Merge pull request #334 from Scale3-Labs/rohit/S3EN-2778-release-main

Rohit/s3 en 2778 release main

Author: rohit-kadhe

src/examples/azureopenai_example/__init__.py

@@ -0,0 +1,9 @@
+from examples.azureopenai_example.completion import chat_completion
+from langtrace_python_sdk import with_langtrace_root_span, langtrace
+
+langtrace.init()
+
+class AzureOpenAIRunner:
+    @with_langtrace_root_span("AzureOpenAI")
+    def run(self):
+        chat_completion()

src/examples/azureopenai_example/completion.py

@@ -0,0 +1,22 @@
+import os
+from langchain_openai import AzureChatOpenAI
+
+from langtrace_python_sdk.utils.with_root_span import with_langtrace_root_span
+
+model = AzureChatOpenAI(
+    azure_endpoint=os.environ['AZURE_OPENAI_ENDPOINT'],
+    azure_deployment=os.environ['AZURE_OPENAI_DEPLOYMENT_NAME'],
+    openai_api_version=os.environ['AZURE_OPENAI_API_VERSION'],
+)
+
+@with_langtrace_root_span()
+def chat_completion():
+    messages = [
+        (
+            "system",
+            "You are a helpful assistant that translates English to French. Translate the user sentence.",
+        ),
+        ("human", "I love programming."),
+    ]
+    result = model.invoke(messages)
+    print(result)

src/examples/langchain_example/__init__.py

@@ -2,6 +2,7 @@
 from langtrace_python_sdk import with_langtrace_root_span
 
 from .groq_example import groq_basic, groq_streaming
+from .langgraph_example_tools import basic_graph_tools
 
 
 class LangChainRunner:
@@ -10,6 +11,7 @@ def run(self):
         basic_app()
         rag()
         load_and_split()
+        basic_graph_tools()
 
 
 class GroqRunner:

src/examples/langchain_example/langgraph_example_tools.py

@@ -0,0 +1,151 @@
+from typing import Annotated
+
+from langchain_anthropic import ChatAnthropic
+from langchain_core.messages import HumanMessage
+from langchain_core.pydantic_v1 import BaseModel
+from typing_extensions import TypedDict
+from langchain_core.pydantic_v1 import BaseModel, Field
+from langchain_core.tools import Tool
+from langgraph.checkpoint.memory import MemorySaver
+from langgraph.graph import StateGraph
+from langgraph.graph.message import add_messages
+from langgraph.prebuilt import ToolNode, tools_condition
+from langchain_core.messages import AIMessage, ToolMessage
+
+from langtrace_python_sdk import langtrace
+
+langtrace.init()
+
+primes = {998: 7901, 999: 7907, 1000: 7919}
+
+
+class PrimeInput(BaseModel):
+    n: int = Field()
+
+
+def is_prime(n: int) -> bool:
+    if n <= 1 or (n % 2 == 0 and n > 2):
+        return False
+    for i in range(3, int(n**0.5) + 1, 2):
+        if n % i == 0:
+            return False
+    return True
+
+
+def get_prime(n: int, primes: dict = primes) -> str:
+    return str(primes.get(int(n)))
+
+
+async def aget_prime(n: int, primes: dict = primes) -> str:
+    return str(primes.get(int(n)))
+
+
+class State(TypedDict):
+    messages: Annotated[list, add_messages]
+    # This flag is new
+    ask_human: bool
+
+
+class RequestAssistance(BaseModel):
+    """Escalate the conversation to an expert. Use this if you are unable to assist directly or if the user requires support beyond your permissions.
+
+    To use this function, relay the user's 'request' so the expert can provide the right guidance.
+    """
+
+    request: str
+
+
+llm = ChatAnthropic(model="claude-3-haiku-20240307")
+# We can bind the llm to a tool definition, a pydantic model, or a json schema
+llm_with_tools = llm.bind_tools([RequestAssistance])
+tools = [
+    Tool(
+        name="GetPrime",
+        func=get_prime,
+        description="A tool that returns the `n`th prime number",
+        args_schema=PrimeInput,
+        coroutine=aget_prime,
+    ),
+]
+
+
+def chatbot(state: State):
+    response = llm_with_tools.invoke(state["messages"])
+    ask_human = False
+    if (
+        response.tool_calls
+        and response.tool_calls[0]["name"] == RequestAssistance.__name__
+    ):
+        ask_human = True
+    return {"messages": [response], "ask_human": ask_human}
+
+
+graph_builder = StateGraph(State)
+
+graph_builder.add_node("chatbot", chatbot)
+graph_builder.add_node("tools", ToolNode(tools=tools))
+
+
+def create_response(response: str, ai_message: AIMessage):
+    return ToolMessage(
+        content=response,
+        tool_call_id=ai_message.tool_calls[0]["id"],
+    )
+
+
+def human_node(state: State):
+    new_messages = []
+    if not isinstance(state["messages"][-1], ToolMessage):
+        # Typically, the user will have updated the state during the interrupt.
+        # If they choose not to, we will include a placeholder ToolMessage to
+        # let the LLM continue.
+        new_messages.append(
+            create_response("No response from human.", state["messages"][-1])
+        )
+    return {
+        # Append the new messages
+        "messages": new_messages,
+        # Unset the flag
+        "ask_human": False,
+    }
+
+
+def select_next_node(state: State):
+    if state["ask_human"]:
+        return "human"
+    # Otherwise, we can route as before
+    return tools_condition(state)
+
+
+def basic_graph_tools():
+    graph_builder.add_node("human", human_node)
+    graph_builder.add_conditional_edges(
+        "chatbot",
+        select_next_node,
+        {"human": "human", "tools": "tools", "__end__": "__end__"},
+    )
+    graph_builder.add_edge("tools", "chatbot")
+    graph_builder.add_edge("human", "chatbot")
+    graph_builder.set_entry_point("chatbot")
+    memory = MemorySaver()
+    graph = graph_builder.compile(
+        checkpointer=memory,
+        interrupt_before=["human"],
+    )
+
+    config = {"configurable": {"thread_id": "1"}}
+    events = graph.stream(
+        {
+            "messages": [
+                (
+                    "user",
+                    "I'm learning LangGraph. Could you do some research on it for me?",
+                )
+            ]
+        },
+        config,
+        stream_mode="values",
+    )
+    for event in events:
+        if "messages" in event:
+            event["messages"][-1]

