rubyjedi-nokogiri_java 1.4.0.20100513161003-java
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- data/.autotest +26 -0
- data/CHANGELOG.ja.rdoc +330 -0
- data/CHANGELOG.rdoc +341 -0
- data/Manifest.txt +277 -0
- data/README.ja.rdoc +105 -0
- data/README.rdoc +125 -0
- data/Rakefile +307 -0
- data/bin/nokogiri +49 -0
- data/deps.rip +5 -0
- data/ext/nokogiri/extconf.rb +149 -0
- data/ext/nokogiri/html_document.c +145 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +272 -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 +92 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/nokogiri.c +96 -0
- data/ext/nokogiri/nokogiri.h +148 -0
- data/ext/nokogiri/xml_attr.c +92 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +67 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +54 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +52 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +386 -0
- data/ext/nokogiri/xml_document.h +24 -0
- data/ext/nokogiri/xml_document_fragment.c +46 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +192 -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 +97 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -0
- data/ext/nokogiri/xml_entity_reference.h +9 -0
- data/ext/nokogiri/xml_io.c +31 -0
- data/ext/nokogiri/xml_io.h +11 -0
- data/ext/nokogiri/xml_namespace.c +82 -0
- data/ext/nokogiri/xml_namespace.h +13 -0
- data/ext/nokogiri/xml_node.c +1080 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +405 -0
- data/ext/nokogiri/xml_node_set.h +9 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +593 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +159 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +283 -0
- data/ext/nokogiri/xml_sax_parser.h +43 -0
- data/ext/nokogiri/xml_sax_parser_context.c +157 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +114 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +156 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +52 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +48 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath.c +53 -0
- data/ext/nokogiri/xml_xpath.h +11 -0
- data/ext/nokogiri/xml_xpath_context.c +239 -0
- data/ext/nokogiri/xml_xpath_context.h +9 -0
- data/ext/nokogiri/xslt_stylesheet.c +131 -0
- data/ext/nokogiri/xslt_stylesheet.h +9 -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.rb +123 -0
- data/lib/nokogiri/css.rb +25 -0
- data/lib/nokogiri/css/generated_parser.rb +659 -0
- data/lib/nokogiri/css/generated_tokenizer.rb +145 -0
- data/lib/nokogiri/css/node.rb +99 -0
- data/lib/nokogiri/css/parser.rb +82 -0
- data/lib/nokogiri/css/parser.y +230 -0
- data/lib/nokogiri/css/syntax_error.rb +7 -0
- data/lib/nokogiri/css/tokenizer.rb +7 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +164 -0
- data/lib/nokogiri/decorators/slop.rb +33 -0
- data/lib/nokogiri/ffi/encoding_handler.rb +42 -0
- data/lib/nokogiri/ffi/html/document.rb +28 -0
- data/lib/nokogiri/ffi/html/element_description.rb +81 -0
- data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
- data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
- data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
- data/lib/nokogiri/ffi/libxml.rb +372 -0
- data/lib/nokogiri/ffi/structs/common_node.rb +26 -0
- data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
- data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
- data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
- data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
- data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
- data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
- data/lib/nokogiri/ffi/structs/xml_char_encoding_handler.rb +11 -0
- data/lib/nokogiri/ffi/structs/xml_document.rb +108 -0
- data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
- data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
