self-unzip.html

This repo contains tools to create self-extracting HTML pages. It works by taking a payload, compressing it, and encoding the results using ASCII85. It then puts the resulting string in a template file, that contains the code to decode and decompress the payload again.

Currently, there are multiple actions implemented, that can be executed, after the payload is decoded:

• --download-link: Download the payload as a file when clicking a button (example use case: bypass antivirus / filters)
• --download-auto: Automatically download the payload as a file when the page is opened (example use case: bypass antivirus / filters)
• --driveby-redirect <REDIRECT_URL>: Perform a drive-by download and redirect to the REDIRECT_URL (example use case: phishing)
• --eval: Execute payload as JavaScript code (example use case: obfuscate malicious JS code)
• --replace: Show payload as HTML page (example use case: compress a big web page)
• --show-text: Show payload as text (example use case: compress a big text file)
• --copy-text: Copy payload as text to the clipboard (example use case: download PowerShell script in remote browsing environment)
• --copy-base64: Copy payload as base64 encoded text and show commands that can be used to decode it (example use case: download executable file in remote browsing environment)
• --custom <YOUR_JAVASCRIPT_CODE>: This allows you to write your own action as JavaScript code

There are both a web and a python version. The web version is not actively developed anymore and very bare bones, but easy to use and requires no installation. The Python version has more features and is generally recommended.

Feature comparison

Feature	Web version	Python version
Base64 encoding	no	yes
Ascii85 encoding	yes	yes
GZIP compression	yes, always	yes, can be disabled
AES-GCM encryption	no	yes
Automatic detection of most efficient algorithms	no	yes
Payload actions	download-link, eval, replace	all

Python version

A Python script to generate self extracting web pages is under python/main.py. It just requires a modern Python version (probably Python3.9+) and has no mandatory external dependencies. But if you want to use the encryption feature, you need to install pycryptodomex with pip.

Installation

You can install it with pip or docker.

pip

To install the latest release with encryption support:

python3 -m pip install -U self-unzip-html[all]

To install the latest release without encryption support:

python3 -m pip install -U self-unzip-html

If you want to have the bleeding edge version (main branch), you can clone the repository and run:

python3 -m pip install .

Example usage of the pip package:

self-unzip-html html --download-auto -o psexec.html -i PsExec.exe

Docker

You can use the image pushed to ghcr.io:

docker run --rm -v "$PWD:/share" ghcr.io/six-two/self-unzip-html html --download-auto -o psexec.html -i ./PsExec.exe 

To use the bleeding edge version (main branch), you can build the Dockerfile:

docker build -t self-unzip-html .

Usage of docker image:

docker run --rm -v "$PWD:/share" self-unzip-html html --download-auto -o psexec.html -i ./PsExec.exe 

Usage

Do a basic HTML smuggling, that will show download link for an executable file:

self-unzip-html html --download-auto -o psexec.html -i PsExec.exe

Or if you wanted to password-protect the output:

self-unzip-html encrypted-html --download-auto -o psexec_encrypted.html -p YourPasswordHere -i PsExec.exe

Instead of HTML pages, you can also embed an HTML smuggling payload in an SVG file:

self-unzip-html svg -i PsExec.exe --download-auto -o psexec.svg

Please note that not all payload actions are available for SVGs due to technical limitations. Use self-unzip-html svg --help to list supported actions.

You can show all flags with the --help flag and can see some example commands in ./test.sh.

Phishing

If you want to use it for phishing (sending a download link to a malicious file), the --driveby-redirect is likely what you want. It starts a download and immediately redirects to any URL you choose, so that it looks like the other site started the download:

Just search for a "thank you for downloading" page that does not start a download. There are many of them for software like Skype, AnyDesk, etc. Then rename your payload file to something a visitor of the download page would expect and create the HTML smuggling page:

self-unzip-html html -i AnyDesk.exe -o anydesk-download.html --driveby-redirect https://anydesk.com/en/downloads/guide/thank-you --obscure-action

Host that HTML file on your server and link to it in your emails.

Encryption

Encryption uses AES-GCM for encryption and tamper detection and PBKDF2 with slightly over 1 million rounds of SHA-256 for password derivation. The code has not been audited, so use it with caution. If data security is very important to you may want to manually encrypt it beforehand (for example using gpg).

You can automatically decrypt a page by adding the password as the hash in a URL like encrypted.html#monkey123!. The hash will not be sent to the server, so your password may only be stored locally (in your browsing history). Otherwise, a prompt will ask you for the password.

