is this an empty element?
|
180
|
+
*/
|
181
|
+
static VALUE
|
182
|
+
empty_eh(VALUE self)
|
183
|
+
{
|
184
|
+
const htmlElemDesc *description;
|
185
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
186
|
+
|
187
|
+
if (description->empty) { return Qtrue; }
|
188
|
+
return Qfalse;
|
189
|
+
}
|
190
|
+
|
191
|
+
/*
|
192
|
+
* call-seq:
|
193
|
+
* save_end_tag?
|
194
|
+
*
|
195
|
+
* Should the end tag be saved?
|
196
|
+
*/
|
197
|
+
static VALUE
|
198
|
+
save_end_tag_eh(VALUE self)
|
199
|
+
{
|
200
|
+
const htmlElemDesc *description;
|
201
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
202
|
+
|
203
|
+
if (description->saveEndTag) { return Qtrue; }
|
204
|
+
return Qfalse;
|
205
|
+
}
|
206
|
+
|
207
|
+
/*
|
208
|
+
* call-seq:
|
209
|
+
* implied_end_tag?
|
210
|
+
*
|
211
|
+
* Can the end tag be implied for this tag?
|
212
|
+
*/
|
213
|
+
static VALUE
|
214
|
+
implied_end_tag_eh(VALUE self)
|
215
|
+
{
|
216
|
+
const htmlElemDesc *description;
|
217
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
218
|
+
|
219
|
+
if (description->endTag) { return Qtrue; }
|
220
|
+
return Qfalse;
|
221
|
+
}
|
222
|
+
|
223
|
+
/*
|
224
|
+
* call-seq:
|
225
|
+
* implied_start_tag?
|
226
|
+
*
|
227
|
+
* Can the start tag be implied for this tag?
|
228
|
+
*/
|
229
|
+
static VALUE
|
230
|
+
implied_start_tag_eh(VALUE self)
|
231
|
+
{
|
232
|
+
const htmlElemDesc *description;
|
233
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
234
|
+
|
235
|
+
if (description->startTag) { return Qtrue; }
|
236
|
+
return Qfalse;
|
237
|
+
}
|
238
|
+
|
239
|
+
/*
|
240
|
+
* call-seq:
|
241
|
+
* name
|
242
|
+
*
|
243
|
+
* Get the tag name for this ElemementDescription
|
244
|
+
*/
|
245
|
+
static VALUE
|
246
|
+
name(VALUE self)
|
247
|
+
{
|
248
|
+
const htmlElemDesc *description;
|
249
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
250
|
+
|
251
|
+
if (NULL == description->name) { return Qnil; }
|
252
|
+
return NOKOGIRI_STR_NEW2(description->name);
|
253
|
+
}
|
254
|
+
|
255
|
+
/*
|
256
|
+
* call-seq:
|
257
|
+
* [](tag_name)
|
258
|
+
*
|
259
|
+
* Get ElemementDescription for +tag_name+
|
260
|
+
*/
|
261
|
+
static VALUE
|
262
|
+
get_description(VALUE klass, VALUE tag_name)
|
263
|
+
{
|
264
|
+
const htmlElemDesc *description = htmlTagLookup(
|
265
|
+
(const xmlChar *)StringValueCStr(tag_name)
|
266
|
+
);
|
267
|
+
|
268
|
+
if (NULL == description) { return Qnil; }
|
269
|
+
return Data_Wrap_Struct(klass, 0, 0, DISCARD_CONST_QUAL(void *, description));
|
270
|
+
}
|
271
|
+
|
272
|
+
void
|
273
|
+
noko_init_html_element_description(void)
|
274
|
+
{
|
275
|
+
cNokogiriHtml4ElementDescription = rb_define_class_under(mNokogiriHtml4, "ElementDescription", rb_cObject);
|
276
|
+
|
277
|
+
rb_undef_alloc_func(cNokogiriHtml4ElementDescription);
|
278
|
+
|
279
|
+
rb_define_singleton_method(cNokogiriHtml4ElementDescription, "[]", get_description, 1);
|
280
|
+
|
281
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "name", name, 0);
|
282
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
|
283
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
|
284
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "save_end_tag?", save_end_tag_eh, 0);
|
285
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "empty?", empty_eh, 0);
|
286
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "deprecated?", deprecated_eh, 0);
|
287
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "inline?", inline_eh, 0);
|
288
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "description", description, 0);
|
289
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "sub_elements", sub_elements, 0);
|
290
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "default_sub_element", default_sub_element, 0);
|
291
