nokogiri 1.10.0 → 1.11.1
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/README.md +178 -90
- data/dependencies.yml +28 -26
- data/ext/nokogiri/depend +476 -357
- data/ext/nokogiri/extconf.rb +508 -354
- data/ext/nokogiri/html_document.c +79 -78
- data/ext/nokogiri/html_sax_parser_context.c +4 -2
- data/ext/nokogiri/html_sax_push_parser.c +14 -8
- data/ext/nokogiri/nokogiri.c +37 -40
- data/ext/nokogiri/nokogiri.h +26 -17
- data/ext/nokogiri/test_global_handlers.c +41 -0
- data/ext/nokogiri/xml_document.c +18 -4
- data/ext/nokogiri/xml_io.c +8 -6
- data/ext/nokogiri/xml_node.c +23 -6
- data/ext/nokogiri/xml_node_set.c +1 -1
- data/ext/nokogiri/xml_reader.c +6 -17
- data/ext/nokogiri/xml_relax_ng.c +29 -11
- data/ext/nokogiri/xml_sax_parser.c +2 -7
- data/ext/nokogiri/xml_sax_parser_context.c +4 -2
- data/ext/nokogiri/xml_sax_push_parser.c +2 -0
- data/ext/nokogiri/xml_schema.c +84 -13
- data/ext/nokogiri/xml_syntax_error.c +23 -0
- data/ext/nokogiri/xml_syntax_error.h +15 -3
- data/ext/nokogiri/xml_xpath_context.c +80 -4
- data/ext/nokogiri/xslt_stylesheet.c +2 -9
- data/lib/nokogiri.rb +22 -22
- data/lib/nokogiri/css.rb +1 -0
- data/lib/nokogiri/css/node.rb +1 -0
- data/lib/nokogiri/css/parser.rb +63 -62
- data/lib/nokogiri/css/parser.y +2 -2
- data/lib/nokogiri/css/parser_extras.rb +39 -36
- data/lib/nokogiri/css/syntax_error.rb +1 -0
- data/lib/nokogiri/css/tokenizer.rb +105 -103
- data/lib/nokogiri/css/xpath_visitor.rb +73 -43
- data/lib/nokogiri/decorators/slop.rb +1 -0
- data/lib/nokogiri/html.rb +1 -0
- data/lib/nokogiri/html/builder.rb +1 -0
- data/lib/nokogiri/html/document.rb +13 -26
- data/lib/nokogiri/html/document_fragment.rb +1 -0
- data/lib/nokogiri/html/element_description.rb +1 -0
- data/lib/nokogiri/html/element_description_defaults.rb +1 -0
- data/lib/nokogiri/html/entity_lookup.rb +1 -0
- data/lib/nokogiri/html/sax/parser.rb +1 -0
- data/lib/nokogiri/html/sax/parser_context.rb +1 -0
- data/lib/nokogiri/html/sax/push_parser.rb +1 -0
- data/lib/nokogiri/jruby/dependencies.rb +20 -0
- data/lib/nokogiri/syntax_error.rb +1 -0
- data/lib/nokogiri/version.rb +3 -109
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +182 -0
- data/lib/nokogiri/xml.rb +1 -0
- data/lib/nokogiri/xml/attr.rb +1 -0
- data/lib/nokogiri/xml/attribute_decl.rb +1 -0
- data/lib/nokogiri/xml/builder.rb +36 -32
- data/lib/nokogiri/xml/cdata.rb +1 -0
- data/lib/nokogiri/xml/character_data.rb +1 -0
- data/lib/nokogiri/xml/document.rb +21 -16
- data/lib/nokogiri/xml/document_fragment.rb +5 -6
- data/lib/nokogiri/xml/dtd.rb +1 -0
- data/lib/nokogiri/xml/element_content.rb +1 -0
- data/lib/nokogiri/xml/element_decl.rb +1 -0
- data/lib/nokogiri/xml/entity_decl.rb +1 -0
- data/lib/nokogiri/xml/entity_reference.rb +1 -0
- data/lib/nokogiri/xml/namespace.rb +1 -0
- data/lib/nokogiri/xml/node.rb +588 -250
- data/lib/nokogiri/xml/node/save_options.rb +1 -0
- data/lib/nokogiri/xml/node_set.rb +43 -8
- data/lib/nokogiri/xml/notation.rb +1 -0
- data/lib/nokogiri/xml/parse_options.rb +10 -3
- data/lib/nokogiri/xml/pp.rb +1 -0
- data/lib/nokogiri/xml/pp/character_data.rb +1 -0
- data/lib/nokogiri/xml/pp/node.rb +1 -0
- data/lib/nokogiri/xml/processing_instruction.rb +1 -0
- data/lib/nokogiri/xml/reader.rb +7 -3
- data/lib/nokogiri/xml/relax_ng.rb +7 -2
- data/lib/nokogiri/xml/sax.rb +1 -0
- data/lib/nokogiri/xml/sax/document.rb +1 -0
- data/lib/nokogiri/xml/sax/parser.rb +1 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +1 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +1 -0
- data/lib/nokogiri/xml/schema.rb +13 -4
