nokogiri 1.11.0.rc3-x86-linux → 1.11.3-x86-linux
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 +4 -4
- data/Gemfile +3 -0
- data/LICENSE-DEPENDENCIES.md +1015 -947
- data/LICENSE.md +1 -1
- data/README.md +168 -91
- data/ext/nokogiri/depend +37 -358
- data/ext/nokogiri/extconf.rb +581 -374
- data/ext/nokogiri/html_document.c +78 -82
- data/ext/nokogiri/html_element_description.c +84 -71
- data/ext/nokogiri/html_entity_lookup.c +21 -16
- data/ext/nokogiri/html_sax_parser_context.c +69 -66
- data/ext/nokogiri/html_sax_push_parser.c +42 -34
- 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/libxml2_backwards_compat.c +121 -0
- data/ext/nokogiri/nokogiri.c +192 -93
- data/ext/nokogiri/nokogiri.h +177 -98
- data/ext/nokogiri/test_global_handlers.c +40 -0
- data/ext/nokogiri/xml_attr.c +15 -15
- data/ext/nokogiri/xml_attribute_decl.c +18 -18
- data/ext/nokogiri/xml_cdata.c +13 -18
- data/ext/nokogiri/xml_comment.c +19 -26
- data/ext/nokogiri/xml_document.c +246 -188
- data/ext/nokogiri/xml_document_fragment.c +13 -15
- data/ext/nokogiri/xml_dtd.c +54 -48
- data/ext/nokogiri/xml_element_content.c +30 -27
- data/ext/nokogiri/xml_element_decl.c +22 -22
- data/ext/nokogiri/xml_encoding_handler.c +17 -11
- data/ext/nokogiri/xml_entity_decl.c +32 -30
- data/ext/nokogiri/xml_entity_reference.c +16 -18
- data/ext/nokogiri/xml_namespace.c +56 -49
- data/ext/nokogiri/xml_node.c +371 -320
- data/ext/nokogiri/xml_node_set.c +168 -156
- data/ext/nokogiri/xml_processing_instruction.c +17 -19
- data/ext/nokogiri/xml_reader.c +191 -157
- data/ext/nokogiri/xml_relax_ng.c +52 -28
- data/ext/nokogiri/xml_sax_parser.c +118 -118
- data/ext/nokogiri/xml_sax_parser_context.c +103 -86
- data/ext/nokogiri/xml_sax_push_parser.c +36 -27
- data/ext/nokogiri/xml_schema.c +95 -47
- data/ext/nokogiri/xml_syntax_error.c +42 -21
- data/ext/nokogiri/xml_text.c +13 -17
- data/ext/nokogiri/xml_xpath_context.c +206 -123
- data/ext/nokogiri/xslt_stylesheet.c +158 -161
- data/lib/nokogiri.rb +3 -7
- data/lib/nokogiri/2.5/nokogiri.so +0 -0
- data/lib/nokogiri/2.6/nokogiri.so +0 -0
- data/lib/nokogiri/2.7/nokogiri.so +0 -0
- data/lib/nokogiri/3.0/nokogiri.so +0 -0
- data/lib/nokogiri/css/parser.rb +3 -3
- data/lib/nokogiri/css/parser.y +2 -2
- data/lib/nokogiri/css/xpath_visitor.rb +70 -42
- data/lib/nokogiri/extension.rb +26 -0
- data/lib/nokogiri/html/document.rb +12 -26
- data/lib/nokogiri/html/document_fragment.rb +15 -15
- data/lib/nokogiri/version.rb +2 -149
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +205 -0
- data/lib/nokogiri/xml/document.rb +91 -35
- data/lib/nokogiri/xml/document_fragment.rb +4 -6
- data/lib/nokogiri/xml/node.rb +89 -69
- data/lib/nokogiri/xml/parse_options.rb +6 -0
- data/lib/nokogiri/xml/reader.rb +2 -9
- data/lib/nokogiri/xml/relax_ng.rb +6 -2
- data/lib/nokogiri/xml/schema.rb +12 -4
