Translate javascript modules - node.js

I was trying to run some NodeJS code using QickJS and NectarJS, and I stumble in an old question I have, how to import modules.
In special, qjs -m <source-js> loads ES6 modules, and expect import export statements,, instead of require, and writing module.exports, or exports.
Also I use an builtin std module (that cannot be resolved by node_module).
For the first example I changed the code manually, but I would like to have a script to do the transformations. I expected this to be doable with webpack, but I can't find how.
Example
main.js
const h = require('./hello.js');
h.sayHello();
hello.js
const h = require('./print.js');
function sayHello(){
h.print("Hello");
}
module.exports = {sayHello};
print.js
module.exports = {
print(s) {
console.log(s);
}
}
I can run this with the command node main.js, but qjs -m main.js, will fail with 'require' is not defined

Related

What is the difference between module.exports=require('other') and the same with a temporary variable?

reproduce link https://stackblitz.com/edit/node-sp5xay?file=index.mjs
Assume we have a project like this:
.
├── dep
│ ├── a.mjs
│ ├── b.js
│ └── c.js
└── entry.mjs
// entry.mjs
import { foo } from "./dep/a.mjs";
console.log(foo);
// dep/a.mjs
export * from './b.js'
// dep/b.js
module.exports = require("./c.js"); // 💯
// why this not working ❌
// const m = require("./c.js");
// module.exports = m;
// dep/c.js
exports.foo = "foo";
and we run in terminal
node entry.mjs
it's very confuse it will throw error if in dep/b.js we use:
// why this not working ❌
const m = require("./c.js");
module.exports = m;
if in dep/b.js we use:
module.exports = require("./c.js");
it will work expect!
module.exports=require have something magic? like symlink? if any docs i miss?
The origin of this problem is that I saw the vue3 source code
vue3 core source code export
module.exports=require have something magic?
Yes, it does have some magic. The problem is that you are importing a CommonJS module into an ES module. The latter require a static declaration of exported names, which the former do not provide. See the node.js documentation:
When importing CommonJS modules, the module.exports object is provided as the default export.
So just don't do export * from './b.js', rather import b from './b.js' then refer to b.foo on the CommonJS module object.
However,
For better compatibility with existing usage in the JS ecosystem, Node.js in addition attempts to determine the CommonJS named exports of every imported CommonJS module to provide them as separate ES module exports using a static analysis process.
[…]
The detection of named exports is based on common syntax patterns but does not always correctly detect named exports. In these cases, using the default import form described above can be a better option.
Named exports detection covers many common export patterns, reexport patterns and build tool and transpiler outputs. See cjs-module-lexer for the exact semantics implemented.
(emphasis mine)
And indeed,
module.exports = require("./c.js");
is one of the "reexport patterns" that is detected, but using a temporary variable
const m = require("./c.js");
module.exports = m;
is not. You just can't use named imports from a CommonJS module that does this. The proper solution to fix this is of course to rewrite the module to ESM syntax and use export * from "./c.js";, not any module.exports assignments.
In Node.js, the module.exports object is used to specify what a module should make available when it is imported using require.
The first statement , module.exports = require('module'), sets the module.exports object to the value of the module object that is imported using require. This means that anything that is exported by the module module will be made available to the code that imports it.
On the other hand, module.exports = {...} sets the module.exports object to an object literal containing the properties and values that should be made available to code that imports the module. This allows you to specify exactly what should be made available from the module, rather than relying on the exports of another module.
For example, suppose you have a module called myModule that exports a single function:
module.exports = {
myFunction: () => {
console.log('Hello, world!');
}
};
This module can then be imported and used like this:
const myModule = require('myModule');
myModule.myFunction(); // prints 'Hello, world!'
On the other hand, if you set module.exports to the value of another module's exports, like this:
module.exports = require('anotherModule');
Then the exports of anotherModule will be made available to code that imports myModule. For example:
const myModule = require('myModule');
myModule.someFunction(); // calls a function exported by anotherModule

How to deal with node modules in the browser?