- data/lib/nokogiri/xml/searchable.rb +25 -16
- data/lib/nokogiri/xml/syntax_error.rb +1 -0
- data/lib/nokogiri/xml/text.rb +1 -0
- data/lib/nokogiri/xml/xpath.rb +1 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +1 -0
- data/lib/nokogiri/xml/xpath_context.rb +1 -0
- data/lib/nokogiri/xslt.rb +1 -0
- data/lib/nokogiri/xslt/stylesheet.rb +1 -0
- data/lib/xsd/xmlparser/nokogiri.rb +1 -0
- data/patches/libxml2/0002-Remove-script-macro-support.patch +40 -0
- data/patches/libxml2/0003-Update-entities-to-remove-handling-of-ssi.patch +44 -0
- data/patches/libxml2/0004-libxml2.la-is-in-top_builddir.patch +25 -0
- data/patches/libxml2/0005-Fix-infinite-loop-in-xmlStringLenDecodeEntities.patch +32 -0
- data/patches/libxml2/0006-htmlParseComment-treat-as-if-it-closed-the-comment.patch +73 -0
- data/patches/libxml2/0007-use-new-htmlParseLookupCommentEnd-to-find-comment-en.patch +103 -0
- data/patches/libxml2/0008-use-glibc-strlen.patch +53 -0
- data/patches/libxml2/0009-avoid-isnan-isinf.patch +81 -0
- data/ports/archives/libxml2-2.9.10.tar.gz +0 -0
- data/ports/archives/libxslt-1.1.34.tar.gz +0 -0
- metadata +106 -99
- data/ports/archives/libxml2-2.9.9.tar.gz +0 -0
- data/ports/archives/libxslt-1.1.33.tar.gz +0 -0
@@ -1,6 +1,7 @@
|
|
1
1
|
#include <html_document.h>
|
2
2
|
|
3
3
|
static ID id_encoding_found;
|
4
|
+
static ID id_to_s;
|
4
5
|
|
5
6
|
/*
|
6
7
|
* call-seq:
|
@@ -8,19 +9,20 @@ static ID id_encoding_found;
|
|
8
9
|
*
|
9
10
|
* Create a new document
|
10
11
|
*/
|
11
|
-
static VALUE
|
12
|
+
static VALUE
|
13
|
+
rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
|
12
14
|
{
|
13
15
|
VALUE uri, external_id, rest, rb_doc;
|
14
16
|
htmlDocPtr doc;
|
15
17
|
|
16
18
|
rb_scan_args(argc, argv, "0*", &rest);
|
17
|
-
uri
|
19
|
+
uri = rb_ary_entry(rest, (long)0);
|
18
20
|
external_id = rb_ary_entry(rest, (long)1);
|
19
21
|
|
20
22
|
doc = htmlNewDoc(
|
21
|
-
|
22
|
-
|
23
|
-
|
23
|
+
RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
|
24
|
+
RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
|
25
|
+
);
|
24
26
|
rb_doc = Nokogiri_wrap_xml_document(klass, doc);
|
25
27
|
rb_obj_call_init(rb_doc, argc, argv);
|
26
28
|
return rb_doc ;
|
@@ -33,60 +35,55 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
|
|
33
35
|
* Read the HTML document from +io+ with given +url+, +encoding+,
|
34
36
|
* and +options+. See Nokogiri::HTML.parse
|
35
37
|
*/
|
36
|
-
static VALUE
|
37
|
-
|
38
|
-
VALUE url,
|
39
|
-
VALUE encoding,
|
40
|
-
VALUE options )
|
38
|
+
static VALUE
|
39
|
+
rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
|
41
40
|
{
|
42
|
-
|
43
|
-
|
44
|
-
|
45
|
-
|
46
|
-
|
41
|
+
VALUE rb_doc;
|
42
|
+
VALUE rb_error_list = rb_ary_new();
|
43
|
+
htmlDocPtr c_doc;
|
44
|
+
const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
|
45
|
+
const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
|
46
|
+
int options = NUM2INT(rb_options);
|
47
|
+
|
48
|
+
xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
|
49
|
+
|
50
|
+
c_doc = htmlReadIO(io_read_callback, io_close_callback, (void *)rb_io, c_url, c_encoding, options);
|
47
51
|
|
48
|
-
xmlResetLastError();
|
49
|
-
xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
|
50
|
-
|
51
|
-
doc = htmlReadIO(
|
52
|
-
io_read_callback,
|
53
|
-
io_close_callback,
|
54
|
-
(void *)io,
|
55
|
-
c_url,
|
56
|
-
c_enc,
|
57
|
-
(int)NUM2INT(options)
|
58
|
-
);
|
59
52
|
xmlSetStructuredErrorFunc(NULL, NULL);
|
60
53
|
|
61
54
|
/*
|
62
55
|
* If EncodingFound has occurred in EncodingReader, make sure to do
|
63
56
|
* a cleanup and propagate the error.
