/**
* Authors: k.inaba
* License: NYSL 0.9982 http://www.kmonos.net/nysl/
*
* Parser for Polemy programming language
*/
module polemy.parse;
import polemy._common;
import polemy.lex;
import polemy.ast;
/// Thrown when encountered a syntax error
class ParseException : Exception
{
mixin ExceptionWithPosition;
}
/// Parse a string and return its AST
/// Throws: ParseException, LexException, UnexpectedEOF
AST parseString(S, T...)(S str, T fn_ln_cn)
{
return parserFromString(str, fn_ln_cn).parse();
}
/// Parse the content of a file and return its AST
/// Throws: ParseException, LexException, UnexpectedEOF
AST parseFile(S, T...)(S filename, T ln_cn)
{
return parserFromFile(filename, ln_cn).parse();
}
// Named Constructors of Parser
private auto parserFromLexer(Lexer)(Lexer lex)
{ return new Parser!Lexer(lex); }
private auto parserFromString(T...)(T params)
{ return parserFromLexer(lexerFromString(params)); }
private auto parserFromFile(T...)(T params)
{ return parserFromLexer(lexerFromFile(params)); }
// Parser
private class Parser(Lexer)
if( isForwardRange!(Lexer) && is(ElementType!(Lexer) == Token) )
{
AST parse()
{
auto e = Body();
if( !lex.empty )
throw genex!ParseException(currentPosition(), "parsing ended but some tokens left");
return e;
}
AST Body()
{
/// Body ::= Declaration
/// | TopLevelExpression
if( closingBracket() )
return doNothingExpression();
auto saved = lex.save;
if( auto e = Declaration() )
return e;
lex = saved;
return TopLevelExpression();
}
AST Declaration() // returns null if it is not a declaration
{
/// Declaration ::=
/// ["@" Layer|"let"|"var"|"def"] Var "=" Expression ([";"|"in"] Body?)?
/// | ["@" Layer|"let"|"var"|"def"] Var "(" Param%"," ")" "{" Body "}" ([";"|"in"] Body?)?
auto pos = currentPosition();
string layer = "";
if( tryEat("@") )
{
layer = "@" ~ eatId("after @", AllowQuoted);
if( tryEat("(") )
return null; // @lay(...) expression, not a declaration
}
string kwd = layer;
if( layer.empty && !tryEat(kwd="let") && !tryEat(kwd="var") && !tryEat(kwd="def") )
return null; // none of {@lay, let, var, def} occurred, it's not a declaration
auto varpos = currentPosition();
string var = eatId("after "~kwd, AllowQuoted); // name of the declared variable
auto e = tryEat("(")
? parseLambdaAfterOpenParen(varpos) // let var ( ...
: (eat("=", "after "~kwd), E(0)); // let var = ...
if( moreDeclarationExists() )
return new LetExpression(pos, var, layer, e, Body());
else
return new LetExpression(pos, var, layer, e, new VarExpression(varpos, var));
}
AST TopLevelExpression()
{
/// TopLevelExpression ::= Expression ([";"|"in"] Body?)?
