661 lines
19 KiB
C++
661 lines
19 KiB
C++
#ifndef __NASAL_PARSER_H__
|
|
#define __NASAL_PARSER_H__
|
|
|
|
#include <stack>
|
|
|
|
enum token_type
|
|
{
|
|
__stack_end,
|
|
__null_end,// $
|
|
__equal,// =
|
|
__cmp_equal,// ==
|
|
__cmp_not_equal,// !=
|
|
__cmp_less,__cmp_less_or_equal,// < <=
|
|
__cmp_more,__cmp_more_or_equal,// > >=
|
|
__and_operator,__or_operator,__nor_operator,// and or !
|
|
__add_operator,__sub_operator,__mul_operator,__div_operator,__link_operator,// + - * / ~
|
|
__add_equal,__sub_equal,__mul_equal,__div_equal,__link_equal,// += -= *= /= ~=
|
|
__left_brace,__right_brace,// {}
|
|
__left_bracket,__right_bracket,// []
|
|
__left_curve,__right_curve,// ()
|
|
__semi,// ;
|
|
__comma,// ,
|
|
__colon,// :
|
|
__dot,// .
|
|
__var,// var reserve word
|
|
__func,// func reserve word
|
|
__id,
|
|
__identifier,__identifiers,__identifier_end,__lacked_identifier,
|
|
__parameter,__parameters,__parameter_end,
|
|
__number,__string,
|
|
__scalar,__scalars,__scalar_end,
|
|
__list,__list_end,
|
|
__hash,__hash_end,
|
|
__hash_member,__hash_members,__hash_member_end,__hash_search,
|
|
__statement,__statements,
|
|
__function,__call_function,__call_function_end,//function(){}
|
|
__definition,
|
|
__assignment,__pre_assignment,
|
|
__calculation,
|
|
__loop,__continue,__break,__for,__forindex,__foreach,__while,// for()while() continue; break;
|
|
__choose,__if,__elsif,__else,// if else if else
|
|
__return
|
|
};
|
|
void print_token_type(int type)
|
|
{
|
|
std::string context="";
|
|
switch(type)
|
|
{
|
|
case __stack_end:
|
|
context="__stack_end";
|
|
break;
|
|
case __equal:
|
|
context="=";
|
|
break;
|
|
case __cmp_equal:
|
|
context="==";
|
|
break;
|
|
case __cmp_not_equal:
|
|
context="!=";
|
|
break;
|
|
case __cmp_less:
|
|
context="<";
|
|
break;
|
|
case __cmp_less_or_equal:
|
|
context="<=";
|
|
break;
|
|
case __cmp_more:
|
|
context=">";
|
|
break;
|
|
case __cmp_more_or_equal:
|
|
context=">=";
|
|
break;
|
|
case __and_operator:
|
|
context="and";
|
|
break;
|
|
case __or_operator:
|
|
context="or";
|
|
break;
|
|
case __nor_operator:
|
|
context="!";
|
|
break;
|
|
case __add_operator:
|
|
context="+";
|
|
break;
|
|
case __sub_operator:
|
|
context="-";
|
|
break;
|
|
case __mul_operator:
|
|
context="*";
|
|
break;
|
|
case __div_operator:
|
|
context="/";
|
|
break;
|
|
case __link_operator:
|
|
context="~";
|
|
break;
|
|
case __add_equal:
|
|
context="+=";
|
|
break;
|
|
case __sub_equal:
|
|
context="-=";
|
|
break;
|
|
case __mul_equal:
|
|
context="*=";
|
|
break;
|
|
case __div_equal:
|
|
context="/=";
|
|
break;
|
|
case __link_equal:
|
|
context="~=";
|
|
break;
|
|
case __left_brace:
|
|
context="{";
|
|
break;
|
|
case __right_brace:
|
|
context="}";
|
|
break;
|
|
case __left_bracket:
|
|
context="[";
|
|
break;
|
|
case __right_bracket:
|
|
context="]";
|
|
break;
|
|
case __left_curve:
|
|
context="(";
|
|
break;
|
|
case __right_curve:
|
|
context=")";
|
|
break;
|
|
case __semi:
|
|
context=";";
|
|
break;