|
64
57
|
*/
|
65
|
-
if (rb_respond_to(
|
66
|
-
VALUE encoding_found = rb_funcall(
|
58
|
+
if (rb_respond_to(rb_io, id_encoding_found)) {
|
59
|
+
VALUE encoding_found = rb_funcall(rb_io, id_encoding_found, 0);
|
67
60
|
if (!NIL_P(encoding_found)) {
|
68
|
-
xmlFreeDoc(
|
61
|
+
xmlFreeDoc(c_doc);
|
69
62
|
rb_exc_raise(encoding_found);
|
70
63
|
}
|
71
64
|
}
|
72
65
|
|
73
|
-
if(
|
74
|
-
|
66
|
+
if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
|
67
|
+
VALUE rb_error ;
|
75
68
|
|
76
|
-
xmlFreeDoc(
|
69
|
+
xmlFreeDoc(c_doc);
|
77
70
|
|
78
|
-
|
79
|
-
if(
|
80
|
-
rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
|
81
|
-
else
|
71
|
+
rb_error = rb_ary_entry(rb_error_list, 0);
|
72
|
+
if (rb_error == Qnil) {
|
82
73
|
rb_raise(rb_eRuntimeError, "Could not parse document");
|
74
|
+
} else {
|
75
|
+
VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
|
76
|
+
exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
|
77
|
+
exception_message);
|
78
|
+
rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
|
79
|
+
}
|
83
80
|
|
84
81
|
return Qnil;
|
85
82
|
}
|
86
83
|
|
87
|
-
|
88
|
-
rb_iv_set(
|
89
|
-
return
|
84
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
|
85
|
+
rb_iv_set(rb_doc, "@errors", rb_error_list);
|
86
|
+
return rb_doc;
|
90
87
|
}
|
91
88
|
|
92
89
|
/*
|
@@ -96,43 +93,45 @@ static VALUE read_io( VALUE klass,
|
|
96
93
|
* Read the HTML document contained in +string+ with given +url+, +encoding+,
|
97
94
|
* and +options+. See Nokogiri::HTML.parse
|
98
95
|
*/
|
99
|
-
static VALUE
|
100
|
-
|
101
|
-
VALUE url,
|
102
|
-
VALUE encoding,
|
103
|
-
VALUE options )
|
96
|
+
static VALUE
|
97
|
+
rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
|
104
98
|
{
|
105
|
-
|
106
|
-
|
107
|
-
|
108
|
-
|
109
|
-
|
110
|
-
|
111
|
-
|
99
|
+
VALUE rb_doc;
|
100
|
+
VALUE rb_error_list = rb_ary_new();
|
101
|
+
htmlDocPtr c_doc;
|
102
|
+
const char *c_buffer = StringValuePtr(rb_html);
|
103
|
+
const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
|
104
|
+
const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
|
105
|
+
int html_len = (int)RSTRING_LEN(rb_html);
|
106
|
+
int options = NUM2INT(rb_options);
|
112
107
|
|
113
|
-
|
114
|
-
|
108
|
+
xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
|
109
|
+
|
110
|
+
c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
|
115
111
|
|
116
|
-
doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
|
117
112
|
xmlSetStructuredErrorFunc(NULL, NULL);
|
118
113
|
|
119
|
-
if(
|
120
|
-
|
114
|
+
if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
|
115
|
+
VALUE rb_error ;
|
121
116
|
|
122
|
-
xmlFreeDoc(
|
117
|
+
xmlFreeDoc(c_doc);
|
123
118
|
|
124
|
-
|
125
|
-
if(
|
126
|
-
rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
