rewrite parser

[clo] / src / index.ts

diff --git a/src/index.ts b/src/index.ts
index 3c3384d520bc8bf8d6e54bd6f00bbf47cf909ee2..157b1607348020eac4b7e8648093eb77ff8059c6 100644 (file)
--- a/src/index.ts
+++ b/src/index.ts
@@ -1,74 +1,145 @@
 var fs = require('fs');
+import jsTokens from "js-tokens";
+import * as util from 'util';
+import * as p from 'typescript-parsec';
+import { Token } from 'typescript-parsec';
+/**
+ * 
+ * # REPRESENTATION
+ */
+/**
+ * convert a `tkTree` AST to S-expr string
+ * @param t the `tkTree`
+ * @returns S-expr String
+ */
+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;
+        }
+    }
 
-type Some<T> = { _tag: "Some"; value: T };
-type None = {_tag: "None"};
+    return str;
+}
 
+/**inspect the inner of the representation. */
+let repr = (x : any)=>{return util.inspect(x, {depth: null})};
 /**
- * @description Like the `Some(a)` and `None` in Rust.
- *
- * @example
- * ```ts
- * let exam1 : Maybe<Number> = { _tag: "Some", value: 12 };
- * let exam2 : Maybe<Number> = None;
- * ```
+ * 
+ * # TYPES
  */
-export type Maybe<T> = Some<T> | None;
 
 
+type tkTree = string | tkTree[];
+
+enum TokenKind {
+    Seperator,
+    Semicolon,
+    Number,
+    Op,
+    ExprMark,
+    Paren,
+    SpaceNL,
+    Id,
+    Str,
+}
+
 /**
- * @description
- * the pair of the string to be matched later and the string that have been matched
- * @param matched : string have been matched
- * @param remained : string will be tested whether it'll be matched.
+ * Parsing
  */
-export type MatcheePair = {matched : string; remained : string};
+const lexer = p.buildLexer([
+    [true, /^\d+(\.\d+)?/g, TokenKind.Number],
+    [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],
+    [false, /^(\s|\n|\r|\t)+/g, TokenKind.SpaceNL]
+]);
 
 /**
- * @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`.
+ * 
+ * # TEST
  */
-export function match1Char(c : string) : (m: MatcheePair) => Maybe<MatcheePair> {
-    return (m : MatcheePair)=>{
-        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 inputTxt=
+`import ast;
+---
+122`;
+
+
+const PROG = p.rule<TokenKind, tkTree>();
+const UNIT = p.rule<TokenKind, tkTree>();
+const IMPORTS = p.rule<TokenKind, tkTree>();
+const SEMICOLON = p.rule<TokenKind, tkTree>();
+
+
+let doubleMinus = { type: 'Punctuator', value: '--' };
+let doubleMinus2 = p.str('--');
+const TERM = p.rule<TokenKind, tkTree>();
+
+function applyUnit(value: Token<TokenKind.Number>): tkTree{
+    return value.text;
+}
+
+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 applyImports(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`.
+ * PROG : IMPORTS '---' UNIT;
  */
-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);
-    }
-}
+PROG.setPattern(
+    p.lrec_sc(IMPORTS, p.seq(p.str('---'), UNIT), applyParts)
+
+)
 
 /**
- *  @description thendo(input, f, ...) like
- * a ==> f
+ * PROG : 'import' Id* SEMICOLON;
  */
-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);
-    }
-}
+IMPORTS.setPattern(
+    p.apply(p.seq(p.str('import'), p.rep_sc(p.tok(TokenKind.Id)), SEMICOLON) , applyImports)
+);
+
+/**
+ * SEMICOLON : ';';
+ */
+SEMICOLON.setPattern(
+    p.apply(p.tok(TokenKind.Semicolon), applySemiColon)
+);
+
+/**
+ * UNIT : Number;
+ */
+UNIT.setPattern(
+    p.apply(p.tok(TokenKind.Number), applyUnit)
+);
+
+let tree = p.expectSingleResult(p.expectEOF(PROG.parse(lexer.parse(inputTxt))));
+
+
+
+console.log("RESULT="+tkTreeToSExp(tree));