nokogiri 1.18.0.rc1-x86_64-linux-gnu
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +38 -0
- data/LICENSE-DEPENDENCIES.md +2224 -0
- data/LICENSE.md +9 -0
- data/README.md +293 -0
- data/bin/nokogiri +131 -0
- data/dependencies.yml +42 -0
- data/ext/nokogiri/depend +38 -0
- data/ext/nokogiri/extconf.rb +1173 -0
- data/ext/nokogiri/gumbo.c +610 -0
- data/ext/nokogiri/html4_document.c +171 -0
- data/ext/nokogiri/html4_element_description.c +299 -0
- data/ext/nokogiri/html4_entity_lookup.c +37 -0
- data/ext/nokogiri/html4_sax_parser.c +40 -0
- data/ext/nokogiri/html4_sax_parser_context.c +98 -0
- data/ext/nokogiri/html4_sax_push_parser.c +96 -0
- data/ext/nokogiri/include/libexslt/exslt.h +108 -0
- data/ext/nokogiri/include/libexslt/exsltconfig.h +70 -0
- data/ext/nokogiri/include/libexslt/exsltexports.h +63 -0
- data/ext/nokogiri/include/libxml2/libxml/HTMLparser.h +336 -0
- data/ext/nokogiri/include/libxml2/libxml/HTMLtree.h +147 -0
- data/ext/nokogiri/include/libxml2/libxml/SAX.h +202 -0
- data/ext/nokogiri/include/libxml2/libxml/SAX2.h +171 -0
- data/ext/nokogiri/include/libxml2/libxml/c14n.h +115 -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 +82 -0
- data/ext/nokogiri/include/libxml2/libxml/encoding.h +244 -0
- data/ext/nokogiri/include/libxml2/libxml/entities.h +166 -0
- data/ext/nokogiri/include/libxml2/libxml/globals.h +41 -0
- data/ext/nokogiri/include/libxml2/libxml/hash.h +251 -0
- data/ext/nokogiri/include/libxml2/libxml/list.h +137 -0
- data/ext/nokogiri/include/libxml2/libxml/nanoftp.h +186 -0
- data/ext/nokogiri/include/libxml2/libxml/nanohttp.h +98 -0
- data/ext/nokogiri/include/libxml2/libxml/parser.h +1390 -0
- data/ext/nokogiri/include/libxml2/libxml/parserInternals.h +671 -0
- data/ext/nokogiri/include/libxml2/libxml/pattern.h +106 -0
- data/ext/nokogiri/include/libxml2/libxml/relaxng.h +219 -0
- data/ext/nokogiri/include/libxml2/libxml/schemasInternals.h +959 -0
- data/ext/nokogiri/include/libxml2/libxml/schematron.h +143 -0
- data/ext/nokogiri/include/libxml2/libxml/threads.h +87 -0
- data/ext/nokogiri/include/libxml2/libxml/tree.h +1382 -0
- data/ext/nokogiri/include/libxml2/libxml/uri.h +106 -0
- data/ext/nokogiri/include/libxml2/libxml/valid.h +477 -0
- data/ext/nokogiri/include/libxml2/libxml/xinclude.h +136 -0
- data/ext/nokogiri/include/libxml2/libxml/xlink.h +189 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlIO.h +438 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlautomata.h +146 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlerror.h +962 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlexports.h +146 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlmemory.h +188 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlmodule.h +57 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlreader.h +436 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlregexp.h +215 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlsave.h +102 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlschemas.h +249 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlschemastypes.h +152 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlstring.h +140 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlunicode.h +366 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlversion.h +347 -0
- data/ext/nokogiri/include/libxml2/libxml/xmlwriter.h +489 -0
- data/ext/nokogiri/include/libxml2/libxml/xpath.h +579 -0
- data/ext/nokogiri/include/libxml2/libxml/xpathInternals.h +633 -0
- data/ext/nokogiri/include/libxml2/libxml/xpointer.h +138 -0
- data/ext/nokogiri/include/libxslt/attributes.h +39 -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 +1995 -0
- data/ext/nokogiri/include/libxslt/xsltconfig.h +146 -0
- data/ext/nokogiri/include/libxslt/xsltexports.h +64 -0
- data/ext/nokogiri/include/libxslt/xsltlocale.h +44 -0
- data/ext/nokogiri/include/libxslt/xsltutils.h +343 -0
- data/ext/nokogiri/libxml2_polyfill.c +114 -0
- data/ext/nokogiri/nokogiri.c +294 -0
- data/ext/nokogiri/nokogiri.h +238 -0
- data/ext/nokogiri/test_global_handlers.c +40 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_cdata.c +62 -0
