jhillacre / highfleet-decoder

Command line interface to decode HighFleet messages, captured via ocr, keep frequency of words in clear text and based of that suggest codes for cipher text.

highfleet-decoder

A command line interface for dealing with HighFleet radio intercepts.

• captures clear text or cipher text via OCR.
• track the frequency of words in clear text.
• generate cipher code differences by matching frequent word differences vs cipher words.
• helps translate a code difference to a code.

---

• highfleet-decoder
  • Primer
  • Install
  • Usage
  • Tips
  • Todo
  • License

---

Primer

What are these messages, and how do they work? Check out this guide on YouTube:

Install

1. download the source or clone the repo
2. For Windows, install tesseract, https://tesseract-ocr.github.io/tessdoc/Installation.html and add C:\Program Files\Tesseract-OCR to your PATH in your environment variables.
3. install python, the pipenv is set up for Python 3.11
4. install pipenv, $ pip install pipenv
5. use pipenv sync to create an environment and install dependencies.

Usage

$ pipenv run python highfleet-decoder.py

• Follow onscreen prompts to
• capture message
• fix OCR
• confirm if the message is clear or cipher text (it knows)
• review code suggestions.

Tips

The project is a rough prototype.

• Valid word characters are strings.ascii_uppercase + strings.digits + "=-". I haven't dealt with dashed locations yet, probably will be treated as multiple words.
• HighFleet messages are in all caps. When correcting OCR text, your input will be transformed into uppercase.
• When correcting senders or receivers, the equals sign must be in the right place to be detected. Receivers and senders are kept separate from body words.
• Senders and receivers must be last and first. If there is OCR text after, you should blank it or swap positions with other text.
• 1 and I look the same in HighFleet's font. The project tries to deal with this by determining if the rest of the word is primarily numbers or letters.
• Are the suggestions actually solutions?
  • Depends on the frequency list being populated somewhat.
  • Assuming that unique words have unique difference patterns

Todo

• Finding edge cases by running campaigns.
• Ability to have different frequency sets for starting clean with a new campaign, if desired.
• Add unit tests.
• Add CI, preferably in CircleCI, since I've used it.
• Refactor code into not one big file.
• Screenshots.
• Colour output.
• Arguments and argument parsing?
• pypi package?

License

BSD-3-Clause license

words_alpha.txt used under "Unlicense license", dwyl/english-words