nokogiri 1.11.0.rc4-arm64-darwin
Sign up to get free protection for your applications and to get access to all the features.
Potentially problematic release.
This version of nokogiri might be problematic. Click here for more details.
- checksums.yaml +7 -0
- data/LICENSE-DEPENDENCIES.md +1682 -0
- data/LICENSE.md +9 -0
- data/README.md +200 -0
- data/bin/nokogiri +118 -0
- data/dependencies.yml +74 -0
- data/ext/nokogiri/depend +477 -0
- data/ext/nokogiri/extconf.rb +819 -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/include/libexslt/exslt.h +102 -0
- data/ext/nokogiri/include/libexslt/exsltconfig.h +70 -0
- data/ext/nokogiri/include/libexslt/exsltexports.h +140 -0
- data/ext/nokogiri/include/libxml2/libxml/DOCBparser.h +96 -0
- data/ext/nokogiri/include/libxml2/libxml/HTMLparser.h +306 -0
- data/ext/nokogiri/include/libxml2/libxml/HTMLtree.h +147 -0
- data/ext/nokogiri/include/libxml2/libxml/SAX.h +173 -0
- data/ext/nokogiri/include/libxml2/libxml/SAX2.h +178 -0
- data/ext/nokogiri/include/libxml2/libxml/c14n.h +126 -0
- data/ext/nokogiri/include/libxml2/libxml/catalog.h +182 -0
- data/ext/nokogiri/include/libxml2/libxml/chvalid.h +230 -0
- data/ext/nokogiri/include/libxml2/libxml/debugXML.h +217 -0
- data/ext/nokogiri/include/libxml2/libxml/dict.h +79 -0
- data/ext/nokogiri/include/libxml2/libxml/encoding.h +245 -0
- data/ext/nokogiri/include/libxml2/libxml/entities.h +151 -0
- data/ext/nokogiri/include/libxml2/libxml/globals.h +508 -0
- data/ext/nokogiri/include/libxml2/libxml/hash.h +236 -0
- data/ext/nokogiri/include/libxml2/libxml/list.h +137 -0
- data/ext/nokogiri/include/libxml2/libxml/nanoftp.h +163 -0
- data/ext/nokogiri/include/libxml2/libxml/nanohttp.h +81 -0
- data/ext/nokogiri/include/libxml2/libxml/parser.h +1241 -0
- data/ext/nokogiri/include/libxml2/libxml/parserInternals.h +644 -0
- data/ext/nokogiri/include/libxml2/libxml/pattern.h +100 -0
- data/ext/nokogiri/include/libxml2/libxml/relaxng.h +217 -0
- data/ext/nokogiri/include/libxml2/libxml/schemasInternals.h +958 -0
- data/ext/nokogiri/include/libxml2/libxml/schematron.h +142 -0
- data/ext/nokogiri/include/libxml2/libxml/threads.h +89 -0
- data/ext/nokogiri/include/libxml2/libxml/tree.h +1311 -0
- data/ext/nokogiri/include/libxml2/libxml/uri.h +94 -0
- data/ext/nokogiri/include/libxml2/libxml/valid.h +458 -0
- data/ext/nokogiri/include/libxml2/libxml/xinclude.h +129 -0
- data/ext/nokogiri/include/libxml2/libxml/xlink.h +189 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlIO.h +366 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlautomata.h +146 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlerror.h +945 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlexports.h +153 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlmemory.h +224 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlmodule.h +57 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlreader.h +428 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlregexp.h +222 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlsave.h +88 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlschemas.h +246 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlschemastypes.h +151 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlstring.h +140 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlunicode.h +202 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlversion.h +485 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlwriter.h +488 -0