|
|
case __comma:
|
|
context=",";
|
|
break;
|
|
case __colon:
|
|
context=":";
|
|
break;
|
|
case __dot:
|
|
context=".";
|
|
break;
|
|
case __var:
|
|
context="var";
|
|
break;
|
|
case __func:
|
|
context="func";
|
|
break;
|
|
case __id:
|
|
context="id";
|
|
break;
|
|
case __identifier:
|
|
context="identifier";
|
|
break;
|
|
case __identifiers:
|
|
context="identifiers";
|
|
break;
|
|
case __scalar:
|
|
context="scalar";
|
|
break;
|
|
case __scalars:
|
|
context="scalars";
|
|
break;
|
|
case __list:
|
|
context="list";
|
|
break;
|
|
case __hash:
|
|
context="hash";
|
|
break;
|
|
case __hash_member:
|
|
context="hash_member";
|
|
break;
|
|
case __hash_members:
|
|
context="hash_members";
|
|
break;
|
|
case __continue:
|
|
context="continue";
|
|
break;
|
|
case __break:
|
|
context="break";
|
|
break;
|
|
case __for:
|
|
context="for";
|
|
break;
|
|
case __forindex:
|
|
context="forindex";
|
|
break;
|
|
case __foreach:
|
|
context="foreach";
|
|
break;
|
|
case __while:
|
|
context="while";
|
|
break;
|
|
case __if:
|
|
context="if";
|
|
break;
|
|
case __elsif:
|
|
context="elsif";
|
|
break;
|
|
case __else:
|
|
context="else";
|
|
break;
|
|
case __return:
|
|
context="return";
|
|
break;
|
|
case __calculation:
|
|
context="calculation";
|
|
break;
|
|
case __function:
|
|
context="function_module";
|
|
break;
|
|
case __definition:
|
|
context="definition";
|
|
break;
|
|
case __number:
|
|
context="number";
|
|
break;
|
|
case __string:
|
|
context="string";
|
|
break;
|
|
}
|
|
std::cout<<context;
|
|
return;
|
|
}
|
|
|
|
const int max_token_len=20;
|
|
struct cmp_seq
|
|
{
|
|
int tokens[max_token_len];
|
|
int res;
|
|
}par[]=
|
|
{
|
|
{{__left_bracket,__list_end}, __list},
|
|
{{__number,__list_end}, __list_end},
|
|
{{__id,__list_end}, __list_end},
|
|
{{__identifier,__list_end}, __list_end},
|
|
{{__string,__list_end}, __list_end},
|
|
{{__list,__list_end}, __list_end},
|
|
{{__hash,__list_end}, __list_end},
|
|
{{__comma,__number,__list_end}, __list_end},
|
|
{{__comma,__id,__list_end}, __list_end},
|
|
{{__comma,__identifier,__list_end}, __list_end},
|
|
{{__comma,__string,__list_end}, __list_end},
|
|
{{__comma,__list,__list_end}, __list_end},
|
|
{{__comma,__hash,__list_end}, __list_end},
|
|
{{__right_bracket}, __list_end},
|
|
|
|
{{__id,__colon,__number}, __hash_member},
|
|
{{__id,__colon,__id}, __hash_member},
|
|
{{__id,__colon,__identifier}, __hash_member},
|
|
{{__id,__colon,__string}, __hash_member},
|
|
{{__id,__colon,__list}, __hash_member},
|
|
{{__id,__colon,__hash}, __hash_member},
|
|
|
|
{{__left_brace,__hash_end}, __hash},
|
|
{{__hash_member,__hash_end}, __hash_end},
|
|
{{__comma,__hash_member,__hash_end}, __hash_end},
|
|
{{__right_brace}, __hash_end},
|
|
|
|
{{__id,__dot,__hash_search}, __hash_search},
|
|
{{__identifier,__dot,__hash_search}, __hash_search},
|
|
{{__identifier}, __hash_search},
|
|
{{__id}, __hash_search},
|
|
|
|
{{__call_function}, __identifier},
|
|
{{__id,__left_bracket,__number,__right_bracket}, __identifier},
