From ebf999520c40b93d377ffe35ddd10a68d649cc1e Mon Sep 17 00:00:00 2001
From: Tanmaypatil123 <tanmaypatil3151@gmail.com>
Date: Tue, 24 Oct 2023 12:39:18 +0530
Subject: [PATCH] Modification in pipeline

---
 pandasai/pipelines/__init__.py |  3 ++
 pandasai/pipelines/base.py     | 59 ++++++++++++++++++++++++++++------
 2 files changed, 52 insertions(+), 10 deletions(-)

diff --git a/pandasai/pipelines/__init__.py b/pandasai/pipelines/__init__.py
index e69de29bb..43464c0d8 100644
--- a/pandasai/pipelines/__init__.py
+++ b/pandasai/pipelines/__init__.py
@@ -0,0 +1,3 @@
+from .base import Pipeline, BaseLogic, PromptRequestLogic
+
+__all__ = ["Pipeline", "BaseLogic", "PromptRequestLogic"]
diff --git a/pandasai/pipelines/base.py b/pandasai/pipelines/base.py
index d56030b59..b7261d808 100644
--- a/pandasai/pipelines/base.py
+++ b/pandasai/pipelines/base.py
@@ -1,9 +1,12 @@
 from ..schemas.df_config import Config
-from typing import Union, Optional
+from typing import Union, Optional, List
 from pandasai.responses.context import Context
 from ..helpers.logger import Logger
 from pandasai.responses.response_parser import ResponseParser
 from abc import ABC, abstractmethod
+import pandas as pd
+import numpy as np
+from pandasai import SmartDataframe
 
 
 class BaseLogic(ABC):
@@ -11,15 +14,18 @@ class BaseLogic(ABC):
     Logic units for pipeline.
     """
 
+    _config: Config = None
+    _logger: Logger = None
+
     def __init__(self):
         pass
 
     @abstractmethod
-    def call(self, input_):
+    def call(self, config: Config, logger: Logger, input_):
         """
         This method will return output according to
         Implementation."""
-        pass
+        raise NotImplementedError("call method is not implemented.")
 
 
 class PromptRequestLogic(BaseLogic):
@@ -27,15 +33,40 @@ class PromptRequestLogic(BaseLogic):
     Logic units for pipeline.
     """
 
-    def __init__(self):
-        pass
+    def __init__(
+        self,
+    ):
+        super().__init__()
 
-    def call(self,input_):
+    def call(self, config: Config, logger: Logger, input_):
         """
         This method will return output according to
         implementation.
         """
-        pass
+        head = input_
+        new_data = []
+        column_data_types = head.dtypes
+
+        ## generate 100 synthetic examples
+        for _ in range(100):
+            new_row = {}
+            for column_name, data_type in column_data_types.iteritems():
+                if np.issubdtype(data_type, np.number):
+                    # Generate random numbers for numerical columns
+                    mean = head[column_name].mean()
+                    std_dev = head[column_name].std()
+                    new_value = np.random.normal(mean, std_dev)
+                else:
+                    # Generate random values for non-numerical columns
+                    unique_values = head[column_name].unique()
+                    new_value = np.random.choice(unique_values)
+                new_row[column_name] = new_value
+
+            new_data.append(new_row)
+
+        ## convert it into smartdataframe
+        df = pd.DataFrame(new_data)
+        return SmartDataframe(df, config=Config)
 
 
 class Pipeline:
@@ -45,13 +76,14 @@ class Pipeline:
 
     _config: Config = None
     _logger: Logger
-    _logics: list(BaseLogic)
+    _logics: List[BaseLogic]
 
     def __init__(
         self,
         config: Union[Config, dict] = None,
         context: Optional[Context] = None,
-        logics: Optional(list(PromptRequestLogic)) = None,
+        logics: Optional[List[BaseLogic]] = None,
+        df: pd.DataFrame = None,
     ):
         """
         Intialize the pipeline with given context and configuration
@@ -78,10 +110,17 @@ def __init__(
         else:
             self._response_parser = ResponseParser(context)
 
+        self._logics = logics
+        self._df = df
+
     @abstractmethod
     def execute(self):
         """
         This functions is responsible to loop through logic and
             Implementation.
         """
-        pass
+        result = self._df
+        for logic in self._logics:
+            result = logic.call(self._config, self._logger, result)
+
+        return result