mirror of
https://github.com/curioustorvald/tsvm.git
synced 2026-03-07 19:51:51 +09:00
396 lines
14 KiB
JavaScript
396 lines
14 KiB
JavaScript
class ParserError extends Error {
|
|
constructor(...args) {
|
|
super(...args);
|
|
Error.captureStackTrace(this, ParserError);
|
|
}
|
|
}
|
|
let bF = {};
|
|
|
|
/** Parses following EBNF rule:
|
|
* stmt =
|
|
* "IF" , if_equation , "THEN" , stmt , ["ELSE" , stmt]
|
|
* | "DEFUN" , [ident] , "(" , [ident , {" , " , ident}] , ")" , "=" , stmt
|
|
* | "ON" , ident , ident , equation , {"," , equation}
|
|
* | "(" , stmt , ")"
|
|
* | function_call ;
|
|
* @return: BasicAST
|
|
*/
|
|
bF._parseStmt = function(lnum, tokens, states, recDepth) {
|
|
let headTkn = tokens[0].toUpperCase();
|
|
let headSta = states[0];
|
|
|
|
let treeHead = new BasicAST();
|
|
treeHead.astDepth = recDepth;
|
|
treeHead.astLnum = lnum;
|
|
|
|
let thenPos = -1;
|
|
let elsePos = -1;
|
|
let parenDepth = 0;
|
|
let parenStart = -1;
|
|
let parenEnd = -1;
|
|
|
|
// scan for parens that will be used for several rules
|
|
// also find nearest THEN and ELSE but also take parens into account
|
|
for (let k = 0; k < tokens.length; k++) {
|
|
// increase paren depth and mark paren start position
|
|
if (tokens[k] == "(" && states[k] != "qot") {
|
|
parenDepth += 1;
|
|
if (parenStart == -1 && parenDepth == 1) parenStart = k;
|
|
}
|
|
// decrease paren depth
|
|
else if (tokens[k] == ")" && states[k] != "qot") {
|
|
if (parenEnd == -1 && parenDepth == 1) parenEnd = k;
|
|
parenDepth -= 1;
|
|
}
|
|
|
|
if (parenDepth == 0) {
|
|
if (-1 == thenPos && "THEN" == tokens[k].toUpperCase() && "lit" == states[k])
|
|
thenPos = k;
|
|
else if (-1 == elsePos && "ELSE" == tokens[k].toUpperCase() && "lit" == states[k])
|
|
elsePos = k;
|
|
}
|
|
}
|
|
|
|
// unmatched brackets, duh!
|
|
if (parenDepth != 0) throw lang.syntaxfehler(lnum, lang.unmatchedBrackets);
|
|
|
|
|
|
// ## case for:
|
|
// "IF" , if_equation , "THEN" , stmt , ["ELSE" , stmt]
|
|
if ("IF" == headTkn && "lit" == headSta) {
|
|
// "THEN" not found, raise error!
|
|
if (thenPos == -1) throw new ParserError("IF without THEN in " + lnum);
|
|
|
|
treeHead.astValue = "IF";
|
|
treeHead.astType = "function";
|
|
|
|
treeHead.astLeaves[0] = bF._parseEquation(lnum,
|
|
tokens.slice(1, thenPos),
|
|
states.slice(1, thenPos),
|
|
recDepth + 1,
|
|
true // if_equation mode
|
|
);
|
|
treeHead.astLeaves[1] = bF._parseStmt(lnum,
|
|
tokens.slice(thenPos + 1, (elsePos != -1) ? elsePos : tokens.length),
|
|
states.slice(thenPos + 1, (elsePos != -1) ? elsePos : tokens.length),
|
|
recDepth + 1
|
|
);
|
|
if (elsePos != -1)
|
|
treeHead.astLeaves[2] = bF._parseStmt(lnum,
|
|
tokens.slice(elsePos + 1, tokens.length),
|
|
states.slice(elsePos + 1, tokens.length),
|
|
recDepth + 1
|
|
);
|
|
|
|
return treeHead;
|
|
}
|
|
|
|
// ## case for:
|
|
// | "(" , stmt , ")"
|
|
if (parenStart == 0 && parenEnd == tokens.length - 1) {
|
|
return bF._parseStmt(lnum,
|
|
tokens.slice(parenStart + 1, parenEnd),
|
|
states.slice(parenStart + 1, parenEnd),
|
|
recDepth
|
|
);
|
|
}
|
|
|
|
// ## case for:
|
|
// | function_call ;
|
|
try {
|
|
return bF._parseFunctionCall(lnum, tokens, states, recDepth);
|
|
}
|
|
catch (e) {
|
|
throw new ParserError("Statement cannot be parsed: "+e+" in "+lnum);
|
|
}
|
|
}
|
|
/** Parses following EBNF rule:
|
|
* equation
|
|
* | ident , "(" , [function_call , {argsep , function_call} , [argsep]] , ")"
|
|
* | ident , function_call , {argsep , function_call} , [argsep]
|
|
* @return: BasicAST
|
|
*/
|
|
bF._parseFunctionCall = function(lnum, tokens, states, recDepth) {
|
|