- data/ext/nokogiri/xml_comment.c +57 -0
- data/ext/nokogiri/xml_document.c +784 -0
- data/ext/nokogiri/xml_document_fragment.c +29 -0
- data/ext/nokogiri/xml_dtd.c +208 -0
- data/ext/nokogiri/xml_element_content.c +131 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_encoding_handler.c +112 -0
- data/ext/nokogiri/xml_entity_decl.c +112 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -0
- data/ext/nokogiri/xml_namespace.c +181 -0
- data/ext/nokogiri/xml_node.c +2459 -0
- data/ext/nokogiri/xml_node_set.c +518 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_reader.c +777 -0
- data/ext/nokogiri/xml_relax_ng.c +149 -0
- data/ext/nokogiri/xml_sax_parser.c +403 -0
- data/ext/nokogiri/xml_sax_parser_context.c +390 -0
- data/ext/nokogiri/xml_sax_push_parser.c +206 -0
- data/ext/nokogiri/xml_schema.c +226 -0
- data/ext/nokogiri/xml_syntax_error.c +93 -0
- data/ext/nokogiri/xml_text.c +59 -0
- data/ext/nokogiri/xml_xpath_context.c +502 -0
- data/ext/nokogiri/xslt_stylesheet.c +421 -0
- data/gumbo-parser/CHANGES.md +63 -0
- data/gumbo-parser/Makefile +129 -0
- data/gumbo-parser/THANKS +27 -0
- data/lib/nokogiri/3.1/nokogiri.so +0 -0
- data/lib/nokogiri/3.2/nokogiri.so +0 -0
- data/lib/nokogiri/3.3/nokogiri.so +0 -0
- data/lib/nokogiri/3.4/nokogiri.so +0 -0
- data/lib/nokogiri/class_resolver.rb +67 -0
- data/lib/nokogiri/css/node.rb +58 -0
- data/lib/nokogiri/css/parser.rb +772 -0
- data/lib/nokogiri/css/parser.y +277 -0
- data/lib/nokogiri/css/parser_extras.rb +36 -0
- data/lib/nokogiri/css/selector_cache.rb +38 -0
- data/lib/nokogiri/css/syntax_error.rb +9 -0
- data/lib/nokogiri/css/tokenizer.rb +155 -0
- data/lib/nokogiri/css/tokenizer.rex +57 -0
- data/lib/nokogiri/css/xpath_visitor.rb +375 -0
- data/lib/nokogiri/css.rb +132 -0
- data/lib/nokogiri/decorators/slop.rb +42 -0
- data/lib/nokogiri/encoding_handler.rb +57 -0
- data/lib/nokogiri/extension.rb +32 -0
- data/lib/nokogiri/gumbo.rb +15 -0
- data/lib/nokogiri/html.rb +48 -0
- data/lib/nokogiri/html4/builder.rb +37 -0
- data/lib/nokogiri/html4/document.rb +235 -0
- data/lib/nokogiri/html4/document_fragment.rb +166 -0
- data/lib/nokogiri/html4/element_description.rb +25 -0
- data/lib/nokogiri/html4/element_description_defaults.rb +2040 -0
- data/lib/nokogiri/html4/encoding_reader.rb +121 -0
- data/lib/nokogiri/html4/entity_lookup.rb +15 -0
- data/lib/nokogiri/html4/sax/parser.rb +48 -0
- data/lib/nokogiri/html4/sax/parser_context.rb +15 -0
- data/lib/nokogiri/html4/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html4.rb +42 -0
- data/lib/nokogiri/html5/builder.rb +40 -0
- data/lib/nokogiri/html5/document.rb +199 -0
- data/lib/nokogiri/html5/document_fragment.rb +200 -0
- data/lib/nokogiri/html5/node.rb +103 -0
- data/lib/nokogiri/html5.rb +368 -0
- data/lib/nokogiri/jruby/dependencies.rb +3 -0
- data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
- data/lib/nokogiri/syntax_error.rb +6 -0
- data/lib/nokogiri/version/constant.rb +6 -0
- data/lib/nokogiri/version/info.rb +224 -0
- data/lib/nokogiri/version.rb +4 -0
- data/lib/nokogiri/xml/attr.rb +66 -0
- data/lib/nokogiri/xml/attribute_decl.rb +22 -0
- data/lib/nokogiri/xml/builder.rb +494 -0
- data/lib/nokogiri/xml/cdata.rb +13 -0
- data/lib/nokogiri/xml/character_data.rb +9 -0
- data/lib/nokogiri/xml/document.rb +514 -0
- data/lib/nokogiri/xml/document_fragment.rb +276 -0
- data/lib/nokogiri/xml/dtd.rb +34 -0
- data/lib/nokogiri/xml/element_content.rb +46 -0
- data/lib/nokogiri/xml/element_decl.rb +17 -0
- data/lib/nokogiri/xml/entity_decl.rb +23 -0
- data/lib/nokogiri/xml/entity_reference.rb +20 -0
- data/lib/nokogiri/xml/namespace.rb +57 -0
- data/lib/nokogiri/xml/node/save_options.rb +76 -0
- data/lib/nokogiri/xml/node.rb +1650 -0
- data/lib/nokogiri/xml/node_set.rb +449 -0
- data/lib/nokogiri/xml/notation.rb +19 -0
- data/lib/nokogiri/xml/parse_options.rb +213 -0
- data/lib/nokogiri/xml/pp/character_data.rb +21 -0
- data/lib/nokogiri/xml/pp/node.rb +73 -0
- data/lib/nokogiri/xml/pp.rb +4 -0
- data/lib/nokogiri/xml/processing_instruction.rb +11 -0
