libxml-ruby 3.2.2-x64-mingw-ucrt
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/HISTORY +848 -0
- data/LICENSE +21 -0
- data/MANIFEST +166 -0
- data/README.rdoc +217 -0
- data/Rakefile +99 -0
- data/ext/libxml/extconf.rb +61 -0
- data/ext/libxml/libxml.c +80 -0
- data/ext/libxml/libxml_ruby.def +35 -0
- data/ext/libxml/ruby_libxml.h +67 -0
- data/ext/libxml/ruby_xml.c +933 -0
- data/ext/libxml/ruby_xml.h +10 -0
- data/ext/libxml/ruby_xml_attr.c +333 -0
- data/ext/libxml/ruby_xml_attr.h +12 -0
- data/ext/libxml/ruby_xml_attr_decl.c +153 -0
- data/ext/libxml/ruby_xml_attr_decl.h +11 -0
- data/ext/libxml/ruby_xml_attributes.c +275 -0
- data/ext/libxml/ruby_xml_attributes.h +15 -0
- data/ext/libxml/ruby_xml_cbg.c +85 -0
- data/ext/libxml/ruby_xml_document.c +1123 -0
- data/ext/libxml/ruby_xml_document.h +11 -0
- data/ext/libxml/ruby_xml_dtd.c +248 -0
- data/ext/libxml/ruby_xml_dtd.h +9 -0
- data/ext/libxml/ruby_xml_encoding.c +250 -0
- data/ext/libxml/ruby_xml_encoding.h +16 -0
- data/ext/libxml/ruby_xml_error.c +996 -0
- data/ext/libxml/ruby_xml_error.h +12 -0
- data/ext/libxml/ruby_xml_html_parser.c +89 -0
- data/ext/libxml/ruby_xml_html_parser.h +10 -0
- data/ext/libxml/ruby_xml_html_parser_context.c +337 -0
- data/ext/libxml/ruby_xml_html_parser_context.h +10 -0
- data/ext/libxml/ruby_xml_html_parser_options.c +46 -0
- data/ext/libxml/ruby_xml_html_parser_options.h +10 -0
- data/ext/libxml/ruby_xml_input_cbg.c +191 -0
- data/ext/libxml/ruby_xml_input_cbg.h +20 -0
- data/ext/libxml/ruby_xml_io.c +47 -0
- data/ext/libxml/ruby_xml_io.h +10 -0
- data/ext/libxml/ruby_xml_namespace.c +153 -0
- data/ext/libxml/ruby_xml_namespace.h +10 -0
- data/ext/libxml/ruby_xml_namespaces.c +293 -0
- data/ext/libxml/ruby_xml_namespaces.h +9 -0
- data/ext/libxml/ruby_xml_node.c +1402 -0
- data/ext/libxml/ruby_xml_node.h +13 -0
- data/ext/libxml/ruby_xml_parser.c +91 -0
- data/ext/libxml/ruby_xml_parser.h +12 -0
- data/ext/libxml/ruby_xml_parser_context.c +999 -0
- data/ext/libxml/ruby_xml_parser_context.h +10 -0
- data/ext/libxml/ruby_xml_parser_options.c +66 -0
- data/ext/libxml/ruby_xml_parser_options.h +12 -0
- data/ext/libxml/ruby_xml_reader.c +1239 -0
- data/ext/libxml/ruby_xml_reader.h +17 -0
- data/ext/libxml/ruby_xml_relaxng.c +110 -0
- data/ext/libxml/ruby_xml_relaxng.h +10 -0
- data/ext/libxml/ruby_xml_sax2_handler.c +326 -0
- data/ext/libxml/ruby_xml_sax2_handler.h +10 -0
- data/ext/libxml/ruby_xml_sax_parser.c +116 -0
- data/ext/libxml/ruby_xml_sax_parser.h +10 -0
- data/ext/libxml/ruby_xml_schema.c +278 -0
- data/ext/libxml/ruby_xml_schema.h +809 -0
- data/ext/libxml/ruby_xml_schema_attribute.c +109 -0
- data/ext/libxml/ruby_xml_schema_attribute.h +15 -0
- data/ext/libxml/ruby_xml_schema_element.c +95 -0
- data/ext/libxml/ruby_xml_schema_element.h +14 -0
- data/ext/libxml/ruby_xml_schema_facet.c +52 -0
- data/ext/libxml/ruby_xml_schema_facet.h +13 -0
- data/ext/libxml/ruby_xml_schema_type.c +232 -0
- data/ext/libxml/ruby_xml_schema_type.h +9 -0
- data/ext/libxml/ruby_xml_version.h +9 -0
- data/ext/libxml/ruby_xml_writer.c +1133 -0
- data/ext/libxml/ruby_xml_writer.h +10 -0
- data/ext/libxml/ruby_xml_xinclude.c +16 -0
- data/ext/libxml/ruby_xml_xinclude.h +11 -0
- data/ext/libxml/ruby_xml_xpath.c +194 -0
- data/ext/libxml/ruby_xml_xpath.h +13 -0
- data/ext/libxml/ruby_xml_xpath_context.c +360 -0
- data/ext/libxml/ruby_xml_xpath_context.h +9 -0
- data/ext/libxml/ruby_xml_xpath_expression.c +81 -0
- data/ext/libxml/ruby_xml_xpath_expression.h +10 -0
- data/ext/libxml/ruby_xml_xpath_object.c +338 -0
- data/ext/libxml/ruby_xml_xpath_object.h +17 -0
- data/ext/libxml/ruby_xml_xpointer.c +99 -0
- data/ext/libxml/ruby_xml_xpointer.h +11 -0
- data/ext/vc/libxml_ruby.sln +28 -0
- data/lib/3.1/libxml_ruby.so +0 -0
- data/lib/libxml/attr.rb +123 -0
- data/lib/libxml/attr_decl.rb +80 -0
- data/lib/libxml/attributes.rb +14 -0
- data/lib/libxml/document.rb +194 -0
- data/lib/libxml/error.rb +95 -0
- data/lib/libxml/hpricot.rb +78 -0
- data/lib/libxml/html_parser.rb +96 -0
- data/lib/libxml/namespace.rb +62 -0
- data/lib/libxml/namespaces.rb +38 -0
- data/lib/libxml/node.rb +323 -0
- data/lib/libxml/parser.rb +101 -0
- data/lib/libxml/sax_callbacks.rb +180 -0
- data/lib/libxml/sax_parser.rb +41 -0
- data/lib/libxml/schema/attribute.rb +19 -0
- data/lib/libxml/schema/element.rb +19 -0
- data/lib/libxml/schema/type.rb +21 -0
- data/lib/libxml/schema.rb +48 -0
- data/lib/libxml/tree.rb +29 -0
- data/lib/libxml-ruby.rb +30 -0
- data/lib/libxml.rb +5 -0
- data/lib/xml/libxml.rb +10 -0
- data/lib/xml.rb +14 -0
- data/libxml-ruby.gemspec +48 -0
- data/script/benchmark/depixelate +634 -0
- data/script/benchmark/hamlet.xml +9055 -0
- data/script/benchmark/parsecount +170 -0
- data/script/benchmark/sock_entries.xml +507 -0
- data/script/benchmark/throughput +41 -0