|
|
{{__id,__left_bracket,__id,__right_bracket}, __identifier},
|
|
{{__id,__left_bracket,__identifier,__right_bracket}, __identifier},
|
|
|
|
|
|
{{__id,__left_curve,__call_function_end}, __call_function},
|
|
{{__number,__call_function_end}, __call_function_end},
|
|
{{__id,__call_function_end}, __call_function_end},
|
|
{{__identifier,__call_function_end}, __call_function_end},
|
|
{{__string,__call_function_end}, __call_function_end},
|
|
{{__list,__call_function_end}, __call_function_end},
|
|
{{__hash,__call_function_end}, __call_function_end},
|
|
{{__function,__call_function_end}, __call_function_end},
|
|
{{__comma,__number,__call_function_end}, __call_function_end},
|
|
{{__comma,__id,__call_function_end}, __call_function_end},
|
|
{{__comma,__identifier,__call_function_end}, __call_function_end},
|
|
{{__comma,__string,__call_function_end}, __call_function_end},
|
|
{{__comma,__list,__call_function_end}, __call_function_end},
|
|
{{__comma,__hash,__call_function_end}, __call_function_end},
|
|
{{__comma,__function,__call_function_end}, __call_function_end},
|
|
{{__right_curve}, __call_function_end},
|
|
|
|
{{__var,__id,__equal,__number,__semi}, __definition},
|
|
{{__var,__id,__equal,__id,__semi}, __definition},
|
|
{{__var,__id,__equal,__identifier,__semi}, __definition},
|
|
{{__var,__id,__equal,__string,__semi}, __definition},
|
|
{{__var,__id,__equal,__list,__semi}, __definition},
|
|
{{__var,__id,__equal,__hash,__semi}, __definition},
|
|
{{__var,__id,__equal,__hash_search,__semi}, __definition},
|
|
|
|
{{__var,__id,__equal,__number,__semi}, __assignment},
|
|
{{__var,__id,__equal,__id,__semi}, __assignment},
|
|
{{__var,__id,__equal,__identifier,__semi}, __assignment},
|
|
{{__var,__id,__equal,__string,__semi}, __assignment},
|
|
{{__var,__id,__equal,__list,__semi}, __assignment},
|
|
{{__var,__id,__equal,__hash,__semi}, __assignment},
|
|
{{__var,__id,__equal,__hash_search,__semi}, __assignment},
|
|
{{__definition}, __statement},
|
|
{{__assignment}, __statement}
|
|
};
|
|
int num_of_par=sizeof(par)/sizeof(cmp_seq);
|
|
|
|
|
|
struct parse_unit
|
|
{
|
|
int line;
|
|
int type;
|
|
};
|
|
class PDA
|
|
{
|
|
private:
|
|
std::stack<parse_unit> main_stack;
|
|
std::stack<parse_unit> error_stack;
|
|
bool stack_running_problem_check;
|
|
public:
|
|
PDA()
|
|
{
|
|
stack_running_problem_check=false;
|
|
//when the DFS searched too deep (over 1024 levels) the stack_running_problem_check
|
|
//will be set to true then return ro the main_progress
|
|
//when main_progress finds that the stack_running_problem_check is true
|
|
//then main_progress ends with an error : stack out of range
|
|
//and return the line where the error occurred
|
|
}
|
|
void stack_input(std::stack<parse_unit>& temp)
|
|
{
|
|
while(!temp.empty())
|
|
{
|
|
main_stack.push(temp.top());
|
|
temp.pop();
|
|
}
|
|
return;
|
|
}
|
|
void print_error()
|
|
{
|
|
if(!error_stack.empty())
|
|
{
|
|
std::stack<parse_unit> temp;
|
|
while(!error_stack.empty())