- data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
- data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
- data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
- data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
- data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
- data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
- data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
- data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
- data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
- data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
- data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +124 -0
- data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
- data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
- data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
- data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +37 -0
- data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
- data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
- data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
- data/lib/nokogiri/ffi/xml/attr.rb +41 -0
- data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
- data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
- data/lib/nokogiri/ffi/xml/comment.rb +18 -0
- data/lib/nokogiri/ffi/xml/document.rb +135 -0
- data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
- data/lib/nokogiri/ffi/xml/dtd.rb +67 -0
- data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
- data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
- data/lib/nokogiri/ffi/xml/entity_decl.rb +27 -0
- data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
- data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
- data/lib/nokogiri/ffi/xml/node.rb +465 -0
- data/lib/nokogiri/ffi/xml/node_set.rb +146 -0
- data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
- data/lib/nokogiri/ffi/xml/reader.rb +227 -0
- data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
- data/lib/nokogiri/ffi/xml/sax/parser.rb +135 -0
- data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
- data/lib/nokogiri/ffi/xml/sax/push_parser.rb +55 -0
- data/lib/nokogiri/ffi/xml/schema.rb +92 -0
- data/lib/nokogiri/ffi/xml/syntax_error.rb +98 -0
- data/lib/nokogiri/ffi/xml/text.rb +18 -0
- data/lib/nokogiri/ffi/xml/xpath.rb +19 -0
- data/lib/nokogiri/ffi/xml/xpath_context.rb +135 -0
- data/lib/nokogiri/ffi/xslt/stylesheet.rb +50 -0
- data/lib/nokogiri/html.rb +36 -0
- data/lib/nokogiri/html/builder.rb +35 -0
- data/lib/nokogiri/html/document.rb +88 -0
- data/lib/nokogiri/html/document_fragment.rb +15 -0
- data/lib/nokogiri/html/element_description.rb +23 -0
- data/lib/nokogiri/html/element_description_defaults.rb +671 -0
- data/lib/nokogiri/html/entity_lookup.rb +13 -0
- data/lib/nokogiri/html/sax/parser.rb +48 -0
- data/lib/nokogiri/html/sax/parser_context.rb +16 -0
- data/lib/nokogiri/nokogiri.jar +0 -0
- data/lib/nokogiri/syntax_error.rb +4 -0
- data/lib/nokogiri/version.rb +33 -0
- data/lib/nokogiri/version_warning.rb +11 -0
- data/lib/nokogiri/xml.rb +67 -0
- data/lib/nokogiri/xml/attr.rb +14 -0
- data/lib/nokogiri/xml/attribute_decl.rb +18 -0
- data/lib/nokogiri/xml/builder.rb +405 -0
- data/lib/nokogiri/xml/cdata.rb +11 -0
- data/lib/nokogiri/xml/character_data.rb +7 -0
- data/lib/nokogiri/xml/document.rb +163 -0
- data/lib/nokogiri/xml/document_fragment.rb +73 -0
- data/lib/nokogiri/xml/dtd.rb +11 -0
- data/lib/nokogiri/xml/element_content.rb +36 -0
- data/lib/nokogiri/xml/element_decl.rb +13 -0
- data/lib/nokogiri/xml/entity_decl.rb +15 -0
- data/lib/nokogiri/xml/fragment_handler.rb +73 -0
- data/lib/nokogiri/xml/namespace.rb +13 -0
- data/lib/nokogiri/xml/node.rb +730 -0
- data/lib/nokogiri/xml/node/save_options.rb +42 -0
- data/lib/nokogiri/xml/node_set.rb +318 -0
- data/lib/nokogiri/xml/notation.rb +6 -0