Custom actions

With the --custom flag you can specify your own JavaScript that should handle the unpacked data. The data is passed via the og_data parameter. The value is an Uint8Array which represents the bytes of the input file. If you want to interpret it as UTF-8 text and convert it to a string, you can use new TextDecoder().decode(og_data). If you want to build your payload based on one of the built-in payloads, you can find them in python/self_unzip_html/static_js.py

When debugging your own payload it can be useful to see go through it step by step. For this I find it easiest to create a page with:

self-unzip-html html -i README.md --custom 'window.og_data = og_data' -o test_custom.html

The resulting page exposes og_data in the global scope. You can then paste your payload into the JavaScript console piece by piece and inspect the output or DOM in between.

Web version

There is a bare-bones page generator written in plain HTML and JavaScript. To use it, just clone the repo and put the contents of the site directory somewhere in your web server directory.

You can of course also use it with Python's built-in web server:

python3 -m http.server --directory ./site/

I also host it via Vercel at https://self-extracting-html.six-two.dev/.

JavaScript code

Python

The JavaScript code is modular to decrease size and allow mixing and matching different algorithms. The unminified files are in code_to_minify/. They are minified by running code_to_minify/build.sh, which updates python/self_unzip_html/minified_js.py.

template.html

This basically just explains, how I generated the obfuscated script in template.html (JavaScript version). In case you are paranoid, you can reproduce the steps. Or if there is an important update to fflate or ascii85, I will have to run them again.

Install with npm:

npm install .

First use rollup to only select the actually used code:

./node_modules/.bin/rollup -c rollup.config.js

This command should create output/main.js

The next step is optional. If you want to skip it, just rename main.min.js to main.js in the output directory. Otherwise, minify the code (may require you to install an external minifier like closure-compiler or https://jscompressor.treblereel.dev/).

closure-compiler output/main.js --js_output_file output/main.min.js

License

The python/self_unzip_html/template.svg image was generated using the ChatGPT prompt make me a really cool looking SVG for a hacking tool. The cooler the better with me just replacing some text. So to my understanding it is not copyrightable and you can do with it what you want :)

The rest of the project is under the MIT license, so you can do whatever as long as you credit my work to me.

Notable changes

HEAD

• Added options for specifying output file names and paths for --copy-base64 and --copy-text. Example usage: --linux-path /tmp/ --mac-path /tmp/ --windows-path C:\\\\Temp\\\\
• Removed textarea from --copy-base64 due to page crashes / reloads with big payloads. To get the textarea instead of the copy button, use the new --show-base64 flag.

Version 0.3.0

• Breaking change: Removed self-unzip-html.py script, use self-unzip-html instead.
• Breaking change: Implemented different subcommands (html, encrypted-html, svg, encrypted-svg) that only have the arguments that they support.
• Breaking change: Renamed --download to --download-link and added --download-auto which automatically starts the download
• Added --copy-text and --copy-base64 options for smuggling files when downloads are not possible (for example in remote browsing setups)
• Added serve subcommand that serves the current directory and allows downloading files directly or via HTML smuggling. Each file in a directory listing has an entry like .DS_Store (HTML, SVG).
• Refactoring of the python code to make it usable as a library
• Allow specifying multiple values for --compression and --encoding like --encoding hex,base64,ascii85
• Added --iterations option the PBKDF2, enabling a custom speed / security tradeoff
• Added --cache-password option
• Added option to supply decryption password via localStorage: localStorage.setItem("self_unzip_pw", "YOUR_PASSWORD_HERE")
• Added --encoding hex option
• Added support for SVG output files

Version 0.2.1

• Added Dockerfile and published image. Download with docker pull ghcr.io/six-two/self-unzip-html
• Added --show-text action, which shows the payload as plain text

Version 0.2.0

• Breaking change: Replaced the --type parameter with specific flags like --download, --driveby-redirect, --eval, --replace, and --custom.
  • --download now has an optional parameter that allows you to change the file's name.
  • --driveby-redirect allows you to perform a driveby download and redirect the visitor immediately to another page.
  • --custom allows you to provide your own JavaScript code, which is useful for not implemented techniques or special usecases.
• Password protection now encrypts the payload action too. This should prevent anyone without the password from inspecting the file and for example seeing that it performs a drive by download and redirects to a specific site.
• The debugging console messages are disabled by default, which results in nicer and shorter code. If enabled with the --console-log flag, the data will be formatted as hex dumps for easier debugging.
• Added --html, --html-file and --title options to change the visual appearence of the initial page without needing a whole custom template.