- data/script/test +6 -0
- data/setup.rb +1584 -0
- data/test/c14n/given/doc.dtd +1 -0
- data/test/c14n/given/example-1.xml +14 -0
- data/test/c14n/given/example-2.xml +11 -0
- data/test/c14n/given/example-3.xml +18 -0
- data/test/c14n/given/example-4.xml +9 -0
- data/test/c14n/given/example-5.xml +12 -0
- data/test/c14n/given/example-6.xml +2 -0
- data/test/c14n/given/example-7.xml +11 -0
- data/test/c14n/given/example-8.xml +11 -0
- data/test/c14n/given/example-8.xpath +10 -0
- data/test/c14n/given/world.txt +1 -0
- data/test/c14n/result/1-1-without-comments/example-1 +4 -0
- data/test/c14n/result/1-1-without-comments/example-2 +11 -0
- data/test/c14n/result/1-1-without-comments/example-3 +14 -0
- data/test/c14n/result/1-1-without-comments/example-4 +9 -0
- data/test/c14n/result/1-1-without-comments/example-5 +3 -0
- data/test/c14n/result/1-1-without-comments/example-6 +1 -0
- data/test/c14n/result/1-1-without-comments/example-7 +1 -0
- data/test/c14n/result/1-1-without-comments/example-8 +1 -0
- data/test/c14n/result/with-comments/example-1 +6 -0
- data/test/c14n/result/with-comments/example-2 +11 -0
- data/test/c14n/result/with-comments/example-3 +14 -0
- data/test/c14n/result/with-comments/example-4 +9 -0
- data/test/c14n/result/with-comments/example-5 +4 -0
- data/test/c14n/result/with-comments/example-6 +1 -0
- data/test/c14n/result/with-comments/example-7 +1 -0
- data/test/c14n/result/without-comments/example-1 +4 -0
- data/test/c14n/result/without-comments/example-2 +11 -0
- data/test/c14n/result/without-comments/example-3 +14 -0
- data/test/c14n/result/without-comments/example-4 +9 -0
- data/test/c14n/result/without-comments/example-5 +3 -0
- data/test/c14n/result/without-comments/example-6 +1 -0
- data/test/c14n/result/without-comments/example-7 +1 -0
- data/test/model/atom.xml +13 -0
- data/test/model/bands.iso-8859-1.xml +5 -0
- data/test/model/bands.utf-8.xml +5 -0
- data/test/model/bands.xml +5 -0
- data/test/model/books.xml +154 -0
- data/test/model/cwm_1_0.xml +11336 -0
- data/test/model/merge_bug_data.xml +58 -0
- data/test/model/ruby-lang.html +238 -0
- data/test/model/rubynet.xml +79 -0
- data/test/model/rubynet_project +1 -0
- data/test/model/shiporder.rnc +28 -0
- data/test/model/shiporder.rng +86 -0
- data/test/model/shiporder.xml +23 -0
- data/test/model/shiporder.xsd +40 -0
- data/test/model/soap.xml +27 -0
- data/test/model/xinclude.xml +5 -0
- data/test/test_attr.rb +181 -0
- data/test/test_attr_decl.rb +132 -0
- data/test/test_attributes.rb +136 -0
- data/test/test_canonicalize.rb +120 -0
- data/test/test_deprecated_require.rb +12 -0
- data/test/test_document.rb +132 -0
- data/test/test_document_write.rb +146 -0
- data/test/test_dtd.rb +129 -0
- data/test/test_encoding.rb +129 -0
- data/test/test_encoding_sax.rb +115 -0
- data/test/test_error.rb +178 -0
- data/test/test_helper.rb +9 -0
- data/test/test_html_parser.rb +162 -0
- data/test/test_html_parser_context.rb +23 -0
- data/test/test_namespace.rb +60 -0
- data/test/test_namespaces.rb +200 -0
- data/test/test_node.rb +237 -0
- data/test/test_node_cdata.rb +50 -0
- data/test/test_node_comment.rb +32 -0
- data/test/test_node_copy.rb +40 -0
- data/test/test_node_edit.rb +158 -0
- data/test/test_node_pi.rb +37 -0
- data/test/test_node_text.rb +69 -0
- data/test/test_node_write.rb +97 -0
- data/test/test_node_xlink.rb +28 -0
- data/test/test_parser.rb +324 -0
- data/test/test_parser_context.rb +198 -0
- data/test/test_properties.rb +38 -0
- data/test/test_reader.rb +364 -0
- data/test/test_relaxng.rb +53 -0
- data/test/test_sax_parser.rb +326 -0
- data/test/test_schema.rb +168 -0
- data/test/test_suite.rb +48 -0
- data/test/test_traversal.rb +152 -0
- data/test/test_writer.rb +468 -0
- data/test/test_xinclude.rb +20 -0
- data/test/test_xml.rb +263 -0
- data/test/test_xpath.rb +244 -0
- data/test/test_xpath_context.rb +88 -0
- data/test/test_xpath_expression.rb +37 -0
- data/test/test_xpointer.rb +72 -0
- metadata +325 -0
@@ -0,0 +1,12 @@
|
|
1
|
+
/* Please see the LICENSE file for copyright and distribution information */
|
2
|
+
|
3
|
+
#ifndef __RXML_ERROR__
|
4
|
+
#define __RXML_ERROR__
|
5
|
+
|
6
|
+
extern VALUE eXMLError;
|
7
|
+
|
8
|
+
void rxml_init_error();
|
9
|
+
VALUE rxml_error_wrap(xmlErrorPtr xerror);
|
10
|
+
NORETURN(void rxml_raise(xmlErrorPtr xerror));
|
11
|
+
|
12
|
+
#endif
|
@@ -0,0 +1,89 @@
|
|
1
|
+
/* Please see the LICENSE file for copyright and distribution information */
|
2
|
+
|
3
|
+
#include "ruby_libxml.h"
|
4
|
+
|
5
|
+
/* Document-class: LibXML::XML::HTMLParser
|
6
|
+
*
|
7
|
+
* The HTML parser implements an HTML 4.0 non-verifying parser with an API
|
8
|
+
* compatible with the XML::Parser. In contrast with the XML::Parser,
|
9
|
+
* it can parse "real world" HTML, even if it severely broken from a
|
10
|
+
* specification point of view.