|
127
|
-
else
|
119
|
+
rb_error = rb_ary_entry(rb_error_list, 0);
|
120
|
+
if (rb_error == Qnil) {
|
128
121
|
rb_raise(rb_eRuntimeError, "Could not parse document");
|
122
|
+
} else {
|
123
|
+
VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
|
124
|
+
exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
|
125
|
+
exception_message);
|
126
|
+
rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
|
127
|
+
}
|
129
128
|
|
130
129
|
return Qnil;
|
131
130
|
}
|
132
131
|
|
133
|
-
|
134
|
-
rb_iv_set(
|
135
|
-
return
|
132
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
|
133
|
+
rb_iv_set(rb_doc, "@errors", rb_error_list);
|
134
|
+
return rb_doc;
|
136
135
|
}
|
137
136
|
|
138
137
|
/*
|
@@ -141,7 +140,8 @@ static VALUE read_memory( VALUE klass,
|
|
141
140
|
*
|
142
141
|
* The type for this document
|
143
142
|
*/
|
144
|
-
static VALUE
|
143
|
+
static VALUE
|
144
|
+
rb_html_document_type(VALUE self)
|
145
145
|
{
|
146
146
|
htmlDocPtr doc;
|
147
147
|
Data_Get_Struct(self, xmlDoc, doc);
|
@@ -149,22 +149,23 @@ static VALUE type(VALUE self)
|
|
149
149
|
}
|
150
150
|
|
151
151
|
VALUE cNokogiriHtmlDocument ;
|
152
|
-
void init_html_document()
|
153
|
-
{
|
154
|
-
VALUE nokogiri = rb_define_module("Nokogiri");
|
155
|
-
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
156
|
-
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
157
|
-
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
158
|
-
VALUE xml_doc = rb_define_class_under(xml, "Document", node);
|
159
|
-
VALUE klass = rb_define_class_under(html, "Document", xml_doc);
|
160
152
|
|
161
|
-
|
153
|
+
void
|
154
|
+
init_html_document()
|
155
|
+
{
|
156
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
157
|
+
VALUE nokogiri_xml = rb_define_module_under(nokogiri, "XML");
|
158
|
+
VALUE nokogiri_xml_node = rb_define_class_under(nokogiri_xml, "Node", rb_cObject);
|
159
|
+
VALUE nokogiri_xml_document = rb_define_class_under(nokogiri_xml, "Document", nokogiri_xml_node);
|
160
|
+
VALUE nokogiri_html = rb_define_module_under(nokogiri, "HTML");
|
161
|
+
cNokogiriHtmlDocument = rb_define_class_under(nokogiri_html, "Document", nokogiri_xml_document);
|
162
162
|
|
163
|
-
rb_define_singleton_method(
|
164
|
-
rb_define_singleton_method(
|
165
|
-
rb_define_singleton_method(
|
163
|
+
rb_define_singleton_method(cNokogiriHtmlDocument, "read_memory", rb_html_document_s_read_memory, 4);
|
164
|
+
rb_define_singleton_method(cNokogiriHtmlDocument, "read_io", rb_html_document_s_read_io, 4);
|
165
|
+
rb_define_singleton_method(cNokogiriHtmlDocument, "new", rb_html_document_s_new, -1);
|
166
166
|
|
167
|
-
rb_define_method(
|
167
|
+
rb_define_method(cNokogiriHtmlDocument, "type", rb_html_document_type, 0);
|
168
168
|
|
169
169
|
