template <typename R> bool extract_control(R& result,tokenizer& tokenizer)
{
result.clear();
//controls
dictionary<string,id<string>> controls=
{
"\r",symbols()._cr,
"\n",symbols()._lf
};
//buffer
array<ascii> buffer;
//any
if(!tokenizer.any(buffer,controls.keys()))
return false;
//commit
result=buffer;
update(controls[buffer.join("")],buffer);
return true;
}
template <typename R> bool extract_number(R& result,tokenizer& tokenizer)
{
result.clear();
//buffer
array<ascii> buffer;
array<ascii> current;
//sign
if(tokenizer.any(current,pack("+","-")))
buffer.append(current);
//integer part
if(tokenizer.digit(current))
buffer.append(current);
//real part
if(tokenizer.delimit(current,"."))
{
buffer.append(current);
if(tokenizer.digit(current))
buffer.append(current);
}
//parse with c api
if(!buffer.join("").is_real())
return false;
//no dot after
if(tokenizer.starts_with("."))
return false;
//delimited
if(!tokenizer.is_delimited())
return false;
//commit
result=buffer;
update(symbols()._number,buffer);
return true;
}
template <typename R> bool extract_escape(R& result,tokenizer& tokenizer)
{
result.clear();
//controls
dictionary<string,ascii> controls=
{
"\\r",'\r',
"\\n",'\n',
"\\t",'\t'
};
//buffer
R content;
array<ascii> buffer;
array<ascii> current;
//escape
if(tokenizer.delimit(current,"\\x"))
{
//hexa
buffer.append(current);
//hexa
if(!tokenizer.xdigit(current,2))
return false;
buffer.append(current);
//parse with c api
int value=0;
if(!current.join("").parse_integer(value,16))
return false;
//byte
if(!between(value,0,256))
return false;
//push
content.push(value);
}
else if(tokenizer.any(current,controls.keys()))
{
//control
buffer.append(current);
content.push(controls[current]);
}
else if(tokenizer.delimit(current,"\\"))
{
//backslash
buffer.append(current);
//next
if(!tokenizer.shift(current,1))
return false;
content.append(current);
buffer.append(current);
}
else
return false;
//commit
result=content;
update(symbols()._escape,buffer);
return true;
}
template <typename R> bool extract_delimiter(R& result,tokenizer& tokenizer)
{
result.clear();
//delimiters
array<string> delimiters=
{
//comment
"//",
"/*",
"*/",
//literal
"\"",
"'",
//block
"(",
")",
"[",
"]",
"{",
"}",
//separator
"/",
"//",
",",
";",
".",
"..",
"...",
":",
"::",
//operator
"->",
"~",
"%",
"^",
"=",
"==",
"===",
">",
">>",
">>>",
">=",
"<",
"<<",
"<<<",
"<=",
"+",
"++",
"-",
"--",
"!",
"&",
"&&",
"|",
"||"
};
//longest first
delimiters.sort_descending();
//buffer
array<ascii> buffer;
//any
if(!tokenizer.any(buffer,delimiters))
return false;
//commit
result=buffer;
update(symbols()._delimiter,buffer);
return true;
}