var fs = require('fs');
-
-export type Some<T> = { _tag: "Some"; value: T };
-export type None = {_tag: "None"};
-
-
+import jsTokens from "js-tokens";
+import * as util from 'util';
+import * as p from 'typescript-parsec';
+import { Token } from 'typescript-parsec';
+import { TokenType } from "./tokenize";
/**
- * wrap a x in a `Some(T)`
- * @param x : variable to be wrapped.
- * @returns wrapped `x`.
+ *
+ * # REPRESENTATION
*/
-function toSome<T>(x: T): Some<T>{
- return { _tag: "Some", value: x};
-}
/**
- * @description Like the `Some(a)` and `None` in Rust.
- *
- * @example
- * ```ts
- * let exam1 : Maybe<Number> = { _tag: "Some", value: 12 };
- * let exam2 : Maybe<Number> = None;
- * ```
+ * convert a `tkTree` AST to S-expr string
+ * @param t the `tkTree`
+ * @returns S-expr String
*/
-export type Maybe<T> = Some<T> | None;
+export function tkTreeToSExp(t: tkTree): string{
+ var str = "";
+ if (Array.isArray(t)){
+ let strArray = t.map((x)=>tkTreeToSExp(x));
+ str = "(" + strArray.join(" ") + ")";
+ }else{
+ if (t=== undefined){
+ str = "%undefined"
+ }else{
+ str = t;
+ }
+ }
-/**
- * @description
- * the pair of the string to be matched later and the string that have been matched
- * @var matched : have been matched
- * @var remained : will be tested whether it'll be matched.
- * @var matched_type (optional): the type of the matched string
-*/
-export interface MatcheePair {
- matched : string
- remained : string
- matched_type?: TokenType
+ return str;
}
+/**inspect the inner of the representation. */
+let repr = (x : any)=>{return util.inspect(x, {depth: null})};
/**
- * The types of Token
- * NL, // newline
- *
- * SP, // half-width space and tab
- *
- * ID, // identifier
- *
- * STR, // string
*
- * OP, // operator or something like it
- *
- * FLO, // float num
- *
- * INT, // Integer
+ * # TYPES
*/
-export enum TokenType{
- NL, // newlinw
- SP, // half-width space and tab
- ID, // identifier
- STR, // string
- OP, // operator
- FLO, // float num
- INT, // integer
-}
-/**
- * tokenized token.
- * @var text : the content text
- * @var type (optional): the type of the token
- * @var col : the column number
- * @var ln : the line number
- */
-export interface Token{
- text: string,
- type?: TokenType,
- col: number,
- ln: number,
+
+type tkTree = string | tkTree[];
+
+enum TokenKind {
+ Seperator, // ---
+ Semicolon, // ;
+ Number,
+ Op,
+ ExprMark, // @
+ ExcapeAt, // \@
+ Paren,
+ SpaceNL, // \s\t\n\r
+ Id,
+ Str,
+ Comment, // /* ooo */
}
/**
- * @description
- * it returns a function which test if the first char of the `remained` part of
- * the argument of the function is `c`, if it's true, update the `MatchedPair` wrapped
- * in `Some`. Otherwise, it returns `None`.
- * * @param c : the char to be test.
- * @returns the updated `MatchedPair` wrapped in `Some(x)` or `None`.
+ * Parsing
*/
-export function match1Char(c : string) : (m: MatcheePair) => Maybe<MatcheePair> {
- return (m : MatcheePair)=>{
- if (m.remained.length == 0){
- return { _tag: "None" };
- }
- const charToBeMatched = m.remained[0];
- if (charToBeMatched === c){
- return {_tag: "Some", value :{
- matched : m.matched + charToBeMatched,
- remained : m.remained.substring(1)}};
- }
- else{
- return {_tag: "None"};
- }
- }
-};
+const lexer = p.buildLexer([
+ [true, /^\d+(\.\d+)?/g, TokenKind.Number],
+ [true, /^\\\@/g, TokenKind.ExcapeAt],
+ [true, /^\/\*([^/]|\/[^*])*\*\//g, TokenKind.Comment],
+ [true, /^\;/g, TokenKind.Semicolon],
+ [true, /^[-][-][-]/g, TokenKind.Seperator],
+ [true, /^[\+\-\*\/\&\|\!\^\<\>\~\=\?]+/g, TokenKind.Op],
+ [true, /^\@/g, TokenKind.ExprMark],
+ [true, /^[()\[\]{}]/g, TokenKind.Paren],
+ [true, /^["]([\"]|[\\].)*["]/g, TokenKind.Str],
+ [true, /^[']([\']|[\\].)*[']/g, TokenKind.Str],
+ [true, /^[()\[\]{}]/g, TokenKind.Paren],
+ [true, /^[^\/\\\@\s\n\t\r;]+/g, TokenKind.Id],
+ [true, /^(\s|\n|\r|\t)+/g, TokenKind.SpaceNL],
+
+]);
/**
*
- * @param m : the `MatcheePair` to be consumed.
