rubyjedi-nokogiri_java 1.4.0.20100513161003-java
Sign up to get free protection for your applications and to get access to all the features.
- 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 +1,145 @@
|
|
1
|
+
#include <html_document.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* new
|
6
|
+
*
|
7
|
+
* Create a new document
|
8
|
+
*/
|
9
|
+
static VALUE new(int argc, VALUE *argv, VALUE klass)
|
10
|
+
{
|
11
|
+
VALUE uri, external_id, rest, rb_doc;
|
12
|
+
|
13
|
+
rb_scan_args(argc, argv, "0*", &rest);
|
14
|
+
uri = rb_ary_entry(rest, (long)0);
|
15
|
+
external_id = rb_ary_entry(rest, (long)1);
|
16
|
+
|
17
|
+
htmlDocPtr doc = htmlNewDoc(
|
18
|
+
RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
|
19
|
+
RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
|
20
|
+
);
|
21
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, doc);
|
22
|
+
rb_obj_call_init(rb_doc, argc, argv);
|
23
|
+
return rb_doc ;
|
24
|
+
}
|
25
|
+
|
26
|
+
/*
|
27
|
+
* call-seq:
|
28
|
+
* read_io(io, url, encoding, options)
|
29
|
+
*
|
30
|
+
* Read the HTML document from +io+ with given +url+, +encoding+,
|
31
|
+
* and +options+. See Nokogiri::HTML.parse
|
32
|
+
*/
|
33
|
+
static VALUE read_io( VALUE klass,
|
34
|
+
VALUE io,
|
35
|
+
VALUE url,
|
36
|
+
VALUE encoding,
|
37
|
+
VALUE options )
|
38
|
+
{
|
39
|
+
const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
|
40
|
+
const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
|
41
|
+
VALUE error_list = rb_ary_new();
|
42
|
+
|
43
|
+
xmlResetLastError();
|
44
|
+
xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
|
45
|
+
|
46
|
+
htmlDocPtr doc = htmlReadIO(
|
47
|
+
io_read_callback,
|
48
|
+
io_close_callback,
|
49
|
+
(void *)io,
|
50
|
+
c_url,
|
51
|
+
c_enc,
|
52
|
+
(int)NUM2INT(options)
|
53
|
+
);
|
54
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
55
|
+
|
56
|
+
if(doc == NULL) {
|
57
|
+
xmlFreeDoc(doc);
|
58
|
+
|
59
|
+
xmlErrorPtr error = xmlGetLastError();
|
60
|
+
if(error)
|
61
|
+
rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
|
62
|
+
else
|
63
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
64
|
+
|
65
|
+
return Qnil;
|
66
|
+
}
|
67
|
+
|
68
|
+
VALUE document = Nokogiri_wrap_xml_document(klass, doc);
|
69
|
+
rb_iv_set(document, "@errors", error_list);
|
70
|
+
return document;
|
71
|
+
}
|
72
|
+
|
73
|
+
/*
|
74
|
+
* call-seq:
|
75
|
+
* read_memory(string, url, encoding, options)
|
76
|
+
*
|
77
|
+
* Read the HTML document contained in +string+ with given +url+, +encoding+,
|
78
|
+
* and +options+. See Nokogiri::HTML.parse
|
79
|
+
*/
|
80
|
+
static VALUE read_memory( VALUE klass,
|
81
|
+
VALUE string,
|
82
|
+
VALUE url,
|
83
|
+
VALUE encoding,
|
84
|
+
VALUE options )
|
85
|
+
{
|
86
|
+
const char * c_buffer = StringValuePtr(string);
|
87
|
+
const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
|
88
|
+
const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
|
89
|
+
int len = RSTRING_LEN(string);
|
90
|
+
VALUE error_list = rb_ary_new();
|
91
|
+
|
92
|
+
xmlResetLastError();
|
93
|
+
xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
|
94
|
+
|
95
|
+
htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
|
96
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
97
|
+
|
98
|
+
if(doc == NULL) {
|
99
|
+
xmlFreeDoc(doc);
|
100
|
+
|
101
|
+
xmlErrorPtr error = xmlGetLastError();
|
102
|
+
if(error)
|
103
|
+
rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
|
104
|
+
else
|
105
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
106
|
+
|
107
|
+
return Qnil;