- data/lib/nokogiri/xml/parse_options.rb +85 -0
- data/lib/nokogiri/xml/pp.rb +2 -0
- data/lib/nokogiri/xml/pp/character_data.rb +18 -0
- data/lib/nokogiri/xml/pp/node.rb +56 -0
- data/lib/nokogiri/xml/processing_instruction.rb +8 -0
- data/lib/nokogiri/xml/reader.rb +74 -0
- data/lib/nokogiri/xml/relax_ng.rb +32 -0
- data/lib/nokogiri/xml/sax.rb +4 -0
- data/lib/nokogiri/xml/sax/document.rb +160 -0
- data/lib/nokogiri/xml/sax/parser.rb +115 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
- data/lib/nokogiri/xml/schema.rb +61 -0
- data/lib/nokogiri/xml/syntax_error.rb +43 -0
- data/lib/nokogiri/xml/xpath.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +8 -0
- data/lib/nokogiri/xml/xpath_context.rb +16 -0
- data/lib/nokogiri/xslt.rb +48 -0
- data/lib/nokogiri/xslt/stylesheet.rb +25 -0
- data/lib/xercesImpl.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +90 -0
- data/tasks/test.rb +100 -0
- data/test/css/test_nthiness.rb +159 -0
- data/test/css/test_parser.rb +282 -0
- data/test/css/test_tokenizer.rb +190 -0
- data/test/css/test_xpath_visitor.rb +76 -0
- data/test/ffi/test_document.rb +35 -0
- data/test/files/2ch.html +108 -0
- data/test/files/address_book.rlx +12 -0
- data/test/files/address_book.xml +10 -0
- data/test/files/bar/bar.xsd +4 -0
- data/test/files/dont_hurt_em_why.xml +422 -0
- data/test/files/exslt.xml +8 -0
- data/test/files/exslt.xslt +35 -0
- data/test/files/foo/foo.xsd +4 -0
- data/test/files/po.xml +32 -0
- data/test/files/po.xsd +66 -0
- data/test/files/shift_jis.html +10 -0
- data/test/files/shift_jis.xml +5 -0
- data/test/files/snuggles.xml +3 -0
- data/test/files/staff.dtd +10 -0
- data/test/files/staff.xml +59 -0
- data/test/files/staff.xslt +32 -0
- data/test/files/tlm.html +850 -0
- data/test/files/valid_bar.xml +2 -0
- data/test/helper.rb +137 -0
- data/test/html/sax/test_parser.rb +83 -0
- data/test/html/sax/test_parser_context.rb +48 -0
- data/test/html/test_builder.rb +164 -0
- data/test/html/test_document.rb +385 -0
- data/test/html/test_document_encoding.rb +77 -0
- data/test/html/test_document_fragment.rb +157 -0
- data/test/html/test_element_description.rb +98 -0
- data/test/html/test_named_characters.rb +14 -0
- data/test/html/test_node.rb +242 -0
- data/test/html/test_node_encoding.rb +27 -0
- data/test/test_convert_xpath.rb +135 -0
- data/test/test_css_cache.rb +45 -0
- data/test/test_encoding_handler.rb +46 -0
- data/test/test_jruby.rb +40 -0
- data/test/test_memory_leak.rb +87 -0
- data/test/test_nokogiri.rb +140 -0
- data/test/test_reader.rb +358 -0
- data/test/test_soap4r_sax.rb +52 -0
- data/test/test_xslt_transforms.rb +150 -0
- data/test/xml/node/test_save_options.rb +20 -0
- data/test/xml/node/test_subclass.rb +44 -0
- data/test/xml/sax/test_parser.rb +314 -0
- data/test/xml/sax/test_parser_context.rb +63 -0
- data/test/xml/sax/test_push_parser.rb +135 -0
- data/test/xml/test_attr.rb +38 -0
- data/test/xml/test_attribute_decl.rb +90 -0
- data/test/xml/test_builder.rb +167 -0
- data/test/xml/test_cdata.rb +38 -0
- data/test/xml/test_comment.rb +29 -0
- data/test/xml/test_document.rb +638 -0
- data/test/xml/test_document_encoding.rb +26 -0
- data/test/xml/test_document_fragment.rb +149 -0
- data/test/xml/test_dtd.rb +92 -0
- data/test/xml/test_dtd_encoding.rb +33 -0
- data/test/xml/test_element_content.rb +56 -0
- data/test/xml/test_element_decl.rb +73 -0
- data/test/xml/test_entity_decl.rb +83 -0
- data/test/xml/test_entity_reference.rb +21 -0
- data/test/xml/test_namespace.rb +70 -0
- data/test/xml/test_node.rb +740 -0
- data/test/xml/test_node_attributes.rb +34 -0
- data/test/xml/test_node_encoding.rb +107 -0
- data/test/xml/test_node_reparenting.rb +279 -0
- data/test/xml/test_node_set.rb +577 -0
- data/test/xml/test_parse_options.rb +52 -0
- data/test/xml/test_processing_instruction.rb +30 -0
- data/test/xml/test_reader_encoding.rb +126 -0