- data/lib/nokogiri/xml/reader.rb +139 -0
- data/lib/nokogiri/xml/relax_ng.rb +75 -0
- data/lib/nokogiri/xml/sax/document.rb +258 -0
- data/lib/nokogiri/xml/sax/parser.rb +199 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +129 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +64 -0
- data/lib/nokogiri/xml/sax.rb +54 -0
- data/lib/nokogiri/xml/schema.rb +140 -0
- data/lib/nokogiri/xml/searchable.rb +297 -0
- data/lib/nokogiri/xml/syntax_error.rb +94 -0
- data/lib/nokogiri/xml/text.rb +11 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +13 -0
- data/lib/nokogiri/xml/xpath.rb +21 -0
- data/lib/nokogiri/xml/xpath_context.rb +49 -0
- data/lib/nokogiri/xml.rb +65 -0
- data/lib/nokogiri/xslt/stylesheet.rb +49 -0
- data/lib/nokogiri/xslt.rb +129 -0
- data/lib/nokogiri.rb +128 -0
- data/lib/xsd/xmlparser/nokogiri.rb +105 -0
- metadata +324 -0
@@ -0,0 +1,96 @@
|
|
1
|
+
#include <nokogiri.h>
|
2
|
+
|
3
|
+
VALUE cNokogiriHtml4SaxPushParser;
|
4
|
+
|
5
|
+
/*
|
6
|
+
* Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
|
7
|
+
*/
|
8
|
+
static VALUE
|
9
|
+
noko_html4_sax_push_parser__native_write(VALUE self, VALUE rb_chunk, VALUE rb_last_chunk)
|
10
|
+
{
|
11
|
+
xmlParserCtxtPtr ctx;
|
12
|
+
const char *chunk = NULL;
|
13
|
+
int size = 0;
|
14
|
+
int status = 0;
|
15
|
+
libxmlStructuredErrorHandlerState handler_state;
|
16
|
+
|
17
|
+
ctx = noko_xml_sax_push_parser_unwrap(self);
|
18
|
+
|
19
|
+
if (Qnil != rb_chunk) {
|
20
|
+
chunk = StringValuePtr(rb_chunk);
|
21
|
+
size = (int)RSTRING_LEN(rb_chunk);
|
22
|
+
}
|
23
|
+
|
24
|
+
noko__structured_error_func_save_and_set(&handler_state, NULL, NULL);
|
25
|
+
|
26
|
+
status = htmlParseChunk(ctx, chunk, size, Qtrue == rb_last_chunk ? 1 : 0);
|
27
|
+
|
28
|
+
noko__structured_error_func_restore(&handler_state);
|
29
|
+
|
30
|
+
if ((status != 0) && !(xmlCtxtGetOptions(ctx) & XML_PARSE_RECOVER)) {
|
31
|
+
// TODO: there appear to be no tests for this block
|
32
|
+
xmlErrorConstPtr e = xmlCtxtGetLastError(ctx);
|
33
|
+
noko__error_raise(NULL, e);
|
34
|
+
}
|
35
|
+
|
36
|
+
return self;
|
37
|
+
}
|
38
|
+
|
39
|
+
/*
|
40
|
+
* Initialize the push parser with +xml_sax+ using +filename+
|
41
|
+
*/
|
42
|
+
static VALUE
|
43
|
+
noko_html4_sax_push_parser__initialize_native(
|
44
|
+
VALUE self,
|
45
|
+
VALUE rb_xml_sax,
|
46
|
+
VALUE rb_filename,
|
47
|
+
VALUE encoding
|
48
|
+
)
|
49
|
+
{
|
50
|
+
htmlSAXHandlerPtr sax;
|
51
|
+
const char *filename = NULL;
|
52
|
+
htmlParserCtxtPtr ctx;
|
53
|
+
xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
|
54
|
+
|
55
|
+
sax = noko_xml_sax_parser_unwrap(rb_xml_sax);
|
56
|
+
|
57
|
+
if (rb_filename != Qnil) { filename = StringValueCStr(rb_filename); }
|
58
|
+
|
59
|
+
if (!NIL_P(encoding)) {
|
60
|
+
enc = xmlParseCharEncoding(StringValueCStr(encoding));
|
61
|
+
if (enc == XML_CHAR_ENCODING_ERROR) {
|
62
|
+
rb_raise(rb_eArgError, "Unsupported Encoding");
|
63
|
+
}
|
64
|
+
}
|
65
|
+
|
66
|
+
ctx = htmlCreatePushParserCtxt(
|
67
|
+
sax,
|
68
|
+
NULL,
|
69
|
+
NULL,
|
70
|
+
0,
|
71
|
+
filename,
|
72
|
+
enc
|
73
|
+
);
|
74
|
+
if (ctx == NULL) {
|
75
|
+
rb_raise(rb_eRuntimeError, "Could not create a parser context");
|
76
|
+
}
|
77
|
+
|
78
|
+
ctx->userData = ctx;
|
79
|
+
ctx->_private = (void *)rb_xml_sax;
|
80
|
+
|
81
|
+
DATA_PTR(self) = ctx;
|
82
|
+
return self;
|
83
|
+
}
|
84
|
+
|
85
|
+
void
|
86
|
+
noko_init_html_sax_push_parser(void)
|
87
|
+
{
|
88
|
+
assert(cNokogiriXmlSaxPushParser);
|
89
|
+
cNokogiriHtml4SaxPushParser =
|
90
|
+
rb_define_class_under(mNokogiriHtml4Sax, "PushParser", cNokogiriXmlSaxPushParser);
|
91
|
+
|
92
|
+
rb_define_private_method(cNokogiriHtml4SaxPushParser, "initialize_native",
|
93
|
+
noko_html4_sax_push_parser__initialize_native, 3);
|
94
|
+
rb_define_private_method(cNokogiriHtml4SaxPushParser, "native_write",
|
95
|
+
noko_html4_sax_push_parser__native_write, 2);
|
96
|
+
}
|
@@ -0,0 +1,108 @@
|
|
1
|
+
/*
|
2
|
+
* Summary: main header file
|
3
|
+
*
|
4
|
+
* Copy: See Copyright for the status of this software.