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "optional_attributes", optional_attributes, 0);
|
292
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "deprecated_attributes", deprecated_attributes, 0);
|
293
|
+
rb_define_method(cNokogiriHtml4ElementDescription, "required_attributes", required_attributes, 0);
|
294
|
+
}
|
@@ -0,0 +1,37 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
static VALUE cNokogiriHtml4EntityLookup;
|
4
|
+
|
5
|
+
/*
|
6
|
+
* call-seq:
|
7
|
+
* get(key)
|
8
|
+
*
|
9
|
+
* Get the HTML4::EntityDescription for +key+
|
10
|
+
*/
|
11
|
+
static VALUE
|
12
|
+
get(VALUE _, VALUE rb_entity_name)
|
13
|
+
{
|
14
|
+
VALUE cNokogiriHtml4EntityDescription;
|
15
|
+
const htmlEntityDesc *c_entity_desc;
|
16
|
+
VALUE rb_constructor_args[3];
|
17
|
+
|
18
|
+
c_entity_desc = htmlEntityLookup((const xmlChar *)StringValueCStr(rb_entity_name));
|
19
|
+
if (NULL == c_entity_desc) {
|
20
|
+
return Qnil;
|
21
|
+
}
|
22
|
+
|
23
|
+
rb_constructor_args[0] = UINT2NUM(c_entity_desc->value);
|
24
|
+
rb_constructor_args[1] = NOKOGIRI_STR_NEW2(c_entity_desc->name);
|
25
|
+
rb_constructor_args[2] = NOKOGIRI_STR_NEW2(c_entity_desc->desc);
|
26
|
+
|
27
|
+
cNokogiriHtml4EntityDescription = rb_const_get_at(mNokogiriHtml4, rb_intern("EntityDescription"));
|
28
|
+
return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtml4EntityDescription);
|
29
|
+
}
|
30
|
+
|
31
|
+
void
|
32
|
+
noko_init_html_entity_lookup(void)
|
33
|
+
{
|
34
|
+
cNokogiriHtml4EntityLookup = rb_define_class_under(mNokogiriHtml4, "EntityLookup", rb_cObject);
|
35
|
+
|
36
|
+
rb_define_method(cNokogiriHtml4EntityLookup, "get", get, 1);
|
37
|
+
}
|
@@ -0,0 +1,116 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtml4SaxParserContext ;
|
4
|
+
|
5
|
+
void xmlSetStructuredErrorFunc(void *aa, void*bb);
|
6
|
+
|
7
|
+
static void
|
8
|
+
deallocate(xmlParserCtxtPtr ctxt)
|
9
|
+
{
|
10
|
+
ctxt->sax = NULL;
|
11
|
+
htmlFreeParserCtxt(ctxt);
|
12
|
+
}
|
13
|
+
|
14
|
+
static VALUE
|
15
|
+
parse_memory(VALUE klass, VALUE data, VALUE encoding)
|
16
|
+
{
|
17
|
+
htmlParserCtxtPtr ctxt;
|
18
|
+
|
19
|
+
Check_Type(data, T_STRING);
|
20
|
+
|
21
|
+
if (!(int)RSTRING_LEN(data)) {
|
22
|
+
rb_raise(rb_eRuntimeError, "data cannot be empty");
|
23
|
+
}
|
24
|
+
|
25
|
+
ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
|
26
|
+
(int)RSTRING_LEN(data));
|
27
|
+
if (ctxt->sax) {
|
28
|
+
xmlFree(ctxt->sax);
|
29
|
+
ctxt->sax = NULL;
|
30
|
+
}
|
31
|
+
|
32
|
+
if (RTEST(encoding)) {
|
33
|
+
xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
|
34
|
+
if (enc != NULL) {
|
35
|
+
xmlSwitchToEncoding(ctxt, enc);
|
36
|
+
if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
|
37
|
+
rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
|
38
|
+
StringValueCStr(encoding));
|
39
|
+
}
|
40
|
+
}
|
41
|
+
}
|
42
|
+
|
43
|
+
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
44
|
+
}
|
45
|
+
|
46
|
+
static VALUE
|
47
|
+
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
|
+
|
73
|
+
NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
|
74
|
+
return Qnil;
|
75
|
+
}
|
76
|
+
|
77
|
+
static VALUE
|
78
|
+
parse_with(VALUE self, VALUE sax_handler)
|
79
|
+
{
|
80
|
+
htmlParserCtxtPtr ctxt;
|
81
|
+
htmlSAXHandlerPtr sax;
|
82
|
+
|
83
|
+
if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser)) {
|
84
|
+
rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
|
85
|
+
}
|
86
|
+
|
87
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
88
|
+
Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
|
89
|
+
|
90
|
+
/* Free the sax handler since we'll assign our own */
|
91
|
+
if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
|
92
|
+
xmlFree(ctxt->sax);
|
93
|
+
}
|
94
|
+
|
95
|
+
ctxt->sax = sax;
|
96
|
+
ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
|
97
|
+
|
98
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
99
|
+
|
100
|
+
rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
|
101
|
+
|
102
|
+
return self;
|
103
|
+
}
|
104
|
+
|
105
|
+
void
|
106
|
+
noko_init_html_sax_parser_context(void)
|
107
|
+
{
|
108
|
+
assert(cNokogiriXmlSaxParserContext);
|
109
|
+
cNokogiriHtml4SaxParserContext = rb_define_class_under(mNokogiriHtml4Sax, "ParserContext",
|
110
|
+
cNokogiriXmlSaxParserContext);
|
111
|
+
|
112
|
+
rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "memory", parse_memory, 2);
|
113
|
+
rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "file", parse_file, 2);
|
114
|
+
|
115
|
+
rb_define_method(cNokogiriHtml4SaxParserContext, "parse_with", parse_with, 1);
|
116
|
+
}
|
@@ -0,0 +1,95 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtml4SaxPushParser;
|
4
|
+
|
5
|
+
/*
|
6
|
+
* call-seq:
|
7
|
+
* native_write(chunk, last_chunk)
|
8
|
+
*
|
9
|
+
* Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
|
10
|
+
*/
|
11
|
+
static VALUE
|
12
|
+
native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
|
13
|
+
{
|
14
|
+
xmlParserCtxtPtr ctx;
|
15
|
+
const char *chunk = NULL;
|
16
|
+
int size = 0;
|
17
|
+
int status = 0;
|
18
|
+
libxmlStructuredErrorHandlerState handler_state;
|
19
|
+
|
20
|
+
Data_Get_Struct(self, xmlParserCtxt, ctx);
|
21
|
+
|
22
|
+
if (Qnil != _chunk) {
|
23
|
+
chunk = StringValuePtr(_chunk);
|
24
|
+
size = (int)RSTRING_LEN(_chunk);
|
25
|
+
}
|
26
|
+
|
27
|
+
Nokogiri_structured_error_func_save_and_set(&handler_state, NULL, NULL);
|
28
|
+
|
29
|
+
status = htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0);
|
30
|
+
|
31
|
+
Nokogiri_structured_error_func_restore(&handler_state);
|
32
|
+
|
33
|
+
if ((status != 0) && !(ctx->options & XML_PARSE_RECOVER)) {
|
34
|
+
// TODO: there appear to be no tests for this block
|
35
|
+
xmlErrorPtr e = xmlCtxtGetLastError(ctx);
|
36
|
+
Nokogiri_error_raise(NULL, e);
|
37
|
+
}
|
38
|
+
|
39
|
+
return self;
|
40
|
+
}
|
41
|
+
|
42
|
+
/*
|
43
|
+
* call-seq:
|
44
|
+
* initialize_native(xml_sax, filename)
|
45
|
+
*
|
46
|
+
* Initialize the push parser with +xml_sax+ using +filename+
|
47
|
+
*/
|
48
|
+
static VALUE
|
49
|
+
initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
|
50
|
+
VALUE encoding)
|
51
|
+
{
|
52
|
+
htmlSAXHandlerPtr sax;
|
53
|
+
const char *filename = NULL;
|
54
|
+
htmlParserCtxtPtr ctx;
|
55
|
+
xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
|
56
|
+
|
57
|
+
Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
|
58
|
+
|
59
|
+
if (_filename != Qnil) { filename = StringValueCStr(_filename); }
|
60
|
+
|
61
|
+
if (!NIL_P(encoding)) {
|
62
|
+
enc = xmlParseCharEncoding(StringValueCStr(encoding));
|
63
|
+
if (enc == XML_CHAR_ENCODING_ERROR) {
|
64
|
+
rb_raise(rb_eArgError, "Unsupported Encoding");
|
65
|
+
}
|
66
|
+
}
|
67
|
+
|
68
|
+
ctx = htmlCreatePushParserCtxt(
|
69
|
+
sax,
|
70
|
+
NULL,
|
71
|
+
NULL,
|
72
|
+
0,
|
73
|
+
filename,
|
74
|
+
enc
|
75
|
+
);
|
76
|
+
if (ctx == NULL) {
|
77
|
+
rb_raise(rb_eRuntimeError, "Could not create a parser context");
|
78
|
+
}
|
79
|
+
|
80
|
+
ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
|
81
|
+
|
82
|
+
ctx->sax2 = 1;
|
83
|
+
DATA_PTR(self) = ctx;
|
84
|
+
return self;
|
85
|
+
}
|
86
|
+
|
87
|
+
void
|
88
|
+
noko_init_html_sax_push_parser(void)
|
89
|
+
{
|
90
|
+
assert(cNokogiriXmlSaxPushParser);
|
91
|
+
cNokogiriHtml4SaxPushParser = rb_define_class_under(mNokogiriHtml4Sax, "PushParser", cNokogiriXmlSaxPushParser);
|
92
|
+
|
93
|
+
rb_define_private_method(cNokogiriHtml4SaxPushParser, "initialize_native", initialize_native, 3);
|
94
|
+
rb_define_private_method(cNokogiriHtml4SaxPushParser, "native_write", native_write, 2);
|
95
|
+
}
|