- data/lib/nokogiri/xml/searchable.rb +3 -1
- data/lib/nokogiri/xml/xpath.rb +1 -3
- data/lib/nokogiri/xml/xpath/syntax_error.rb +1 -1
- metadata +150 -171
- data/ext/nokogiri/html_document.h +0 -10
- data/ext/nokogiri/html_element_description.h +0 -10
- data/ext/nokogiri/html_entity_lookup.h +0 -8
- data/ext/nokogiri/html_sax_parser_context.h +0 -11
- data/ext/nokogiri/html_sax_push_parser.h +0 -9
- data/ext/nokogiri/xml_attr.h +0 -9
- data/ext/nokogiri/xml_attribute_decl.h +0 -9
- data/ext/nokogiri/xml_cdata.h +0 -9
- data/ext/nokogiri/xml_comment.h +0 -9
- data/ext/nokogiri/xml_document.h +0 -23
- data/ext/nokogiri/xml_document_fragment.h +0 -10
- data/ext/nokogiri/xml_dtd.h +0 -10
- data/ext/nokogiri/xml_element_content.h +0 -10
- data/ext/nokogiri/xml_element_decl.h +0 -9
- data/ext/nokogiri/xml_encoding_handler.h +0 -8
- data/ext/nokogiri/xml_entity_decl.h +0 -10
- data/ext/nokogiri/xml_entity_reference.h +0 -9
- data/ext/nokogiri/xml_io.c +0 -63
- data/ext/nokogiri/xml_io.h +0 -11
- data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
- data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
- data/ext/nokogiri/xml_namespace.h +0 -14
- data/ext/nokogiri/xml_node.h +0 -13
- data/ext/nokogiri/xml_node_set.h +0 -12
- data/ext/nokogiri/xml_processing_instruction.h +0 -9
- data/ext/nokogiri/xml_reader.h +0 -10
- data/ext/nokogiri/xml_relax_ng.h +0 -9
- data/ext/nokogiri/xml_sax_parser.h +0 -39
- data/ext/nokogiri/xml_sax_parser_context.h +0 -10
- data/ext/nokogiri/xml_sax_push_parser.h +0 -9
- data/ext/nokogiri/xml_schema.h +0 -9
- data/ext/nokogiri/xml_syntax_error.h +0 -13
- data/ext/nokogiri/xml_text.h +0 -9
- data/ext/nokogiri/xml_xpath_context.h +0 -10
- data/ext/nokogiri/xslt_stylesheet.h +0 -14
- data/lib/nokogiri/2.4/nokogiri.so +0 -0
@@ -1,4 +1,6 @@
|
|
1
|
-
#include <
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
static VALUE cNokogiriHtmlEntityLookup;
|
2
4
|
|
3
5
|
/*
|
4
6
|
* call-seq:
|
@@ -6,27 +8,30 @@
|
|
6
8
|
*
|
7
9
|
* Get the HTML::EntityDescription for +key+
|
8
10
|
*/
|
9
|
-
static VALUE
|
11
|
+
static VALUE
|
12
|
+
get(VALUE _, VALUE rb_entity_name)
|
10
13
|
{
|
11
|
-
|
12
|
-
|
13
|
-
VALUE
|
14
|
+
VALUE cNokogiriHtmlEntityDescription;
|
15
|
+
const htmlEntityDesc *c_entity_desc;
|
16
|
+
VALUE rb_constructor_args[3];
|
14
17
|
|
15
|
-
|
16
|
-
|
18
|
+
c_entity_desc = htmlEntityLookup((const xmlChar *)StringValueCStr(rb_entity_name));
|
19
|
+
if (NULL == c_entity_desc) {
|
20
|
+
return Qnil;
|
21
|
+
}
|
17
22
|
|
18
|
-
|
19
|
-
|
20
|
-
|
23
|
+
rb_constructor_args[0] = INT2NUM((long)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);
|
21
26
|
|
22
|
-
|
27
|
+
cNokogiriHtmlEntityDescription = rb_const_get_at(mNokogiriHtml, rb_intern("EntityDescription"));
|
28
|
+
return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtmlEntityDescription);
