path: root/postrun.cpp

#include <iostream>
#include <fstream>
#include <sstream>
#include <vector>
#include <unordered_map>
#include <functional>
#include <cstdlib>
#include <cstring>
#include <cmath>

using namespace std;

inline bool isword(char c){return isalpha(c)||c=='_'||c=='@'||c=='$';}
inline bool isextword(char c){return isword(c)||isdigit(c);}
inline bool isoperator(char c){return (bool)strchr("+-*/=><!%^{}",c);}

vector<string> tokenise(istream &stream){
	vector<string> tokens;
	string token;
	char c,stringmode;
	bool gotminus=false;
	while(true){
		c=stream.get();
		if(!stream)break;
		if(c=='#'){
			while((c=stream.get())!='\n'&&stream);
			c=stream.get();
			if(!stream)break;
		}
		if(isdigit(c)){
			if(token.size()==0){
				token="#";
				if(gotminus)token+="-";
				gotminus=false;
			}
			token+=c;
		} else if(isword(c)||(token.size()>0&&isextword(c))){
			token+=c;
		} else {
			if(token.size())tokens.push_back(move(token));
			if(isoperator(c)){
				if(c=='-'&&isdigit(stream.peek()))gotminus=true;
				else tokens.push_back(string(1,c));
			} else if(c=='"'||c=='\''){
				stringmode=c;
				while(true){
					c=stream.get();
					if(!stream)throw string("Non-terminated string at end of file");
					if(c==stringmode){
						token="'"+token;
						tokens.push_back(move(token));
						break;
					}
					if(c=='\\'){
						c=stream.get();
						if(!stream)throw string("Non-terminated escape sequence in string at end of file");
						switch(c){
							case 'n':c='\n';break;
							case 'r':c='\r';break;
							case 't':c='\t';break;
							case '0':c='\0';break;
							case '"':c='"';break;
							case '\'':c='\'';break;
							case 'x':{
								int res;
								c=stream.get();
								if(!stream)throw string("Non-terminated hexadecimal escape sequence in string at end of file");
								res=c>='0'&&c<='9'?c-'0':(c>='a'&&c<='f')||(c>='A'&&c<='F')?(c&~32)-'A'+10:-1;
								if(res==-1)throw "Invalid character '"+string(1,c)+"'' in hexadecimal escape sequence in string";
								res*=16;
								c=stream.get();
								if(!stream)throw string("Non-terminated hexadecimal escape sequence in string at end of file");
								res+=c>='0'&&c<='9'?c-'0':(c>='a'&&c<='f')||(c>='A'&&c<='F')?(c&~32)-'A'+10:-241;
								if(res<0)throw "Invalid character '"+string(1,c)+"'' in hexadecimal escape sequence in string";
								c=(char)res;
								break;
							}
						}
					}
					token+=c;
				}
			} else if(isspace(c));
			else throw string("Invalid character found in source: '")+c+'\'';
		}
	}
	if(token.size())tokens.push_back(move(token));
	return tokens;
}

string readfile(ifstream stream){
	stream.seekg(0,ios::end);
	size_t len=stream.tellg();
	stream.seekg(0,ios::beg);
	string contents;
	contents.resize(len);
	stream.read(&*contents.begin(),len);
	return contents;
}

struct Stackitem{
	bool isstr;
	string strval;
	int intval;

	Stackitem(void):isstr(false),intval(0){}
	Stackitem(int _i):isstr(false),intval(_i){}
	Stackitem(const string &_s):isstr(true),strval(_s),intval(0){}
	Stackitem(const Stackitem&)=default;
	Stackitem(Stackitem &&other):isstr(other.isstr),strval(move(other.strval)),intval(other.intval){
		other.isstr=false;
	}
	Stackitem& operator=(const Stackitem&)=default;
	Stackitem& operator=(Stackitem &&other){
		isstr=other.isstr;
		strval=move(other.strval);
		intval=other.intval;
		other.isstr=false;
		return *this;
	}
	explicit operator bool(void) const{return (isstr&&strval.size())||(!isstr&&intval);}
	bool operator==(const Stackitem &other) const{return isstr==other.isstr&&(isstr?strval==other.strval:intval==other.intval);}
};

