class Nokogiri::HTML::SAX::PushParser
Attributes
document[RW]
The Nokogiri::HTML::SAX::Document on which the PushParser will be operating
Public Class Methods
new(doc = HTML::SAX::Document.new, file_name = nil, encoding = 'UTF-8')
click to toggle source
# File lib/nokogiri/html/sax/push_parser.rb, line 10 def initialize(doc = HTML::SAX::Document.new, file_name = nil, encoding = 'UTF-8') @document = doc @encoding = encoding @sax_parser = HTML::SAX::Parser.new(doc, @encoding) ## Create our push parser context initialize_native(@sax_parser, file_name, encoding) end
Public Instance Methods
finish()
click to toggle source
Finish the parsing. This method is only necessary for Nokogiri::HTML::SAX::Document#end_document to be called.
# File lib/nokogiri/html/sax/push_parser.rb, line 30 def finish write '', true end
write(chunk, last_chunk = false)
click to toggle source
Write a chunk of HTML to the PushParser. Any callback methods that can be
called will be called immediately.
# File lib/nokogiri/html/sax/push_parser.rb, line 22 def write chunk, last_chunk = false native_write(chunk, last_chunk) end
Also aliased as: <<
Private Instance Methods
initialize_native(xml_sax, filename)
click to toggle source
Initialize the push parser with xml_sax using
filename
static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
VALUE encoding)
{
htmlSAXHandlerPtr sax;
const char * filename = NULL;
htmlParserCtxtPtr ctx;
xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
if(_filename != Qnil) filename = StringValueCStr(_filename);
if (!NIL_P(encoding)) {
enc = xmlParseCharEncoding(StringValueCStr(encoding));
if (enc == XML_CHAR_ENCODING_ERROR)
rb_raise(rb_eArgError, "Unsupported Encoding");
}
ctx = htmlCreatePushParserCtxt(
sax,
NULL,
NULL,
0,
filename,
enc
);
if(ctx == NULL)
rb_raise(rb_eRuntimeError, "Could not create a parser context");
ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
ctx->sax2 = 1;
DATA_PTR(self) = ctx;
return self;
}
native_write(chunk, last_chunk)
click to toggle source
Write chunk to PushParser.
last_chunk triggers the end_document handle
static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
{
xmlParserCtxtPtr ctx;
const char * chunk = NULL;
int size = 0;
Data_Get_Struct(self, xmlParserCtxt, ctx);
if(Qnil != _chunk) {
chunk = StringValuePtr(_chunk);
size = (int)RSTRING_LEN(_chunk);
}
if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
if (!(ctx->options & XML_PARSE_RECOVER)) {
xmlErrorPtr e = xmlCtxtGetLastError(ctx);
Nokogiri_error_raise(NULL, e);
}
}
return self;
}