|
108
|
+
}
|
109
|
+
|
110
|
+
VALUE document = Nokogiri_wrap_xml_document(klass, doc);
|
111
|
+
rb_iv_set(document, "@errors", error_list);
|
112
|
+
return document;
|
113
|
+
}
|
114
|
+
|
115
|
+
/*
|
116
|
+
* call-seq:
|
117
|
+
* type
|
118
|
+
*
|
119
|
+
* The type for this document
|
120
|
+
*/
|
121
|
+
static VALUE type(VALUE self)
|
122
|
+
{
|
123
|
+
htmlDocPtr doc;
|
124
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
125
|
+
return INT2NUM((long)doc->type);
|
126
|
+
}
|
127
|
+
|
128
|
+
VALUE cNokogiriHtmlDocument ;
|
129
|
+
void init_html_document()
|
130
|
+
{
|
131
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
132
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
133
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
134
|
+
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
135
|
+
VALUE xml_doc = rb_define_class_under(xml, "Document", node);
|
136
|
+
VALUE klass = rb_define_class_under(html, "Document", xml_doc);
|
137
|
+
|
138
|
+
cNokogiriHtmlDocument = klass;
|
139
|
+
|
140
|
+
rb_define_singleton_method(klass, "read_memory", read_memory, 4);
|
141
|
+
rb_define_singleton_method(klass, "read_io", read_io, 4);
|
142
|
+
rb_define_singleton_method(klass, "new", new, -1);
|
143
|
+
|
144
|
+
rb_define_method(klass, "type", type, 0);
|
145
|
+
}
|
@@ -0,0 +1,272 @@
|
|
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
|
+
htmlElemDesc * description;
|
12
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
13
|
+
|
14
|
+
VALUE list = rb_ary_new();
|
15
|
+
|
16
|
+
if(NULL == description->attrs_req) return list;
|
17
|
+
|
18
|
+
int i = 0;
|
19
|
+
while(description->attrs_req[i]) {
|
20
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
|
21
|
+
i++;
|
22
|
+
}
|
23
|
+
|
24
|
+
return list;
|
25
|
+
}
|
26
|
+
|
27
|
+
/*
|
28
|
+
* call-seq:
|
29
|
+
* deprecated_attributes
|
30
|
+
*
|
31
|
+
* A list of deprecated attributes for this element
|
32
|
+
*/
|
33
|
+
static VALUE deprecated_attributes(VALUE self)
|
34
|
+
{
|
35
|
+
htmlElemDesc * description;
|
36
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
37
|
+
|
38
|
+
VALUE list = rb_ary_new();
|
39
|
+
|
40
|
+
if(NULL == description->attrs_depr) return list;
|
41
|
+
|
42
|
+
int i = 0;
|
43
|
+
while(description->attrs_depr[i]) {
|
44
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
|
45
|
+
i++;
|
46
|
+
}
|
47
|
+
|
48
|
+
return list;
|
49
|
+
}
|
50
|
+
|
51
|
+
/*
|
52
|
+
* call-seq:
|
53
|
+
* optional_attributes
|
54
|
+
*
|
55
|
+
* A list of optional attributes for this element
|
56
|
+
*/
|
57
|
+
static VALUE optional_attributes(VALUE self)
|
58
|
+
{
|
59
|
+
htmlElemDesc * description;
|
60
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
61
|
+
|
62
|
+
VALUE list = rb_ary_new();
|
63
|
+
|
64
|
+
if(NULL == description->attrs_opt) return list;
|
65
|
+
|
66
|
+
int i = 0;
|
67
|
+
while(description->attrs_opt[i]) {
|
68
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
|
69
|
+
i++;
|
70
|
+
}
|
71
|
+
|
72
|
+
return list;
|
73
|
+
}
|
74
|
+
|
75
|
+
/*
|
76
|
+
* call-seq:
|
77
|
+
* default_sub_element
|
78
|
+
*
|
79
|
+
* The default sub element for this element
|
80
|
+
*/
|
81
|
+
static VALUE default_sub_element(VALUE self)
|
82
|
+
{
|
83
|
+
htmlElemDesc * description;
|
84
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
85
|
+
|
86
|
+
return NOKOGIRI_STR_NEW2(description->defaultsubelt);
|
87
|
+
}
|
88
|
+
|
89
|
+
/*
|
90
|
+
* call-seq:
|
91
|
+
* sub_elements
|
92
|
+
*
|
93
|
+
* A list of allowed sub elements for this element.