auto pos = currentPosition();
auto e = E(0);
if( moreDeclarationExists() )
return new LetExpression(pos, "_", "", e, Body());
else
return e;
}
private bool moreDeclarationExists()
{
return (tryEat(";") || tryEat("in")) && !closingBracket();
}
private bool closingBracket()
{
return lex.empty || !lex.front.quoted && ["}",")","]"].canFind(lex.front.str);
}
// [TODO] make this customizable from program
private static string[][] operator_perferences = [
["||"],
["&&"],
["!="],
["=="],
["<","<=",">",">="],
["|"],
["^"],
["&"],
["<<", ">>"],
["+","-"],
["~"],
["*","/","%"],
["^^","**"]
];
AST E(size_t level)
{
/// Expression ::= (Binary left-associative operators over) Funcall
AST rec(AST lhs)
{
if( closingBracket() )
return lhs;
auto pos = currentPosition();
foreach(op; operator_perferences[level])
if( tryEat(op) )
return rec(
new FuncallExpression(lhs.pos, new VarExpression(pos, op), lhs, E(level+1)));
return lhs;
}
if( operator_perferences.length <= level )
return Funcall();
else
return rec(E(level+1));
}
AST Funcall()
{
/// Funcall ::= BaseExpression ["(" Expression%"," ")"]*
auto e = BaseExpression();
while( tryEat("(") )
{
auto pos = currentPosition();
AST[] args;
while( !tryEat(")") ) {
if( lex.empty )
throw genex!UnexpectedEOF(pos, "closing ')' for arguments not found");
args ~= E(0);
if( !tryEat(",") ) {
eat(")", "after function parameters");
break;
}
}
e = new FuncallExpression(e.pos, e, args);
}
return e;
}
AST BaseExpression()
{
if( lex.empty )
throw genex!UnexpectedEOF(currentPosition(), "Reached EOF when tried to parse an expression");
auto pos = lex.front.pos;
if( lex.front.quoted )
{
scope(exit) lex.popFront;
return new StrLiteral(pos, lex.front.str);
}
if( isNumber(lex.front.str) )
{
scope(exit) lex.popFront;
return new IntLiteral(pos, BigInt(cast(string)lex.front.str));
}
if( tryEat("@") )
{
auto lay = "@"~eatId("for layer ID");
eat("(", "for layered execution");
auto e = Body();
eat(")", "after "~lay~"(...");
return new LayeredExpression(pos, lay, e);
}
if( tryEat("(") )
{
auto e = Body();
eat(")", "after parenthesized expression");
return e;
}
if( tryEat("if") )
{
eat("(", "after if");
auto cond = E(0);
eat(")", "after if condition");
auto thenPos = lex.front.pos;
eat("{", "after if condition");
auto th = Body();
eat("}", "after if-then body");
auto el = doNothingExpression();
auto elsePos = (lex.empty ? LexPosition.dummy : lex.front.pos);
if( tryEat("else") ) {
eat("{", "after else");
el = Body();
eat("}", "after else body");
}
return new FuncallExpression(pos,
new VarExpression(pos, "if"),
cond,
new FunLiteral(thenPos, [], th),
new FunLiteral(elsePos, [], el)
);
}
if( tryEat("fun") || tryEat("\u03BB") ) // lambda!!
{
eat("(", "after fun");
return parseLambdaAfterOpenParen(pos);
}
scope(exit) lex.popFront;
return new VarExpression(pos, lex.front.str);
}
AST parseLambdaAfterOpenParen(immutable LexPosition pos)
{
Parameter[] params;
while( !tryEat(")") )
{
params ~= parseParam();
if( !tryEat(",") ) {
eat(")", "after function parameters");
break;
}
}
eat("{", "after function parameters");
auto funbody = Body();
eat("}", "after function body");
return new FunLiteral(pos, params, funbody);
}
Parameter parseParam()
{
string var;
string[] lay;
while( !closingBracket() && !lex.empty && lex.front.str!="," )
{
auto pos = currentPosition();
string p = eatId("for function parameter", AllowQuoted);
if( p == "@" )
lay ~= "@" ~ eatId("after @", AllowQuoted);
else if( var.empty )
var = p;
else
throw genex!ParseException(pos, "one parameter has two names");
}
return new Parameter(var, lay);
}
private:
Lexer lex;
this(Lexer lex) { this.lex = lex; }
bool isNumber(string s)
{
return find!(`a<'0' || '9'<a`)(s).empty;
}