- data/test/xml/test_relax_ng.rb +60 -0
- data/test/xml/test_schema.rb +89 -0
- data/test/xml/test_syntax_error.rb +12 -0
- data/test/xml/test_text.rb +30 -0
- data/test/xml/test_unparented_node.rb +381 -0
- data/test/xml/test_xpath.rb +169 -0
- metadata +477 -0
@@ -0,0 +1,92 @@
|
|
1
|
+
#include <html_sax_parser_context.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtmlSaxParserContext ;
|
4
|
+
|
5
|
+
static void deallocate(xmlParserCtxtPtr ctxt)
|
6
|
+
{
|
7
|
+
NOKOGIRI_DEBUG_START(handler);
|
8
|
+
|
9
|
+
ctxt->sax = NULL;
|
10
|
+
|
11
|
+
htmlFreeParserCtxt(ctxt);
|
12
|
+
|
13
|
+
NOKOGIRI_DEBUG_END(handler);
|
14
|
+
}
|
15
|
+
|
16
|
+
static VALUE parse_memory(VALUE klass, VALUE data, VALUE encoding)
|
17
|
+
{
|
18
|
+
if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
|
19
|
+
if(!(int)RSTRING_LEN(data))
|
20
|
+
rb_raise(rb_eRuntimeError, "data cannot be empty");
|
21
|
+
|
22
|
+
htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(
|
23
|
+
StringValuePtr(data),
|
24
|
+
(int)RSTRING_LEN(data)
|
25
|
+
);
|
26
|
+
|
27
|
+
if(RTEST(encoding)) {
|
28
|
+
xmlCharEncoding enc = xmlParseCharEncoding(StringValuePtr(encoding));
|
29
|
+
if(enc != XML_CHAR_ENCODING_ERROR) {
|
30
|
+
xmlSwitchEncoding(ctxt, enc);
|
31
|
+
if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
|
32
|
+
rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
|
33
|
+
StringValuePtr(encoding));
|
34
|
+
}
|
35
|
+
}
|
36
|
+
}
|
37
|
+
|
38
|
+
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
39
|
+
}
|
40
|
+
|
41
|
+
static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
|
42
|
+
{
|
43
|
+
htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
|
44
|
+
StringValuePtr(filename),
|
45
|
+
StringValuePtr(encoding)
|
46
|
+
);
|
47
|
+
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
48
|
+
}
|
49
|
+
|
50
|
+
static VALUE parse_with(VALUE self, VALUE sax_handler)
|
51
|
+
{
|
52
|
+
if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
|
53
|
+
rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
|
54
|
+
|
55
|
+
htmlParserCtxtPtr ctxt;
|
56
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
57
|
+
|
58
|
+
htmlSAXHandlerPtr sax;
|
59
|
+
Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
|
60
|
+
|
61
|
+
// Free the sax handler since we'll assign our own
|
62
|
+
if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
|
63
|
+
xmlFree(ctxt->sax);
|
64
|
+
|
65
|
+
ctxt->sax = sax;
|
66
|
+
ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
|
67
|
+
|
68
|
+
htmlParseDocument(ctxt);
|
69
|
+
|
70
|
+
if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
|
71
|
+
|
72
|
+
NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
|
73
|
+
return self;
|
74
|
+
}
|
75
|
+
|
76
|
+
void init_html_sax_parser_context()
|
77
|
+
{
|
78
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
79
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
80
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
81
|
+
VALUE sax = rb_define_module_under(xml, "SAX");
|
82
|
+
VALUE hsax = rb_define_module_under(html, "SAX");
|
83
|
+
VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
|
84
|
+
VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
|
85
|
+
|
86
|
+
cNokogiriHtmlSaxParserContext = klass;
|
87
|
+
|
88
|
+
rb_define_singleton_method(klass, "memory", parse_memory, 2);
|
89
|
+
rb_define_singleton_method(klass, "file", parse_file, 2);
|
90
|
+
|
91
|
+
rb_define_method(klass, "parse_with", parse_with, 1);
|
92
|
+
}
|
@@ -0,0 +1,96 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE mNokogiri ;
|
4
|
+
VALUE mNokogiriXml ;
|
5
|
+
VALUE mNokogiriHtml ;
|
6
|
+
VALUE mNokogiriXslt ;
|
7
|
+
VALUE mNokogiriXmlSax ;
|
8
|
+
VALUE mNokogiriHtmlSax ;
|
9
|
+
|
10
|
+
#ifdef USE_INCLUDED_VASPRINTF
|
11
|
+
/*
|
12
|
+
* I srsly hate windows. it doesn't have vasprintf.