My title is a bit vague, here is what I'm trying to do:
I have a typescript npm package
I want it to be useable on both node and browser.
I'm building it using a simple tsc command (no bundling), in order to get proper typings
My module has 1 entry point, an index.ts file, which exposes (re-exports) everything.
Some functions in this module are meant to be used on node-only, so there are node imports in some files, like:
import { fileURLToPath } from 'url'
import { readFile } from 'fs/promises'
import { resolve } from 'path'
// ...
I would like to find a way to:
Not trip-up bundlers with this
Not force users of this package to add "hacks" to their bundler config, like mentioned here: Node cannot find module "fs" when using webpack
Throw sensible Errors in case they are trying to use node-only features
Use proper typings inside my module, utilizing #types/node in my code
My main problem is, that no matter what, I have to import or require the node-only modules, which breaks requirement 1 (trips up bundlers, or forces the user to add some polyfill).
The only way I found that's working, is what isomorphic packages use, which is to have 2 different entry points, and mark it in my package.json like so:
{
// The entry point for node modules
"main": "lib/index.node.js",
// The entry point for bundlers
"browser": "lib/index.browser.js",
// Common typings
"typings": "lib/index.browser.d.ts"
}
This is however very impractical, and forces me to do a lots of repetition, as I don't have 2 different versions of the package, just some code that should throw in the browser when used.
Is there a way to make something like this work?
// create safe-fs.ts locally and use it instead of the real "fs" module
import * as fs from 'fs'
function createModuleProxy(moduleName: string): any {
return new Proxy(
{},
{
get(target, property) {
return () => {
throw new Error(`Function "${String(property)}" from module "${moduleName}" should only be used on node.js`)
}
},
},
)
}
const isNode = typeof window === undefined && typeof process === 'object'
const safeFs: typeof fs = isNode ? fs : createModuleProxy('fs')
export default safeFs
As it stands, this trips up bundlers, as I'm still importing fs.

Transforming UMD modules to ES modules in RollupJS ("The requested module X does not provide an export named 'default'")

I am currently having a TypeScript project that includes Ethers.js, which in turn includes bn.js.
The problem is
SyntaxError: The requested module './../../../bn.js/lib/bn.js' does not provide an export named 'default'
It appears to me this is because BN is in UMD format (see at https://github.com/indutny/bn.js/blob/master/lib/bn.js#L1)
(function (module, exports) {
'use strict';
// Utils
function assert (val, msg) {
if (!val) throw new Error(msg || 'Assertion failed');
}
and the correponding .ts declaration is
"use strict";
/**
* BigNumber
*
* A wrapper around the BN.js object. We use the BN.js library
* because it is used by elliptic, so it is required regardless.
*
*/
import _BN from "bn.js";
import BN = _BN.BN;
import { Bytes, Hexable, hexlify, isBytes, isHexString } from "#ethersproject/bytes";
import { Logger } from "#ethersproject/logger";
import { version } from "./_version";
const logger = new Logger(version);
It may be there's something that could be done at importing (source) or in Rollup. Difficult to tell!
Here is a screenshot of the build errors (one variation, depending on if building or running directly)
Question: Is there a way to transform this format to an ESM format in application Rollup pipeline?
I have tried using #rollup/plugin-commonjs and #rollup/plugin-node-resolve as in
resolve({ browser: true, preferBuiltins: false }), commonjs()]
(or see the project as whole at https://github.com/veikkoeeva/erc1155sample/blob/main/web/rollup.config.js, the error shows with npm run test or npm run start (in console log)).
Thus far I've had no luck cracking this, though. Hence coming here wondering if there's a dumb issue I don't see or if this is a genuinely tougher issue.
Edit: indeed, following https://rollupjs.org/guide/en/#error-name-is-not-exported-by-module and maybe named exports is the key here...

How to use module.exports of Nodejs [duplicate]