|
5
|
+
*/
|
6
|
+
|
7
|
+
|
8
|
+
#ifndef __EXSLT_H__
|
9
|
+
#define __EXSLT_H__
|
10
|
+
|
11
|
+
#include <libxml/tree.h>
|
12
|
+
#include <libxml/xpath.h>
|
13
|
+
#include "exsltexports.h"
|
14
|
+
#include <libexslt/exsltconfig.h>
|
15
|
+
|
16
|
+
#ifdef __cplusplus
|
17
|
+
extern "C" {
|
18
|
+
#endif
|
19
|
+
|
20
|
+
EXSLTPUBVAR const char *exsltLibraryVersion;
|
21
|
+
EXSLTPUBVAR const int exsltLibexsltVersion;
|
22
|
+
EXSLTPUBVAR const int exsltLibxsltVersion;
|
23
|
+
EXSLTPUBVAR const int exsltLibxmlVersion;
|
24
|
+
|
25
|
+
/**
|
26
|
+
* EXSLT_COMMON_NAMESPACE:
|
27
|
+
*
|
28
|
+
* Namespace for EXSLT common functions
|
29
|
+
*/
|
30
|
+
#define EXSLT_COMMON_NAMESPACE ((const xmlChar *) "http://exslt.org/common")
|
31
|
+
/**
|
32
|
+
* EXSLT_CRYPTO_NAMESPACE:
|
33
|
+
*
|
34
|
+
* Namespace for EXSLT crypto functions
|
35
|
+
*/
|
36
|
+
#define EXSLT_CRYPTO_NAMESPACE ((const xmlChar *) "http://exslt.org/crypto")
|
37
|
+
/**
|
38
|
+
* EXSLT_MATH_NAMESPACE:
|
39
|
+
*
|
40
|
+
* Namespace for EXSLT math functions
|
41
|
+
*/
|
42
|
+
#define EXSLT_MATH_NAMESPACE ((const xmlChar *) "http://exslt.org/math")
|
43
|
+
/**
|
44
|
+
* EXSLT_SETS_NAMESPACE:
|
45
|
+
*
|
46
|
+
* Namespace for EXSLT set functions
|
47
|
+
*/
|
48
|
+
#define EXSLT_SETS_NAMESPACE ((const xmlChar *) "http://exslt.org/sets")
|
49
|
+
/**
|
50
|
+
* EXSLT_FUNCTIONS_NAMESPACE:
|
51
|
+
*
|
52
|
+
* Namespace for EXSLT functions extension functions
|
53
|
+
*/
|
54
|
+
#define EXSLT_FUNCTIONS_NAMESPACE ((const xmlChar *) "http://exslt.org/functions")
|
55
|
+
/**
|
56
|
+
* EXSLT_STRINGS_NAMESPACE:
|
57
|
+
*
|
58
|
+
* Namespace for EXSLT strings functions
|
59
|
+
*/
|
60
|
+
#define EXSLT_STRINGS_NAMESPACE ((const xmlChar *) "http://exslt.org/strings")
|
61
|
+
/**
|
62
|
+
* EXSLT_DATE_NAMESPACE:
|
63
|
+
*
|
64
|
+
* Namespace for EXSLT date functions
|
65
|
+
*/
|
66
|
+
#define EXSLT_DATE_NAMESPACE ((const xmlChar *) "http://exslt.org/dates-and-times")
|
67
|
+
/**
|
68
|
+
* EXSLT_DYNAMIC_NAMESPACE:
|
69
|
+
*
|
70
|
+
* Namespace for EXSLT dynamic functions
|
71
|
+
*/
|
72
|
+
#define EXSLT_DYNAMIC_NAMESPACE ((const xmlChar *) "http://exslt.org/dynamic")
|
73
|
+
|
74
|
+
/**
|
75
|
+
* SAXON_NAMESPACE:
|
76
|
+
*
|
77
|
+
* Namespace for SAXON extensions functions
|
78
|
+
*/
|
79
|
+
#define SAXON_NAMESPACE ((const xmlChar *) "http://icl.com/saxon")
|
80
|
+
|
81
|
+
EXSLTPUBFUN void EXSLTCALL exsltCommonRegister (void);
|
82
|
+
#ifdef EXSLT_CRYPTO_ENABLED
|
83
|
+
EXSLTPUBFUN void EXSLTCALL exsltCryptoRegister (void);
|
84
|
+
#endif
|
85
|
+
EXSLTPUBFUN void EXSLTCALL exsltMathRegister (void);
|
86
|
+
EXSLTPUBFUN void EXSLTCALL exsltSetsRegister (void);
|
87
|
+
EXSLTPUBFUN void EXSLTCALL exsltFuncRegister (void);
|
88
|
+
EXSLTPUBFUN void EXSLTCALL exsltStrRegister (void);
|
89
|
+
EXSLTPUBFUN void EXSLTCALL exsltDateRegister (void);
|
90
|
+
EXSLTPUBFUN void EXSLTCALL exsltSaxonRegister (void);
|
91
|
+
EXSLTPUBFUN void EXSLTCALL exsltDynRegister(void);
|
92
|
+
|
93
|
+
EXSLTPUBFUN void EXSLTCALL exsltRegisterAll (void);
|
94
|
+
|
95
|
+
EXSLTPUBFUN int EXSLTCALL exsltDateXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
96
|
+
const xmlChar *prefix);
|
97
|
+
EXSLTPUBFUN int EXSLTCALL exsltMathXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
98
|
+
const xmlChar *prefix);
|
99
|
+
EXSLTPUBFUN int EXSLTCALL exsltSetsXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
100
|
+
const xmlChar *prefix);
|
101
|
+
EXSLTPUBFUN int EXSLTCALL exsltStrXpathCtxtRegister (xmlXPathContextPtr ctxt,
|
102
|
+
const xmlChar *prefix);
|
103
|
+
|
104
|
+
#ifdef __cplusplus
|
105
|
+
}
|
106
|
+
#endif
|
107
|
+
#endif /* __EXSLT_H__ */
|
108
|
+
|
@@ -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 "0.8.23"
|
22
|
+
|
23
|
+
/**
|
24
|
+
* LIBEXSLT_VERSION:
|
25
|
+
*
|
26
|
+
* the version number: 1.2.3 value is 10203
|
27
|
+
*/
|
28
|
+
#define LIBEXSLT_VERSION 823
|
29
|
+
|
30
|
+
/**
|
31
|
+
* LIBEXSLT_VERSION_STRING:
|
32
|
+
*
|
33
|
+
* the version number string, 1.2.3 value is "10203"
|
34
|
+
*/
|
35
|
+
#define LIBEXSLT_VERSION_STRING "823"
|
36
|
+
|
37
|
+
/**
|
38
|
+
* LIBEXSLT_VERSION_EXTRA:
|
39
|
+
*
|
40
|
+
* extra version information, used to show a Git commit description
|
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__ */
|
@@ -0,0 +1,63 @@
|
|
1
|
+
/*
|
2
|
+
* Summary: macros for marking symbols as exportable/importable.