#define BUILTIN_GUARD_STACKSIZE(nm,sz) {if(S.size()<(sz))throw string("Builtin '" nm "' needs " #sz " stack item")+(sz==1?"":"s");}

const unordered_map<string,function<void(vector<Stackitem>&,unordered_map<string,Stackitem>&)>> builtins={
	{"+",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("+",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(!a.isstr&&!b.isstr){
			S.emplace_back(a.intval+b.intval);
			return;
		}
		if(!a.isstr){
			a.strval=to_string(a.intval);
			a.isstr=true;
		}
		if(!b.isstr){
			b.strval=to_string(b.intval);
			b.isstr=true;
		}
		S.emplace_back(a.strval+b.strval);
	}},
	{"-",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("-",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr||b.isstr)throw string("Builtin '-' cannot accept a string argument");
		S.emplace_back(a.intval-b.intval);
	}},
	{"*",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("*",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr||b.isstr)throw string("Builtin '*' cannot accept a string argument");
		S.emplace_back(a.intval*b.intval);
	}},
	{"/",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("/",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr||b.isstr)throw string("Builtin '/' cannot accept a string argument");
		S.emplace_back(a.intval/b.intval);
	}},
	{"%",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("%",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr||b.isstr){
			cerr<<"modding ";
			if(a.isstr)cerr<<'"'<<a.strval<<'"'; else cerr<<a.intval;
			cerr<<" and ";
			if(b.isstr)cerr<<'"'<<b.strval<<'"'; else cerr<<b.intval;
			cerr<<endl;
		}
		if(a.isstr||b.isstr)throw string("Builtin '%' cannot accept a string argument");
		S.emplace_back(a.intval%b.intval);
	}},
	{"^",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("^",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr||b.isstr)throw string("Builtin '^' cannot accept a string argument");
		S.emplace_back((int)pow(a.intval,b.intval));
	}},
	{"!",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("!",1)
		Stackitem v=move(S.back()); S.pop_back();
		if(v.isstr)S.emplace_back(!v.strval.size());
		else S.emplace_back(!v.intval);
	}},
	{"=",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("=",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		S.emplace_back(a==b);
	}},
	{"<",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("<",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr&&b.isstr)S.emplace_back(a.strval<b.strval);
		else if(!a.isstr&&!b.isstr)S.emplace_back(a.intval<b.intval);
		else if(a.isstr)S.emplace_back(a.strval<to_string(b.intval));
		else S.emplace_back(to_string(a.intval)<b.strval);
	}},
	{">",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE(">",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		if(a.isstr&&b.isstr)S.emplace_back(a.strval>b.strval);
		else if(!a.isstr&&!b.isstr)S.emplace_back(a.intval>b.intval);
		else if(a.isstr)S.emplace_back(a.strval>to_string(b.intval));
		else S.emplace_back(to_string(a.intval)>b.strval);
	}},
	{"print",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("print",1)
		Stackitem v=move(S.back()); S.pop_back();
		if(v.isstr)cout<<v.strval<<flush;
		else cout<<v.intval<<flush;
	}},
	{"lf",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		cout<<endl;
	}},
	{"dup",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("dup",1)
		S.push_back(S.back());
	}},
	{"pop",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("pop",1)
		S.pop_back();
	}},
	{"swap",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("swap",2)
		Stackitem b=move(S.back()); S.pop_back();
		Stackitem a=move(S.back()); S.pop_back();
		S.push_back(b);
		S.push_back(a);
	}},
	{"store",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("store",2)
		Stackitem name=move(S.back()); S.pop_back();
		if(!name.isstr)throw string("Second argument to 'store' not a string");
		variables[name.strval]=move(S.back());
		S.pop_back();
	}},