// ## case for:
|
|
// equation
|
|
try {
|
|
return bF._parseEquation(lnum, tokens, states, recDepth);
|
|
}
|
|
// if ParserError is raised, continue to apply other rules
|
|
catch (e) {
|
|
if (!(e instanceof ParserError)) throw e;
|
|
}
|
|
|
|
let parenDepth = 0;
|
|
let parenStart = -1;
|
|
let parenEnd = -1;
|
|
let _argsepsOnLevelZero = []; // argseps collected when parenDepth == 0
|
|
let _argsepsOnLevelOne = []; // argseps collected when parenDepth == 1
|
|
|
|
// Scan for unmatched parens and mark off the right operator we must deal with
|
|
for (let k = 0; k < tokens.length; k++) {
|
|
// increase paren depth and mark paren start position
|
|
if (tokens[k] == "(" && states[k] != "qot") {
|
|
parenDepth += 1;
|
|
if (parenStart == -1 && parenDepth == 1) parenStart = k;
|
|
}
|
|
// decrease paren depth
|
|
else if (tokens[k] == ")" && states[k] != "qot") {
|
|
if (parenEnd == -1 && parenDepth == 1) parenEnd = k;
|
|
parenDepth -= 1;
|
|
}
|
|
|
|
if (parenDepth == 0 && states[k] == "sep")
|
|
_argsepsOnLevelZero.push(k);
|
|
if (parenDepth == 1 && states[k] == "sep")
|
|
_argsepsOnLevelOne.push(k);
|
|
}
|
|
|
|
// unmatched brackets, duh!
|
|
if (parenDepth != 0) throw lang.syntaxfehler(lnum, lang.unmatchedBrackets);
|
|
let parenUsed = (parenStart == 1 && parenEnd == states.length - 1);
|
|
|
|
// ## case for:
|
|
// | ident , "(" , [function_call , {argsep , function_call} , [argsep]] , ")"
|
|
// | ident , function_call , {argsep , function_call} , [argsep]
|
|
let treeHead = new BasicAST();
|
|
treeHead.astDepth = recDepth;
|
|
treeHead.astLnum = lnum;
|
|
|
|
// set function name and also check for syntax by deliberately parsing the word
|
|
treeHead.astValue = bF._parseIdent(lnum, [tokens[0]], [states[0]], recDepth + 1).astValue; // always UPPERCASE
|
|
|
|
// 5 8 11 [end]
|
|
let argSeps = parenUsed ? _argsepsOnLevelOne : _argsepsOnLevelZero; // choose which "sep tray" to use
|
|
// 1 6 9 12
|
|
let argStartPos = [1 + (parenUsed)].concat(argSeps.map(k => k+1));
|
|
// [1,5) [6,8) [9,11) [12,end)
|
|
let argPos = argStartPos.map((s,i) => {return{start:s, end:(argSeps[i] || tokens.length - (parenUsed))}}); // use end of token position as separator position
|
|
|
|
// check for trailing separator
|
|
let hasTrailingSep = (states[states.length - 1 - (parenUsed)] == "sep");
|
|
// exclude last separator from recursion if input tokens has trailing separator
|
|
if (hasTrailingSep) argPos.pop();
|
|
|
|
// recursively parse function arguments
|
|
treeHead.astLeaves = argPos.map(x => bF._parseFunctionCall(lnum,
|
|
tokens.slice(x.start, x.end),
|
|
states.slice(x.start, x.end),
|
|
recDepth + 1
|
|
));
|
|
treeHead.astType = "function";
|
|
treeHead.astSeps = argSeps;
|
|
|
|
return treeHead;
|
|
}
|
|
bF._parseIdent = function(lnum, tokens, states, recDepth) {
|
|
if (!Array.isArray(tokens) && !Array.isArray(states)) throw new ParserError("Tokens and states are not array");
|
|
if (tokens.length > 1 || states[0] != "lit") throw new ParserError(`illegal token count '${tokens.length}' with states '${states}' in ${lnum}`);
|
|
|
|
let treeHead = new BasicAST();
|
|
treeHead.astDepth = recDepth;
|
|
treeHead.astLnum = lnum;
|
|
treeHead.astValue = tokens[0].toUpperCase();
|
|
treeHead.astType = "lit";
|
|
|
|
return treeHead;
|
|
}
|
|
/**
|
|
* @return: BasicAST
|
|
*/
|
|
bF._parseLit = function(lnum, tokens, states, recDepth) {
|
|
if (!Array.isArray(tokens) && !Array.isArray(states)) throw new ParserError("Tokens and states are not array");
|
|
if (tokens.length > 1) throw new ParserError("parseLit 1");
|
|
|
|
let treeHead = new BasicAST();
|