|
3
|
+
*
|
4
|
+
* Copy: See Copyright for the status of this software.
|
5
|
+
*/
|
6
|
+
|
7
|
+
#ifndef __EXSLT_EXPORTS_H__
|
8
|
+
#define __EXSLT_EXPORTS_H__
|
9
|
+
|
10
|
+
#if defined(_WIN32) || defined(__CYGWIN__)
|
11
|
+
/** DOC_DISABLE */
|
12
|
+
|
13
|
+
#ifdef LIBEXSLT_STATIC
|
14
|
+
#define EXSLTPUBLIC
|
15
|
+
#elif defined(IN_LIBEXSLT)
|
16
|
+
#define EXSLTPUBLIC __declspec(dllexport)
|
17
|
+
#else
|
18
|
+
#define EXSLTPUBLIC __declspec(dllimport)
|
19
|
+
#endif
|
20
|
+
|
21
|
+
#define EXSLTCALL __cdecl
|
22
|
+
|
23
|
+
/** DOC_ENABLE */
|
24
|
+
#else /* not Windows */
|
25
|
+
|
26
|
+
/**
|
27
|
+
* EXSLTPUBLIC:
|
28
|
+
*
|
29
|
+
* Macro which declares a public symbol
|
30
|
+
*/
|
31
|
+
#define EXSLTPUBLIC
|
32
|
+
|
33
|
+
/**
|
34
|
+
* EXSLTCALL:
|
35
|
+
*
|
36
|
+
* Macro which declares the calling convention for exported functions
|
37
|
+
*/
|
38
|
+
#define EXSLTCALL
|
39
|
+
|
40
|
+
#endif /* platform switch */
|
41
|
+
|
42
|
+
/*
|
43
|
+
* EXSLTPUBFUN:
|
44
|
+
*
|
45
|
+
* Macro which declares an exportable function
|
46
|
+
*/
|
47
|
+
#define EXSLTPUBFUN EXSLTPUBLIC
|
48
|
+
|
49
|
+
/**
|
50
|
+
* EXSLTPUBVAR:
|
51
|
+
*
|
52
|
+
* Macro which declares an exportable variable
|
53
|
+
*/
|
54
|
+
#define EXSLTPUBVAR EXSLTPUBLIC extern
|
55
|
+
|
56
|
+
/* Compatibility */
|
57
|
+
#if !defined(LIBEXSLT_PUBLIC)
|
58
|
+
#define LIBEXSLT_PUBLIC EXSLTPUBVAR
|
59
|
+
#endif
|
60
|
+
|
61
|
+
#endif /* __EXSLT_EXPORTS_H__ */
|
62
|
+
|
63
|
+
|
@@ -0,0 +1,336 @@
|
|
1
|
+
/*
|
2
|
+
* Summary: interface for an HTML 4.0 non-verifying parser
|
3
|
+
* Description: this module implements an HTML 4.0 non-verifying parser
|
4
|
+
* with API compatible with the XML parser ones. It should
|
5
|
+
* be able to parse "real world" HTML, even if severely
|
6
|
+
* broken from a specification point of view.
|
7
|
+
*
|
8
|
+
* Copy: See Copyright for the status of this software.
|
9
|
+
*
|
10
|
+
* Author: Daniel Veillard
|
11
|
+
*/
|
12
|
+
|
13
|
+
#ifndef __HTML_PARSER_H__
|
14
|
+
#define __HTML_PARSER_H__
|
15
|
+
#include <libxml/xmlversion.h>
|
16
|
+
#include <libxml/parser.h>
|
17
|
+
|
18
|
+
#ifdef LIBXML_HTML_ENABLED
|
19
|
+
|
20
|
+
#ifdef __cplusplus
|
21
|
+
extern "C" {
|
22
|
+
#endif
|
23
|
+
|
24
|
+
/*
|
25
|
+
* Most of the back-end structures from XML and HTML are shared.
|
26
|
+
*/
|
27
|
+
typedef xmlParserCtxt htmlParserCtxt;
|
28
|
+
typedef xmlParserCtxtPtr htmlParserCtxtPtr;
|
29
|
+
typedef xmlParserNodeInfo htmlParserNodeInfo;
|
30
|
+
typedef xmlSAXHandler htmlSAXHandler;
|
31
|
+
typedef xmlSAXHandlerPtr htmlSAXHandlerPtr;
|
32
|
+
typedef xmlParserInput htmlParserInput;
|
33
|
+
typedef xmlParserInputPtr htmlParserInputPtr;
|
34
|
+
typedef xmlDocPtr htmlDocPtr;
|
35
|
+
typedef xmlNodePtr htmlNodePtr;
|
36
|
+
|
37
|
+
/*
|
38
|
+
* Internal description of an HTML element, representing HTML 4.01
|
39
|
+
* and XHTML 1.0 (which share the same structure).
