adding first agent

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+_data/** filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -12,6 +12,9 @@ RUN pip install --no-cache-dir -r requirements.txt
 
 COPY . .
 
+# Install agents 
+RUN pip install -e _agents/pandas_expression_generator
+
 # Make the script executable
 RUN chmod +x start.sh
 

_agents/pandas_expression_generator/pyproject.toml

@@ -0,0 +1,18 @@
+[build-system]
+build-backend = "setuptools.build_meta"
+requires = ["setuptools >= 64"]
+
+[project]
+name = "pandas_expression_generator"
+version = "0.1.0"
+dependencies = [
+  "aiqtoolkit[langchain]",
+]
+requires-python = ">=3.11,<3.13"
+description = "Custom AIQ Toolkit Workflow"
+classifiers = ["Programming Language :: Python"]
+
+
+
+[project.entry-points.'aiq.components']
+pandas_expression_generator = "pandas_expression_generator.register"

_agents/pandas_expression_generator/src/pandas_expression_generator.egg-info/PKG-INFO

@@ -0,0 +1,7 @@
+Metadata-Version: 2.4
+Name: pandas_expression_generator
+Version: 0.1.0
+Summary: Custom AIQ Toolkit Workflow
+Classifier: Programming Language :: Python
+Requires-Python: <3.13,>=3.11
+Requires-Dist: aiqtoolkit[langchain]

_agents/pandas_expression_generator/src/pandas_expression_generator.egg-info/SOURCES.txt

@@ -0,0 +1,10 @@
+pyproject.toml
+src/pandas_expression_generator/__init__.py
+src/pandas_expression_generator/pandas_expression_generator_function.py
+src/pandas_expression_generator/register.py
+src/pandas_expression_generator.egg-info/PKG-INFO
+src/pandas_expression_generator.egg-info/SOURCES.txt
+src/pandas_expression_generator.egg-info/dependency_links.txt
+src/pandas_expression_generator.egg-info/entry_points.txt
+src/pandas_expression_generator.egg-info/requires.txt
+src/pandas_expression_generator.egg-info/top_level.txt

_agents/pandas_expression_generator/src/pandas_expression_generator.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

_agents/pandas_expression_generator/src/pandas_expression_generator.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[aiq.components]
+pandas_expression_generator = pandas_expression_generator.register

_agents/pandas_expression_generator/src/pandas_expression_generator.egg-info/requires.txt

@@ -0,0 +1 @@
+aiqtoolkit[langchain]

_agents/pandas_expression_generator/src/pandas_expression_generator.egg-info/top_level.txt

@@ -0,0 +1 @@
+pandas_expression_generator

_agents/pandas_expression_generator/src/pandas_expression_generator/configs/config.yml

@@ -0,0 +1,25 @@
+general:
+  use_uvloop: true
+  logging:
+    console:
+      _type: console
+      level: WARN
+
+  front_end:
+    _type: fastapi
+
+  front_end:
+    _type: console
+
+llms:
+  nim_llm:
+    _type: nim
+    model_name : meta/llama-3.1-405b-instruct
+    temperature: 0.0
+
+workflow:
+  _type: pandas_expression_generator
+  llm: nim_llm
+  csv_file: "/workspaces/do_aiq2/_data/retail_customer/Retail_Customer_Cluster_Data.xlsx"
+  csv_metafile: "/workspaces/do_aiq2/_data/retail_customer/meta.md"
+  description: "Use this agent to handles retail customers data"

_agents/pandas_expression_generator/src/pandas_expression_generator/pandas_expression_generator_function.py