|
11
|
+
*
|
12
|
+
* The HTML parser creates an in-memory document object
|
13
|
+
* that consist of any number of XML::Node instances. This is simple
|
14
|
+
* and powerful model, but has the major limitation that the size of
|
15
|
+
* the document that can be processed is limited by the amount of
|
16
|
+
* memory available.
|
17
|
+
*
|
18
|
+
* Using the html parser is simple:
|
19
|
+
*
|
20
|
+
* parser = XML::HTMLParser.file('my_file')
|
21
|
+
* doc = parser.parse
|
22
|
+
*
|
23
|
+
* You can also parse documents (see XML::HTMLParser.document),
|
24
|
+
* strings (see XML::HTMLParser.string) and io objects (see
|
25
|
+
* XML::HTMLParser.io).
|
26
|
+
*/
|
27
|
+
|
28
|
+
VALUE cXMLHtmlParser;
|
29
|
+
static ID CONTEXT_ATTR;
|
30
|
+
|
31
|
+
|
32
|
+
/* call-seq:
|
33
|
+
* XML::HTMLParser.initialize -> parser
|
34
|
+
*
|
35
|
+
* Initializes a new parser instance with no pre-determined source.
|
36
|
+
*/
|
37
|
+
static VALUE rxml_html_parser_initialize(int argc, VALUE *argv, VALUE self)
|
38
|
+
{
|
39
|
+
VALUE context = Qnil;
|
40
|
+
|
41
|
+
rb_scan_args(argc, argv, "01", &context);
|
42
|
+
|
43
|
+
if (context == Qnil)
|
44
|
+
{
|
45
|
+
rb_raise(rb_eArgError, "An instance of a XML::Parser::Context must be passed to XML::HTMLParser.new");
|
46
|
+
}
|
47
|
+
|
48
|
+
rb_ivar_set(self, CONTEXT_ATTR, context);
|
49
|
+
return self;
|
50
|
+
}
|
51
|
+
|
52
|
+
/*
|
53
|
+
* call-seq:
|
54
|
+
* parser.parse -> XML::Document
|
55
|
+
*
|
56
|
+
* Parse the input XML and create an XML::Document with
|
57
|
+
* it's content. If an error occurs, XML::Parser::ParseError
|
58
|
+
* is thrown.
|
59
|
+
*/
|
60
|
+
static VALUE rxml_html_parser_parse(VALUE self)
|
61
|
+
{
|
62
|
+
xmlParserCtxtPtr ctxt;
|
63
|
+
VALUE context = rb_ivar_get(self, CONTEXT_ATTR);
|
64
|
+
|
65
|
+
Data_Get_Struct(context, xmlParserCtxt, ctxt);
|
66
|
+
|
67
|
+
if (htmlParseDocument(ctxt) == -1 && ! ctxt->recovery)
|
68
|
+
{
|
69
|
+
rxml_raise(&ctxt->lastError);
|
70
|
+
}
|
71
|
+
|
72
|
+
rb_funcall(context, rb_intern("close"), 0);
|
73
|
+
|
74
|
+
return rxml_document_wrap(ctxt->myDoc);
|
75
|
+
}
|
76
|
+
|
77
|
+
void rxml_init_html_parser(void)
|
78
|
+
{
|
79
|
+
CONTEXT_ATTR = rb_intern("@context");
|
80
|
+
|
81
|
+
cXMLHtmlParser = rb_define_class_under(mXML, "HTMLParser", rb_cObject);
|
82
|
+
|
83
|
+
/* Atributes */
|
84
|
+
rb_define_attr(cXMLHtmlParser, "input", 1, 0);
|
85
|
+
|
86
|
+
/* Instance methods */
|
87
|
+
rb_define_method(cXMLHtmlParser, "initialize", rxml_html_parser_initialize, -1);
|
88
|
+
rb_define_method(cXMLHtmlParser, "parse", rxml_html_parser_parse, 0);
|
89
|
+
}
|
@@ -0,0 +1,337 @@
|
|
1
|
+
/* Please see the LICENSE file for copyright and distribution information */
|
2
|
+
|
3
|
+
#include "ruby_libxml.h"
|
4
|
+
#include "ruby_xml_html_parser_context.h"
|
5
|
+
|
6
|
+
|
7
|
+
/*
|
8
|
+
* Document-class: LibXML::XML::HTMLParser::Context
|
9
|
+
*
|
10
|
+
* The XML::HTMLParser::Context class provides in-depth control over how
|
11
|
+
* a document is parsed.
