RangeError: Maximum call stack size exceeded

[hatRiot]

Seems related to issue #86 , but not on Android.

> npm view javascript-obfuscator version
0.11.0
> node --version
v8.4.0

The error seems to manifest itself with long strings:

     BinaryExpression: function (expr, precedence, flags) {
                                   ^
RangeError: Maximum call stack size exceeded
    at CodeGenerator.BinaryExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:1837:36)
    at CodeGenerator.generateExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:2648:28)
    at CodeGenerator.BinaryExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:1845:29)
    at CodeGenerator.generateExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:2648:28)
    at CodeGenerator.BinaryExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:1845:29)
    at CodeGenerator.generateExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:2648:28)
    at CodeGenerator.BinaryExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:1845:29)
    at CodeGenerator.generateExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:2648:28)
    at CodeGenerator.BinaryExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:1845:29)
    at CodeGenerator.generateExpression (/node_modules/javascript-obfuscator/node_modules/escodegen-wallaby/escodegen.js:2648:28)

Oddly, this same code works on Node v8.0, but not Node v8.4, so perhaps it isn't related to javascript-obfuscator directly?

Config:

var config = {
    compact: true,
    controlFlowFlattening: false,
    controlFlowFlatteningThreshold: 0.75,
    deadCodeInjection: true,
    deadCodeInjectionThreshold: 0.4,
    debugProtection: false,
    debugProtectionInterval: false,
    disableConsoleOutput: true,
    mangle: false,
    reservedNames: [],
    rotateStringArray: true,
    seed: 0,
    selfDefending: false,
    sourceMap: false,
    sourceMapBaseUrl: '',
    sourceMapFileName: '',
    sourceMapMode: 'separate',
    stringArray: true,
    stringArrayEncoding: true,
    stringArrayThreshold: 0.75,
    unicodeEscapeSequence: false
}

You can reproduce this by generating a script with a large, multiline string:

> cat generate.py 
data = "A"*200000
data = [data[i:i+75] for i in range(0, len(data), 75)]
chunked = '+\n'.join(["\"{0}\"".format(x) for x in data])
with open('output.js', 'w+') as f:
    f.write("var test = {0};".format(chunked))

And then running output.js through javascript-obfuscator.

[sanex3339]

Hi! Can you provide input source code that will passed to obfuscator (generated by script above)?

[hatRiot]

The output of the script is just a long, multiline string of the form:

var test = "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"+
"AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"+
"AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"+
"AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"+
"AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"+
"AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"+
..............

It would be too long to reproduce in its entirety here. This script is fed into a small script that reads it and invokes Javascript-Obfuscator:

var jsob = require('javascript-obfuscator');
var fs = require('fs');

var config = {
    compact: true,
    controlFlowFlattening: false,
    controlFlowFlatteningThreshold: 0.75,
    deadCodeInjection: true,
    deadCodeInjectionThreshold: 0.4,
    debugProtection: false,
    debugProtectionInterval: false,
    disableConsoleOutput: true,
    mangle: false,
    reservedNames: [],
    rotateStringArray: true,
    seed: 0,
    selfDefending: false,
    sourceMap: false,
    sourceMapBaseUrl: '',
    sourceMapFileName: '',
    sourceMapMode: 'separate',
    stringArray: true,
    stringArrayEncoding: true,
    stringArrayThreshold: 0.75,
    unicodeEscapeSequence: false
}

if(process.argv.length <= 1){
    process.exit();
}

var unobf_data = fs.readFileSync(process.argv[2], 'utf8');
var obf = jsob.obfuscate(unobf_data, config);
fs.writeFileSync(process.argv[2], obf);

[sanex3339]

How many lines i should create to reproduce?

[hatRiot]

My script generates 200k characters and breaks them into 75 character lines, so about 2667 lines should trigger the issue.

[hatRiot]

FYI it looks like this may be a bug in escodegen. There's an open issue from 2016 here that looks to be the exact same thing.

Feel free to close this if you'd like.

[sanex3339]

Yep, look's like it related to escodegen.

[D00Movenok]

Hi! Using node v20.3.1 and javascript-obfuscator 4.0.2, got the same issue.
As I see, javascript-obfuscator now uses personal fork of escodegen. Will It be fixed in the future?