void eat(string kwd, lazy string msg)
{
if( !tryEat(kwd) )
if( lex.empty )
throw genex!UnexpectedEOF(
currentPosition(), sprintf!"%s is expected %s but not found"(kwd,msg));
else
throw genex!ParseException(
currentPosition(), sprintf!"%s is expected for %s but not found"(kwd,msg));
}
bool tryEat(string kwd)
{
if( lex.empty || lex.front.quoted || lex.front.str!=kwd )
return false;
lex.popFront;
return true;
}
enum {AllowQuoted=true, DisallowQuoted=false};
string eatId(lazy string msg, bool aq=DisallowQuoted)
{
if( lex.empty )
throw genex!UnexpectedEOF(currentPosition(), "identifier is expected but not found "~msg);
if( !aq && lex.front.quoted )
throw genex!ParseException(currentPosition(), "identifier is expected but not found "~msg);
scope(exit) lex.popFront;
return lex.front.str;
}
AST doNothingExpression()
{
return new IntLiteral(currentPosition(), BigInt(178));
}
immutable(LexPosition) currentPosition()
{
return lex.empty ? null : lex.front.pos;
}
}
unittest
{
mixin EasyAST;
assert_eq(parseString(`123`), intl(123));
assert_eq(parseString(`"foo"`), strl("foo"));
assert_eq(parseString(`fun(){1}`), fun([],intl(1)));
assert_eq(parseString(`fun(x){1}`), fun(["x"],intl(1)));
assert_eq(parseString("\u03BB(){1}"), fun([],intl(1)));
assert_eq(parseString("\u03BB(x){1}"), fun(["x"],intl(1)));
assert_eq(parseString(`1;2`), let("_","",intl(1),intl(2)));
assert_eq(parseString(`1;2;`), let("_","",intl(1),intl(2)));
assert_eq(parseString(`let x=1 in 2`), let("x","",intl(1),intl(2)));
assert_eq(parseString(`var x=1;2;`), let("x","",intl(1),intl(2)));
assert_eq(parseString(`def x=1`), let("x","",intl(1),var("x")));
assert_eq(parseString(`@val x=1;`), let("x","@val",intl(1),var("x")));
assert_eq(parseString(`@typ x="#int";`), let("x","@typ",strl("#int"),var("x")));
assert_eq(parseString(`f(1,2)`), call(var("f"),intl(1),intl(2)));
assert_eq(parseString(`if(1){2}`), call(var("if"),intl(1),fun([],intl(2)),fun([],intl(178))));
assert_eq(parseString(`if(1){2}else{3}`), call(var("if"),intl(1),fun([],intl(2)),fun([],intl(3))));
assert_eq(parseString(`if(1){}else{3}()()`),
call(call(call(var("if"),intl(1),fun([],intl(178)),fun([],intl(3))))));
assert_eq(parseString(`1+2*3`), call(var("+"),intl(1),call(var("*"),intl(2),intl(3))));
assert_eq(parseString(`(1+2)*3`), call(var("*"),call(var("+"),intl(1),intl(2)),intl(3)));
assert_eq(parseString(`1*(2+3)`), call(var("*"),intl(1),call(var("+"),intl(2),intl(3))));
assert_eq(parseString(`1*2+3`), call(var("+"),call(var("*"),intl(1),intl(2)),intl(3)));
assert_eq(parseString(`@x(1)`), lay("@x", intl(1)));
assert_eq(parseString(`fun(x @v @t, y, z @t){}`),
funp([param("x",["@v","@t"]), param("y",[]), param("z",["@t"])], intl(178)));
assert_eq(parseString(`
let x = 100; #comment
let y = 200; #comment!!!!!
x+y
`),
let("x", "", intl(100), let("y", "", intl(200), call(var("+"), var("x"), var("y"))))
);
assert_eq(parseString(`
var fac = fun(x){ if(x <= 1) {1} else {x*fac(x-1)} };
fac(10)
`),
let("fac", "", fun(["x"],
call(var("if"),
call(var("<="), var("x"), intl(1)),
fun([], intl(1)),
fun([], call(var("*"), var("x"), call(var("fac"),call(var("-"),var("x"),intl(1)))))
)),
call(var("fac"),intl(10))
)
);
}
unittest
{
assert_throw!UnexpectedEOF(parseString(`1+`));
assert_throw!ParseException(parseString(`1+2}`));
assert_throw!UnexpectedEOF(parseString(`let "x"`));
assert_throw!UnexpectedEOF(parseString(`var`));
assert_throw!ParseException(parseString(`@val x ==`));
assert_throw!ParseException(parseString(`if(){1}`));
assert_throw!UnexpectedEOF(parseString(`f(`));
}
unittest
{
mixin EasyAST;
assert_eq(parseString(`def foo(x) { x+1 }; foo`),
let("foo", "",
fun(["x"], call(var("+"), var("x"), intl(1))),
var("foo"))
);
}