|
12
|
+
*/
|
13
|
+
|
14
|
+
VALUE cXMLHtmlParserContext;
|
15
|
+
static ID IO_ATTR;
|
16
|
+
|
17
|
+
/* OS X 10.5 ships with libxml2 version 2.6.16 which does not expose the
|
18
|
+
htmlNewParserCtxt (or htmlInitParserCtxt which it uses) method. htmlNewParserCtxt
|
19
|
+
wasn't added to the libxml2 header files until 2.6.27. So the next two
|
20
|
+
methods are simply copied from a newer version of libxml2 (2.7.2). */
|
21
|
+
#if LIBXML_VERSION < 20627
|
22
|
+
#define XML_CTXT_FINISH_DTD_0 0xabcd1234
|
23
|
+
static int htmlInitParserCtxt(htmlParserCtxtPtr ctxt)
|
24
|
+
{
|
25
|
+
htmlSAXHandler *sax;
|
26
|
+
if (ctxt == NULL) return(-1);
|
27
|
+
|
28
|
+
memset(ctxt, 0, sizeof(htmlParserCtxt));
|
29
|
+
ctxt->dict = xmlDictCreate();
|
30
|
+
if (ctxt->dict == NULL) {
|
31
|
+
rb_raise(rb_eNoMemError, "htmlInitParserCtxt: out of memory\n");
|
32
|
+
return(-1);
|
33
|
+
}
|
34
|
+
sax = (htmlSAXHandler *) xmlMalloc(sizeof(htmlSAXHandler));
|
35
|
+
if (sax == NULL) {
|
36
|
+
rb_raise(rb_eNoMemError, "htmlInitParserCtxt: out of memory\n");
|
37
|
+
return(-1);
|
38
|
+
}
|
39
|
+
else
|
40
|
+
memset(sax, 0, sizeof(htmlSAXHandler));
|
41
|
+
|
42
|
+
ctxt->inputTab = (htmlParserInputPtr *) xmlMalloc(5 * sizeof(htmlParserInputPtr));
|
43
|
+
if (ctxt->inputTab == NULL) {
|
44
|
+
rb_raise(rb_eNoMemError, "htmlInitParserCtxt: out of memory\n");
|
45
|
+
ctxt->inputNr = 0;
|
46
|
+
ctxt->inputMax = 0;
|
47
|
+
ctxt->input = NULL;
|
48
|
+
return(-1);
|
49
|
+
}
|
50
|
+
ctxt->inputNr = 0;
|
51
|
+
ctxt->inputMax = 5;
|
52
|
+
ctxt->input = NULL;
|
53
|
+
ctxt->version = NULL;
|
54
|
+
ctxt->encoding = NULL;
|
55
|
+
ctxt->standalone = -1;
|
56
|
+
ctxt->instate = XML_PARSER_START;
|
57
|
+
|
58
|
+
ctxt->nodeTab = (htmlNodePtr *) xmlMalloc(10 * sizeof(htmlNodePtr));
|
59
|
+
if (ctxt->nodeTab == NULL) {
|
60
|
+
rb_raise(rb_eNoMemError, "htmlInitParserCtxt: out of memory\n");
|
61
|
+
ctxt->nodeNr = 0;
|
62
|
+
ctxt->nodeMax = 0;
|
63
|
+
ctxt->node = NULL;
|
64
|
+
ctxt->inputNr = 0;
|
65
|
+
ctxt->inputMax = 0;
|
66
|
+
ctxt->input = NULL;
|
67
|
+
return(-1);
|
68
|
+
}
|
69
|
+
ctxt->nodeNr = 0;
|
70
|
+
ctxt->nodeMax = 10;
|
71
|
+
ctxt->node = NULL;
|
72
|
+
|
73
|
+
ctxt->nameTab = (const xmlChar **) xmlMalloc(10 * sizeof(xmlChar *));
|
74
|
+
if (ctxt->nameTab == NULL) {
|
75
|
+
rb_raise(rb_eNoMemError, "htmlInitParserCtxt: out of memory\n");
|
76
|
+
ctxt->nameNr = 0;
|
77
|
+
ctxt->nameMax = 10;
|
78
|
+
ctxt->name = NULL;
|
79
|
+
ctxt->nodeNr = 0;
|
80
|
+
ctxt->nodeMax = 0;
|
81
|
+
ctxt->node = NULL;
|
82
|
+
ctxt->inputNr = 0;
|
83
|
+
ctxt->inputMax = 0;
|
84
|
+
ctxt->input = NULL;
|
85
|
+
return(-1);
|
86
|
+
}
|
87
|
+
ctxt->nameNr = 0;
|
88
|
+
ctxt->nameMax = 10;
|
89
|
+
ctxt->name = NULL;
|
90
|
+
|
91
|
+
if (sax == NULL) ctxt->sax = (xmlSAXHandlerPtr) &htmlDefaultSAXHandler;
|
92
|
+
else {
|
93
|
+
ctxt->sax = sax;
|
94
|
+
memcpy(sax, &htmlDefaultSAXHandler, sizeof(xmlSAXHandlerV1));
|
95
|
+
}
|
96
|
+
ctxt->userData = ctxt;
|
97
|
+
ctxt->myDoc = NULL;
|
98
|
+
ctxt->wellFormed = 1;
|
99
|
+
ctxt->replaceEntities = 0;
|
100
|
+
ctxt->linenumbers = xmlLineNumbersDefaultValue;
|
101
|
+
ctxt->html = 1;
|
102
|
+
ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_0;
|
103
|
+
ctxt->vctxt.userData = ctxt;
|
104
|
+
ctxt->vctxt.error = xmlParserValidityError;
|
105
|
+
ctxt->vctxt.warning = xmlParserValidityWarning;
|
106
|
+
ctxt->record_info = 0;
|
107
|
+
ctxt->validate = 0;
|
108
|
+
ctxt->nbChars = 0;
|
109
|
+
ctxt->checkIndex = 0;
|
110
|
+
ctxt->catalogs = NULL;
|
111
|
+
xmlInitNodeInfoSeq(&ctxt->node_seq);
|
112
|
+
return(0);
|
113
|
+
}
|
114
|
+
|
115
|
+
static htmlParserCtxtPtr htmlNewParserCtxt(void)
|
116
|
+
{
|
117
|
+
xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) xmlMalloc(sizeof(xmlParserCtxt));
|
118
|
+
if (ctxt == NULL) {
|
119
|
+
rb_raise(rb_eNoMemError, "NewParserCtxt: out of memory\n");
|
120
|
+
return(NULL);
|
121
|
+
}
|
122
|
+
memset(ctxt, 0, sizeof(xmlParserCtxt));
|
123
|
+
if (htmlInitParserCtxt(ctxt) < 0) {
|
124
|
+
htmlFreeParserCtxt(ctxt);
|
125
|
+
return(NULL);
|
126
|
+
}
|
127
|
+
return(ctxt);
|
128
|
+
}
|
129
|
+
#endif
|
130
|
+
|
131
|
+
static void rxml_html_parser_context_free(htmlParserCtxtPtr ctxt)
|
132
|
+
{
|
133
|
+
htmlFreeParserCtxt(ctxt);
|
134
|
+
}
|
135
|
+
|
136
|
+
static VALUE rxml_html_parser_context_wrap(htmlParserCtxtPtr ctxt)
|
137
|
+
{
|
138
|
+
return Data_Wrap_Struct(cXMLHtmlParserContext, NULL, rxml_html_parser_context_free, ctxt);
|
139
|
+
}
|
140
|
+
|
141
|
+
/* call-seq:
|
142
|
+
* XML::HTMLParser::Context.file(file) -> XML::HTMLParser::Context
|
143
|
+
*
|
144
|
+
* Creates a new parser context based on the specified file or uri.