|
94
|
+
*/
|
95
|
+
static VALUE sub_elements(VALUE self)
|
96
|
+
{
|
97
|
+
htmlElemDesc * description;
|
98
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
99
|
+
|
100
|
+
VALUE list = rb_ary_new();
|
101
|
+
|
102
|
+
if(NULL == description->subelts) return list;
|
103
|
+
|
104
|
+
int i = 0;
|
105
|
+
while(description->subelts[i]) {
|
106
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
|
107
|
+
i++;
|
108
|
+
}
|
109
|
+
|
110
|
+
return list;
|
111
|
+
}
|
112
|
+
|
113
|
+
/*
|
114
|
+
* call-seq:
|
115
|
+
* description
|
116
|
+
*
|
117
|
+
* The description for this element
|
118
|
+
*/
|
119
|
+
static VALUE description(VALUE self)
|
120
|
+
{
|
121
|
+
htmlElemDesc * description;
|
122
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
123
|
+
|
124
|
+
return NOKOGIRI_STR_NEW2(description->desc);
|
125
|
+
}
|
126
|
+
|
127
|
+
/*
|
128
|
+
* call-seq:
|
129
|
+
* inline?
|
130
|
+
*
|
131
|
+
* Is this element an inline element?
|
132
|
+
*/
|
133
|
+
static VALUE inline_eh(VALUE self)
|
134
|
+
{
|
135
|
+
htmlElemDesc * description;
|
136
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
137
|
+
|
138
|
+
if(description->isinline) return Qtrue;
|
139
|
+
return Qfalse;
|
140
|
+
}
|
141
|
+
|
142
|
+
/*
|
143
|
+
* call-seq:
|
144
|
+
* deprecated?
|
145
|
+
*
|
146
|
+
* Is this element deprecated?
|
147
|
+
*/
|
148
|
+
static VALUE deprecated_eh(VALUE self)
|
149
|
+
{
|
150
|
+
htmlElemDesc * description;
|
151
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
152
|
+
|
153
|
+
if(description->depr) return Qtrue;
|
154
|
+
return Qfalse;
|
155
|
+
}
|
156
|
+
|
157
|
+
/*
|
158
|
+
* call-seq:
|
159
|
+
* empty?
|
160
|
+
*
|
161
|
+
* Is this an empty element?
|
162
|
+
*/
|
163
|
+
static VALUE empty_eh(VALUE self)
|
164
|
+
{
|
165
|
+
htmlElemDesc * description;
|
166
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
167
|
+
|
168
|
+
if(description->empty) return Qtrue;
|
169
|
+
return Qfalse;
|
170
|
+
}
|
171
|
+
|
172
|
+
/*
|
173
|
+
* call-seq:
|
174
|
+
* save_end_tag?
|
175
|
+
*
|
176
|
+
* Should the end tag be saved?
|
177
|
+
*/
|
178
|
+
static VALUE save_end_tag_eh(VALUE self)
|
179
|
+
{
|
180
|
+
htmlElemDesc * description;
|
181
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
182
|
+
|
183
|
+
if(description->saveEndTag) return Qtrue;
|
184
|
+
return Qfalse;
|
185
|
+
}
|
186
|
+
|
187
|
+
/*
|
188
|
+
* call-seq:
|
189
|
+
* implied_end_tag?
