code转tokens

/*
* code 转tokens
* */
//判断字符
function isEqual(test,code,pos,end) {
if(pos<end&&code.charCodeAt(pos-1)!==92&&test===code.substr(pos,test.length)){
return true;
}
return false;
}
//找到字符
function find(test,code,pos,end){
while (pos<end&&!isEqual(test,code,pos,end)) {pos++}
return pos;
}

function getToken(obj,code) {
const token={};
Object.assign(token,obj)
token.value=code.substr(token.start,token.end-token.start)
return token;
}
//分隔符
const Punctuation=[
'...',
'{|',
'|}',
'::',
'=>',
'${',
'(',
')',
',',
'@',
':',
'[',
'.',
'?',
']',
'{',
'`',
'}',
';'
];
//运算符
const Operators=[
'!==',
'===',
'!=',
'**',
'-=',
'++',
'--',
'||',
'&&',
'>>',
'<<',
'+=',
'==',
'+',
'&',
'>',
'^',
'|',
'<',
'-',
'%',
'*',
'/',
'='
];
//空白
const Emptys=[
' ',
' ',
' ',
]
//注释
const Comments=[
['//',' '],
['/*','*/'],
];
//文本
const notes=[
['"','"'],
["'","'"],
["`","`"],
];
const rockArr=[].concat(Emptys,Punctuation,Operators)
rockArr.sort(function (p1,p2) {
return p2.length>p1.length;
})
//获取'无子'元素
function getRock(code,start,end,preChild) {
if(start>=end){return }
//碰撞空白
while (code[start]===' '||code[start]===' '||code[start]===' '||code[start]===' '){
start++;
}
let pos=start;
//数字和变量
if( (/[_$a-z0-9]/i.test(code[pos])||/.d/.test(code.substr(pos,2)))&&code.charCodeAt(pos-1)!==92 ){
pos++;
while (pos<end&&(/[_$a-z0-9]/i.test(code[pos])||/.d/.test(code.substr(pos,2)))){
pos++;
}
return getToken({start:start,end:pos,type:'variable'},code)
}
//注释
for(let i=0;i<Comments.length;i++){
const arr=Comments[i];
if(isEqual(arr[0],code,pos,end)){
pos=find(arr[1],code,pos+arr[0].length,end)+arr[1].length;
break;
}
}
if(pos>start){
return getToken({start:start,end:pos,type:'comment'},code)
}
if(!preChild||(preChild.type==='symbol'&&code.substring(preChild.start,preChild.end)!==')')){
//文本
for(let i=0;i<notes.length;i++){
const arr=notes[i];
if(isEqual(arr[0],code,pos,end)){
pos=find(arr[1],code,pos+arr[0].length,end)+arr[1].length;
break;
}
}

if(pos>start&&pos<end){
return getToken({start:start,end:pos,type:'string'},code)
}


}

//symbol
for(let i=0;i<rockArr.length;i++){
if(isEqual(rockArr[i],code,pos,end)){
pos=pos+rockArr[i].length;
break;
}
}
if(pos>start){
return getToken({start:start,end:pos,type:'symbol'},code)
}
throw code[pos];
}
//获取子元素
function getChild(code,pos,end,preChild) {
const cTok=getRock(code,pos,end,preChild);
//template
if(isEqual('`',code,pos,end)){
const templateTree={start:pos,end:0,type:'Template',child:[]}
let isEnd=false;
let pos3=pos+1;

while (pos3<end&&!isEnd) {
if(isEqual('${',code,pos3,end)){
const start=pos3;
pos3=find('}',code,pos3+2,end);
let token={
start:start,
end:pos3+1,
type:'TemplateExpression',
child:getAllChild(code,start+2,pos3)
};
templateTree.child.push(token)
}else if(isEqual("`",code,pos3,end)){
isEnd=true;
}
pos3++
}

templateTree.end=pos3;
if(templateTree.child.length>0){
templateTree.quais=[]
for(let i=0;i<templateTree.child.length;i++){
const tok=templateTree.child[i];
if(i===0){
templateTree.quais.push({start:templateTree.start+1,end:tok.start})
}else{
templateTree.quais.push({start:templateTree.child[i-1].end,end:tok.start})
}
if(i===templateTree.child.length-1){
templateTree.quais.push({start:tok.end,end:templateTree.end-1})
}