|
23
29
|
}
|
24
30
|
|
25
|
-
void
|
31
|
+
void
|
32
|
+
noko_init_html_entity_lookup()
|
26
33
|
{
|
27
|
-
|
28
|
-
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
29
|
-
VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
|
34
|
+
cNokogiriHtmlEntityLookup = rb_define_class_under(mNokogiriHtml, "EntityLookup", rb_cObject);
|
30
35
|
|
31
|
-
rb_define_method(
|
36
|
+
rb_define_method(cNokogiriHtmlEntityLookup, "get", get, 1);
|
32
37
|
}
|
@@ -1,116 +1,119 @@
|
|
1
|
-
#include <
|
1
|
+
#include <nokogiri.h>
|
2
2
|
|
3
3
|
VALUE cNokogiriHtmlSaxParserContext ;
|
4
4
|
|
5
|
-
static void
|
5
|
+
static void
|
6
|
+
deallocate(xmlParserCtxtPtr ctxt)
|
6
7
|
{
|
7
|
-
NOKOGIRI_DEBUG_START(
|
8
|
+
NOKOGIRI_DEBUG_START(ctxt);
|
8
9
|
|
9
10
|
ctxt->sax = NULL;
|
10
11
|
|
11
12
|
htmlFreeParserCtxt(ctxt);
|
12
13
|
|
13
|
-
NOKOGIRI_DEBUG_END(
|
14
|
+
NOKOGIRI_DEBUG_END(ctxt);
|
14
15
|
}
|
15
16
|
|
16
17
|
static VALUE
|
17
18
|
parse_memory(VALUE klass, VALUE data, VALUE encoding)
|
18
19
|
{
|
19
|
-
|
20
|
-
|
21
|
-
|
22
|
-
|
23
|
-
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
28
|
-
|
29
|
-
|
30
|
-
|
20
|
+
htmlParserCtxtPtr ctxt;
|
21
|
+
|
22
|
+
if (NIL_P(data)) {
|
23
|
+
rb_raise(rb_eArgError, "data cannot be nil");
|
24
|
+
}
|
25
|
+
if (!(int)RSTRING_LEN(data)) {
|
26
|
+
rb_raise(rb_eRuntimeError, "data cannot be empty");
|
27
|
+
}
|
28
|
+
|
29
|
+
ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
|
30
|
+
(int)RSTRING_LEN(data));
|
31
|
+
if (ctxt->sax) {
|
32
|
+
xmlFree(ctxt->sax);
|
33
|
+
ctxt->sax = NULL;
|
34
|
+
}
|
35
|
+
|
36
|
+
if (RTEST(encoding)) {
|
37
|
+
xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
|
38
|
+
if (enc != NULL) {
|
39
|
+
xmlSwitchToEncoding(ctxt, enc);
|
40
|
+
if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
|
41
|
+
rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
|
42
|
+
StringValueCStr(encoding));
|
43
|
+
}
|
31
44
|
}
|
45
|
+
}
|
32
46
|
|
33
|
-
|
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);
|
47
|
+
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
45
48
|
}
|
46
49
|
|
47
|
-
static VALUE
|
50
|
+
static VALUE
|
51
|
+
parse_file(VALUE klass, VALUE filename, VALUE encoding)
|
48
52
|
{
|
49
53
|
htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
|
50
|
-
|
51
|
-
|
52
|
-
|
54
|
+
StringValueCStr(filename),
|
55
|
+
StringValueCStr(encoding)
|
56
|
+
);
|
53
57
|
return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
|
54
58
|
}
|
55
59
|
|
56
60
|
static VALUE
|
57
61
|
parse_doc(VALUE ctxt_val)
|
58
62
|
{
|
59
|
-
|
60
|
-
|
61
|
-
|
63
|
+
htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
|
64
|
+
htmlParseDocument(ctxt);
|
65
|
+
return Qnil;
|
62
66
|
}
|
63
67
|
|
64
68
|
static VALUE
|
65
69