@@ -0,0 +1,150 @@
+import logging
+
+from pydantic import Field
+
+from aiq.builder.builder import Builder
+from aiq.builder.function_info import FunctionInfo
+from aiq.cli.register_workflow import register_function
+from aiq.data_models.function import FunctionBaseConfig
+from aiq.data_models.component_ref import LLMRef
+from aiq.builder.framework_enum import LLMFrameworkEnum
+from langchain_core.messages import AIMessage
+import pandas as pd 
+from pydantic import BaseModel
+import numpy as np
+
+logger = logging.getLogger(__name__)
+
+
+class PandasExpressionGeneratorFunctionConfig(FunctionBaseConfig, name="pandas_expression_generator"):
+    """
+    AIQ Toolkit function template. Please update the description.
+    """
+    # Add your custom configuration parameters here
+    # parameter: str = Field(default="default_value", description="Notional description for this parameter")
+    llm: LLMRef
+    csv_file: str
+    csv_metafile: str
+    description: str = Field(default="A dataset query engine", description="Describe for which data the engine is used for")
+    max_retries: int = Field(default=2)
+
+class FunctionInput(BaseModel):
+    original_user_query: str = Field(description="The original user query as-is")
+    past_step: str = Field(description="Information/ output/ data from the previous action(s) if any else leave it empty string")
+    current_step: str = Field(description="Part of the intermediate step when resovling the big user query.")
+    next_step: str = Field(description="What next are we plannning to do once we perform the current step.")
+
+@register_function(config_type=PandasExpressionGeneratorFunctionConfig)
+async def pandas_expression_generator_function(
+    config: PandasExpressionGeneratorFunctionConfig, builder: Builder
+):
+    df: pd.DataFrame = pd.read_excel(config.csv_file)
+    with open(config.csv_metafile, "r", encoding="utf-8") as f:
+        df_meta: str = f.read()
+
+
+    def extract_bracket_content(text):
+        start = text.find('[')
+        end = text.rfind(']') + 1  # include the last ']'
+        if start == -1 or end == -1 or start >= end:
+            return ""  # return empty if brackets not found properly
+        return text[start:end]
+
+    import ast
+    def expression_executor(input_exp: str, df=None) -> tuple:
+        # Convert the input string to a list
+        commands = ast.literal_eval(input_exp)
+
+        # local_vars = dict([(str.strip, command.split('=', 1)) for command in commands])
+        local_vars:dict = dict([list(map(str.strip, command.split('=', 1))) for command in commands])
+
+        # Dictionary to simulate local scope
+        local_vars.update({'df': df, 'pd': pd, 'np': np})
+        # local_vars = {}
+        for command in commands:
+            exec(command, globals(), local_vars)
+
+        # for command in commands:
+        #     # Split at the first `=`
+        #     var, expr = map(str.strip, command.split('=', 1))
+        #     # Evaluate the expression in the local_vars context and assign to var
+        #     local_vars[var] = eval(expr, {}, local_vars)
+        return local_vars.get("output", None)
+        try:
+            output
+            # output = scope.get("output")
+        except Exception as e: 
+            logger.error("[expression_executor] ", e)
+            output = None
+        return output  # return all evaluated variables
+    
+
+
+    # Implement your function logic here
+    async def _response_fn(input_query: FunctionInput) -> str:
+
+        # Create LLM 
+        llm_ = await builder.get_llm(config.llm, wrapper_type=LLMFrameworkEnum.LANGCHAIN)
+
+        # Prompt
+        prompt_ = f'''You are a python expression generator. 
+        
+        Given the dataframe `df` and pandas import as `pd` which has the following metadata
+
+        CSV Meta
+        ---
+        {df_meta}
+        
+        Generate python expression to solve the query:
+        current_step: {input_query.current_step}
+
+        You can reformulate(extend or reduce) the current step based on the following information 
+        original_user_query: {input_query.original_user_query}
+        past_step: {input_query.past_step}
+        next_step: {input_query.next_step}
+
+        You must generate list of python expressions output format must as follows:
+
+        [
+            "key_var = df[...]"                 // intermediate step 
+            "key_var2 = key_var[df[...]...]"    // intermediate step 
+
+                                                // reset the index of df if requrired
+                                                // convert the grouped dataframe to dictionaries
+                                                // 
+            "output" = sting.  Have the string as list (records) of dictionaries (columns). Ensure to reset index if required
+        ]
+
+        DONT add prefix or suffix the text in output, just output as described format.
+
+        Certain template questions and expected approach from you
+        1. When user wants to what is present in the data
+        > you should respond back with names of columns 
+        '''
+
+        # just var declaration
+        df 
+
+        try:
+            ai_message: AIMessage = await llm_.ainvoke(prompt_)
+            structured_text_str = extract_bracket_content(ai_message.content)
+            output_message = str(expression_executor(structured_text_str, df))[:250]
+            #
+            # Loop break 
+        except Exception as e: 
+            logger.error(str(e))
+            logger.info("Retrying...")
+            output_message = "Unable to serve the request."
+            # output_message = f"Hello from pandas_expression_generator workflow! You said: {input_message}"
+
+        return output_message
+
+    try:
+        yield FunctionInfo.from_fn(
+            _response_fn,
+            description=config.description
+        )
+    except GeneratorExit:
+        print("Function exited early!")
+    finally:
+        print("Cleaning up pandas_expression_generator workflow.")

_agents/pandas_expression_generator/src/pandas_expression_generator/register.py

@@ -0,0 +1,5 @@
+# pylint: disable=unused-import
+# flake8: noqa
+
+# Import any tools which need to be automatically registered here
+from pandas_expression_generator import pandas_expression_generator_function

_data/retail_customer/Retail_Customer_Cluster_Data.xlsx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:643249aed77417875d9de227e9278f19a6fa015a27cccdcc1bfa6f30f96300ca
+size 14884

_data/retail_customer/meta.md

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acc3e3133f0223c972115dc4513b6f44d0e59cf33330b05cf036b38e23df1d25
+size 2881

_data/retail_customer/readme.md

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ca0862d59bca2daf655928d57abc9684c9f844e2715bc868157d89809cb1477
+size 51

config.yaml

@@ -20,10 +20,19 @@ llms:
 functions: 
   current_datetime:
     _type: current_datetime
+    
+  retail_customer_data: 
+    _type: pandas_expression_generator
+    llm: nim_llm
+    csv_file: "/app/_data/retail_customer/Retail_Customer_Cluster_Data.xlsx"
+    csv_metafile: "/app/_data/retail_customer/meta.md"
+    description: "Use this tool to get data related to Customer Retails."
+
 
 workflow:
   _type: react_agent
   tool_names: 
+    - retail_customer_data
     - current_datetime
   llm_name: nim_llm
   verbose: true