}
}
return templateTree
}

if(!preChild||(preChild.type==='symbol'&&code.substring(preChild.start,preChild.end)!==')')){
//jsx
if(cTok.value==='<'){
const tagName=getRock(code,cTok.end,end).value;
if(/^[a-z][a-z0-9]*$/.test(tagName)){
let isEnd=false;
let isCloseI=1;
let pos3=pos+1;
while (pos3<end&&!isEnd) {
if(isEqual('"',code,pos3,end)){
pos3=find('"',code,pos3+1,end)
}
if(isEqual("'",code,pos3,end)){
pos3=find("'",code,pos3+1,end)
}
if(isEqual("{",code,pos3,end)){
pos3=find("}",code,pos3+1,end)
}
if(isEqual(">",code,pos3,end)){
if(isCloseI===0){
isEnd=true
}
}else if(isEqual("<",code,pos3,end)){
const tok=getRock(code,pos3+1,end);
if(code.substring(tok.start,tok.end)==='/'){
isCloseI--;
pos3=tok.end-1;
}else{
isCloseI++;

}
}else if(isEqual("/",code,pos3,end)){
const tok=getRock(code,pos3+1,end);
if(code[tok.start]==='>'){
isCloseI--;
pos3=tok.end-1;
if(isCloseI===0){
isEnd=true
}
}
}
pos3++
}
if(code[pos3-1]==='>'){
return {start:pos,end:pos3,type:'JSXElement'}
}
}

}


//正则
if(isEqual('/',code,pos,end)){
let pos3=pos+2;
while (pos3<end&&!isEqual('/',code,pos3,end)) {
if(isEqual('[',code,pos3,end)){
pos3=find(']',code,pos3+1,end)
}
pos3++
}
if(pos3<end){
pos3++;
while (pos3<end&&/[igmsUxaDe]/.test(code[pos3])) {pos3++;}
return getToken({start:pos,end:pos3},code)
}
}
}

return cTok

}
function transform(token,code) {
if(token.type==='JSXElement'){
return getTagChild(code,token.start,token.end)
}
return token;
}
//获取所有子元素
function getAllChild(code,start,end) {
const childs=[]
let token=getChild(code,start,end)
while (token){
childs.push(transform(token,code))
token=getChild(code,token.end,end,token)
}
return childs;
}
function CodeToTokens(code){
const token={
start:0,
end:code.length,
child:getAllChild(code,0,code.length)
};
return token;
}
module.exports=CodeToTokens;

//解析jsx
function getTagChild(code,start,end) {
if(start>=end){return }
let pos=start;
//碰撞空白
while (!(isEqual('<',code,pos,end)&&!isEqual('<',code,pos-1,end))&&pos<end){
pos++;
}
if(pos>start){
return {start:start,end:pos}
}
//jsx
const json={
start:pos,
end:0,
openingElement:null,
closingElement:null,
}
let child=null;
let isEnd=false;
let isCloseI=1;
let pos3=pos+1;
while (pos3<end&&!isEnd) {
if(isEqual('"',code,pos3,end)){
pos3=find('"',code,pos3+1,end)
}
if(isEqual("'",code,pos3,end)){
pos3=find("'",code,pos3+1,end)
}
if(isEqual("{",code,pos3,end)){
pos3=find("}",code,pos3+1,end)
}
if(isEqual(">",code,pos3,end)){
if(isCloseI===0){
isEnd=true
json.closingElement={
type:'JSXClosingElement',
start:child.end,
end:pos3+1,
}
}
if(isCloseI===1){
if(!json.openingElement){
json.openingElement={
type:'JSXOpeningElement',
start:pos,
end:pos3+1,
}
}

}
}else if(isEqual("<",code,pos3,end)){
const tok=getRock(code,pos3+1,end);
if(code.substring(tok.start,tok.end)==='/'){
isCloseI--;
if(isCloseI===0){
console.log(pos3,json.openingElement.end)
if(pos3>json.openingElement.end){
child={
start:json.openingElement.end,
end:pos3,
}
}

}
pos3=tok.start;
}else{
isCloseI++;

}
}else if(isEqual("/",code,pos3,end)){
const tok=getRock(code,pos3+1,end);

if(code[tok.start]==='>'){
isCloseI--;
pos3=tok.end-1;
if(isCloseI===0){
isEnd=true
json.openingElement={
type:'JSXOpeningElement',
start:pos,
end:pos3+1,
}
}
}
}
pos3++
}
if(code[pos3-1]==='>'){
json.end=pos3;
if(child){
json.child=getAllTagChild(code,child.start,child.end)
}

return getToken(json,code)
}
}
//获取所有子元素
function getAllTagChild(code,start,end) {
const childs=[]
let token=getTagChild(code,start,end)

while (token){
childs.push(token)
token=getTagChild(code,token.end,end,token)
}
return childs;
}

//test
const fs=require('fs');
const code=fs.readFileSync('./code.js').toString()
const tree=CodeToTokens(code)
console.log(tree)
原文地址:https://www.cnblogs.com/caoke/p/11159643.html