|
|
{
|
|
temp.push(error_stack.top());
|
|
error_stack.pop();
|
|
}
|
|
std::cout<<">>[Error] Parse error:";
|
|
int now_line=0;
|
|
while(!temp.empty())
|
|
{
|
|
if(temp.top().line!=now_line)
|
|
{
|
|
now_line=temp.top().line;
|
|
std::cout<<std::endl<<"line "<<now_line<<" ";
|
|
}
|
|
print_token_type(temp.top().type);
|
|
std::cout<<" ";
|
|
temp.pop();
|
|
}
|
|
std::cout<<std::endl;
|
|
}
|
|
else
|
|
std::cout<<">>[Parse] 0 error occurred."<<std::endl;
|
|
return;
|
|
}
|
|
bool extend_comp_progress(const int type)
|
|
{
|
|
static int depth=0;
|
|
++depth;
|
|
if(stack_running_problem_check)
|
|
return false;
|
|
if(depth==1024)
|
|
{
|
|
stack_running_problem_check=true;
|
|
return false;
|
|
}
|
|
std::stack<parse_unit> recognized_stack;
|
|
std::stack<parse_unit> comp_stack;
|
|
parse_unit temp;
|
|
temp.line=0;
|
|
temp.type=0;
|
|
for(int i=0;i<num_of_par;++i)
|
|
{
|
|
if(par[i].res==type)
|
|
{
|
|
while(!comp_stack.empty())
|
|
comp_stack.pop();
|
|
for(int j=max_token_len-1;j>=0;--j)
|
|
{
|
|
if(par[i].tokens[j])
|
|
{
|
|
temp.type=par[i].tokens[j];
|
|
comp_stack.push(temp);
|
|
}
|
|
}
|
|
while((!comp_stack.empty()) && (!main_stack.empty()))
|
|
{
|
|
if(comp_stack.top().type==main_stack.top().type)
|
|
{
|
|
comp_stack.pop();
|
|
recognized_stack.push(main_stack.top());
|
|
main_stack.pop();
|
|
}
|
|
else if(comp_stack.top().type!=main_stack.top().type)
|
|
{
|
|
if(!extend_comp_progress(comp_stack.top().type))
|
|
{
|
|
//if after all search the real type
|
|
//can not be found,then recognized_stack
|
|
//returns all the tokens in it
|
|
//to make sure the next comparation works
|
|
while(!recognized_stack.empty())
|
|
{
|
|
main_stack.push(recognized_stack.top());
|
|
recognized_stack.pop();
|
|
}
|
|
break;
|
|
}
|
|
else
|
|
comp_stack.pop();
|
|
}
|
|
}
|
|
if(comp_stack.empty())
|
|
{
|
|
--depth;
|
|
return true;
|
|
}
|
|
}
|
|
}
|
|
--depth;
|
|
return false;
|
|
}
|
|
void main_comp_progress()
|
|
{
|
|
std::stack<parse_unit> recognized_stack;
|
|
std::stack<parse_unit> comp_stack;
|
|
parse_unit temp;
|
|
temp.line=0;
|
|
temp.type=0;
|
|
|
|
while(!main_stack.empty())
|
|
{
|
|
bool ispar=false;
|
|
for(int i=0;i<num_of_par;++i)
|
|
{
|
|
while(!comp_stack.empty())
|
|
comp_stack.pop();
|
|
for(int j=max_token_len-1;j>=0;--j)
|
|
{
|
|
if(par[i].tokens[j])
|
|
{
|
|
temp.type=par[i].tokens[j];
|
|
comp_stack.push(temp);
|
|
}
|
|
}
|
|
while((!comp_stack.empty()) && (!main_stack.empty()))
|
|
{
|
|
//because of the check !omp_stack.empty()
|
|
//every token sequence should not be null
|
|
if(comp_stack.top().type==main_stack.top().type)
|
|
{
|
|
comp_stack.pop();
|
|
recognized_stack.push(main_stack.top());
|
|
main_stack.pop();
|
|
}
|
|
else if(comp_stack.top().type!=main_stack.top().type)
|
|
{
|
|
if(!extend_comp_progress(comp_stack.top().type))
|
|
{
|
|
if(stack_running_problem_check)
|
|
{
|
|
std::cout<<">>[Parse] Stack out of range in line "<<main_stack.top().line<<std::endl;
|
|
return;
|
|
}
|
|
