ai-server-sdk 0.0.30

Utility package to connect to AI Server instances.

AI Server

ai-server-sdk is a python client SDK to connect to the AI Server

Using this package you can:

• Inference with Models you have acces to within the server
• Create Pandas DataFrame from Databases connections
• Pull Storage objects
• Run pixel and get the direct output or full json response.
• Pull data products from an existing insight using REST API.

Install

pip install ai-server-sdk

or

pip install ai-server-sdk[full]

Note: The full option installs optional dependencies for langchain support.

Usage

To interract with an ai-server instance, import the ai_server package and connect via ServerClient.

Setup

from ai_server import ServerClient

# define access keys
loginKeys = {"secretKey":"<your_secret_key>","accessKey":"<your_access_key>"}

# create connection object by passing in the secret key, access key and base url for the api
server_connection = ServerClient(base='<Your deployed server Monolith URL>', access_key=loginKeys['accessKey'], secret_key=loginKeys['secretKey'])

# if you are logged in with OAuth, a bearer token can be provided
server_connection = ServerClient(base='<Your deployed server Monolith URL>', bearer_token="bearer_token_value")

Inference with different Model Engines

# import the model engine class for the ai_server package
from ai_server import ModelEngine

model = ModelEngine(engine_id="2c6de0ff-62e0-4dd0-8380-782ac4d40245", insight_id=server_connection.cur_insight)

# if your model is for text-generation, ask a question
model.ask(command = 'What is the capital of France?')
# example output
# {'response': 'The capital of France is Paris.',
#  'messageId': '0a80c2ce-76f9-4466-b2a2-8455e4cab34a',
#  'messageType': 'CHAT',
#  'roomId': '28261853-0e41-49b0-8a50-df34e8c62a19',
#  'numberOfTokensInResponse': 6, 'numberOfTokensInPrompt': 6}

# stream the response
for chunk in model.stream_ask(command=command):
    print(chunk, end="", flush=True)

# instantiate a different model for embeddings, get embeddings for some text
model = ModelEngine(engine_id="e4449559-bcff-4941-ae72-0e3f18e06660", insight_id=server_connection.cur_insight)
model.embeddings(strings_to_embed=['text1','text2'])
# example output
# {'response': [[0.007663827, -0.030877046, ... -0.035327386]],
#  'numberOfTokensInPrompt': 8, 'numberOfTokensInResponse': 0}

# Integrate with langchain
model = ModelEngine(engine_id="2c6de0ff-62e0-4dd0-8380-782ac4d40245", insight_id=server_connection.cur_insight)
langchain_llm = model.to_langchain_chat_model()
command = 'What is the capital of France?'
output = langchain_llm.invoke(input = command)
# example output
# AIMessage(content='The capital of France is Paris.', additional_kwargs={}, response_metadata={'numberOfTokensInResponse': 6, 'numberOfTokensInPrompt': 6, 'messageType': 'CHAT', 'messageId': 'bd4f54fe-fd9b-4538-8531-696c4cdae01f', 'roomId': '57c03aae-5c10-498e-9a25-027201daa917'}, id='run-e9672e53-0cfd-4cb6-b9e4-3d5304314f73-0')

# stream the response
for chunk in langchain_llm.stream(command):
    print(chunk.content, end="", flush=True)

Interact with a Vector Database by adding document(s), querying, and removing document(s)

# import the vector engine class for the ai_server package
from ai_server import VectorEngine

# initialize the connection to the vector database
vectorEngine = VectorEngine(engine_id="221a50a4-060c-4aa8-8b7c-e2bc97ee3396", insight_id=server_connection.cur_insight)

# Add document(s) that have been uploaded to the insight
vectorEngine.addDocument(file_paths = ['fileName1.pdf', 'fileName2.pdf', ..., 'fileNameX.pdf'])

# Add Vector CSV File document(s) that have been uploaded to the insight
vectorEngine.addVectorCSVFile(file_paths = ['fileName1.csv', 'fileName2csv', ..., 'fileNameX.csv'])

# Perform a nearest neighbor search on the embedded documents
vectorEngine.nearestNeighbor(search_statement = 'Sample Search Statement', limit = 5)

# List all the documents the vector database currently comprises of
vectorEngine.listDocuments()

# Remove document(s) from the vector database
vectorEngine.removeDocument(file_names = ['fileName1.pdf', 'fileName2.pdf', ..., 'fileNameX.pdf'])

# integrate with langchain
vector = VectorEngine(engine_id = "221a50a4-060c-4aa8-8b7c-e2bc97ee3396", insight_id=server_connection.cur_insight)
langhchain_vector = vector.to_langchain_vector_store()
langhchain_vector.listDocs()
langhchain_vector.addDocs(file_paths = ['file1.pdf','file2.pdf',...])
langhchain_vector.removeDocs(file_names = ['file1.pdf','file2.pdf',...])
langhchain_vector.similaritySearch(query = 'Sample Search Statement', k=5)

Connect to Databases and execute create, read, and delete operations

Run the passed string query against the engine. The query passed must be in the structure that the specific engine implementation.