	{"get",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("get",1)
		Stackitem name=move(S.back()); S.pop_back();
		if(!name.isstr)throw string("Argument to 'get' not a string");
		S.push_back(variables[name.strval]);
	}},
	//leaves variable with value 0
	{"swapoutvar",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("swapoutvar",1)
		Stackitem name=move(S.back()); S.pop_back();
		if(!name.isstr)throw string("Second argument to 'swapoutvar' not a string");
		S.push_back(move(variables[name.strval]));
		variables[name.strval].intval=0;
	}},
	//positive n: roll OFF top to bottom; negative n: roll off bottom ONTO top
	{"roll",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("roll",1)
		Stackitem ns=move(S.back()); S.pop_back();
		if(ns.isstr)throw string("Argument to 'roll' not a number");
		int n=ns.intval;
		if(S.size()<2)return;
		bool negative=n<0;
		if(negative)n=-n;
		n%=S.size();
		if(n==0)return;
		if(negative){
			vector<Stackitem> tempstore;
			tempstore.reserve(n);
			int ssz=S.size();
			for(int i=ssz-n;i<ssz;i++)tempstore.push_back(move(S[i]));
			S.erase(S.end()-n,S.end());
			S.insert(S.begin(),n,Stackitem());
			for(int i=0;i<n;i++)S[i]=move(tempstore[i]);
		} else {
			//cerr<<"rolling back n="<<n<<endl;
			vector<Stackitem> tempstore;
			tempstore.reserve(n);
			for(int i=0;i<n;i++)tempstore.push_back(move(S[i]));
			S.erase(S.begin(),S.begin()+n);
			for(int i=0;i<n;i++)S.push_back(move(tempstore[i]));
		}
	}},
	//leaves the string on the stack
	{"stridx",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("stridx",2)
		Stackitem idx=move(S.back()); S.pop_back();
		const Stackitem &str=S.back();
		if(!str.isstr)throw string("First argument to 'stridx' not a string");
		if(idx.isstr)throw string("Second argument to 'stridx' not a number");
		if(idx.intval<0||idx.intval>=(int)str.strval.size())throw string("Index argument to 'stridx' out of range");
		S.emplace_back(string(1,str.strval[idx.intval]));
	}},
	//leaves the string on the stack
	{"strlen",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("strlen",1)
		const Stackitem &str=S.back();
		if(!str.isstr)throw string("Argument to 'strlen' not a string");
		S.emplace_back(str.strval.size());
	}},
	//leaves the string on the stack
	{"substr",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("substr",3)
		const Stackitem to=move(S.back()); S.pop_back();
		const Stackitem from=move(S.back()); S.pop_back();
		const Stackitem &str=S.back();
		if(!str.isstr)throw string("First argument to 'substr' not a string");
		if(from.isstr)throw string("Second argument to 'substr' not a number");
		if(to.isstr)throw string("Third argument to 'substr' not a number");
		S.emplace_back(str.strval.substr(from.intval,to.intval-from.intval));
	}},
	{"chr",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("chr",1)
		Stackitem ascii=move(S.back()); S.pop_back();
		if(ascii.isstr)throw string("Argument to 'chr' not a number");
		S.emplace_back(string(1,ascii.intval));
	}},
	{"ord",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("ord",1)
		if(!S.back().isstr)throw string("Argument to 'ord' not a string");
		S.back().isstr=false;
		S.back().intval=(int)S.back().strval[0];
		S.back().strval="";
	}},
	{"getline",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("getline",0)
		string line;
		getline(cin,line);
		if(!cin)S.emplace_back(-1);
		else S.emplace_back(line);
	}},
	{"getc",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("getc",0)
		if(!cin)S.emplace_back(-1);
		else S.emplace_back(string(1,cin.get()));
	}},
	{"stackdump",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("stackdump",0)
		cerr<<"STACKDUMP: ";
		for(const Stackitem &si : S){
			if(si.isstr)cerr<<'"'<<si.strval<<"\" ";
			else cerr<<si.intval<<' ';
		}
		cerr<<endl;
	}},
	{"exit",[](vector<Stackitem> &S,unordered_map<string,Stackitem> &variables){
		BUILTIN_GUARD_STACKSIZE("exit",0)
		exit(0);
	}},
};
#undef BUILTIN_GUARD_STACKSIZE