|
treeHead.astDepth = recDepth;
|
|
treeHead.astLnum = lnum;
|
|
if (_debugSyntaxAnalysis) serial.println("literal/number: "+tokens[0]);
|
|
treeHead.astValue = ("qot" == states[0]) ? tokens[0] : tokens[0].toUpperCase();
|
|
treeHead.astType = ("qot" == states[0]) ? "string" : ("num" == states[0]) ? "num" : "lit";
|
|
|
|
return treeHead;
|
|
}
|
|
bF._EquationIllegalTokens = ["IF","THEN","ELSE","DEFUN","ON"];
|
|
bF.isSemanticLiteral = function(token, state) {
|
|
return "]" == token || ")" == token ||
|
|
"qot" == state || "num" == state || "bool" == state || "lit" == state;
|
|
}
|
|
/** Parses following EBNF rule:
|
|
* equation =
|
|
* lit
|
|
* | "(" , equation , ")"
|
|
* | equation , op , equation
|
|
* | op_uni , equation
|
|
* @return: BasicAST
|
|
*/
|
|
bF._parseEquation = function(lnum, tokens, states, recDepth, ifMode) {
|
|
|
|
// ## case for:
|
|
// lit
|
|
let headTkn = tokens[0].toUpperCase();
|
|
if (!bF._EquationIllegalTokens.includes(headTkn) && tokens.length == 1) {
|
|
return bF._parseLit(lnum, tokens, states, recDepth);
|
|
}
|
|
|
|
// scan for operators with highest precedence, use rightmost one if multiple were found
|
|
let topmostOp;
|
|
let topmostOpPrc = 0;
|
|
let operatorPos = -1;
|
|
|
|
// find and mark position of parentheses
|
|
// properly deal with the nested function calls
|
|
let parenDepth = 0;
|
|
let parenStart = -1;
|
|
let parenEnd = -1;
|
|
|
|
// Scan for unmatched parens and mark off the right operator we must deal with
|
|
for (let k = 0; k < tokens.length; k++) {
|
|
// increase paren depth and mark paren start position
|
|
if (tokens[k] == "(" && states[k] != "qot") {
|
|
parenDepth += 1;
|
|
if (parenStart == -1 && parenDepth == 1) parenStart = k;
|
|
}
|
|
// decrease paren depth
|
|
else if (tokens[k] == ")" && states[k] != "qot") {
|
|
if (parenEnd == -1 && parenDepth == 1) parenEnd = k;
|
|
parenDepth -= 1;
|
|
}
|
|
|
|
// determine the right operator to deal with
|
|
if (parenDepth == 0) {
|
|
if (states[k] == "op" && bF.isSemanticLiteral(tokens[k-1], states[k-1]) &&
|
|
((bF._opPrc[tokens[k].toUpperCase()] > topmostOpPrc) ||
|
|
(!bF._opRh[tokens[k].toUpperCase()] && bF._opPrc[tokens[k].toUpperCase()] == topmostOpPrc))
|
|
) {
|
|
topmostOp = tokens[k].toUpperCase();
|
|
topmostOpPrc = bF._opPrc[tokens[k].toUpperCase()];
|
|
operatorPos = k;
|
|
}
|
|
}
|
|
}
|
|
|
|
// unmatched brackets, duh!
|
|
if (parenDepth != 0) throw lang.syntaxfehler(lnum, lang.unmatchedBrackets);
|
|
if (_debugSyntaxAnalysis) serial.println("Equation NEW Paren position: "+parenStart+", "+parenEnd);
|
|
|
|
// ## case for:
|
|
// "(" , equation , ")"
|
|
if (parenStart == 0 && parenEnd == tokens.length - 1) {
|
|
return bF._parseEquation(lnum,
|
|
tokens.slice(parenStart + 1, parenEnd),
|
|
states.slice(parenStart + 1, parenEnd),
|
|
recDepth
|
|
);
|
|
}
|
|
// ## case for:
|
|
// equation , op, equation
|
|
// | op_uni , equation
|
|
// if operator is found, split by the operator and recursively parse the LH and RH
|
|
if (topmostOp !== undefined) {
|
|
if (_debugSyntaxAnalysis) serial.println("operator: "+topmostOp+", pos: "+operatorPos);
|
|
|
|
if (ifMode && topmostOp == "=") throw lang.syntaxfehler(lnum, "'=' used on IF, did you mean '=='?");
|
|
if (ifMode && topmostOp == ":") throw lang.syntaxfehler(lnum, "':' used on IF");
|
|
|
|
|
|
// this is the AST we're going to build up and return
|
|
// (other IF clauses don't use this)
|
|
let treeHead = new BasicAST();
|
|
treeHead.astDepth = recDepth;
|
|
treeHead.astLnum = lnum;
|
|
treeHead.astValue = topmostOp;
|
|
treeHead.astType = "op";
|
|
|
|
// BINARY_OP?