# import the database engine class for the ai_server package
from ai_server import DatabaseEngine

# Create an relation to database based on the engine identifier
database = DatabaseEngine(engine_id="4a1f9466-4e6d-49cd-894d-7d22182344cd", insight_id=server_connection.cur_insight)
database.execQuery(query='SELECT PATIENT, HEIGHT, WEIGHT FROM diab LIMIT 4')

	PATIENT	HEIGHT	WEIGHT
0	20337	64	114
1	3750	64	161
2	40785	67	187
3	12778	72	145

Run query operations against the engine. Query must be in the structure that the specific engine implementation

# insert statement
database.insertData(query = 'INSERT INTO table_name (column1, column2, column3, ...) VALUES (value1, value2, value3, ...)')
# update statement
database.updateData(query = 'UPDATE table_name set column1=value1 where age=19')
# delete statement
database.removeData(query='DELETE FROM diab WHERE age=19')

# integrate with langchain
database = DatabaseEngine(engine_id="4a1f9466-4e6d-49cd-894d-7d22182344cd", insight_id=server_connection.cur_insight)
langhchain_db = database.to_langchain_database()
langhchain_db.executeQuery(query = 'SELECT * FROM table_name')
langhchain_db.insertQuery(query = 'INSERT INTO table_name (column1, column2, column3, ...) VALUES (value1, value2, value3, ...)')
langhchain_db.updateQuery(query = 'UPDATE table_name set column1=value1 WHERE condition')
langhchain_db.removeQuery(query = 'DELETE FROM table_name WHERE condition')

Run Function Engines

# import the function engine class for the ai_server package
from ai_server import FunctionEngine

# initialize the connection ot the function engine
function = FunctionEngine(engine_id="f3a4c8b2-7f3e-4d04-8c1f-2b0e3dabf5e9", insight_id=server_connection.cur_insight)
function.execute({"lat":"37.540","lon":"77.4360"})
# example output
# '{"cloud_pct": 2, "temp": 28, "feels_like": 27, "humidity": 20, "min_temp": 28, "max_temp": 28, "wind_speed": 5, "wind_degrees": 352, "sunrise": 1716420915, "sunset": 1716472746}'

Using REST API to pull data product from an Insight

# define the Project ID
projectId = '30991037-1e73-49f5-99d3-f28210e6b95c'

# define the Insight ID
inishgtId = '26b373b3-cd52-452c-a987-0adb8817bf73'

# define the SQL for the data product you want to query within the insight
sql = 'select * FROM DATA_PRODUCT_123'

# if you dont provide one of the following, it will ask you to provide it via prompt
diabetes_df = server_connection.import_data_product(project_id = projectId, insight_id = inishgtId, sql = sql)
diabetes_df.head()

	AGE	PATIENT	WEIGHT
0	19	4823	119
1	19	17790	135
2	20	1041	159
3	20	2763	274
4	20	3750	161

Get the output or JSON response of any pixel

# run the pixel and get the output
server_connection.run_pixel('1+1')
2

# run the pixel and get the entire json response
server_connection.run_pixel('1+1', full_response=True)
# example output
# {'insightID': '8b419eaf-df7d-4a7f-869e-8d7d59bbfde8',
# 'pixelReturn': [{'pixelId': '3',
#   'pixelExpression': '1 + 1 ;',
#   'isMeta': False,
#   'output': 2,
#   'operationType': ['OPERATION']}]}

Upload / Download files to an Insight

from ai_server import ServerClient

# define access keys
loginKeys = {"secretKey":"<your_secret_key>","accessKey":"<your_access_key>"}

# create connection object by passing in the secret key, access key and base url for the api
server_connection = ServerClient(access_key=loginKeys['accessKey'], secret_key=loginKeys['secretKey'], base='<Your deployed server Monolith URL>')

server_connection.upload_files(files=["path_to_local_file1", "path_to_local_file2"], project_id="your_project_id", insight_id="your_insight_id", path="path_to_upload_files_in_insight")

server_connection.download_file(file=["path_to_insight_file"], project_id="your_project_id", insight_id="your_insight_id",custom_filename="filename_for_download")

Using tools via langchain

import ai_server
from ai_server import ServerClient
from ai_server import ModelEngine
from langchain_core.messages import HumanMessage
from langchain_core.tools import tool

loginKeys = {"secretKey":"<your_secret_key>","accessKey":"<your_access_key>"}

# create connection object by passing in the secret key, access key and base url for the api
server_connection = ServerClient(base='<Your deployed server Monolith URL>', access_key=loginKeys['accessKey'], secret_key=loginKeys['secretKey'])

model = ModelEngine(
    engine_id="4acbe913-df40-4ac0-b28a-daa5ad91b172",
    insight_id=server_connection.cur_insight,
)

langchain_model = model.to_langchain_chat_model()

@tool
def multiply(a: int, b: int) -> int:
    """Multiply a and b.

    Args:
        a: first int
        b: second int
    """
    return a * b


@tool
def add(a: int, b: int) -> int:
    """Adds a and b.

    Args:
        a: first int
        b: second int
    """
    return a + b


@tool
def divide(a: int, b: int) -> float:
    """Divide a and b.

    Args:
        a: first int
        b: second int
    """
    return a / b


tools = [add, multiply, divide]

query = "What is 3 * 12?"
messages = [HumanMessage(query)]

langchain_chat_with_tools = langchain_model.bind_tools(tools)
result = langchain_chat_with_tools.invoke(messages)
messages.append(result)

for tool_call in result.tool_calls:
    selected_tool = {"add": add, "multiply": multiply}[tool_call["name"].lower()]
    tool_msg = selected_tool.invoke(tool_call)
    messages.append(tool_msg)

final_output = langchain_chat_with_tools.invoke(messages)
print(final_output)

---