- data/ext/nokogiri/include/libxml2/libxml/xpath.h +566 -0
- data/ext/nokogiri/include/libxml2/libxml/xpathInternals.h +632 -0
- data/ext/nokogiri/include/libxml2/libxml/xpointer.h +114 -0
- data/ext/nokogiri/include/libxslt/attributes.h +38 -0
- data/ext/nokogiri/include/libxslt/documents.h +93 -0
- data/ext/nokogiri/include/libxslt/extensions.h +262 -0
- data/ext/nokogiri/include/libxslt/extra.h +72 -0
- data/ext/nokogiri/include/libxslt/functions.h +78 -0
- data/ext/nokogiri/include/libxslt/imports.h +75 -0
- data/ext/nokogiri/include/libxslt/keys.h +53 -0
- data/ext/nokogiri/include/libxslt/namespaces.h +68 -0
- data/ext/nokogiri/include/libxslt/numbersInternals.h +73 -0
- data/ext/nokogiri/include/libxslt/pattern.h +84 -0
- data/ext/nokogiri/include/libxslt/preproc.h +43 -0
- data/ext/nokogiri/include/libxslt/security.h +104 -0
- data/ext/nokogiri/include/libxslt/templates.h +77 -0
- data/ext/nokogiri/include/libxslt/transform.h +207 -0
- data/ext/nokogiri/include/libxslt/variables.h +118 -0
- data/ext/nokogiri/include/libxslt/xslt.h +110 -0
- data/ext/nokogiri/include/libxslt/xsltInternals.h +1978 -0
- data/ext/nokogiri/include/libxslt/xsltconfig.h +180 -0
- data/ext/nokogiri/include/libxslt/xsltexports.h +142 -0
- data/ext/nokogiri/include/libxslt/xsltlocale.h +76 -0
- data/ext/nokogiri/include/libxslt/xsltutils.h +313 -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/nokogiri.rb +127 -0
- data/lib/nokogiri/2.5/nokogiri.bundle +0 -0
- data/lib/nokogiri/2.6/nokogiri.bundle +0 -0
- data/lib/nokogiri/2.7/nokogiri.bundle +0 -0
- data/lib/nokogiri/3.0/nokogiri.bundle +0 -0
- data/lib/nokogiri/css.rb +28 -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/decorators/slop.rb +43 -0
- data/lib/nokogiri/html.rb +38 -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/jruby/dependencies.rb +20 -0
- data/lib/nokogiri/syntax_error.rb +5 -0
- data/lib/nokogiri/version.rb +3 -0
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +182 -0
- data/lib/nokogiri/xml.rb +76 -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.rb +1240 -0
- data/lib/nokogiri/xml/node/save_options.rb +62 -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.rb +3 -0
- data/lib/nokogiri/xml/pp/character_data.rb +19 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -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.rb +5 -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/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.rb +11 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +12 -0
- data/lib/nokogiri/xml/xpath_context.rb +17 -0
- data/lib/nokogiri/xslt.rb +57 -0
- data/lib/nokogiri/xslt/stylesheet.rb +26 -0
- data/lib/xsd/xmlparser/nokogiri.rb +103 -0
- metadata +565 -0
@@ -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
|
+
}
|
@@ -0,0 +1,87 @@
|
|
1
|
+
#include <html_sax_push_parser.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* native_write(chunk, last_chunk)
|
6
|
+
*
|
7
|
+
* Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
|
8
|
+
*/
|
9
|
+
static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
|
10
|
+
{
|
11
|
+
xmlParserCtxtPtr ctx;
|
12
|
+
const char * chunk = NULL;
|
13
|
+
int size = 0;
|
14
|
+
|
15
|
+
|
16
|
+
Data_Get_Struct(self, xmlParserCtxt, ctx);
|
17
|
+
|
18
|
+
if(Qnil != _chunk) {
|
19
|
+
chunk = StringValuePtr(_chunk);
|
20
|
+
size = (int)RSTRING_LEN(_chunk);
|
21
|
+
}
|
22
|
+
|
23
|
+
if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
|
24
|
+
if (!(ctx->options & XML_PARSE_RECOVER)) {
|
25
|
+
xmlErrorPtr e = xmlCtxtGetLastError(ctx);
|
26
|
+
Nokogiri_error_raise(NULL, e);
|
27
|
+
}
|
28
|
+
}
|
29
|
+
|
30
|
+
return self;
|
31
|
+
}
|
32
|
+
|
33
|
+
/*
|
34
|
+
* call-seq:
|
35
|
+
* initialize_native(xml_sax, filename)
|
36
|
+
*
|
37
|
+
* Initialize the push parser with +xml_sax+ using +filename+
|
38
|
+
*/
|
39
|
+
static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
|
40
|
+
VALUE encoding)
|
41
|
+
{
|
42
|
+
htmlSAXHandlerPtr sax;
|
43
|
+
const char * filename = NULL;
|
44
|
+
htmlParserCtxtPtr ctx;
|
45
|
+
xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
|
46
|
+
|
47
|
+
Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
|
48
|
+
|
49
|
+
if(_filename != Qnil) filename = StringValueCStr(_filename);
|
50
|
+
|
51
|
+
if (!NIL_P(encoding)) {
|
52
|
+
enc = xmlParseCharEncoding(StringValueCStr(encoding));
|
53
|
+
if (enc == XML_CHAR_ENCODING_ERROR)
|
54
|
+
rb_raise(rb_eArgError, "Unsupported Encoding");
|
55
|
+
}
|
56
|
+
|
57
|
+
ctx = htmlCreatePushParserCtxt(
|
58
|
+
sax,
|
59
|
+
NULL,
|
60
|
+
NULL,
|
61
|
+
0,
|
62
|
+
filename,
|
63
|
+
enc
|
64
|
+
);
|
65
|
+
if(ctx == NULL)
|
66
|
+
rb_raise(rb_eRuntimeError, "Could not create a parser context");
|
67
|
+
|
68
|
+
ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
|
69
|
+
|
70
|
+
ctx->sax2 = 1;
|
71
|
+
DATA_PTR(self) = ctx;
|
72
|
+
return self;
|
73
|
+
}
|
74
|
+
|
75
|
+
VALUE cNokogiriHtmlSaxPushParser;
|
76
|
+
void init_html_sax_push_parser()
|
77
|
+
{
|
78
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
79
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
80
|
+
VALUE sax = rb_define_module_under(html, "SAX");
|
81
|
+
VALUE klass = rb_define_class_under(sax, "PushParser", cNokogiriXmlSaxPushParser);
|
82
|
+
|
83
|
+
cNokogiriHtmlSaxPushParser = klass;
|
84
|
+
|
85
|
+
rb_define_private_method(klass, "initialize_native", initialize_native, 3);
|
86
|
+
rb_define_private_method(klass, "native_write", native_write, 2);
|
87
|
+
}
|
@@ -0,0 +1,102 @@
|
|
1
|
+
|
2
|
+
#ifndef __EXSLT_H__
|
3
|
+
#define __EXSLT_H__
|
4
|
+
|
5
|
+
#include <libxml/tree.h>
|
6
|
+
#include <libxml/xpath.h>
|
7
|
+
#include "exsltexports.h"
|
8
|
+
#include <libexslt/exsltconfig.h>
|
9
|
+
|
10
|
+
#ifdef __cplusplus
|
11
|
+
extern "C" {
|
12
|
+
#endif
|
13
|
+
|
14
|
+
EXSLTPUBVAR const char *exsltLibraryVersion;
|
15
|
+
EXSLTPUBVAR const int exsltLibexsltVersion;
|
16
|
+
EXSLTPUBVAR const int exsltLibxsltVersion;
|
17
|
+
EXSLTPUBVAR const int exsltLibxmlVersion;
|
18
|
+
|
19
|
+
/**
|
20
|
+
* EXSLT_COMMON_NAMESPACE:
|
21
|
+
*
|
22
|
+
* Namespace for EXSLT common functions
|
23
|
+
*/
|
24
|
+
