Prompt-Horizon

Prompt-Horizon is a Python library that enables developers to anonymize JSON objects by creating placeholders for keys and values, while generating a reversible mapping to restore the original JSON data. The purpose of this library is to facilitate data sharing while preserving privacy and allowing for the de-anonymization of the data when required.

Note: I had the idea for this library when I was trying to find a way to obfuscate data objects such as JSON to include them as contextual information in prompts (hence the name).

Prompt-Horizon is not designed to specifically identify and anonymize PII (personally identifiable information) or other sensitive data within the JSON objects.

Installation

pip install prompt-horizon

Usage

Anonymize JSON

from prompt_horizon import anonymize

input_json = {
    "name": "John",
    "age": 30,
    "city": "New York",
    "favorites": ["pizza", "basketball"]
}

#This line will return the anonymized JSON object and the mapping object.
anonymized_json, map_object = anonymize(input_json)

Original JSON:

{
    "name": "John",
    "age": 30,
    "city": "New York",
    "favorites": ["pizza", "basketball"]
}

Anonymized JSON:

{
    "a1": "a2",
    "a3": "a4",
    "a5": "a6",
    "a7": ["a8", "a9"]
}

input and output file handling is supported

from prompt_horizon import anonymize

#This line will read the JSON data from "input.json", anonymize it, and save the anonymized JSON to "anonymized.json".
anonymized_json, map_object = anonymize("input.json", output_file_path="anonymized.json")

anonymize only the values

from prompt_horizon import anonymize

#This line will read the JSON data from "input.json" and anonymize only the values.
anonymized_json, map_object = anonymize("input.json", anonymize_keys=False)

De-anonymize JSON

from prompt_horizon import de_anonymize

#This line will return the de-anonymized JSON object, which should be the same as the original input JSON.
deanonymized_json = de_anonymize(anonymized_json, map_object=map_object)

input and output files are supported

from prompt_horizon import de_anonymize

#This will read the anonymized JSON data from "anonymized.json", the mapping object from "map_file.json", de-anonymize the JSON data, and save the de-anonymized JSON to "deanonymized.json".
deanonymized_json = de_anonymize("anonymized.json", map_file_path="map_file.json", output_file_path="deanonymized.json")

License

This project is licensed under the MIT License.