|
145
|
+
*
|
146
|
+
* Parameters:
|
147
|
+
*
|
148
|
+
* file - A filename or uri.
|
149
|
+
*/
|
150
|
+
static VALUE rxml_html_parser_context_file(VALUE klass, VALUE file)
|
151
|
+
{
|
152
|
+
htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(StringValuePtr(file), NULL);
|
153
|
+
if (!ctxt)
|
154
|
+
rxml_raise(&xmlLastError);
|
155
|
+
|
156
|
+
/* This is annoying, but xmlInitParserCtxt (called indirectly above) and
|
157
|
+
xmlCtxtUseOptionsInternal (called below) initialize slightly different
|
158
|
+
context options, in particular XML_PARSE_NODICT which xmlInitParserCtxt
|
159
|
+
sets to 0 and xmlCtxtUseOptionsInternal sets to 1. So we have to call both. */
|
160
|
+
htmlCtxtUseOptions(ctxt, rxml_libxml_default_options());
|
161
|
+
|
162
|
+
return rxml_html_parser_context_wrap(ctxt);
|
163
|
+
}
|
164
|
+
|
165
|
+
/* call-seq:
|
166
|
+
* XML::HTMLParser::Context.io(io) -> XML::HTMLParser::Context
|
167
|
+
*
|
168
|
+
* Creates a new parser context based on the specified io object.
|
169
|
+
*
|
170
|
+
* Parameters:
|
171
|
+
*
|
172
|
+
* io - A ruby IO object.
|
173
|
+
*/
|
174
|
+
static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
|
175
|
+
{
|
176
|
+
VALUE result;
|
177
|
+
htmlParserCtxtPtr ctxt;
|
178
|
+
xmlParserInputBufferPtr input;
|
179
|
+
xmlParserInputPtr stream;
|
180
|
+
|
181
|
+
if (NIL_P(io))
|
182
|
+
rb_raise(rb_eTypeError, "Must pass in an IO object");
|
183
|
+
|
184
|
+
input = xmlParserInputBufferCreateIO((xmlInputReadCallback) rxml_read_callback, NULL,
|
185
|
+
(void*)io, XML_CHAR_ENCODING_NONE);
|
186
|
+
|
187
|
+
ctxt = htmlNewParserCtxt();
|
188
|
+
if (!ctxt)
|
189
|
+
{
|
190
|
+
xmlFreeParserInputBuffer(input);
|
191
|
+
rxml_raise(&xmlLastError);
|
192
|
+
}
|
193
|
+
|
194
|
+
/* This is annoying, but xmlInitParserCtxt (called indirectly above) and
|
195
|
+
xmlCtxtUseOptionsInternal (called below) initialize slightly different
|
196
|
+
context options, in particular XML_PARSE_NODICT which xmlInitParserCtxt
|
197
|
+
sets to 0 and xmlCtxtUseOptionsInternal sets to 1. So we have to call both. */
|
198
|
+
htmlCtxtUseOptions(ctxt, rxml_libxml_default_options());
|
199
|
+
|
200
|
+
stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
|
201
|
+
|
202
|
+
if (!stream)
|
203
|
+
{
|
204
|
+
xmlFreeParserInputBuffer(input);
|
205
|
+
xmlFreeParserCtxt(ctxt);
|
206
|
+
rxml_raise(&xmlLastError);
|
207
|
+
}
|
208
|
+
inputPush(ctxt, stream);
|
209
|
+
result = rxml_html_parser_context_wrap(ctxt);
|
210
|
+
|
211
|
+
/* Attach io object to parser so it won't get freed.*/
|
212
|
+
rb_ivar_set(result, IO_ATTR, io);
|
213
|
+
|
214
|
+
return result;
|
215
|
+
}
|
216
|
+
|
217
|
+
/* call-seq:
|
218
|
+
* XML::HTMLParser::Context.string(string) -> XML::HTMLParser::Context
|
219
|
+
*
|
220
|
+
* Creates a new parser context based on the specified string.
|
221
|
+
*
|
222
|
+
* Parameters:
|
223
|
+
*
|
224
|
+
* string - A string that contains the data to parse.