- * @returns if the length of `m.remained` >= 1; consumes the matchee by 1 char and wraps it in `Some`,
- * otherwise, returns `None`.
+ * # TEST
*/
-export function matchAny(m : MatcheePair) : Maybe<MatcheePair>{
- if (m.remained.length >= 1){
- return {_tag: "Some", value :{
- matched : m.matched + m.remained[0],
- remained : m.remained.substring(1)}};
- }else{
- return {_tag: "None"};
- }
+const inputTxt=
+`import a as b; /*bacourt*/
+/* ba choir
+ipsum lorem*/
+
+import you as john;
+---
+
+臺中市\\\@
+
+公園
+@1+2==3;
+
+console.log("122");@
+
+山頂
+`;
+
+
+const PROG = p.rule<TokenKind, tkTree>();
+const SEGMENT = p.rule<TokenKind, tkTree>();
+const IMPORT = p.rule<TokenKind, tkTree>();
+const IMPORTS = p.rule<TokenKind, tkTree>();
+const SEMICOLON = p.rule<TokenKind, tkTree>();
+const EXCAPE_AT = p.rule<TokenKind, tkTree>();
+const NOT_AT_TEXT = p.rule<TokenKind, tkTree>();
+const CONTENT = p.rule<TokenKind, tkTree>();
+
+let doubleMinus = { type: 'Punctuator', value: '--' };
+let doubleMinus2 = p.str('--');
+const TERM = p.rule<TokenKind, tkTree>();
+
+function applySegment(input: [Token<TokenKind>, Token<TokenKind>[],
+ Token<TokenKind>]): tkTree[]{
+ let unpackedInnerExprs = input[1].map((x)=>{return x.text});
+ return ["%exprs", unpackedInnerExprs];
}
-/**
- * @description
- * it returns a function which test if the first char of the `remained` part of
- * the argument of the function is between `l` and `u`, if it's true, update the `MatchedPair` wrapped
- * in `Some`. Otherwise, it returns `None`.
- * * @param l : lower bound char, 1-char string
- * * @param u : upper bound char, 1-char string
- * @returns the updated `MatchedPair` wrapped in `Some(x)` or `None`.
- */
-export function matchRange(l : string, u : string) : (m: MatcheePair) => Maybe<MatcheePair> {
- let lCodepoint = charToCodepoint(l);
- let uCodepoint = charToCodepoint(u);
- if (l > u){
- throw new Error("Error: the codepoint of `"+l+"` is not smaller than `"+u+"`)");
- }
- return (m : MatcheePair)=>{
- if (m.remained.length < 1){
- return {_tag : "None"};
- }
- const charToBeMatched = m.remained[0];
- const codePointToBeMatched = charToCodepoint(charToBeMatched);
- if (codePointToBeMatched >= lCodepoint && codePointToBeMatched <= uCodepoint){
- return {_tag: "Some", value :{
- matched : m.matched + charToBeMatched,
- remained : m.remained.substring(1)}};
- }
- else{
- return {_tag: "None"};
- }
- }
+function applySemiColon(value: Token<TokenKind.Semicolon>): tkTree{
+ return value.text;
+}
+
+function applyParts(first: tkTree,
+ second: [Token<TokenKind>, tkTree]):tkTree {
+ return ["%clo", first , second[1]];
+}
+
+
+function applyComment(value: Token<TokenKind.Comment>): tkTree[]{
+ return [value.text];
+}
+
+
+function applyImport(input: [Token<TokenKind>,Token<TokenKind>[], tkTree]) : tkTree{
+ let importTail = input[1].map(x=>x.text);
+ return ["import"].concat(importTail);
};
-/**
- * convert the one-char string to codepoint.
- * @param s : the string to code point.
- * @returns if `s.length > 1` return error; otherwise, return the codepoint of `s`.
