X-Git-Url: https://git.kianting.info/?a=blobdiff_plain;f=src%2Findex.ts;h=883d94b742145129ff948e0325d94457b274b8f1;hb=5fbb467bdf501dee03c78817242581d962547c4f;hp=26cd10df0e081ec2ce445b72c26aeef17bdde0bb;hpb=4cfa9c7cb47795674b6f2c2f2d0f63fe24100c49;p=clo

diff --git a/src/index.ts b/src/index.ts
index 26cd10d..883d94b 100644
--- a/src/index.ts
+++ b/src/index.ts
@@ -1,8 +1,17 @@
 var fs = require('fs');
 
-type Some<T> = { _tag: "Some"; value: T };
-type None = {_tag: "None"};
+export type Some<T> = { _tag: "Some"; value: T };
+export type None = {_tag: "None"};
+
 
+/**
+ * wrap a x in a `Some(T)`
+ * @param x : variable to be wrapped.
+ * @returns wrapped `x`.
+ */
+function toSome<T>(x: T): Some<T>{
+    return  { _tag: "Some", value: x};  
+}
 /**
  * @description Like the `Some(a)` and `None` in Rust.
  *
@@ -18,10 +27,55 @@ export type Maybe<T> = Some<T> | None;
 /**
  * @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.
+ * @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 
+}
+
+/**
+ * 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
  */
-export type MatcheePair = {matched : string; remained : string};
+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,
+}
 
 /**
  * @description
@@ -33,6 +87,9 @@ export type MatcheePair = {matched : string; remained : string};
  */
 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 :{
@@ -45,6 +102,22 @@ export function match1Char(c : string) : (m: MatcheePair) => Maybe<MatcheePair>
     }
 };
 
+/**
+ * 
+ * @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`.
+ */
+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"};
+    }
+}
+
 /**
  * @description
  * it returns a function which test if the first char of the `remained` part of
@@ -61,7 +134,9 @@ export function matchRange(l : string, u : string) : (m: MatcheePair) => Maybe<M
         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){
@@ -92,6 +167,10 @@ export function charToCodepoint(s : string): number{
 /**
  *  @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`.
  */
 export function thenDo<T>(input : Maybe<T>, f : Function) : Maybe<T>{
     if (input._tag == "None"){
@@ -102,3 +181,182 @@ export function thenDo<T>(input : Maybe<T>, f : Function) : Maybe<T>{
         return f(inner);
     }
 }
+
+/**
+ * @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
+ */
+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;
+            }
+        }
+    };
+}
+
+
+/**
+* @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;
+    };
+}
+
+/**
+* @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);
+        }
+    };
+}
+
+/**
+ * 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)
+ */
+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);
+
+        if (wrapped_new_x._tag != "None"){
+            return wrapped_new_x;
+        }else{
+        return wrapped_old_x;
+        }
+    };
+}
+
+
+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));
+        }
+
+        return token_list;
+    }
+
+    console.log(term([], input_matchee_pair));
+
+    // TODO: id, string, space, basic operator, 3 marks: @, {, }.
+
+}
+
+