What is the purpose of Node.js module.exports and how do you use it?
I can't seem to find any information on this, but it appears to be a rather important part of Node.js as I often see it in source code.
According to the Node.js documentation:
module
A reference to the current
module. In particular module.exports
is the same as the exports object. See
src/node.js for more information.
But this doesn't really help.
What exactly does module.exports do, and what would a simple example be?
module.exports is the object that's actually returned as the result of a require call.
The exports variable is initially set to that same object (i.e. it's a shorthand "alias"), so in the module code you would usually write something like this:
let myFunc1 = function() { ... };
let myFunc2 = function() { ... };
exports.myFunc1 = myFunc1;
exports.myFunc2 = myFunc2;
to export (or "expose") the internally scoped functions myFunc1 and myFunc2.
And in the calling code you would use:
const m = require('./mymodule');
m.myFunc1();
where the last line shows how the result of require is (usually) just a plain object whose properties may be accessed.
NB: if you overwrite exports then it will no longer refer to module.exports. So if you wish to assign a new object (or a function reference) to exports then you should also assign that new object to module.exports
It's worth noting that the name added to the exports object does not have to be the same as the module's internally scoped name for the value that you're adding, so you could have:
let myVeryLongInternalName = function() { ... };
exports.shortName = myVeryLongInternalName;
// add other objects, functions, as required
followed by:
const m = require('./mymodule');
m.shortName(); // invokes module.myVeryLongInternalName
This has already been answered but I wanted to add some clarification...
You can use both exports and module.exports to import code into your application like this:
var mycode = require('./path/to/mycode');
The basic use case you'll see (e.g. in ExpressJS example code) is that you set properties on the exports object in a .js file that you then import using require()
So in a simple counting example, you could have:
(counter.js):
var count = 1;
exports.increment = function() {
count++;
};
exports.getCount = function() {
return count;
};
... then in your application (web.js, or really any other .js file):
var counting = require('./counter.js');
console.log(counting.getCount()); // 1
counting.increment();
console.log(counting.getCount()); // 2
In simple terms, you can think of required files as functions that return a single object, and you can add properties (strings, numbers, arrays, functions, anything) to the object that's returned by setting them on exports.
Sometimes you'll want the object returned from a require() call to be a function you can call, rather than just an object with properties. In that case you need to also set module.exports, like this:
(sayhello.js):
module.exports = exports = function() {
console.log("Hello World!");
};
(app.js):
var sayHello = require('./sayhello.js');
sayHello(); // "Hello World!"
The difference between exports and module.exports is explained better in this answer here.
Note that the NodeJS module mechanism is based on CommonJS modules which are supported in many other implementations like RequireJS, but also SproutCore, CouchDB, Wakanda, OrientDB, ArangoDB, RingoJS, TeaJS, SilkJS, curl.js, or even Adobe Photoshop (via PSLib).
You can find the full list of known implementations here.
Unless your module use node specific features or module, I highly encourage you then using exports instead of module.exports which is not part of the CommonJS standard, and then mostly not supported by other implementations.
Another NodeJS specific feature is when you assign a reference to a new object to exports instead of just adding properties and methods to it like in the last example provided by Jed Watson in this thread. I would personally discourage this practice as this breaks the circular reference support of the CommonJS modules mechanism. It is then not supported by all implementations and Jed example should then be written this way (or a similar one) to provide a more universal module:
(sayhello.js):
exports.run = function() {
console.log("Hello World!");
}
(app.js):
var sayHello = require('./sayhello');
sayHello.run(); // "Hello World!"
Or using ES6 features
(sayhello.js):
Object.assign(exports, {
// Put all your public API here
sayhello() {
console.log("Hello World!");
}
});
(app.js):
const { sayHello } = require('./sayhello');
sayHello(); // "Hello World!"
PS: It looks like Appcelerator also implements CommonJS modules, but without the circular reference support (see: Appcelerator and CommonJS modules (caching and circular references))
Some few things you must take care if you assign a reference to a new object to exports and /or modules.exports:
1. All properties/methods previously attached to the original exports or module.exports are of course lost because the exported object will now reference another new one
This one is obvious, but if you add an exported method at the beginning of an existing module, be sure the native exported object is not referencing another object at the end
exports.method1 = function () {}; // exposed to the original exported object
exports.method2 = function () {}; // exposed to the original exported object
module.exports.method3 = function () {}; // exposed with method1 & method2
var otherAPI = {
// some properties and/or methods
}
exports = otherAPI; // replace the original API (works also with module.exports)
2. In case one of exports or module.exports reference a new value, they don't reference to the same object any more
exports = function AConstructor() {}; // override the original exported object
exports.method2 = function () {}; // exposed to the new exported object
// method added to the original exports object which not exposed any more
module.exports.method3 = function () {};
3. Tricky consequence. If you change the reference to both exports and module.exports, hard to say which API is exposed (it looks like module.exports wins)
// override the original exported object
module.exports = function AConstructor() {};
// try to override the original exported object
// but module.exports will be exposed instead
exports = function AnotherConstructor() {};
the module.exports property or the exports object allows a module to select what should be shared with the application
I have a video on module_export available here
When dividing your program code over multiple files, module.exports is used to publish variables and functions to the consumer of a module. The require() call in your source file is replaced with corresponding module.exports loaded from the module.
Remember when writing modules
Module loads are cached, only initial call evaluates JavaScript.
It's possible to use local variables and functions inside a module, not everything needs to be exported.
The module.exports object is also available as exports shorthand. But when returning a sole function, always use module.exports.
According to: "Modules Part 2 - Writing modules".
the refer link is like this:
exports = module.exports = function(){
//....
}
the properties of exports or module.exports ,such as functions or variables , will be exposed outside