while(!recognized_stack.empty())
|
|
{
|
|
main_stack.push(recognized_stack.top());
|
|
recognized_stack.pop();
|
|
}
|
|
break;
|
|
}
|
|
else
|
|
comp_stack.pop();
|
|
}
|
|
}
|
|
if(comp_stack.empty())
|
|
{
|
|
while(!recognized_stack.empty())
|
|
recognized_stack.pop();
|
|
ispar=true;
|
|
break;
|
|
}
|
|
}
|
|
if(!ispar && !main_stack.empty())
|
|
{
|
|
error_stack.push(main_stack.top());
|
|
main_stack.pop();
|
|
}
|
|
}
|
|
std::cout<<"[Parse] Complete checking."<<std::endl;
|
|
print_error();
|
|
return;
|
|
}
|
|
};
|
|
|
|
class nasal_parser
|
|
{
|
|
private:
|
|
std::stack<parse_unit> parser;
|
|
public:
|
|
void parse_process(std::list<token>& lexer)
|
|
{
|
|
while(!parser.empty())
|
|
parser.pop();
|
|
for(std::list<token>::iterator i=lexer.begin();i!=lexer.end();++i)
|
|
{
|
|
parse_unit temp_parse;
|
|
temp_parse.line=(*i).line;
|
|
if(((*i).content=="var") || ((*i).content=="func") || ((*i).content=="return") || ((*i).content=="nil") || ((*i).content=="continue") || ((*i).content=="break") || ((*i).content=="and") || ((*i).content=="or"))
|
|
{
|
|
if((*i).content=="var")
|
|
temp_parse.type=__var;
|
|
else if((*i).content=="func")
|
|
temp_parse.type=__func;
|
|
else if((*i).content=="return")
|
|
temp_parse.type=__return;
|
|
else if((*i).content=="nil")
|
|
temp_parse.type=__scalar;
|
|
else if((*i).content=="continue")
|
|
temp_parse.type=__continue;
|
|
else if((*i).content=="break")
|
|
temp_parse.type=__break;
|
|
else if((*i).content=="and")
|
|
temp_parse.type=__and_operator;
|
|
else if((*i).content=="or")
|
|
temp_parse.type=__or_operator;
|
|
}
|
|
else if((*i).type==IDENTIFIER)
|
|
{
|
|
temp_parse.type=__id;
|
|
}
|
|
else if(((*i).content=="for") || ((*i).content=="foreach") || ((*i).content=="while") || ((*i).content=="forindex"))
|
|
{
|
|
if((*i).content=="for")
|
|
temp_parse.type=__for;
|
|
else if((*i).content=="forindex")
|
|
temp_parse.type=__forindex;
|
|
else if((*i).content=="foreach")
|
|
temp_parse.type=__foreach;
|
|
else if((*i).content=="while")
|
|
temp_parse.type=__while;
|
|
}
|
|
else if(((*i).content=="if") || ((*i).content=="else") || ((*i).content=="elsif"))
|
|
{
|
|
if((*i).content=="if")
|
|
temp_parse.type=__if;
|
|
else if((*i).content=="else")
|
|
temp_parse.type=__else;
|
|
else if((*i).content=="elsif")
|
|
temp_parse.type=__elsif;
|
|
}
|
|
else if(((*i).content=="==") || ((*i).content=="!=") || ((*i).content==">") || ((*i).content==">=") || ((*i).content=="<") || ((*i).content=="<="))
|
|
{
|
|
if((*i).content=="==")
|
|
temp_parse.type=__cmp_equal;
|
|
else if((*i).content=="!=")
|
|
temp_parse.type=__cmp_not_equal;
|
|
else if((*i).content==">")
|
|
temp_parse.type=__cmp_more;
|
|
else if((*i).content==">=")
|
|
temp_parse.type=__cmp_more_or_equal;
|
|
else if((*i).content=="<")
|
|
temp_parse.type=__cmp_less;
|
|
else if((*i).content=="<=")
|
|
temp_parse.type=__cmp_less_or_equal;
|
|
}
|
|
else if(((*i).content==";") || ((*i).content==",") || ((*i).content=="=") || ((*i).content==":") || ((*i).content=="."))