|
13
|
+
* Thank you Geoffroy Couprie for this implementation of vasprintf!
|
14
|
+
*/
|
15
|
+
int vasprintf (char **strp, const char *fmt, va_list ap)
|
16
|
+
{
|
17
|
+
int len = vsnprintf (NULL, 0, fmt, ap) + 1;
|
18
|
+
char *res = (char *)malloc((unsigned int)len);
|
19
|
+
if (res == NULL)
|
20
|
+
return -1;
|
21
|
+
*strp = res;
|
22
|
+
return vsnprintf(res, (unsigned int)len, fmt, ap);
|
23
|
+
}
|
24
|
+
#endif
|
25
|
+
|
26
|
+
int is_2_6_16(void)
|
27
|
+
{
|
28
|
+
return (strcmp(xmlParserVersion, "20616") <= 0) ? 1 : 0 ;
|
29
|
+
}
|
30
|
+
|
31
|
+
void Init_nokogiri()
|
32
|
+
{
|
33
|
+
xmlMemSetup(
|
34
|
+
(xmlFreeFunc)ruby_xfree,
|
35
|
+
(xmlMallocFunc)ruby_xmalloc,
|
36
|
+
(xmlReallocFunc)ruby_xrealloc,
|
37
|
+
strdup
|
38
|
+
);
|
39
|
+
|
40
|
+
mNokogiri = rb_define_module("Nokogiri");
|
41
|
+
mNokogiriXml = rb_define_module_under(mNokogiri, "XML");
|
42
|
+
mNokogiriHtml = rb_define_module_under(mNokogiri, "HTML");
|
43
|
+
mNokogiriXslt = rb_define_module_under(mNokogiri, "XSLT");
|
44
|
+
mNokogiriXmlSax = rb_define_module_under(mNokogiriXml, "SAX");
|
45
|
+
mNokogiriHtmlSax = rb_define_module_under(mNokogiriHtml, "SAX");
|
46
|
+
|
47
|
+
rb_const_set( mNokogiri,
|
48
|
+
rb_intern("LIBXML_VERSION"),
|
49
|
+
NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
|
50
|
+
);
|
51
|
+
rb_const_set( mNokogiri,
|
52
|
+
rb_intern("LIBXML_PARSER_VERSION"),
|
53
|
+
NOKOGIRI_STR_NEW2(xmlParserVersion)
|
54
|
+
);
|
55
|
+
|
56
|
+
#ifdef LIBXML_ICONV_ENABLED
|
57
|
+
rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
|
58
|
+
#else
|
59
|
+
rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
|
60
|
+
#endif
|
61
|
+
|
62
|
+
xmlInitParser();
|
63
|
+
|
64
|
+
init_xml_document();
|
65
|
+
init_html_document();
|
66
|
+
init_xml_node();
|
67
|
+
init_xml_document_fragment();
|
68
|
+
init_xml_text();
|
69
|
+
init_xml_cdata();
|
70
|
+
init_xml_processing_instruction();
|
71
|
+
init_xml_attr();
|
72
|
+
init_xml_entity_reference();
|
73
|
+
init_xml_comment();
|
74
|
+
init_xml_node_set();
|
75
|
+
init_xml_xpath_context();
|
76
|
+
init_xml_xpath();
|
77
|
+
init_xml_sax_parser_context();
|
78
|
+
init_xml_sax_parser();
|
79
|
+
init_xml_sax_push_parser();
|
80
|
+
init_xml_reader();
|
81
|
+
init_xml_dtd();
|
82
|
+
init_xml_element_content();
|
83
|
+
init_xml_attribute_decl();
|
84
|
+
init_xml_element_decl();
|
85
|
+
init_xml_entity_decl();
|
86
|
+
init_xml_namespace();
|
87
|
+
init_html_sax_parser_context();
|
88
|
+