there is something you must pay more attention : don't override exports .
why ?
because exports just the reference of module.exports , you can add the properties onto the exports ,but if you override the exports , the reference link will be broken .
good example :
exports.name = 'william';
exports.getName = function(){
console.log(this.name);
}
bad example :
exports = 'william';
exports = function(){
//...
}
If you just want to exposed only one function or variable , like this:
// test.js
var name = 'william';
module.exports = function(){
console.log(name);
}
// index.js
var test = require('./test');
test();
this module only exposed one function and the property of name is private for the outside .
There are some default or existing modules in node.js when you download and install node.js like http, sys etc.
Since they are already in node.js, when we want to use these modules we basically do like import modules, but why? because they are already present in the node.js. Importing is like taking them from node.js and putting them into your program. And then using them.
Whereas Exports is exactly the opposite, you are creating the module you want, let's say the module addition.js and putting that module into the node.js, you do it by exporting it.
Before I write anything here, remember, module.exports.additionTwo is same as exports.additionTwo
Huh, so that's the reason, we do like
exports.additionTwo = function(x)
{return x+2;};
Be careful with the path
Lets say you have created an addition.js module,
exports.additionTwo = function(x){
return x + 2;
};
When you run this on your NODE.JS command prompt:
node
var run = require('addition.js');
This will error out saying
Error: Cannot find module addition.js
This is because the node.js process is unable the addition.js since we didn't mention the path. So, we have can set the path by using NODE_PATH
set NODE_PATH = path/to/your/additon.js
Now, this should run successfully without any errors!!
One more thing, you can also run the addition.js file by not setting the NODE_PATH, back to your nodejs command prompt:
node
var run = require('./addition.js');
Since we are providing the path here by saying it's in the current directory ./ this should also run successfully.
A module encapsulates related code into a single unit of code. When creating a module, this can be interpreted as moving all related functions into a file.
Suppose there is a file Hello.js which include two functions
sayHelloInEnglish = function() {
return "Hello";
};
sayHelloInSpanish = function() {
return "Hola";
};
We write a function only when utility of the code is more than one call.
Suppose we want to increase utility of the function to a different file say World.js,in this case exporting a file comes into picture which can be obtained by module.exports.
You can just export both the function by the code given below
var anyVariable={
sayHelloInEnglish = function() {
return "Hello";
};
sayHelloInSpanish = function() {
return "Hola";
};
}
module.export=anyVariable;
Now you just need to require the file name into World.js inorder to use those functions
var world= require("./hello.js");
The intent is:
Modular programming is a software design technique that emphasizes
separating the functionality of a program into independent,
interchangeable modules, such that each contains everything necessary
to execute only one aspect of the desired functionality.
Wikipedia
I imagine it becomes difficult to write a large programs without modular / reusable code. In nodejs we can create modular programs utilising module.exports defining what we expose and compose our program with require.
Try this example:
fileLog.js
function log(string) { require('fs').appendFileSync('log.txt',string); }
module.exports = log;
stdoutLog.js
function log(string) { console.log(string); }
module.exports = log;
program.js
const log = require('./stdoutLog.js')
log('hello world!');
execute
$ node program.js
hello world!
Now try swapping ./stdoutLog.js for ./fileLog.js.
What is the purpose of a module system?
It accomplishes the following things:
Keeps our files from bloating to really big sizes. Having files with e.g. 5000 lines of code in it are usually real hard to deal with during development.
Enforces separation of concerns. Having our code split up into multiple files allows us to have appropriate file names for every file. This way we can easily identify what every module does and where to find it (assuming we made a logical directory structure which is still your responsibility).
Having modules makes it easier to find certain parts of code which makes our code more maintainable.
How does it work?
NodejS uses the CommomJS module system which works in the following manner:
If a file wants to export something it has to declare it using module.export syntax
If a file wants to import something it has to declare it using require('file') syntax
Example:
test1.js
const test2 = require('./test2'); // returns the module.exports object of a file
test2.Func1(); // logs func1
test2.Func2(); // logs func2
test2.js
module.exports.Func1 = () => {console.log('func1')};
exports.Func2 = () => {console.log('func2')};
Other useful things to know:
Modules are getting cached. When you are loading the same module in 2 different files the module only has to be loaded once. The second time a require() is called on the same module the is pulled from the cache.
Modules are loaded in synchronous. This behavior is required, if it was asynchronous we couldn't access the object retrieved from require() right away.
ECMAScript modules - 2022
From Node 14.0 ECMAScript modules are no longer experimental and you can use them instead of classic Node's CommonJS modules.
ECMAScript modules are the official standard format to package JavaScript code for reuse. Modules are defined using a variety of import and export statements.
You can define an ES module that exports a function:
// my-fun.mjs
function myFun(num) {
// do something
}
export { myFun };
Then, you can import the exported function from my-fun.mjs:
// app.mjs
import { myFun } from './my-fun.mjs';
myFun();
.mjs is the default extension for Node.js ECMAScript modules.
But you can configure the default modules extension to lookup when resolving modules using the package.json "type" field, or the --input-type flag in the CLI.
Recent versions of Node.js fully supports both ECMAScript and CommonJS modules. Moreover, it provides interoperability between them.
module.exports
ECMAScript and CommonJS modules have many differences but the most relevant difference - to this question - is that there are no more requires, no more exports, no more module.exports
In most cases, the ES module import can be used to load CommonJS modules.
If needed, a require function can be constructed within an ES module using module.createRequire().
ECMAScript modules releases history
Release
Changes
v15.3.0, v14.17.0, v12.22.0
Stabilized modules implementation
v14.13.0, v12.20.0
Support for detection of CommonJS named exports
v14.0.0, v13.14.0, v12.20.0
Remove experimental modules warning
v13.2.0, v12.17.0
Loading ECMAScript modules no longer requires a command-line flag
v12.0.0
Add support for ES modules using .js file extension via package.json "type" field
v8.5.0
Added initial ES modules implementation
You can find all the changelogs in Node.js repository
let test = function() {
return "Hello world"
};
exports.test = test;