|
40
|
+
*/
|
41
|
+
typedef struct _htmlElemDesc htmlElemDesc;
|
42
|
+
typedef htmlElemDesc *htmlElemDescPtr;
|
43
|
+
struct _htmlElemDesc {
|
44
|
+
const char *name; /* The tag name */
|
45
|
+
char startTag; /* Whether the start tag can be implied */
|
46
|
+
char endTag; /* Whether the end tag can be implied */
|
47
|
+
char saveEndTag; /* Whether the end tag should be saved */
|
48
|
+
char empty; /* Is this an empty element ? */
|
49
|
+
char depr; /* Is this a deprecated element ? */
|
50
|
+
char dtd; /* 1: only in Loose DTD, 2: only Frameset one */
|
51
|
+
char isinline; /* is this a block 0 or inline 1 element */
|
52
|
+
const char *desc; /* the description */
|
53
|
+
|
54
|
+
/* NRK Jan.2003
|
55
|
+
* New fields encapsulating HTML structure
|
56
|
+
*
|
57
|
+
* Bugs:
|
58
|
+
* This is a very limited representation. It fails to tell us when
|
59
|
+
* an element *requires* subelements (we only have whether they're
|
60
|
+
* allowed or not), and it doesn't tell us where CDATA and PCDATA
|
61
|
+
* are allowed. Some element relationships are not fully represented:
|
62
|
+
* these are flagged with the word MODIFIER
|
63
|
+
*/
|
64
|
+
const char** subelts; /* allowed sub-elements of this element */
|
65
|
+
const char* defaultsubelt; /* subelement for suggested auto-repair
|
66
|
+
if necessary or NULL */
|
67
|
+
const char** attrs_opt; /* Optional Attributes */
|
68
|
+
const char** attrs_depr; /* Additional deprecated attributes */
|
69
|
+
const char** attrs_req; /* Required attributes */
|
70
|
+
};
|
71
|
+
|
72
|
+
/*
|
73
|
+
* Internal description of an HTML entity.
|
74
|
+
*/
|
75
|
+
typedef struct _htmlEntityDesc htmlEntityDesc;
|
76
|
+
typedef htmlEntityDesc *htmlEntityDescPtr;
|
77
|
+
struct _htmlEntityDesc {
|
78
|
+
unsigned int value; /* the UNICODE value for the character */
|
79
|
+
const char *name; /* The entity name */
|
80
|
+
const char *desc; /* the description */
|
81
|
+
};
|
82
|
+
|
83
|
+
#ifdef LIBXML_SAX1_ENABLED
|
84
|
+
|
85
|
+
XML_DEPRECATED
|
86
|
+
XMLPUBVAR const xmlSAXHandlerV1 htmlDefaultSAXHandler;
|
87
|
+
|
88
|
+
#ifdef LIBXML_THREAD_ENABLED
|
89
|
+
XML_DEPRECATED
|
90
|
+
XMLPUBFUN const xmlSAXHandlerV1 *__htmlDefaultSAXHandler(void);
|
91
|
+
#endif
|
92
|
+
|
93
|
+
#endif /* LIBXML_SAX1_ENABLED */
|
94
|
+
|
95
|
+
/*
|
96
|
+
* There is only few public functions.
|
97
|
+
*/
|
98
|
+
XML_DEPRECATED
|
99
|
+
XMLPUBFUN void
|
100
|
+
htmlInitAutoClose (void);
|
101
|
+
XMLPUBFUN const htmlElemDesc *
|
102
|
+
htmlTagLookup (const xmlChar *tag);
|
103
|
+
XMLPUBFUN const htmlEntityDesc *
|
104
|
+
htmlEntityLookup(const xmlChar *name);
|
105
|
+
XMLPUBFUN const htmlEntityDesc *
|
106
|
+
htmlEntityValueLookup(unsigned int value);
|
107
|
+
|
108
|
+
XMLPUBFUN int
|
109
|
+
htmlIsAutoClosed(htmlDocPtr doc,
|
110
|
+
htmlNodePtr elem);
|
111
|
+
XMLPUBFUN int
|
112
|
+
htmlAutoCloseTag(htmlDocPtr doc,
|
113
|
+
const xmlChar *name,
|
114
|
+
htmlNodePtr elem);
|
115
|
+
XML_DEPRECATED
|
116
|
+
XMLPUBFUN const htmlEntityDesc *
|
117
|
+
htmlParseEntityRef(htmlParserCtxtPtr ctxt,
|
118
|
+
const xmlChar **str);
|
119
|
+
XML_DEPRECATED
|
120
|
+
XMLPUBFUN int
|
121
|
+
htmlParseCharRef(htmlParserCtxtPtr ctxt);
|
122
|
+
XML_DEPRECATED
|
123
|
+
XMLPUBFUN void
|
124
|
+
htmlParseElement(htmlParserCtxtPtr ctxt);
|
125
|
+
|
126
|
+
XMLPUBFUN htmlParserCtxtPtr
|
127
|
+
htmlNewParserCtxt(void);
|
128
|
+
XMLPUBFUN htmlParserCtxtPtr
|
129
|
+
htmlNewSAXParserCtxt(const htmlSAXHandler *sax,
|
130
|
+
void *userData);
|
131
|
+
|
132
|
+
XMLPUBFUN htmlParserCtxtPtr
