Update pydantic to 2.0 (#9121)

* bump pydantic version

* Bump dataprep

* Update nixtla

* update readme

* remove log

* Missed renames

* Fix nixtla forecast

mindsdb/integrations/handlers/timegpt_handler/README.md

@@ -1,5 +1,5 @@
 # Briefly describe what ML framework does this handler integrate to MindsDB, and how?
-TimeGPT is a zero-shot forecasting model developed by Nixtla, offered through their `nixtlats` Python package.
+TimeGPT is a zero-shot forecasting model developed by Nixtla, offered through their `nixtla` Python package.
 
 This handler provides a simple wrapper around TimeGPT, and provides easy ingestion of time series data from other MindsDB data sources into the TimeGPT API. User requires a Nixtla API key to use this handler.
 

mindsdb/integrations/handlers/timegpt_handler/requirements.txt

@@ -1 +1 @@
-nixtlats>=0.1.10
+nixtla==0.5.0

mindsdb/integrations/handlers/timegpt_handler/timegpt_handler.py

@@ -1,7 +1,7 @@
 from typing import Optional, Dict
 
 import pandas as pd
-from nixtlats import TimeGPT
+from nixtla import NixtlaClient
 
 from mindsdb.integrations.libs.base import BaseMLEngine
 from mindsdb.integrations.utilities.handler_utils import get_api_key
@@ -34,8 +34,8 @@ def create(self, target: str, df: Optional[pd.DataFrame] = None, args: Optional[
             assert time_settings["is_timeseries"], "Specify time series settings in your query"
 
         timegpt_token =  get_api_key('timegpt', using_args, self.engine_storage, strict=True)
-        timegpt = TimeGPT(token=timegpt_token)
-        assert timegpt.validate_token(), "Invalid TimeGPT token provided."
+        timegpt = NixtlaClient(api_key=timegpt_token)
+        assert timegpt.validate_api_key(), "Invalid TimeGPT token provided."
 
         model_args = {
             'token': timegpt_token,
@@ -74,8 +74,8 @@ def predict(self, df, args={}):
         args = args['predict_params']
         prediction_df = self._transform_to_nixtla_df(df, model_args)
 
-        timegpt = TimeGPT(token=model_args['token'])
-        assert timegpt.validate_token(), "Invalid TimeGPT token provided."
+        timegpt = NixtlaClient(api_key=model_args['token'])
+        assert timegpt.validate_api_key(), "Invalid TimeGPT token provided."
 
         forecast_df = timegpt.forecast(
             prediction_df,
@@ -85,7 +85,7 @@ def predict(self, df, args={}):
             freq=args.get("freq", model_args["freq"]),  # automatically infers correct frequency if not provided by user
             level=model_args["level"],
             finetune_steps=args.get('finetune_steps', model_args['finetune_steps']),
-            validate_token=args.get('validate_token', model_args['validate_token']),
+            validate_api_key=args.get('validate_token', model_args['validate_token']),
             date_features=args.get('date_features', model_args['date_features']),
             date_features_to_one_hot=args.get('date_features_to_one_hot', model_args['date_features_to_one_hot']),
             clean_ex_first=args.get('clean_ex_first', model_args['clean_ex_first']),

requirements/requirements.txt

@@ -18,13 +18,13 @@ walrus==0.9.3
 flask-compress >= 1.0.0
 appdirs >= 1.0.0
 mindsdb-sql ~= 0.15.0
-pydantic >= 1.8.2
+pydantic >= 2.0.3
 mindsdb-evaluator >= 0.0.7, < 0.1.0
 checksumdir >= 1.2.0
 duckdb == 0.9.1
 requests >= 2.30.0
 pydateinfer==0.3.0
-dataprep_ml==0.0.22
+dataprep_ml==0.0.23
 dill == 0.3.6
 numpy
 pytz