id_encoding_found = rb_intern("encoding_found");
|
170
|
+
id_to_s = rb_intern("to_s");
|
170
171
|
}
|
@@ -4,13 +4,13 @@ VALUE cNokogiriHtmlSaxParserContext ;
|
|
4
4
|
|
5
5
|
static void deallocate(xmlParserCtxtPtr ctxt)
|
6
6
|
{
|
7
|
-
NOKOGIRI_DEBUG_START(
|
7
|
+
NOKOGIRI_DEBUG_START(ctxt);
|
8
8
|
|
9
9
|
ctxt->sax = NULL;
|
10
10
|
|
11
11
|
htmlFreeParserCtxt(ctxt);
|
12
12
|
|
13
|
-
NOKOGIRI_DEBUG_END(
|
13
|
+
NOKOGIRI_DEBUG_END(ctxt);
|
14
14
|
}
|
15
15
|
|
16
16
|
static VALUE
|
@@ -92,6 +92,8 @@ parse_with(VALUE self, VALUE sax_handler)
|
|
92
92
|
ctxt->sax = sax;
|
93
93
|
ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
|
94
94
|
|
95
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
96
|
+
|
95
97
|
rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
|
96
98
|
|
97
99
|
return self;
|
@@ -9,9 +9,10 @@
|
|
9
9
|
static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
|
10
10
|
{
|
11
11
|
xmlParserCtxtPtr ctx;
|
12
|
-
const char * chunk
|
13
|
-
int size
|
14
|
-
|
12
|
+
const char * chunk = NULL;
|
13
|
+
int size = 0;
|
14
|
+
int status = 0;
|
15
|
+
libxmlStructuredErrorHandlerState handler_state;
|
15
16
|
|
16
17
|
Data_Get_Struct(self, xmlParserCtxt, ctx);
|
17
18
|
|
@@ -20,11 +21,16 @@ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
|
|
20
21
|
size = (int)RSTRING_LEN(_chunk);
|
21
22
|
}
|
22
23
|
|
23
|
-
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
24
|
+
Nokogiri_structured_error_func_save_and_set(&handler_state, NULL, NULL);
|
25
|
+
|
26
|
+
status = htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0);
|
27
|
+
|
28
|
+
Nokogiri_structured_error_func_restore(&handler_state);
|
29
|
+
|
30
|
+
if ((status != 0) && !(ctx->options & XML_PARSE_RECOVER)) {
|
31
|
+
// TODO: there appear to be no tests for this block
|
32
|
+
xmlErrorPtr e = xmlCtxtGetLastError(ctx);
|
33
|
+
Nokogiri_error_raise(NULL, e);
|
28
34
|
}
|
29
35
|
|
30
36
|
return self;
|
data/ext/nokogiri/nokogiri.c
CHANGED
@@ -1,5 +1,7 @@
|
|
1
1
|
#include <nokogiri.h>
|
2
2
|
|
3
|
+
void init_test_global_handlers(); /* in lieu of test_global_handlers.h */
|
4
|
+
|
3
5
|
VALUE mNokogiri ;
|
4
6
|
VALUE mNokogiriXml ;
|
5
7
|
VALUE mNokogiriHtml ;
|
@@ -7,12 +9,11 @@ VALUE mNokogiriXslt ;
|
|
7
9
|
VALUE mNokogiriXmlSax ;
|
8
10
|
VALUE mNokogiriHtmlSax ;
|
9
11
|
|
10
|
-
#
|
12
|
+
#ifndef HAVE_VASPRINTF
|
11
13
|
/*
|
12
|
-
* I srsly hate windows. it doesn't have vasprintf.
|
13
14
|
* Thank you Geoffroy Couprie for this implementation of vasprintf!
|
14
15
|
*/
|
15
|
-
int vasprintf
|
16
|
+
int vasprintf(char **strp, const char *fmt, va_list ap)
|
16
17
|
{
|
17
18
|
/* Mingw32/64 have a broken vsnprintf implementation that fails when
|
18
19
|
* using a zero-byte limit in order to retrieve the required size for malloc.
