gretelai / gretel-python-client Goto Github PK

For FPE, allow explicit char mask patterns to control what portions of a string are encrypted. For this to work, a full mask must be provided that is equal to the string len of the target string.

Add a blueprint for using FPE w/ char masking + format transformation to normalize strings

• Fix tests, ln 244 of test_client

• Update sync record writing to also raise the exceptions, and handle automatic backoff on getting throttle errors

• build out docstrings for all transformer config modules

• add the modules to the sphinx documentation as separate pages for each one

When iam trying to upload to UI using import option for a csv file it is giving your file is being uploaded and below it another message stating that couldnot upload only json and csv are to be uploaded

Include all xform configs in gretel_client.transformers top level package so everything can be imported from that name space

So in talking with John, I think we need to solve some issues:

1. currently we can specify float_precision as "only encrypt, n decimal places beyond the decimal point" so that lat-lon's can preserve basic features depending on n.
   For example:
   lat 172.12345 with precision = 2
   may yield: 172.12975643
   But, if the incoming value is of type string, it does not get treated like a float.
   To solve this, I am thinking, if float_precision is NOT None, then it is assumed that the data values are floats, and if the incoming value is of type float, it will do what it currently does, if it is of type string, then we figure out where the decimal point is in the string, and work out which digits we need to encrypt from there. So, for the above example as a string we would get:
   "172.12345" -> "172.12975" notice how the number of digits behind the decimal does not need to change in this example, as we do not need to take extra care of outputting a native float value.
2. Make mask and float precision mutually exclusive. Given point 1., it would no longer make sense to specify both, a mask and a float precision, as it would not be clear which one of the 2 should be applied.
   We still need a way to specify masks for things like credit card numbers where we might want to preserve the first digit to retain which credit card was used, or even the first 6 digits to retain which bank the credit card belongs to.
3. I have been thinking about expressing precision in a form closer to sci notation in powers of 10. That way we could specify precision for the integer part as well.

Add Client method for using the real time labeling API endpoint

The new blueprints are self-contained, but as a matter of precaution we are adding assertions into what their expected output is. We should run all blueprint samples during tests as a safeguard.

Getting build wheel error while installing gretel client, running locally in windows 11.
Is it possible to run gretel client locally in windows?
OS: windows 11
Python version : 3.11.3

Update requirements to install tqdm

Update Sphinx to match new Gretel style

Hi sorry the delay in the response
iam uploading safecast data what do we have in the ui as sample data i downloaded it and tried importing again in the ui which is when iam facing the issue

Is GPU mandatory for Ubuntu with the AWS to run gretel locally?

Whether we can run AWS without the GPU, like using normal CPU with Ubuntu?