#define EXSLT_COMMON_NAMESPACE ((const xmlChar *) "http://exslt.org/common")
|
25
|
+
/**
|
26
|
+
* EXSLT_CRYPTO_NAMESPACE:
|
27
|
+
*
|
28
|
+
* Namespace for EXSLT crypto functions
|
29
|
+
*/
|
30
|
+
#define EXSLT_CRYPTO_NAMESPACE ((const xmlChar *) "http://exslt.org/crypto")
|
31
|
+
/**
|
32
|
+
* EXSLT_MATH_NAMESPACE:
|
33
|
+
*
|
34
|
+
* Namespace for EXSLT math functions
|
35
|
+
*/
|
36
|
+
#define EXSLT_MATH_NAMESPACE ((const xmlChar *) "http://exslt.org/math")
|
37
|
+
/**
|
38
|
+
* EXSLT_SETS_NAMESPACE:
|
39
|
+
*
|
40
|
+
* Namespace for EXSLT set functions
|
41
|
+
*/
|
42
|
+
#define EXSLT_SETS_NAMESPACE ((const xmlChar *) "http://exslt.org/sets")
|
43
|
+
/**
|
44
|
+
* EXSLT_FUNCTIONS_NAMESPACE:
|
45
|
+
*
|
46
|
+
* Namespace for EXSLT functions extension functions
|
47
|
+
*/
|
48
|
+
#define EXSLT_FUNCTIONS_NAMESPACE ((const xmlChar *) "http://exslt.org/functions")
|
49
|
+
/**
|
50
|
+
* EXSLT_STRINGS_NAMESPACE:
|
51
|
+
*
|
52
|
+
* Namespace for EXSLT strings functions
|
53
|
+
*/
|
54
|
+
#define EXSLT_STRINGS_NAMESPACE ((const xmlChar *) "http://exslt.org/strings")
|
55
|
+
/**
|
56
|
+
* EXSLT_DATE_NAMESPACE:
|
57
|
+
*
|
58
|
+
* Namespace for EXSLT date functions
|
59
|
+
*/
|
60
|
+
#define EXSLT_DATE_NAMESPACE ((const xmlChar *) "http://exslt.org/dates-and-times")
|
61
|
+
/**
|
62
|
+
* EXSLT_DYNAMIC_NAMESPACE:
|
63
|
+
*
|
64
|
+
* Namespace for EXSLT dynamic functions
|
65
|
+
*/
|
66
|
+
#define EXSLT_DYNAMIC_NAMESPACE ((const xmlChar *) "http://exslt.org/dynamic")
|
67
|
+
|
68
|
+
/**
|
69
|
+
* SAXON_NAMESPACE:
|
70
|
+
*
|
71
|
+
* Namespace for SAXON extensions functions
|
72
|
+
*/
|
73
|
+
#define SAXON_NAMESPACE ((const xmlChar *) "http://icl.com/saxon")
|
74
|
+
|
75
|
+
EXSLTPUBFUN void EXSLTCALL exsltCommonRegister (void);
|
76
|
+
#ifdef EXSLT_CRYPTO_ENABLED
|
77
|
+
EXSLTPUBFUN void EXSLTCALL exsltCryptoRegister (void);
|
78
|
+
#endif
|
79
|
+
EXSLTPUBFUN void EXSLTCALL exsltMathRegister (void);
|
80
|
+
EXSLTPUBFUN void EXSLTCALL exsltSetsRegister (void);
|
81
|
+
EXSLTPUBFUN void EXSLTCALL exsltFuncRegister (void);
|
82
|
+
EXSLTPUBFUN void EXSLTCALL exsltStrRegister (void);
|
83
|
+
EXSLTPUBFUN void EXSLTCALL exsltDateRegister (void);
|
84
|
+
EXSLTPUBFUN void EXSLTCALL exsltSaxonRegister (void);
|
85
|
+
EXSLTPUBFUN void EXSLTCALL exsltDynRegister(void);
|
86
|
+
|
87
|
+
EXSLTPUBFUN void EXSLTCALL exsltRegisterAll (void);
|
88
|
+
|
89
|
+
EXSLTPUBFUN int EXSLTCALL exsltDateXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
90
|
+
const xmlChar *prefix);
|
91
|
+
EXSLTPUBFUN int EXSLTCALL exsltMathXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
92
|
+
const xmlChar *prefix);
|
93
|
+
EXSLTPUBFUN int EXSLTCALL exsltSetsXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
94
|
+
const xmlChar *prefix);
|
95
|
+
EXSLTPUBFUN int EXSLTCALL exsltStrXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
96
|
+
const xmlChar *prefix);
|
97
|
+
|
98
|
+
#ifdef __cplusplus
|
99
|
+
}
|
100
|
+
#endif
|
101
|
+
#endif /* __EXSLT_H__ */
|
102
|
+
|
@@ -0,0 +1,70 @@
|
|
1
|
+
/*
|
2
|
+
* exsltconfig.h: compile-time version information for the EXSLT library
|
3
|
+
*
|
4
|
+
* See Copyright for the status of this software.