|
@@ -21,23 +22,15 @@ int vasprintf (char **strp, const char *fmt, va_list ap)
|
|
21
22
|
char tmp[1];
|
22
23
|
int len = vsnprintf (tmp, 1, fmt, ap) + 1;
|
23
24
|
char *res = (char *)malloc((unsigned int)len);
|
24
|
-
if (res == NULL)
|
25
|
-
|
25
|
+
if (res == NULL) {
|
26
|
+
return -1;
|
27
|
+
}
|
26
28
|
*strp = res;
|
27
29
|
return vsnprintf(res, (unsigned int)len, fmt, ap);
|
28
30
|
}
|
29
31
|
#endif
|
30
32
|
|
31
|
-
void vasprintf_free (void *p)
|
32
|
-
{
|
33
|
-
free(p);
|
34
|
-
}
|
35
|
-
|
36
|
-
#ifdef HAVE_RUBY_UTIL_H
|
37
33
|
#include "ruby/util.h"
|
38
|
-
#else
|
39
|
-
#include "util.h"
|
40
|
-
#endif
|
41
34
|
|
42
35
|
void nokogiri_root_node(xmlNodePtr node)
|
43
36
|
{
|
@@ -45,7 +38,7 @@ void nokogiri_root_node(xmlNodePtr node)
|
|
45
38
|
nokogiriTuplePtr tuple;
|
46
39
|
|
47
40
|
doc = node->doc;
|
48
|
-
if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
|
41
|
+
if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
|
49
42
|
tuple = (nokogiriTuplePtr)doc->_private;
|
50
43
|
st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
|
51
44
|
}
|
@@ -54,7 +47,7 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
|
|
54
47
|
{
|
55
48
|
nokogiriTuplePtr tuple;
|
56
49
|
|
57
|
-
if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
|
50
|
+
if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
|
58
51
|
tuple = (nokogiriTuplePtr)doc->_private;
|
59
52
|
st_insert(tuple->unlinkedNodes, (st_data_t)ns, (st_data_t)ns);
|
60
53
|
}
|
@@ -62,10 +55,10 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
|
|
62
55
|
void Init_nokogiri()
|
63
56
|
{
|
64
57
|
xmlMemSetup(
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
58
|
+
(xmlFreeFunc)ruby_xfree,
|
59
|
+
(xmlMallocFunc)ruby_xmalloc,
|
60
|
+
(xmlReallocFunc)ruby_xrealloc,
|
61
|
+
ruby_strdup
|
69
62
|
);
|
70
63
|
|
71
64
|
mNokogiri = rb_define_module("Nokogiri");
|
@@ -75,27 +68,26 @@ void Init_nokogiri()
|
|
75
68
|
mNokogiriXmlSax = rb_define_module_under(mNokogiriXml, "SAX");
|
76
69
|
mNokogiriHtmlSax = rb_define_module_under(mNokogiriHtml, "SAX");
|
77
70
|
|
78
|
-
rb_const_set(
|
79
|
-
|
80
|
-
|
81
|
-
|
82
|
-
rb_const_set(
|
83
|
-
|
84
|
-
|
85
|
-
|
86
|
-
|
87
|
-
|
88
|
-
|
89
|
-
rb_const_set(mNokogiri, rb_intern("
|
90
|
-
|
91
|
-
rb_const_set(mNokogiri, rb_intern("
|
92
|
-
rb_const_set(mNokogiri, rb_intern("
|
71
|
+
rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
|
72
|
+
rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
|
73
|
+
|
74
|
+
rb_const_set(mNokogiri, rb_intern("LIBXSLT_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXSLT_DOTTED_VERSION));
|
75
|
+
rb_const_set(mNokogiri, rb_intern("LIBXSLT_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xsltEngineVersion));
|
76
|
+
|
77
|
+
#ifdef NOKOGIRI_PACKAGED_LIBRARIES
|
78
|
+
rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qtrue);
|
79
|
+
#ifdef NOKOGIRI_PRECOMPILED_LIBRARIES
|
80
|
+
rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qtrue);
|
81
|
+
#else
|
82
|
+
rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
|
83
|
+
#endif
|
84
|
+
rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
|
85
|
+
rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
|
93
86
|
#else
|
94
|
-
rb_const_set(mNokogiri, rb_intern("
|
95
|
-
rb_const_set(mNokogiri, rb_intern("
|
96
|
-
rb_const_set(mNokogiri, rb_intern("
|
97
|
-
rb_const_set(mNokogiri, rb_intern("
|
98
|
-
rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), Qnil);