|
225
|
+
*/
|
226
|
+
static VALUE rxml_html_parser_context_string(VALUE klass, VALUE string)
|
227
|
+
{
|
228
|
+
htmlParserCtxtPtr ctxt;
|
229
|
+
Check_Type(string, T_STRING);
|
230
|
+
|
231
|
+
if (RSTRING_LEN(string) == 0)
|
232
|
+
rb_raise(rb_eArgError, "Must specify a string with one or more characters");
|
233
|
+
|
234
|
+
ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(string),
|
235
|
+
(int)RSTRING_LEN(string));
|
236
|
+
if (!ctxt)
|
237
|
+
rxml_raise(&xmlLastError);
|
238
|
+
|
239
|
+
/* This is annoying, but xmlInitParserCtxt (called indirectly above) and
|
240
|
+
xmlCtxtUseOptionsInternal (called below) initialize slightly different
|
241
|
+
context options, in particular XML_PARSE_NODICT which xmlInitParserCtxt
|
242
|
+
sets to 0 and xmlCtxtUseOptionsInternal sets to 1. So we have to call both. */
|
243
|
+
htmlCtxtUseOptions(ctxt, rxml_libxml_default_options());
|
244
|
+
|
245
|
+
htmlDefaultSAXHandlerInit();
|
246
|
+
if (ctxt->sax != NULL)
|
247
|
+
memcpy(ctxt->sax, &htmlDefaultSAXHandler, sizeof(xmlSAXHandlerV1));
|
248
|
+
|
249
|
+
return rxml_html_parser_context_wrap(ctxt);
|
250
|
+
}
|
251
|
+
|
252
|
+
/*
|
253
|
+
* call-seq:
|
254
|
+
* context.close -> nil
|
255
|
+
*
|
256
|
+
* Closes the underlying input streams. This is useful when parsing a large amount of
|
257
|
+
* files and you want to close the files without relying on Ruby's garbage collector
|
258
|
+
* to run.
|
259
|
+
*/
|
260
|
+
static VALUE rxml_html_parser_context_close(VALUE self)
|
261
|
+
{
|
262
|
+
htmlParserCtxtPtr ctxt;
|
263
|
+
xmlParserInputPtr xinput;
|
264
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
265
|
+
|
266
|
+
while ((xinput = inputPop(ctxt)) != NULL)
|
267
|
+
{
|
268
|
+
xmlFreeInputStream(xinput);
|
269
|
+
}
|
270
|
+
return Qnil;
|
271
|
+
}
|
272
|
+
|
273
|
+
/*
|
274
|
+
* call-seq:
|
275
|
+
* context.disable_cdata = (true|false)
|
276
|
+
*
|
277
|
+
* Control whether the CDATA nodes will be created in this context.
|
278
|
+
*/
|
279
|
+
static VALUE rxml_html_parser_context_disable_cdata_set(VALUE self, VALUE value)
|
280
|
+
{
|
281
|
+
htmlParserCtxtPtr ctxt;
|
282
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
283
|
+
|
284
|
+
if (ctxt->sax == NULL)
|
285
|
+
rb_raise(rb_eRuntimeError, "Sax handler is not yet set");
|
286
|
+
|
287
|
+
/* LibXML controls this internally with the default SAX handler. */
|
288
|
+
if (value)
|
289
|
+
ctxt->sax->cdataBlock = NULL;
|
290
|
+
else
|
291
|
+
ctxt->sax->cdataBlock = htmlDefaultSAXHandler.cdataBlock;
|
292
|
+
|
293
|
+
return value;
|
294
|
+
}
|
295
|
+
|
296
|
+
/*
|
297
|
+
* call-seq:
|
298
|
+
* context.options = XML::Parser::Options::NOENT |
|
299
|
+
XML::Parser::Options::NOCDATA
|
300
|
+
*
|
301
|
+
* Provides control over the execution of a parser. Valid values
|
302
|
+
* are the constants defined on XML::Parser::Options. Multiple
|
303
|
+
* options can be combined by using Bitwise OR (|).
|
304
|
+
*/
|
305
|
+
static VALUE rxml_html_parser_context_options_set(VALUE self, VALUE options)
|
306
|
+
{
|
307
|
+
int xml_options = NUM2INT(options);
|
308
|
+
htmlParserCtxtPtr ctxt;
|
309
|
+
Check_Type(options, T_FIXNUM);
|
310
|
+
|
311
|
+
Data_Get_Struct(self, htmlParserCtxt, ctxt);
|
312
|
+
htmlCtxtUseOptions(ctxt, xml_options);
|
313
|
+
|
314
|
+
#if LIBXML_VERSION >= 20707
|
315
|
+
/* Big hack here, but htmlCtxtUseOptions doens't support HTML_PARSE_NOIMPLIED.
|
316
|
+
So do it ourselves. There must be a better way??? */
|
317
|
+
if (xml_options & HTML_PARSE_NOIMPLIED)
|
318
|
+
{
|
319
|
+
ctxt->options |= HTML_PARSE_NOIMPLIED;
|
320
|
+
}
|
321
|
+
#endif
|
322
|
+
|
323
|
+
return self;
|
324
|
+
}
|
325
|
+
|
326
|
+
void rxml_init_html_parser_context(void)
|
327
|
+
{
|
328
|
+
IO_ATTR = ID2SYM(rb_intern("@io"));
|
329
|
+
cXMLHtmlParserContext = rb_define_class_under(cXMLHtmlParser, "Context", cXMLParserContext);
|
330
|
+
|
331
|
+
rb_define_singleton_method(cXMLHtmlParserContext, "file", rxml_html_parser_context_file, 1);
|
332
|
+
rb_define_singleton_method(cXMLHtmlParserContext, "io", rxml_html_parser_context_io, 1);
|
333
|
+
rb_define_singleton_method(cXMLHtmlParserContext, "string", rxml_html_parser_context_string, 1);
|
334
|
+
rb_define_method(cXMLHtmlParserContext, "close", rxml_html_parser_context_close, 0);
|
335
|
+
rb_define_method(cXMLHtmlParserContext, "disable_cdata=", rxml_html_parser_context_disable_cdata_set, 1);
|
336
|
+
rb_define_method(cXMLHtmlParserContext, "options=", rxml_html_parser_context_options_set, 1);
|
337
|
+
}
|
@@ -0,0 +1,46 @@
|
|
1
|
+
/* Please see the LICENSE file for copyright and distribution information */
|
2
|
+
|
3
|
+
#include "ruby_libxml.h"
|
4
|
+
|
5
|
+
/* Document-class: LibXML::XML::HTMLParser::Options
|
6
|
+
*
|
7
|
+
* Options to control the operation of the HTMLParser. The easiest
|
8
|
+
* way to set a parser's options is via the methods
|
9
|
+
* XML::HTMLParser.file, XML::HTMLParser.io or XML::HTMLParser.string.