|
133
|
+
htmlCreateMemoryParserCtxt(const char *buffer,
|
134
|
+
int size);
|
135
|
+
|
136
|
+
XMLPUBFUN int
|
137
|
+
htmlParseDocument(htmlParserCtxtPtr ctxt);
|
138
|
+
XML_DEPRECATED
|
139
|
+
XMLPUBFUN htmlDocPtr
|
140
|
+
htmlSAXParseDoc (const xmlChar *cur,
|
141
|
+
const char *encoding,
|
142
|
+
htmlSAXHandlerPtr sax,
|
143
|
+
void *userData);
|
144
|
+
XMLPUBFUN htmlDocPtr
|
145
|
+
htmlParseDoc (const xmlChar *cur,
|
146
|
+
const char *encoding);
|
147
|
+
XMLPUBFUN htmlParserCtxtPtr
|
148
|
+
htmlCreateFileParserCtxt(const char *filename,
|
149
|
+
const char *encoding);
|
150
|
+
XML_DEPRECATED
|
151
|
+
XMLPUBFUN htmlDocPtr
|
152
|
+
htmlSAXParseFile(const char *filename,
|
153
|
+
const char *encoding,
|
154
|
+
htmlSAXHandlerPtr sax,
|
155
|
+
void *userData);
|
156
|
+
XMLPUBFUN htmlDocPtr
|
157
|
+
htmlParseFile (const char *filename,
|
158
|
+
const char *encoding);
|
159
|
+
XMLPUBFUN int
|
160
|
+
UTF8ToHtml (unsigned char *out,
|
161
|
+
int *outlen,
|
162
|
+
const unsigned char *in,
|
163
|
+
int *inlen);
|
164
|
+
XMLPUBFUN int
|
165
|
+
htmlEncodeEntities(unsigned char *out,
|
166
|
+
int *outlen,
|
167
|
+
const unsigned char *in,
|
168
|
+
int *inlen, int quoteChar);
|
169
|
+
XMLPUBFUN int
|
170
|
+
htmlIsScriptAttribute(const xmlChar *name);
|
171
|
+
XML_DEPRECATED
|
172
|
+
XMLPUBFUN int
|
173
|
+
htmlHandleOmittedElem(int val);
|
174
|
+
|
175
|
+
#ifdef LIBXML_PUSH_ENABLED
|
176
|
+
/**
|
177
|
+
* Interfaces for the Push mode.
|
178
|
+
*/
|
179
|
+
XMLPUBFUN htmlParserCtxtPtr
|
180
|
+
htmlCreatePushParserCtxt(htmlSAXHandlerPtr sax,
|
181
|
+
void *user_data,
|
182
|
+
const char *chunk,
|
183
|
+
int size,
|
184
|
+
const char *filename,
|
185
|
+
xmlCharEncoding enc);
|
186
|
+
XMLPUBFUN int
|
187
|
+
htmlParseChunk (htmlParserCtxtPtr ctxt,
|
188
|
+
const char *chunk,
|
189
|
+
int size,
|
190
|
+
int terminate);
|
191
|
+
#endif /* LIBXML_PUSH_ENABLED */
|
192
|
+
|
193
|
+
XMLPUBFUN void
|
194
|
+
htmlFreeParserCtxt (htmlParserCtxtPtr ctxt);
|
195
|
+
|
196
|
+
/*
|
197
|
+
* New set of simpler/more flexible APIs
|
198
|
+
*/
|
199
|
+
/**
|
200
|
+
* xmlParserOption:
|
201
|
+
*
|
202
|
+
* This is the set of XML parser options that can be passed down
|
203
|
+
* to the xmlReadDoc() and similar calls.
|
204
|
+
*/
|
205
|
+
typedef enum {
|
206
|
+
HTML_PARSE_RECOVER = 1<<0, /* Relaxed parsing */
|
207
|
+
HTML_PARSE_NODEFDTD = 1<<2, /* do not default a doctype if not found */
|
208
|
+
HTML_PARSE_NOERROR = 1<<5, /* suppress error reports */
|
209
|
+
HTML_PARSE_NOWARNING= 1<<6, /* suppress warning reports */
|
210
|
+
HTML_PARSE_PEDANTIC = 1<<7, /* pedantic error reporting */
|
211
|
+
HTML_PARSE_NOBLANKS = 1<<8, /* remove blank nodes */
|
212
|
+
HTML_PARSE_NONET = 1<<11,/* Forbid network access */
|
213
|
+
HTML_PARSE_NOIMPLIED= 1<<13,/* Do not add implied html/body... elements */
|
214
|
+
HTML_PARSE_COMPACT = 1<<16,/* compact small text nodes */
|
215
|
+
HTML_PARSE_IGNORE_ENC=1<<21 /* ignore internal document encoding hint */
|
216
|
+
} htmlParserOption;
|
217
|
+
|
218
|
+
XMLPUBFUN void
|
219
|
+
htmlCtxtReset (htmlParserCtxtPtr ctxt);
|
220
|
+
XMLPUBFUN int
|
221
|
+
htmlCtxtUseOptions (htmlParserCtxtPtr ctxt,
|
222
|
+
int options);
|
223
|
+
XMLPUBFUN htmlDocPtr
|
224
|
+
htmlReadDoc (const xmlChar *cur,
|
225
|
+
const char *URL,
|
226
|
+
const char *encoding,
|
227
|
+
int options);
|
228
|
+
XMLPUBFUN htmlDocPtr
|
229
|
+
htmlReadFile (const char *URL,
|
230
|
+
const char *encoding,
|
231
|
+
int options);
|
232
|
+
XMLPUBFUN htmlDocPtr
|
233
|
+