Can I use a custom module resolution function (like "proxyquire") in place of require() with TypeScript?

I have a TypeScript file config.ts that will be run with node:
import myDependency = require('my-dependency');
export = {
doSomething = () => {
...
}
}
In other TypeScript file, I can import this file with full type safety:
import config = require('./config');
config.doSomething();
config.doSomethingElse(); // compiler error, this method doesn't exist
Now I want to unit test this script. In order to mock out the dependencies that this script require()s I'm using proxyquire, which lets me provide the values that my script will get when it makes calls to require(). Here's what my test might look like:
import proxyquire = require('proxyquire');
const config = proxyquire('./config', {
'my-dependency': {} // this mocked object will be provided when config.ts asks for `my-dependency`
});
expect(config.doSomething()).to.do.something();
This works fine, except that my config variable is of type any because I'm using proxyquire() in place of require(). TypeScript must give the require() function special treatment to allow it to perform module resolution. Is there a way to tell the TypeScript compiler that proxyquire() should also do module resolution, similar to require()?
I could rewrite config.ts as a class or make it use an interface. Then I would be able to explicitly type the variables in my tests by importing the class/interface definition. But allowing proxyquire() to implicitly type things for me would be far be the easier solution.
There is a workaround - you can get the type of config.ts module by importing actual module and using typeof in a type cast:
import proxyquire = require('proxyquire');
import configType = require('./config');
const config = <typeof configType> proxyquire('./config', {
'my-dependency': {} // this mocked object will be provided when config.ts asks for `my-dependency`
});
config.doSomething();
// config.noSuchMethod(); // does not compile
This is not ideal because you have to import the same module in your test twice - the real one just to get at the type of it and "proxiquired" one to actually use in your tests, and you have to be careful not to mix up the two. But it's pretty simple compared to the task of implementing another variant of module resolution for typescript. Also, when configType is used in this way - for typing only - its import will not even appear in generated javacsript code.

Resources