|
10
|
+
* For additional control, see XML::HTMLParser::Context#options=.
|
11
|
+
*/
|
12
|
+
|
13
|
+
VALUE mXMLHtmlParserOptions;
|
14
|
+
|
15
|
+
void rxml_init_html_parser_options(void)
|
16
|
+
{
|
17
|
+
mXMLHtmlParserOptions = rb_define_module_under(cXMLHtmlParser, "Options");
|
18
|
+
|
19
|
+
|
20
|
+
#if LIBXML_VERSION >= 20621
|
21
|
+
/* 1: Relax parsing. */
|
22
|
+
rb_define_const(mXMLHtmlParserOptions, "RECOVER", INT2NUM(HTML_PARSE_RECOVER));
|
23
|
+
#endif
|
24
|
+
#if LIBXML_VERSION >= 20708
|
25
|
+
/* 2: Do not default a doctype if not found */
|
26
|
+
rb_define_const(mXMLHtmlParserOptions, "NODEFDTD", INT2NUM(HTML_PARSE_NODEFDTD));
|
27
|
+
#endif
|
28
|
+
/* 32: Suppress error reports. */
|
29
|
+
rb_define_const(mXMLHtmlParserOptions, "NOERROR", INT2NUM(HTML_PARSE_NOERROR));
|
30
|
+
/* 64: Suppress warning reports. */
|
31
|
+
rb_define_const(mXMLHtmlParserOptions, "NOWARNING", INT2NUM(HTML_PARSE_NOWARNING));
|
32
|
+
/* 128: Enable pedantic error reporting. */
|
33
|
+
rb_define_const(mXMLHtmlParserOptions, "PEDANTIC", INT2NUM(HTML_PARSE_PEDANTIC));
|
34
|
+
/* 256: Remove blank nodes. */
|
35
|
+
rb_define_const(mXMLHtmlParserOptions, "NOBLANKS", INT2NUM(HTML_PARSE_NOBLANKS));
|
36
|
+
#if LIBXML_VERSION >= 20621
|
37
|
+
/* 2048: Forbid network access. */
|
38
|
+
rb_define_const(mXMLHtmlParserOptions, "NONET", INT2NUM(HTML_PARSE_NONET));
|
39
|
+
/* 65536: Compact small text nodes. */
|
40
|
+
rb_define_const(mXMLHtmlParserOptions, "COMPACT", INT2NUM(HTML_PARSE_COMPACT));
|
41
|
+
#endif
|
42
|
+
#if LIBXML_VERSION >= 20707
|
43
|
+
/* 8192: Do not add implied html/body... elements */
|
44
|
+
rb_define_const(mXMLHtmlParserOptions, "NOIMPLIED", INT2NUM(HTML_PARSE_NOIMPLIED));
|
45
|
+
#endif
|
46
|
+
}
|
@@ -0,0 +1,191 @@
|
|
1
|
+
/* Author: Martin Povolny (xpovolny@fi.muni.cz) */
|
2
|
+
|
3
|
+
#include "ruby_libxml.h"
|
4
|
+
#include "ruby_xml_input_cbg.h"
|
5
|
+
|
6
|
+
/* Document-class: LibXML::XML::InputCallbacks
|
7
|
+
*
|
8
|
+
* Support for adding custom scheme handlers. */
|
9
|
+
|
10
|
+
static ic_scheme *first_scheme = 0;
|
11
|
+
|
12
|
+
int ic_match(char const *filename)
|
13
|
+
{
|
14
|
+
ic_scheme *scheme;
|
15
|
+
|
16
|
+
//fprintf( stderr, "ic_match: %s\n", filename );
|
17
|
+
|
18
|
+
scheme = first_scheme;
|
19
|
+
while (0 != scheme)
|
20
|
+
{
|
21
|
+
if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST scheme->scheme_name, scheme->name_len))
|
22
|
+
{
|
23
|
+
return 1;
|
24
|
+
}
|
25
|
+
scheme = scheme->next_scheme;
|
26
|
+
}
|
27
|
+
return 0;
|
28
|
+
}
|
29
|
+
|
30
|
+
void* ic_open(char const *filename)
|
31
|
+
{
|
32
|
+
ic_doc_context *ic_doc;
|
33
|
+
ic_scheme *scheme;
|
34
|
+
VALUE res;
|
35
|
+
|
36
|
+
scheme = first_scheme;
|
37
|
+
while (0 != scheme)
|
38
|
+
{
|
39
|
+
if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST scheme->scheme_name, scheme->name_len))
|
40
|
+
{
|
41
|
+
ic_doc = (ic_doc_context*) malloc(sizeof(ic_doc_context));
|
42
|
+
|
43
|
+
res = rb_funcall(scheme->class, rb_intern("document_query"), 1,
|
44
|
+
rb_str_new2(filename));
|
45
|
+
|
46
|
+
ic_doc->buffer = strdup(StringValuePtr(res));
|
47
|
+
|
48
|
+
ic_doc->bpos = ic_doc->buffer;
|
49
|
+
ic_doc->remaining = (int)strlen(ic_doc->buffer);
|
50
|
+
return ic_doc;
|
51
|
+
}
|
52
|
+
scheme = scheme->next_scheme;
|
53
|
+
}
|
54
|
+
return 0;
|
55
|
+
}
|
56
|
+
|
57
|
+
int ic_read(void *context, char *buffer, int len)
|
58
|
+
{
|
59
|
+
ic_doc_context *ic_doc;
|
60
|
+
int ret_len;
|
61
|
+
ic_doc = (ic_doc_context*) context;
|
62
|
+
|
63
|
+
if (len >= ic_doc->remaining)
|
64
|
+
{
|
65
|
+
ret_len = ic_doc->remaining;
|
66
|
+
}
|
67
|
+
else
|
68
|
+
{
|
69
|
+
ret_len = len;
|
70
|
+
}
|
71
|
+
ic_doc->remaining -= ret_len;
|
72
|
+
strncpy(buffer, ic_doc->bpos, ret_len);
|
73
|
+
ic_doc->bpos += ret_len;
|
74
|
+
|
75
|
+
return ret_len;
|
76
|
+
}
|
77
|
+
|
78
|
+
int ic_close(void *context)
|
79
|
+
{
|
80
|
+
ruby_xfree(((ic_doc_context*) context)->buffer);
|
81
|
+
ruby_xfree(context);
|
82
|
+
return 1;
|
83
|
+
}
|
84
|
+
|
85
|
+
/*
|
86
|
+
* call-seq:
|
87
|
+
* register
|
88
|
+
*
|
89
|
+
* Register a new set of I/O callback for handling parser input.