htmlReadMemory (const char *buffer,
|
234
|
+
int size,
|
235
|
+
const char *URL,
|
236
|
+
const char *encoding,
|
237
|
+
int options);
|
238
|
+
XMLPUBFUN htmlDocPtr
|
239
|
+
htmlReadFd (int fd,
|
240
|
+
const char *URL,
|
241
|
+
const char *encoding,
|
242
|
+
int options);
|
243
|
+
XMLPUBFUN htmlDocPtr
|
244
|
+
htmlReadIO (xmlInputReadCallback ioread,
|
245
|
+
xmlInputCloseCallback ioclose,
|
246
|
+
void *ioctx,
|
247
|
+
const char *URL,
|
248
|
+
const char *encoding,
|
249
|
+
int options);
|
250
|
+
XMLPUBFUN htmlDocPtr
|
251
|
+
htmlCtxtParseDocument (htmlParserCtxtPtr ctxt,
|
252
|
+
xmlParserInputPtr input);
|
253
|
+
XMLPUBFUN htmlDocPtr
|
254
|
+
htmlCtxtReadDoc (xmlParserCtxtPtr ctxt,
|
255
|
+
const xmlChar *cur,
|
256
|
+
const char *URL,
|
257
|
+
const char *encoding,
|
258
|
+
int options);
|
259
|
+
XMLPUBFUN htmlDocPtr
|
260
|
+
htmlCtxtReadFile (xmlParserCtxtPtr ctxt,
|
261
|
+
const char *filename,
|
262
|
+
const char *encoding,
|
263
|
+
int options);
|
264
|
+
XMLPUBFUN htmlDocPtr
|
265
|
+
htmlCtxtReadMemory (xmlParserCtxtPtr ctxt,
|
266
|
+
const char *buffer,
|
267
|
+
int size,
|
268
|
+
const char *URL,
|
269
|
+
const char *encoding,
|
270
|
+
int options);
|
271
|
+
XMLPUBFUN htmlDocPtr
|
272
|
+
htmlCtxtReadFd (xmlParserCtxtPtr ctxt,
|
273
|
+
int fd,
|
274
|
+
const char *URL,
|
275
|
+
const char *encoding,
|
276
|
+
int options);
|
277
|
+
XMLPUBFUN htmlDocPtr
|
278
|
+
htmlCtxtReadIO (xmlParserCtxtPtr ctxt,
|
279
|
+
xmlInputReadCallback ioread,
|
280
|
+
xmlInputCloseCallback ioclose,
|
281
|
+
void *ioctx,
|
282
|
+
const char *URL,
|
283
|
+
const char *encoding,
|
284
|
+
int options);
|
285
|
+
|
286
|
+
/* NRK/Jan2003: further knowledge of HTML structure
|
287
|
+
*/
|
288
|
+
typedef enum {
|
289
|
+
HTML_NA = 0 , /* something we don't check at all */
|
290
|
+
HTML_INVALID = 0x1 ,
|
291
|
+
HTML_DEPRECATED = 0x2 ,
|
292
|
+
HTML_VALID = 0x4 ,
|
293
|
+
HTML_REQUIRED = 0xc /* VALID bit set so ( & HTML_VALID ) is TRUE */
|
294
|
+
} htmlStatus ;
|
295
|
+
|
296
|
+
/* Using htmlElemDesc rather than name here, to emphasise the fact
|
297
|
+
that otherwise there's a lookup overhead
|
298
|
+
*/
|
299
|
+
XMLPUBFUN htmlStatus htmlAttrAllowed(const htmlElemDesc*, const xmlChar*, int) ;
|
300
|
+
XMLPUBFUN int htmlElementAllowedHere(const htmlElemDesc*, const xmlChar*) ;
|
301
|
+
XMLPUBFUN htmlStatus htmlElementStatusHere(const htmlElemDesc*, const htmlElemDesc*) ;
|
302
|
+
XMLPUBFUN htmlStatus htmlNodeStatus(htmlNodePtr, int) ;
|
303
|
+
/**
|
304
|
+
* htmlDefaultSubelement:
|
305
|
+
* @elt: HTML element
|
306
|
+
*
|
307
|
+
* Returns the default subelement for this element
|
308
|
+
*/
|
309
|
+
#define htmlDefaultSubelement(elt) elt->defaultsubelt
|
310
|
+
/**
|
311
|
+
* htmlElementAllowedHereDesc:
|
312
|
+
* @parent: HTML parent element
|
313
|
+
* @elt: HTML element
|
314
|
+
*
|
315
|
+
* Checks whether an HTML element description may be a
|
316
|
+
* direct child of the specified element.
|
317
|
+
*
|
318
|
+
* Returns 1 if allowed; 0 otherwise.
|
319
|
+
*/
|
320
|
+
#define htmlElementAllowedHereDesc(parent,elt) \
|
321
|
+
htmlElementAllowedHere((parent), (elt)->name)
|
322
|
+
/**
|
323
|
+
* htmlRequiredAttrs:
|
324
|
+
* @elt: HTML element
|
325
|
+
*
|
326
|
+
* Returns the attributes required for the specified element.
|
327
|
+
*/
|
328
|
+
#define htmlRequiredAttrs(elt) (elt)->attrs_req
|
329
|
+
|
330
|
+
|
331
|
+
#ifdef __cplusplus
|
332
|
+
}
|
333
|
+
#endif
|
334
|
+
|
335
|
+
#endif /* LIBXML_HTML_ENABLED */
|
336
|
+
#endif /* __HTML_PARSER_H__ */
|