|
parse_doc_finalize(VALUE ctxt_val)
|
66
70
|
{
|
67
|
-
|
71
|
+
htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
|
68
72
|
|
69
|
-
|
70
|
-
|
73
|
+
if (ctxt->myDoc) {
|
74
|
+
xmlFreeDoc(ctxt->myDoc);
|
75
|
+
}
|
71
76
|
|
72
|
-
|
73
|
-
|
77
|
+
NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
|
78
|
+
return Qnil;
|
74
79
|
}
|
75
80
|
|
76
81
|
static VALUE
|
77
82
|
parse_with(VALUE self, VALUE sax_handler)
|
78
83
|
{
|
79
|
-
|
80
|
-
|
84
|
+
htmlParserCtxtPtr ctxt;
|
85
|
+
htmlSAXHandlerPtr sax;
|
86
|
+
|
87
|
+
if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser)) {
|
88
|
+
rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
|
89
|
+
}
|
81
90
|
|
82
|
-
|
83
|
-
|
91
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
92
|
+
Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
|
84
93
|
|
85
|
-
|
86
|
-
|
94
|
+
/* Free the sax handler since we'll assign our own */
|
95
|
+
if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
|
96
|
+
xmlFree(ctxt->sax);
|
97
|
+
}
|
87
98
|
|
88
|
-
|
89
|
-
|
90
|
-
xmlFree(ctxt->sax);
|
99
|
+
ctxt->sax = sax;
|
100
|
+
ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
|
91
101
|
|
92
|
-
|
93
|
-
ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
|
102
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
94
103
|
|
95
|
-
|
104
|
+
rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
|
96
105
|
|
97
|
-
|
106
|
+
return self;
|
98
107
|
}
|
99
108
|
|
100
|
-
void
|
109
|
+
void
|
110
|
+
noko_init_html_sax_parser_context()
|
101
111
|
{
|
102
|
-
|
103
|
-
|
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;
|
112
|
+
assert(cNokogiriXmlSaxParserContext);
|
113
|
+
cNokogiriHtmlSaxParserContext = rb_define_class_under(mNokogiriHtmlSax, "ParserContext", cNokogiriXmlSaxParserContext);
|
111
114
|
|
112
|
-
rb_define_singleton_method(
|
113
|
-
rb_define_singleton_method(
|
115
|
+
rb_define_singleton_method(cNokogiriHtmlSaxParserContext, "memory", parse_memory, 2);
|
116
|
+
rb_define_singleton_method(cNokogiriHtmlSaxParserContext, "file", parse_file, 2);
|
114
117
|
|
115
|
-
rb_define_method(
|
118
|
+
rb_define_method(cNokogiriHtmlSaxParserContext, "parse_with", parse_with, 1);
|
116
119
|
}
|
@@ -1,4 +1,6 @@
|
|
1
|
-
#include <
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtmlSaxPushParser;
|
2
4
|
|
3
5
|
/*
|
4
6
|
* call-seq:
|
@@ -6,25 +8,32 @@
|
|
6
8
|
*
|
7
9
|
* Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
|
8
10
|
*/
|
9
|
-
static VALUE
|
11
|
+
static VALUE
|
12
|
+
native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
|
10
13
|
{
|
11
14
|
xmlParserCtxtPtr ctx;
|
12
|
-
const char *
|
13
|
-
int size
|
14
|
-
|
15
|
+
const char *chunk = NULL;
|
16
|
+
int size = 0;
|
17
|
+
int status = 0;
|
18
|
+
libxmlStructuredErrorHandlerState handler_state;