void runpasseval(const vector<string> &T,
                 vector<Stackitem> &S,
                 const unordered_map<string,pair<vector<string>,unordered_map<unsigned int,unsigned int>>> &functions,
                 unordered_map<string,Stackitem> &variables,
                 const unordered_map<unsigned int,unsigned int> &jumpmap){
	unsigned int cursor;
	for(cursor=0;cursor<T.size();cursor++){
		const string &word=T[cursor];
		//cerr<<"Executing word <"<<word<<'>'<<endl;
		if(word[0]=='#'){
			S.emplace_back((int)strtol(word.data()+1,NULL,10));
		} else if(word[0]=='\''){
			S.push_back(word.substr(1));
		} else {
			auto it=variables.find(word);
			if(it!=variables.end()){
				S.push_back(it->second);
			} else {
				auto it=functions.find(word);
				if(it!=functions.end()){
					runpasseval(it->second.first,S,functions,variables,it->second.second);
				} else {
					if(word=="while"){
						if(S.size()<1)throw string("Keyword 'while' needs 1 stack item");
						Stackitem v=S.back(); S.pop_back();
						if(!v){
							auto it=jumpmap.find(cursor);
							if(it==jumpmap.end()){
								cerr<<"NO JUMPMAP ENTRY FOR "<<word<<" AT INDEX "<<cursor<<endl;
							}
							cursor=jumpmap.find(cursor)->second-1; //jump to corresponding end
						}
						continue;
					} else if(word=="if"){
						if(S.size()<1)throw string("Keyword 'if' needs 1 stack item");
						Stackitem v=S.back(); S.pop_back();
						if(!v){
							auto it=jumpmap.find(cursor);
							if(it==jumpmap.end()){
								cerr<<"NO JUMPMAP ENTRY FOR "<<word<<" AT INDEX "<<cursor<<endl;
							}
							cursor=jumpmap.find(cursor)->second-1; //jump to corresponding else/end
						}
						continue;
					} else if(word=="else"||word=="end"){
						auto it=jumpmap.find(cursor);
						if(it==jumpmap.end()){
							cerr<<"NO JUMPMAP ENTRY FOR "<<word<<" AT INDEX "<<cursor<<endl;
						}
						cursor=jumpmap.find(cursor)->second-1;
						continue;
					}

					auto it=builtins.find(word);
					if(it==builtins.end())throw "Unknown variable or function '"+word+"'";
					it->second(S,variables);
				}
			}
		}
	}
}

void populatejumpmap(const vector<string> &T,unordered_map<unsigned int,unsigned int> &jumpmap){
	//first, index flow control keywords
	vector<unsigned int> flowindexes;
	for(unsigned int cursor=0;cursor<T.size();cursor++){
		const string &word=T[cursor];
		if(word=="end"||word=="while"||word=="if"||word=="else")flowindexes.push_back(cursor);
	}

	//then, index jump targets in jumpmap
	for(unsigned int i=0;i<flowindexes.size();i++){
		const string &word=T[flowindexes[i]];
		if(word=="while"){
			if(i==flowindexes.size()-1)throw string("No 'end' after 'while' in source");
			int depth=1;
			unsigned int end;
			for(end=i+1;end<flowindexes.size();end++){
				if(T[flowindexes[end]]=="end")depth--;
				else if(T[flowindexes[end]]=="while"||T[flowindexes[end]]=="if")depth++;
				if(depth==0)break;
			}
			if(end==flowindexes.size())throw string("No 'end' after 'while' in source");
			jumpmap[flowindexes[i]]=flowindexes[end]+1;
			jumpmap[flowindexes[end]]=flowindexes[i];
		} else if(word=="if"){
			if(i==flowindexes.size()-1)throw string("No 'end' after 'if' in source");
			int depth=1;
			unsigned int els,end;
			for(els=i+1;els<flowindexes.size();els++){
				if(depth==1&&T[flowindexes[els]]=="else")break;
				else if(T[flowindexes[els]]=="end")depth--;
				else if(T[flowindexes[els]]=="while"||T[flowindexes[els]]=="if")depth++;
				if(depth==0)break;
			}
			if(els==flowindexes.size())throw string("No 'else' or 'end' after 'if' in source");
			if(depth!=0){ //we found an else
				for(end=els+1;end<flowindexes.size();end++){
					if(T[flowindexes[end]]=="end")depth--;
					else if(T[flowindexes[end]]=="while"||T[flowindexes[end]]=="if")depth++;
					if(depth==0)break;
				}
				if(end==flowindexes.size())throw string("No 'end' after 'if' in source");
				jumpmap[flowindexes[i]]=flowindexes[els]+1;
				jumpmap[flowindexes[els]]=flowindexes[end]+1;
				jumpmap[flowindexes[end]]=flowindexes[end]+1;
			} else { //we found an end in the first place
				//cerr<<"coupling if-end indexes "<<flowindexes[i]<<" and "<<flowindexes[els]<<endl;
				jumpmap[flowindexes[i]]=flowindexes[els]+1;
				jumpmap[flowindexes[els]]=flowindexes[els]+1;
			}
		}
	}
}