init_xslt_stylesheet();
|
89
|
+
init_xml_syntax_error();
|
90
|
+
init_html_entity_lookup();
|
91
|
+
init_html_element_description();
|
92
|
+
init_xml_schema();
|
93
|
+
init_xml_relax_ng();
|
94
|
+
init_nokogiri_io();
|
95
|
+
init_xml_encoding_handler();
|
96
|
+
}
|
@@ -0,0 +1,148 @@
|
|
1
|
+
#ifndef NOKOGIRI_NATIVE
|
2
|
+
#define NOKOGIRI_NATIVE
|
3
|
+
|
4
|
+
#include <stdlib.h>
|
5
|
+
#include <assert.h>
|
6
|
+
#include <libxml/parser.h>
|
7
|
+
#include <libxml/parserInternals.h>
|
8
|
+
#include <libxml/xpath.h>
|
9
|
+
#include <libxml/xpathInternals.h>
|
10
|
+
#include <libxml/xmlreader.h>
|
11
|
+
#include <libxml/xmlsave.h>
|
12
|
+
#include <libxml/xmlschemas.h>
|
13
|
+
#include <libxml/HTMLparser.h>
|
14
|
+
#include <libxml/HTMLtree.h>
|
15
|
+
#include <libxml/relaxng.h>
|
16
|
+
#include <ruby.h>
|
17
|
+
|
18
|
+
#ifdef USE_INCLUDED_VASPRINTF
|
19
|
+
int vasprintf (char **strp, const char *fmt, va_list ap);
|
20
|
+
#else
|
21
|
+
|
22
|
+
#ifndef _GNU_SOURCE
|
23
|
+
#define _GNU_SOURCE
|
24
|
+
#endif
|
25
|
+
|
26
|
+
# include <stdio.h>
|
27
|
+
|
28
|
+
#endif
|
29
|
+
|
30
|
+
int is_2_6_16(void) ;
|
31
|
+
|
32
|
+
#ifndef UNUSED
|
33
|
+
# if defined(__GNUC__)
|
34
|
+
# define MAYBE_UNUSED(name) name __attribute__((unused))
|
35
|
+
# define UNUSED(name) MAYBE_UNUSED(UNUSED_ ## name)
|
36
|
+
# else
|
37
|
+
# define MAYBE_UNUSED(name) name
|
38
|
+
# define UNUSED(name) name
|
39
|
+
# endif
|
40
|
+
#endif
|
41
|
+
|
42
|
+
#ifdef HAVE_RUBY_ENCODING_H
|
43
|
+
|
44
|
+
#include <ruby/encoding.h>
|
45
|
+
|
46
|
+
#define NOKOGIRI_STR_NEW2(str) \
|
47
|
+
({ \
|
48
|
+
VALUE _string = rb_str_new2((const char *)str); \
|
49
|
+
int _enc = rb_enc_find_index("UTF-8"); \
|
50
|
+
rb_enc_associate_index(_string, _enc); \
|
51
|
+
_string; \
|
52
|
+
})
|
53
|
+
|
54
|
+
#define NOKOGIRI_STR_NEW(str, len) \
|
55
|
+
({ \
|
56
|
+
VALUE _string = rb_str_new((const char *)str, (long)len); \
|
57
|
+
int _enc = rb_enc_find_index("UTF-8"); \
|
58
|
+
rb_enc_associate_index(_string, _enc); \
|
59
|
+
_string; \
|
60
|
+
})
|
61
|
+
|
62
|
+
#else
|
63
|
+
|
64
|
+
#define NOKOGIRI_STR_NEW2(str) \
|
65
|
+
rb_str_new2((const char *)str)
|
66
|
+
|
67
|
+
#define NOKOGIRI_STR_NEW(str, len) \
|
68
|
+
rb_str_new((const char *)str, (long)len)
|
69
|
+
#endif
|
70
|
+
|
71
|
+
#define RBSTR_OR_QNIL(_str) \
|
72
|
+
(_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
|
73
|
+
|
74
|
+
#include <xml_io.h>
|
75
|
+
#include <xml_document.h>
|
76
|
+