|
90
|
+
*/
|
91
|
+
static VALUE input_callbacks_register_input_callbacks()
|
92
|
+
{
|
93
|
+
xmlRegisterInputCallbacks(ic_match, ic_open, ic_read, ic_close);
|
94
|
+
return (Qtrue);
|
95
|
+
}
|
96
|
+
|
97
|
+
/*
|
98
|
+
* call-seq:
|
99
|
+
* add_scheme
|
100
|
+
*
|
101
|
+
* No documentation available.
|
102
|
+
*/
|
103
|
+
static VALUE input_callbacks_add_scheme(VALUE self, VALUE scheme_name,
|
104
|
+
VALUE class)
|
105
|
+
{
|
106
|
+
ic_scheme *scheme;
|
107
|
+
|
108
|
+
Check_Type(scheme_name, T_STRING);
|
109
|
+
|
110
|
+
scheme = (ic_scheme*) malloc(sizeof(ic_scheme));
|
111
|
+
scheme->next_scheme = 0;
|
112
|
+
scheme->scheme_name = strdup(StringValuePtr(scheme_name)); /* TODO alloc, dealloc */
|
113
|
+
scheme->name_len = (int)strlen(scheme->scheme_name);
|
114
|
+
scheme->class = class; /* TODO alloc, dealloc */
|
115
|
+
|
116
|
+
//fprintf( stderr, "registered: %s, %d, %s\n", scheme->scheme_name, scheme->name_len, scheme->class );
|
117
|
+
|
118
|
+
if (0 == first_scheme)
|
119
|
+
first_scheme = scheme;
|
120
|
+
else
|
121
|
+
{
|
122
|
+
ic_scheme *pos;
|
123
|
+
pos = first_scheme;
|
124
|
+
while (0 != pos->next_scheme)
|
125
|
+
pos = pos->next_scheme;
|
126
|
+
pos->next_scheme = scheme;
|
127
|
+
}
|
128
|
+
|
129
|
+
return (Qtrue);
|
130
|
+
}
|
131
|
+
|
132
|
+
/*
|
133
|
+
* call-seq:
|
134
|
+
* remove_scheme
|
135
|
+
*
|
136
|
+
* No documentation available.
|
137
|
+
*/
|
138
|
+
static VALUE input_callbacks_remove_scheme(VALUE self, VALUE scheme_name)
|
139
|
+
{
|
140
|
+
char *name;
|
141
|
+
ic_scheme *save_scheme, *scheme;
|
142
|
+
|
143
|
+
Check_Type(scheme_name, T_STRING);
|
144
|
+
name = StringValuePtr(scheme_name);
|
145
|
+
|
146
|
+
if (0 == first_scheme)
|
147
|
+
return Qfalse;
|
148
|
+
|
149
|
+
if (!strncmp(name, first_scheme->scheme_name, first_scheme->name_len))
|
150
|
+
{
|
151
|
+
save_scheme = first_scheme->next_scheme;
|
152
|
+
|
153
|
+
ruby_xfree(first_scheme->scheme_name);
|
154
|
+
ruby_xfree(first_scheme);
|
155
|
+
|
156
|
+
first_scheme = save_scheme;
|
157
|
+
return Qtrue;
|
158
|
+
}
|
159
|
+
|
160
|
+
scheme = first_scheme;
|
161
|
+
while (0 != scheme->next_scheme)
|
162
|
+
{
|
163
|
+
if (!strncmp(name, scheme->next_scheme->scheme_name,
|
164
|
+
scheme->next_scheme->name_len))
|
165
|
+
{
|
166
|
+
save_scheme = scheme->next_scheme->next_scheme;
|
167
|
+
|
168
|
+
ruby_xfree(scheme->next_scheme->scheme_name);
|
169
|
+
ruby_xfree(scheme->next_scheme);
|
170
|
+
|
171
|
+
scheme->next_scheme = save_scheme;
|
172
|
+
return Qtrue;
|
173
|
+
}
|
174
|
+
scheme = scheme->next_scheme;
|
175
|
+
}
|
176
|
+
return Qfalse;
|
177
|
+
}
|
178
|
+
|
179
|
+
void rxml_init_input_callbacks(void)
|
180
|
+
{
|
181
|
+
VALUE cInputCallbacks;
|
182
|
+
cInputCallbacks = rb_define_class_under(mXML, "InputCallbacks", rb_cObject);
|
183
|
+
|
184
|
+
/* Class Methods */
|
185
|
+
rb_define_singleton_method(cInputCallbacks, "register",
|
186
|
+
input_callbacks_register_input_callbacks, 0);
|
187
|
+
rb_define_singleton_method(cInputCallbacks, "add_scheme",
|
188
|
+
input_callbacks_add_scheme, 2);
|
189
|
+
rb_define_singleton_method(cInputCallbacks, "remove_scheme",
|
190
|
+
input_callbacks_remove_scheme, 1);
|
191
|
+
}
|