|
5
|
+
*
|
6
|
+
* daniel@veillard.com
|
7
|
+
*/
|
8
|
+
|
9
|
+
#ifndef __XML_EXSLTCONFIG_H__
|
10
|
+
#define __XML_EXSLTCONFIG_H__
|
11
|
+
|
12
|
+
#ifdef __cplusplus
|
13
|
+
extern "C" {
|
14
|
+
#endif
|
15
|
+
|
16
|
+
/**
|
17
|
+
* LIBEXSLT_DOTTED_VERSION:
|
18
|
+
*
|
19
|
+
* the version string like "1.2.3"
|
20
|
+
*/
|
21
|
+
#define LIBEXSLT_DOTTED_VERSION "1.1.34"
|
22
|
+
|
23
|
+
/**
|
24
|
+
* LIBEXSLT_VERSION:
|
25
|
+
*
|
26
|
+
* the version number: 1.2.3 value is 10203
|
27
|
+
*/
|
28
|
+
#define LIBEXSLT_VERSION 820
|
29
|
+
|
30
|
+
/**
|
31
|
+
* LIBEXSLT_VERSION_STRING:
|
32
|
+
*
|
33
|
+
* the version number string, 1.2.3 value is "10203"
|
34
|
+
*/
|
35
|
+
#define LIBEXSLT_VERSION_STRING "820"
|
36
|
+
|
37
|
+
/**
|
38
|
+
* LIBEXSLT_VERSION_EXTRA:
|
39
|
+
*
|
40
|
+
* extra version information, used to show a CVS compilation
|
41
|
+
*/
|
42
|
+
#define LIBEXSLT_VERSION_EXTRA ""
|
43
|
+
|
44
|
+
/**
|
45
|
+
* WITH_CRYPTO:
|
46
|
+
*
|
47
|
+
* Whether crypto support is configured into exslt
|
48
|
+
*/
|
49
|
+
#if 0
|
50
|
+
#define EXSLT_CRYPTO_ENABLED
|
51
|
+
#endif
|
52
|
+
|
53
|
+
/**
|
54
|
+
* ATTRIBUTE_UNUSED:
|
55
|
+
*
|
56
|
+
* This macro is used to flag unused function parameters to GCC
|
57
|
+
*/
|
58
|
+
#ifdef __GNUC__
|
59
|
+
#ifndef ATTRIBUTE_UNUSED
|
60
|
+
#define ATTRIBUTE_UNUSED __attribute__((unused))
|
61
|
+
#endif
|
62
|
+
#else
|
63
|
+
#define ATTRIBUTE_UNUSED
|
64
|
+
#endif
|
65
|
+
|
66
|
+
#ifdef __cplusplus
|
67
|
+
}
|
68
|
+
#endif
|
69
|
+
|
70
|
+
#endif /* __XML_EXSLTCONFIG_H__ */
|