- */
-export function charToCodepoint(s : string): number{
- if (s.length > 1){
- throw new Error("Error: the length of input string for "+s+ "is "+s.length+`,
- however, it should be 1.`);
- }else{
- return s.charCodeAt(0);
+
+/*
+function applyImportComment(input: [Token<TokenKind>,Token<TokenKind>[],
+ tkTree, Token<TokenKind.Comment>]) : tkTree{
+ let importTail = input[1].map(x=>x.text);
+ let comment = [input[3].text];
+ return ["import"].concat(importTail).concat(comment);
+};*/
+
+function applyImports(input : [tkTree, tkTree[]]): tkTree{
+ let resultBody = [input[0]].concat(input[1]);
+ let resultWrapper = ["%import", resultBody];
+ return resultWrapper;
+};
+
+
+
+
+function applyNotAtText(value : Token<TokenKind>): tkTree{
+ if (value.text == "\\\@"){
+ return '@';
}
+ else{return value.text;}
+};
+
+function applyText (input : tkTree): tkTree[]{
+ return ["%text", input];
+};
+
+function applyContent(input : tkTree[]): tkTree[]{
+ return ["%content", input];
+};
+
+function applySpaceNL(value : Token<TokenKind.SpaceNL>): tkTree{
+ return value.text;
}
/**
- * @description thendo(input, f, ...) like
- * a ==> f
- * @param input: the wrapped input.
- * @param f: the function to be applied.
- *
- * @returns:the applied wrapped result `MatcheePair`.
+ * IMPORTEE: Number, Op, Paren, Id, Str, Comment,
*/
-export function thenDo<T>(input : Maybe<T>, f : Function) : Maybe<T>{
- if (input._tag == "None"){
- return input;
- }
- else{
- let inner = input.value;
- return f(inner);
- }
-}
+let IMPORTEE = p.alt(p.tok(TokenKind.Number),
+ p.tok(TokenKind.Op),
+ p.tok(TokenKind.Paren),
+ p.tok(TokenKind.Id),
+ p.tok(TokenKind.Str),
+ p.tok(TokenKind.SpaceNL),
+ p.tok(TokenKind.Comment));
+
+let NOT_AT = p.alt(p.tok(TokenKind.Seperator),
+ p.tok(TokenKind.Semicolon),
+ p.tok(TokenKind.Number),
+ p.tok(TokenKind.ExcapeAt),
+ p.tok(TokenKind.Op),
+ p.tok(TokenKind.Paren),
+ p.tok(TokenKind.SpaceNL),
+ p.tok(TokenKind.Id),
+ p.tok(TokenKind.Str),
+ p.tok(TokenKind.Comment),
+ );
/**
- * @description "or", like the regex `( f1 | f2 )` .
- * It returns a function `f` of which the argument is`x`.
- * if `f1(x)` is None, then `f` returns `f2(x)`. Otherwise,
- * `F` returns `f1(x)`.
- * @param f1 : 1st function to be compared
- * @param f2 : 2nd function to be compared
- * @returns:the combined function
+ * PROG : IMPORTS '---' CONTENT;
*/
-export function orDo<T>(f1 : Function, f2: Function) : (x : T ) => Maybe<T>{
- return (x) => {
- let f1x : Maybe<T> = (f1(x));
- {
- if (f1x._tag == "None"){
- return f2(x);
- }
- else{
- return f1x;
- }
- }
- };
-}
+PROG.setPattern(
+ p.lrec_sc(IMPORTS, p.seq(p.str('---'), CONTENT), applyParts)
+)
/**
-* @description repeating matching function `f`
-* zero or more times, like the asterisk `*` in regex `f*` .
-* @param f : the function to be repeated 0+ times.
-* @returns:the combined function
-*/
-export function zeroOrMoreDo<T>(f : Function): (x : T) => Maybe<T>{
- return (x)=>{
- var wrapped_old_x : Maybe<T> = {_tag: "Some", value : x};
- var wrapped_new_x : Maybe<T> = wrapped_old_x;
-
- while (wrapped_new_x._tag != "None"){
- wrapped_old_x = wrapped_new_x;
- wrapped_new_x = thenDo(wrapped_old_x, f);
- };
-
- return wrapped_old_x;
- };
-}
+ * NOT_AT_TEXT : NOT_AT
+ */
+NOT_AT_TEXT.setPattern(
+ p.apply(NOT_AT, applyNotAtText)
+);
-/**
-* @description Not. like the `^` inside regex of [^f].
-* returns a function `F(x)` such that if `f(x)` is `None`,
-* returns the x consuming a char; if `f(x)` is not None, F(x)
-* returns `None`.
-* @param f: the function forbidden to be matched.
-* @returns: combined function `F`.
-*/
-export function notDo<T>(f : Function): (x : T) => Maybe<T>{
- return (x)=>{
- let wrapped_x : Maybe<T> = {
- _tag : "Some",
- value : x
- };
- let f_x = thenDo(wrapped_x, f);
-
- if (f_x._tag != "None"){
- return {_tag:"None"};
- }else{
- return thenDo(wrapped_x, matchAny);
- }
- };
-}
+IMPORTS.setPattern(
+ p.apply( p.seq(IMPORT, p.rep(IMPORT)), applyImports)
+);
/**
- * if `x` is matched by `f` once, returns `f(x)`. Otherwise,
- * returns x
- * similar to `?` in regex `f?`.