|
190
|
+
*
|
191
|
+
* Can the end tag be implied for this tag?
|
192
|
+
*/
|
193
|
+
static VALUE implied_end_tag_eh(VALUE self)
|
194
|
+
{
|
195
|
+
htmlElemDesc * description;
|
196
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
197
|
+
|
198
|
+
if(description->endTag) return Qtrue;
|
199
|
+
return Qfalse;
|
200
|
+
}
|
201
|
+
|
202
|
+
/*
|
203
|
+
* call-seq:
|
204
|
+
* implied_start_tag?
|
205
|
+
*
|
206
|
+
* Can the start tag be implied for this tag?
|
207
|
+
*/
|
208
|
+
static VALUE implied_start_tag_eh(VALUE self)
|
209
|
+
{
|
210
|
+
htmlElemDesc * description;
|
211
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
212
|
+
|
213
|
+
if(description->startTag) return Qtrue;
|
214
|
+
return Qfalse;
|
215
|
+
}
|
216
|
+
|
217
|
+
/*
|
218
|
+
* call-seq:
|
219
|
+
* name
|
220
|
+
*
|
221
|
+
* Get the tag name for this ElemementDescription
|
222
|
+
*/
|
223
|
+
static VALUE name(VALUE self)
|
224
|
+
{
|
225
|
+
htmlElemDesc * description;
|
226
|
+
Data_Get_Struct(self, htmlElemDesc, description);
|
227
|
+
|
228
|
+
if(NULL == description->name) return Qnil;
|
229
|
+
return NOKOGIRI_STR_NEW2(description->name);
|
230
|
+
}
|
231
|
+
|
232
|
+
/*
|
233
|
+
* call-seq:
|
234
|
+
* [](tag_name)
|
235
|
+
*
|
236
|
+
* Get ElemementDescription for +tag_name+
|
237
|
+
*/
|
238
|
+
static VALUE get_description(VALUE klass, VALUE tag_name)
|
239
|
+
{
|
240
|
+
const htmlElemDesc * description = htmlTagLookup(
|
241
|
+
(const xmlChar *)StringValuePtr(tag_name)
|
242
|
+
);
|
243
|
+
|
244
|
+
if(NULL == description) return Qnil;
|
245
|
+
return Data_Wrap_Struct(klass, 0, 0, (void *)description);
|
246
|
+
}
|
247
|
+
|
248
|
+
VALUE cNokogiriHtmlElementDescription ;
|
249
|
+
void init_html_element_description()
|
250
|
+
{
|
251
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
252
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
253
|
+
VALUE klass = rb_define_class_under(html, "ElementDescription",rb_cObject);
|
254
|
+
|
255
|
+
cNokogiriHtmlElementDescription = klass;
|
256
|
+
|
257
|
+
rb_define_singleton_method(klass, "[]", get_description, 1);
|
258
|
+
|
259
|
+
rb_define_method(klass, "name", name, 0);
|
260
|
+
rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
|
261
|
+
rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
|
262
|
+
rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
|
263
|
+
rb_define_method(klass, "empty?", empty_eh, 0);
|
264
|
+
rb_define_method(klass, "deprecated?", deprecated_eh, 0);
|
265
|
+
rb_define_method(klass, "inline?", inline_eh, 0);
|
266
|
+
rb_define_method(klass, "description", description, 0);
|
267
|
+
rb_define_method(klass, "sub_elements", sub_elements, 0);
|
268
|
+
rb_define_method(klass, "default_sub_element", default_sub_element, 0);
|
269
|
+
rb_define_method(klass, "optional_attributes", optional_attributes, 0);
|
270
|
+
rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
|
271
|
+
rb_define_method(klass, "required_attributes", required_attributes, 0);
|
272
|
+
}
|
@@ -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 *)StringValuePtr(key));
|
13
|
+
|
14
|
+
if(NULL == desc) return Qnil;
|
15
|
+
VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
|
16
|
+
|
17
|
+
VALUE args[3];
|
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
|
+
}
|