#include <html_entity_lookup.h>
|
77
|
+
#include <html_document.h>
|
78
|
+
#include <xml_node.h>
|
79
|
+
#include <xml_text.h>
|
80
|
+
#include <xml_cdata.h>
|
81
|
+
#include <xml_attr.h>
|
82
|
+
#include <xml_processing_instruction.h>
|
83
|
+
#include <xml_entity_reference.h>
|
84
|
+
#include <xml_document_fragment.h>
|
85
|
+
#include <xml_comment.h>
|
86
|
+
#include <xml_node_set.h>
|
87
|
+
#include <xml_xpath.h>
|
88
|
+
#include <xml_dtd.h>
|
89
|
+
#include <xml_attribute_decl.h>
|
90
|
+
#include <xml_element_decl.h>
|
91
|
+
#include <xml_entity_decl.h>
|
92
|
+
#include <xml_xpath_context.h>
|
93
|
+
#include <xml_element_content.h>
|
94
|
+
#include <xml_sax_parser_context.h>
|
95
|
+
#include <xml_sax_parser.h>
|
96
|
+
#include <xml_sax_push_parser.h>
|
97
|
+
#include <xml_reader.h>
|
98
|
+
#include <html_sax_parser_context.h>
|
99
|
+
#include <xslt_stylesheet.h>
|
100
|
+
#include <xml_syntax_error.h>
|
101
|
+
#include <xml_schema.h>
|
102
|
+
#include <xml_relax_ng.h>
|
103
|
+
#include <html_element_description.h>
|
104
|
+
#include <xml_namespace.h>
|
105
|
+
#include <xml_encoding_handler.h>
|
106
|
+
|
107
|
+
extern VALUE mNokogiri ;
|
108
|
+
extern VALUE mNokogiriXml ;
|
109
|
+
extern VALUE mNokogiriXmlSax ;
|
110
|
+
extern VALUE mNokogiriHtml ;
|
111
|
+
extern VALUE mNokogiriHtmlSax ;
|
112
|
+
extern VALUE mNokogiriXslt ;
|
113
|
+
|
114
|
+
#define NOKOGIRI_ROOT_NODE(_node) \
|
115
|
+
({ \
|
116
|
+
nokogiriTuplePtr tuple = (nokogiriTuplePtr)(_node->doc->_private); \
|
117
|
+
st_insert(tuple->unlinkedNodes, (st_data_t)_node, (st_data_t)_node); \
|
118
|
+
})
|
119
|
+
|
120
|
+
#ifdef DEBUG
|
121
|
+
|
122
|
+
#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
|
123
|
+
#define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
|
124
|
+
|
125
|
+
#else
|
126
|
+
|
127
|
+
#define NOKOGIRI_DEBUG_START(p)
|
128
|
+
#define NOKOGIRI_DEBUG_END(p)
|
129
|
+
|
130
|
+
#ifndef RSTRING_PTR
|
131
|
+
#define RSTRING_PTR(s) (RSTRING(s)->ptr)
|
132
|
+
#endif
|
133
|
+
|
134
|
+
#ifndef RSTRING_LEN
|
135
|
+
#define RSTRING_LEN(s) (RSTRING(s)->len)
|
136
|
+
#endif
|
137
|
+
|
138
|
+
#ifndef RARRAY_PTR
|
139
|
+
#define RARRAY_PTR(a) RARRAY(a)->ptr
|
140
|
+
#endif
|
141
|
+
|
142
|
+
#ifndef RARRAY_LEN
|
143
|
+
#define RARRAY_LEN(a) RARRAY(a)->len
|
144
|
+
#endif
|
145
|
+
|
146
|
+
#endif
|
147
|
+
|
148
|
+
#endif
|
@@ -0,0 +1,92 @@
|
|
1
|
+
#include <xml_attr.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* value=(content)
|
6
|
+
*