|
15
19
|
|
16
20
|
Data_Get_Struct(self, xmlParserCtxt, ctx);
|
17
21
|
|
18
|
-
if(Qnil != _chunk) {
|
22
|
+
if (Qnil != _chunk) {
|
19
23
|
chunk = StringValuePtr(_chunk);
|
20
24
|
size = (int)RSTRING_LEN(_chunk);
|
21
25
|
}
|
22
26
|
|
23
|
-
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
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);
|
28
37
|
}
|
29
38
|
|
30
39
|
return self;
|
@@ -36,34 +45,37 @@ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
|
|
36
45
|
*
|
37
46
|
* Initialize the push parser with +xml_sax+ using +filename+
|
38
47
|
*/
|
39
|
-
static VALUE
|
40
|
-
|
48
|
+
static VALUE
|
49
|
+
initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
|
50
|
+
VALUE encoding)
|
41
51
|
{
|
42
52
|
htmlSAXHandlerPtr sax;
|
43
|
-
const char *
|
53
|
+
const char *filename = NULL;
|
44
54
|
htmlParserCtxtPtr ctx;
|
45
55
|
xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
|
46
56
|
|
47
57
|
Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
|
48
58
|
|
49
|
-
if(_filename != Qnil) filename = StringValueCStr(_filename);
|
59
|
+
if (_filename != Qnil) { filename = StringValueCStr(_filename); }
|
50
60
|
|
51
61
|
if (!NIL_P(encoding)) {
|
52
62
|
enc = xmlParseCharEncoding(StringValueCStr(encoding));
|
53
|
-
if (enc == XML_CHAR_ENCODING_ERROR)
|
63
|
+
if (enc == XML_CHAR_ENCODING_ERROR) {
|
54
64
|
rb_raise(rb_eArgError, "Unsupported Encoding");
|
65
|
+
}
|
55
66
|
}
|
56
67
|
|
57
68
|
ctx = htmlCreatePushParserCtxt(
|
58
|
-
|
59
|
-
|
60
|
-
|
61
|
-
|
62
|
-
|
63
|
-
|
64
|
-
|
65
|
-
if(ctx == NULL)
|
69
|
+
sax,
|
70
|
+
NULL,
|
71
|
+
NULL,
|
72
|
+
0,
|
73
|
+
filename,
|
74
|
+
enc
|
75
|
+
);
|
76
|
+
if (ctx == NULL) {
|
66
77
|
rb_raise(rb_eRuntimeError, "Could not create a parser context");
|
78
|
+
}
|
67
79
|
|
68
80
|
ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
|
69
81
|
|
@@ -72,16 +84,12 @@ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
|
|
72
84
|
return self;
|
73
85
|
}
|
74
86
|
|
75
|
-
|
76
|
-
|
87
|
+
void
|
88
|
+
noko_init_html_sax_push_parser()
|
77
89
|
{
|
78
|
-
|
79
|
-
|
80
|
-
VALUE sax = rb_define_module_under(html, "SAX");
|
81
|
-
VALUE klass = rb_define_class_under(sax, "PushParser", cNokogiriXmlSaxPushParser);
|
82
|
-
|
83
|
-
cNokogiriHtmlSaxPushParser = klass;
|
90
|
+
assert(cNokogiriXmlSaxPushParser);
|
91
|
+
cNokogiriHtmlSaxPushParser = rb_define_class_under(mNokogiriHtmlSax, "PushParser", cNokogiriXmlSaxPushParser);
|
84
92
|
|
85
|
-
rb_define_private_method(
|
86
|
-
rb_define_private_method(
|
93
|
+
rb_define_private_method(cNokogiriHtmlSaxPushParser, "initialize_native", initialize_native, 3);
|
94
|
+
rb_define_private_method(cNokogiriHtmlSaxPushParser, "native_write", native_write, 2);
|
87
95
|
}
|
@@ -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
|
+
|