|
|
{
|
|
if((*i).content==";")
|
|
temp_parse.type=__semi;
|
|
else if((*i).content==",")
|
|
temp_parse.type=__comma;
|
|
else if((*i).content=="=")
|
|
temp_parse.type=__equal;
|
|
else if((*i).content==":")
|
|
temp_parse.type=__colon;
|
|
else if((*i).content==".")
|
|
temp_parse.type=__dot;
|
|
}
|
|
else if(((*i).type==NUMBER) || ((*i).type==STRING))
|
|
{
|
|
if((*i).type==NUMBER)
|
|
temp_parse.type=__number;
|
|
else
|
|
temp_parse.type=__string;
|
|
}
|
|
else if(((*i).content=="+") || ((*i).content=="-") || ((*i).content=="*") || ((*i).content=="/") || ((*i).content=="~") || ((*i).content=="!"))
|
|
{
|
|
if((*i).content=="+")
|
|
temp_parse.type=__add_operator;
|
|
else if((*i).content=="-")
|
|
temp_parse.type=__sub_operator;
|
|
else if((*i).content=="*")
|
|
temp_parse.type=__mul_operator;
|
|
else if((*i).content=="/")
|
|
temp_parse.type=__div_operator;
|
|
else if((*i).content=="~")
|
|
temp_parse.type=__link_operator;
|
|
else if((*i).content=="!")
|
|
temp_parse.type=__nor_operator;
|
|
}
|
|
else if(((*i).content=="+=") || ((*i).content=="-=") || ((*i).content=="*=") || ((*i).content=="/=") || ((*i).content=="~="))
|
|
{
|
|
if((*i).content=="+=")
|
|
temp_parse.type=__add_equal;
|
|
else if((*i).content=="-=")
|
|
temp_parse.type=__sub_equal;
|
|
else if((*i).content=="*=")
|
|
temp_parse.type=__mul_equal;
|
|
else if((*i).content=="/=")
|
|
temp_parse.type=__div_equal;
|
|
else if((*i).content=="~=")
|
|
temp_parse.type=__link_equal;
|
|
}
|
|
else if(((*i).content=="(") || ((*i).content==")") || ((*i).content=="[") || ((*i).content=="]") || ((*i).content=="{") || ((*i).content=="}"))
|
|
{
|
|
char c=(*i).content[0];
|
|
switch(c)
|
|
{
|
|
case '(':
|
|
temp_parse.type=__left_curve;
|
|
break;
|
|
case ')':
|
|
temp_parse.type=__right_curve;
|
|
break;
|
|
case '[':
|
|
temp_parse.type=__left_bracket;
|
|
break;
|
|
case ']':
|
|
temp_parse.type=__right_bracket;
|
|
break;
|
|
case '{':
|
|
temp_parse.type=__left_brace;
|
|
break;
|
|
case '}':
|
|
temp_parse.type=__right_brace;
|
|
break;
|
|
}
|
|
}
|
|
parser.push(temp_parse);//push this into stack
|
|
}
|
|
PDA automata;
|
|
automata.stack_input(parser);
|
|
automata.main_comp_progress();
|
|
return;
|
|
}
|
|
};
|
|
|
|
#endif
|