|
7
|
+
* Set the value for this Attr to +content+
|
8
|
+
*/
|
9
|
+
static VALUE set_value(VALUE self, VALUE content)
|
10
|
+
{
|
11
|
+
xmlAttrPtr attr;
|
12
|
+
Data_Get_Struct(self, xmlAttr, attr);
|
13
|
+
|
14
|
+
if(attr->children) xmlFreeNodeList(attr->children);
|
15
|
+
|
16
|
+
attr->children = attr->last = NULL;
|
17
|
+
|
18
|
+
if(content) {
|
19
|
+
xmlChar *buffer;
|
20
|
+
xmlNode *tmp;
|
21
|
+
|
22
|
+
// Encode our content
|
23
|
+
buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValuePtr(content));
|
24
|
+
|
25
|
+
attr->children = xmlStringGetNodeList(attr->doc, buffer);
|
26
|
+
attr->last = NULL;
|
27
|
+
tmp = attr->children;
|
28
|
+
|
29
|
+
// Loop through the children
|
30
|
+
for(tmp = attr->children; tmp; tmp = tmp->next) {
|
31
|
+
tmp->parent = (xmlNode *)attr;
|
32
|
+
tmp->doc = attr->doc;
|
33
|
+
if(tmp->next == NULL) attr->last = tmp;
|
34
|
+
}
|
35
|
+
|
36
|
+
// Free up memory
|
37
|
+
xmlFree(buffer);
|
38
|
+
}
|
39
|
+
|
40
|
+
return content;
|
41
|
+
}
|
42
|
+
|
43
|
+
/*
|
44
|
+
* call-seq:
|
45
|
+
* new(document, name)
|
46
|
+
*
|
47
|
+
* Create a new Attr element on the +document+ with +name+
|
48
|
+
*/
|
49
|
+
static VALUE new(int argc, VALUE *argv, VALUE klass)
|
50
|
+
{
|
51
|
+
xmlDocPtr xml_doc;
|
52
|
+
VALUE document;
|
53
|
+
VALUE name;
|
54
|
+
VALUE rest;
|
55
|
+
|
56
|
+
rb_scan_args(argc, argv, "2*", &document, &name, &rest);
|
57
|
+
|
58
|
+
Data_Get_Struct(document, xmlDoc, xml_doc);
|
59
|
+
|
60
|
+
xmlAttrPtr node = xmlNewDocProp(
|
61
|
+
xml_doc,
|
62
|
+
(const xmlChar *)StringValuePtr(name),
|
63
|
+
NULL
|
64
|
+
);
|
65
|
+
|
66
|
+
NOKOGIRI_ROOT_NODE((xmlNodePtr)node);
|
67
|
+
|
68
|
+
VALUE rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
|
69
|
+
rb_obj_call_init(rb_node, argc, argv);
|
70
|
+
|
71
|
+
if(rb_block_given_p()) rb_yield(rb_node);
|
72
|
+
|
73
|
+
return rb_node;
|
74
|
+
}
|
75
|
+
|
76
|
+
VALUE cNokogiriXmlAttr;
|
77
|
+
void init_xml_attr()
|
78
|
+
{
|
79
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
80
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
81
|
+
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
82
|
+
|
83
|
+
/*
|
84
|
+
* Attr represents a Attr node in an xml document.
|
85
|
+
*/
|
86
|
+
VALUE klass = rb_define_class_under(xml, "Attr", node);
|
87
|
+
|
88
|
+
cNokogiriXmlAttr = klass;
|
89
|
+
|
90
|
+
rb_define_singleton_method(klass, "new", new, -1);
|
91
|
+
rb_define_method(klass, "value=", set_value, 1);
|
92
|
+
}
|