A compressed dictionary application for the TI-Nspire calculator using Ndless. This project provides a complete offline dictionary with efficient storage using Huffman compression. (Also, this was partially vibe-coded)

• Python 3.7+ with pip
• Wiktextract data file (wiktextract.jsonl)

• Ndless SDK (nspire-gcc toolchain)
• make utility
• Linux/WSL environment (recommended)

cd process_words
pip install -r requirements.txt

Download the English wiktextract JSONL file from Kaikki.org:

# Example: Download English wiktextract data
wget https://kaikki.org/dictionary/English/kaikki.org-dictionary-English.jsonl
mv kaikki.org-dictionary-English.jsonl process_words/wiktextract.jsonl

export WIKTEXTRACT_PATH="process_words/wiktextract.jsonl"

From the main project directory:

make

This will:

1. Process the wiktextract data to generate compressed dictionary files
2. Compile the C++ application for TI-Nspire
3. Copy all files to the calcbin/ folder

Copy the generated files from calcbin/ to your TI-Nspire calculator. huffman_table.csv.tns, index.csv.tns, and dictionary.bin.tns must be in the root directory of the calculator, but nDictionary.tns can be in any folder.

The dictionary uses three main files:

Binary file containing Huffman-encoded word definitions:

• uint16: Size of encoded data (big-endian)
• uint8: Number of padding bits
• bytes[]: Huffman-encoded text

CSV index for word lookups:

word,offset,size
apple,0,156
banana,156,203
...

Character encoding table:

char,code
32,101
97,001
101,11
...

1. Copy files from calcbin/ to your TI-Nspire calculator. huffman_table.csv.tns, index.csv.tns, and dictionary.bin.tns must be in the root directory of the calculator.
2. Launch nDictionary.tns using Ndless
3. Type a word and press Enter to search
4. Use DEL key to delete characters
5. Press ESC or use calculator controls to exit

Ensure the wiktextract file exists and set the environment variable:

export WIKTEXTRACT_PATH="/full/path/to/wiktextract.jsonl"

Install required packages:

pip install -r process_words/requirements.txt

Ensure you have the Ndless SDK properly installed and in your PATH:

which nspire-gcc  # Should return a path

This project is licensed under the MIT License. See the LICENSE file for details.

• add a lua-based GUI
• use LZ-based compression instead of Huffman
• store the index file, huffman file and dictionary file together
• allow the words' files to be stored in a subfolder alongside the app file
• add more words, and come up with a better way to filter duplicate words than pick the longest definition
• Don't load the entire index file into memory, instead use a binary search on the file directly?
• multi-threading for building the dictionary