- * @param f : the function to be matched
- * @returns return wrapped f(x)
+ * IMPORT :
+ * 'import' IMPORTEE* SEMICOLON |
+ * COMMENT |
*/
-export function zeroOrOnceDo<T>(f : Function): (x : T) => Maybe<T>{
- return (x)=>{
- var wrapped_old_x : Maybe<T> = {_tag: "Some", value : x};
- var wrapped_new_x = thenDo(wrapped_old_x, f);
+IMPORT.setPattern(
+ p.alt(
+ p.apply(p.seq(p.str('import'), p.rep_sc(IMPORTEE), SEMICOLON),
+ applyImport),
+ p.apply(p.tok(TokenKind.Comment), applyComment),
+ p.apply(p.tok(TokenKind.SpaceNL), applySpaceNL)
- if (wrapped_new_x._tag != "None"){
- return wrapped_new_x;
- }else{
- return wrapped_old_x;
- }
- };
-}
+ )
+);
+/**
+ * SEMICOLON : ';';
+ */
+SEMICOLON.setPattern(
+ p.apply(p.tok(TokenKind.Semicolon), applySemiColon)
+);
-export function tokenize(input : string){
- var input_matchee_pair : Maybe<MatcheePair> = toSome(
- {matched:"",
- remained: input});
-
- // integer = ([+]|[-])?\d\d*
- let integer = (x : MatcheePair) =>
- { let wrapped_x = toSome(x);
- let plusMinus = orDo(match1Char('+'), match1Char('-')); // ([+]|[-])
- let d = matchRange('0','9'); // \d
- var result = thenDo(thenDo(thenDo(wrapped_x,
- zeroOrOnceDo(plusMinus)),d),
- zeroOrMoreDo(d));
-
- if (result._tag == "Some"){
- result.value.matched_type = TokenType.INT;
- }
- return result;
- }
- let space = (x : MatcheePair) =>{
- let wrapped_x = toSome(x);
- let s_aux = orDo(match1Char(' '), match1Char('\t')); // (" " | "\t")
- var result = thenDo(thenDo(wrapped_x, s_aux), zeroOrMoreDo(s_aux));
- if (result._tag == "Some"){
- result.value.matched_type = TokenType.SP;
- }
- return result;
- }
- let newline = (x : MatcheePair) =>{
- let wrapped_x = toSome(x);
- // nl = \r?\n
- let result = thenDo(thenDo(wrapped_x,
- zeroOrOnceDo(match1Char('\r'))), match1Char('\n'));
- if (result._tag == "Some"){
- result.value.matched_type = TokenType.NL;
- }
- return result;
- }
- let term = (token_list : Array<Token>, x : Some<MatcheePair>)=>{
- var ln = 1;
- var col = 0;
- var old_x = x;
- let term_list = [newline, space, integer];
- let term_aux = term_list.reduce((x,y)=> orDo(x,y));
-
- var new_x : Maybe<MatcheePair> = thenDo(old_x, term_aux);
- while (new_x._tag != "None"){
- if (new_x.value.matched_type != TokenType.NL){
- col += new_x.value.matched.length;
- token_list.push({text : new_x.value.matched,
- type: new_x.value.matched_type,
- ln : ln,
- col : col});
-
- }
- else{
- col = 0;
- ln += 1;
-
- token_list.push({text : new_x.value.matched,
- type: new_x.value.matched_type,
- ln : ln,
- col : col});
-
- }
-
-
- old_x = toSome({matched : "",
- remained : new_x.value.remained});
- new_x = thenDo(old_x, term_aux);
- }
- if (old_x.value.remained.length){
- console.log(token_list);
- throw new Error("the code can't be tokenized is near Ln. "+ln+", Col."+col
- +", starting with "+ old_x.value.remained.substring(0,10));
- }
+/**
+ * SEGMENT : '@' NOT_AT* '@' |
+ * (NOT_AT_TEXT | EXCAPE_AT)*
+ */
+SEGMENT.setPattern(
+ p.alt(
+ p.apply(p.rep_sc(NOT_AT_TEXT), applyText),
+ p.apply(p.seq(p.str('@'), p.rep(NOT_AT), p.str('@')), applySegment),
+ )
+);
- return token_list;
- }
+/**
+ * CONTENT : SEGMENT*
+ */
+CONTENT.setPattern(
+ p.apply(p.rep(SEGMENT), applyContent)
+);
- console.log(term([], input_matchee_pair));
- // TODO: id, string, space, basic operator, 3 marks: @, {, }.
+let tree = p.expectSingleResult(p.expectEOF(PROG.parse(lexer.parse(inputTxt))));
-}
+console.log("RESULT="+tkTreeToSExp(tree));