|
|
if (operatorPos > 0) {
|
|
let subtknL = tokens.slice(0, operatorPos);
|
|
let substaL = states.slice(0, operatorPos);
|
|
let subtknR = tokens.slice(operatorPos + 1, tokens.length);
|
|
let substaR = states.slice(operatorPos + 1, tokens.length);
|
|
|
|
treeHead.astLeaves[0] = bF._parseEquation(lnum, subtknL, substaL, recDepth + 1);
|
|
treeHead.astLeaves[1] = bF._parseEquation(lnum, subtknR, substaR, recDepth + 1);
|
|
}
|
|
else {
|
|
treeHead.astValue = (topmostOp === "-") ? "UNARYMINUS" : "UNARYPLUS";
|
|
treeHead.astLeaves[0] = bF._parseEquation(lnum,
|
|
tokens.slice(operatorPos + 1, tokens.length),
|
|
states.slice(operatorPos + 1, states.length),
|
|
recDepth + 1
|
|
);
|
|
}
|
|
|
|
return treeHead;
|
|
}
|
|
|
|
throw new ParserError(`Equation - illegal token "${headTkn}" in ${lnum}`);
|
|
|
|
}
|
|
|
|
|
|
/////// TEST/////////
|
|
let astToString = function(ast) {
|
|
if (ast === undefined || ast.astType === undefined) return "";
|
|
var sb = "";
|
|
var marker = ("lit" == ast.astType) ? "i" :
|
|
("op" == ast.astType) ? String.fromCharCode(177) :
|
|
("string" == ast.astType) ? String.fromCharCode(182) :
|
|
("num" == ast.astType) ? String.fromCharCode(162) :
|
|
("array" == ast.astType) ? "[" : String.fromCharCode(163);
|
|
sb += "| ".repeat(ast.astDepth) + marker+" Line "+ast.astLnum+" ("+ast.astType+")\n";
|
|
sb += "| ".repeat(ast.astDepth+1) + "leaves: "+(ast.astLeaves.length)+"\n";
|
|
sb += "| ".repeat(ast.astDepth+1) + "value: "+ast.astValue+" (type: "+typeof ast.astValue+")\n";
|
|
for (var k = 0; k < ast.astLeaves.length; k++) {
|
|
if (k > 0)
|
|
sb += "| ".repeat(ast.astDepth+1) + " " + ast.astSeps[k - 1] + "\n";
|
|
sb += astToString(ast.astLeaves[k]);
|
|
}
|
|
sb += "| ".repeat(ast.astDepth) + "`-----------------\n";
|
|
return sb;
|
|
}
|
|
let BasicAST = function() {
|
|
this.astLnum = 0;
|
|
this.astDepth = 0;
|
|
this.astLeaves = [];
|
|
this.astSeps = [];
|
|
this.astValue = undefined;
|
|
this.astType = "null"; // literal, operator, string, number, array, function, null, defun_args (! NOT usrdefun !)
|
|
}
|
|
bF._opPrc = {
|
|
// function call in itself has highest precedence
|
|
"^":1,
|
|
"*":2,"/":2,
|
|
"MOD":3,
|
|
"+":4,"-":4,
|
|
//";":5,
|
|
"<<":6,">>":6,
|
|
"<":7,">":7,"<=":7,"=<":7,">=":7,"=>":7,
|
|
"==":8,"<>":8,"><":8,
|
|
"BAND":8,
|
|
"BXOR":9,
|
|
"BOR":10,
|
|
"AND":11,
|
|
"OR":12,
|
|
"TO":13,
|
|
"STEP":14,
|
|
"!":15,"~":15, // array CONS and PUSH
|
|
"#": 16, // array concat
|
|
"=":999,
|
|
"IN":1000
|
|
};
|
|
bF._opRh = {"^":1,"=":1,"!":1,"IN":1};
|
|
let lnum = 10;
|
|
let tokens = ["print","2","+","5","*","3"];
|
|
let states = ["lit","num","op","num","op","num"];
|
|
let _debugSyntaxAnalysis = false;
|
|
|
|
try {
|
|
let tree = bF._parseStmt(lnum, tokens, states, 0);
|
|
serial.println(astToString(tree));
|
|
}
|
|
catch (e) {
|
|
serial.printerr(e);
|
|
serial.printerr(e.stack || "stack trace undefined");
|
|
}
|