|
87
|
+
rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qfalse);
|
88
|
+
rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
|
89
|
+
rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), Qnil);
|
90
|
+
rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), Qnil);
|
99
91
|
#endif
|
100
92
|
|
101
93
|
#ifdef LIBXML_ICONV_ENABLED
|
@@ -104,6 +96,10 @@ void Init_nokogiri()
|
|
104
96
|
rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
|
105
97
|
#endif
|
106
98
|
|
99
|
+
#ifdef NOKOGIRI_OTHER_LIBRARY_VERSIONS
|
100
|
+
rb_const_set(mNokogiri, rb_intern("OTHER_LIBRARY_VERSIONS"), NOKOGIRI_STR_NEW2(NOKOGIRI_OTHER_LIBRARY_VERSIONS));
|
101
|
+
#endif
|
102
|
+
|
107
103
|
xmlInitParser();
|
108
104
|
|
109
105
|
init_xml_document();
|
@@ -138,4 +134,5 @@ void Init_nokogiri()
|
|
138
134
|
init_xml_relax_ng();
|
139
135
|
init_nokogiri_io();
|
140
136
|
init_xml_encoding_handler();
|
137
|
+
init_test_global_handlers();
|
141
138
|
}
|
data/ext/nokogiri/nokogiri.h
CHANGED
@@ -1,20 +1,25 @@
|
|
1
1
|
#ifndef NOKOGIRI_NATIVE
|
2
2
|
#define NOKOGIRI_NATIVE
|
3
3
|
|
4
|
+
#if _MSC_VER
|
5
|
+
# ifndef WIN32_LEAN_AND_MEAN
|
6
|
+
# define WIN32_LEAN_AND_MEAN
|
7
|
+
# endif /* WIN32_LEAN_AND_MEAN */
|
8
|
+
|
9
|
+
# ifndef WIN32
|
10
|
+
# define WIN32
|
11
|
+
# endif /* WIN32 */
|
12
|
+
|
13
|
+
# include <winsock2.h>
|
14
|
+
# include <ws2tcpip.h>
|
15
|
+
# include <windows.h>
|
16
|
+
#endif
|
17
|
+
|
4
18
|
#include <stdlib.h>
|
5
19
|
#include <string.h>
|
6
20
|
#include <assert.h>
|
7
21
|
#include <stdarg.h>
|
8
|
-
|
9
|
-
#ifdef USE_INCLUDED_VASPRINTF
|
10
|
-
int vasprintf (char **strp, const char *fmt, va_list ap);
|
11
|
-
#else
|
12
|
-
|
13
|
-
#define _GNU_SOURCE
|
14
|
-
# include <stdio.h>
|
15
|
-
#undef _GNU_SOURCE
|
16
|
-
|
17
|
-
#endif
|
22
|
+
#include <stdio.h>
|
18
23
|
|
19
24
|
#include <libxml/parser.h>
|
20
25
|
#include <libxml/entities.h>
|
@@ -29,7 +34,9 @@ int vasprintf (char **strp, const char *fmt, va_list ap);
|
|
29
34
|
#include <libxml/relaxng.h>
|
30
35
|
#include <libxml/xinclude.h>
|
31
36
|
#include <libxslt/extensions.h>
|
37
|
+
#include <libxslt/xsltconfig.h>
|
32
38
|
#include <libxml/c14n.h>
|
39
|
+
|
33
40
|
#include <ruby.h>
|
34
41
|
#include <ruby/st.h>
|
35
42
|
#include <ruby/encoding.h>
|
@@ -93,25 +100,27 @@ extern VALUE mNokogiriHtml ;
|
|
93
100
|
extern VALUE mNokogiriHtmlSax ;
|
94
101
|
extern VALUE mNokogiriXslt ;
|
95
102
|
|
103
|
+
int vasprintf(char **strp, const char *fmt, va_list ap);
|
104
|
+
|
96
105
|
void nokogiri_root_node(xmlNodePtr);
|
97
106
|
void nokogiri_root_nsdef(xmlNsPtr, xmlDocPtr);
|
98
107
|
|
99
108
|
#ifdef DEBUG
|
100
109
|
|
101
|
-
#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);
|
102
|
-
#define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
|
110
|
+
# 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);
|
111
|
+
# define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
|
103
112
|
|
104
113
|
#else
|
105
114
|
|
106
|
-
#define NOKOGIRI_DEBUG_START(p)
|
107
|
-
#define NOKOGIRI_DEBUG_END(p)
|
115
|
+
# define NOKOGIRI_DEBUG_START(p)
|
116
|
+
# define NOKOGIRI_DEBUG_END(p)
|
108
117
|
|
109
118
|
#endif
|
110
119
|
|
111
120
|
#ifndef __builtin_expect
|
112
|
-
#
|
113
|
-
#
|
114
|
-
#
|
121
|
+
# if defined(__GNUC__)
|
122
|
+
# define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
|
123
|
+
# endif
|
115
124
|
#endif
|
116
125
|
|
117
126
|
#define XMLNS_PREFIX "xmlns"
|