void run(vector<string> T){
	unordered_map<string,pair<vector<string>,unordered_map<unsigned int,unsigned int>>> functions;
	//functions is a pair of tokens and jumpmap
	unordered_map<string,Stackitem> variables;
	unordered_map<unsigned int,unsigned int> jumpmap;
	vector<Stackitem> S;
	unsigned int cursor=0;
	//first pass, handle functions and includes
	while(cursor<T.size()){
		const string &word=T[cursor];
		if(word=="@defun"){
			if(cursor+3>=T.size())throw string("Unterminated @defun statement at end of file");
			string name=T[cursor+1];
			if(name[0]!='\'')throw string("@defun expected a string as name, but got '")+name+"'";
			name.erase(0,1);
			if(T[cursor+2]!="{")throw string("Invalid @defun statement of function ")+name;
			unsigned int start=cursor+3;
			unsigned int end;
			int depth=1;
			for(end=start;end<T.size();end++){
				if(T[end]=="{")depth++;
				else if(T[end]=="}")depth--;
				else if(T[end]=="\""){
					while(++end<T.size()&&T[end]!="\"")if(T[end]=="\\")end++;
				} else if(T[end]=="'"){
					while(++end<T.size()&&T[end]!="'")if(T[end]=="\\")end++;
				}
				if(depth==0)break;
			}
			if(depth!=0)throw string("Non-terminated @defun statement at end of file");
			vector<string> &functoks=functions[name].first;
			functoks.resize(end-start);
			for(unsigned int i=start;i<end;i++)functoks[i-start]=move(T[i]);
			T.erase(T.begin()+cursor,T.begin()+end+1);
		} else if(word=="@include"){
			if(cursor+1>=T.size())throw string("Unterminated @include statement at end of file");
			string name=T[cursor+1];
			if(name[0]!='\'')throw string("@include expected a string as file, but got '")+name+"'";
			name.erase(0,1);
			ifstream file(name);
			if(!file)throw string("Could not open file '")+name+"' specified by @include statement";
			vector<string> included=tokenise(file);
			file.close();
			T.erase(T.begin()+cursor,T.begin()+cursor+2);
			T.insert(T.begin()+cursor,included.begin(),included.end());
		} else {
			cursor++;
		}
	}

	//second pass, populate jumpmaps
	populatejumpmap(T,jumpmap);
	for(auto &p : functions){
		populatejumpmap(p.second.first,p.second.second);
	}

	//third pass, evaluate code
	runpasseval(T,S,functions,variables,jumpmap);
}

int main(int argc,char **argv){
	if(argc!=2){
		cerr<<"Call this interpreter with the Postrun source file"<<endl;
		return 1;
	}
	ifstream srcf(argv[1]);
	if(!srcf){
		cerr<<"Could not open file '"<<argv[1]<<"'"<<endl;
		return 1;
	}
	//string source=readfile(srcf);
	try {
		const vector<string> tokens=tokenise(srcf);
		//for(const string &tok : tokens)cerr<<'['<<tok<<"] "; cerr<<endl;
		run(tokens);
		srcf.close();
	} catch(string e){
		cerr<<"ERROR: "<<e<<endl;
		return 1;
	}
}