src/langtrace_python_sdk/instrumentation/anthropic/patch.py

@@ -18,6 +18,7 @@
 from langtrace.trace_attributes import Event, SpanAttributes, LLMSpanAttributes
 from langtrace_python_sdk.utils import set_span_attribute
 from langtrace_python_sdk.utils.silently_fail import silently_fail
+import json
 
 from langtrace_python_sdk.utils.llm import (
     StreamWrapper,
@@ -99,15 +100,20 @@ def set_response_attributes(
                 set_span_attribute(
                     span, SpanAttributes.LLM_RESPONSE_MODEL, result.model
                 )
-                content_item = result.content[0]
-                completion = [
-                    {
-                        "role": result.role or "assistant",
-                        "content": content_item.text,
-                        "type": content_item.type,
-                    }
-                ]
-                set_event_completion(span, completion)
+                if hasattr(result, "content") and result.content[0] is not None:
+                    content = result.content[0]
+                    typ = content.type
+                    role = result.role if result.role else "assistant"
+                    if typ == "tool_result" or typ == "tool_use":
+                        content = content.json()  # type: ignore
+                        set_span_attribute(
+                            span, SpanAttributes.LLM_TOOL_RESULTS, json.dumps(content)
+                        )
+                    if typ == "text":
+                        content = result.content[0].text
+                        set_event_completion(
+                            span, [{"type": typ, role: role, content: content}]
+                        )
 
             if (
                 hasattr(result, "system_fingerprint")

src/langtrace_python_sdk/instrumentation/anthropic/types.py

@@ -68,7 +68,6 @@ def __init__(
 
 class ContentItem:
     role: str
-    content: str
     text: str
     type: str
 

src/langtrace_python_sdk/utils/llm.py

@@ -126,7 +126,7 @@ def get_llm_request_attributes(kwargs, prompts=None, model=None, operation_name=
     tools = kwargs.get("tools", None)
     return {
         SpanAttributes.LLM_OPERATION_NAME: operation_name,
-        SpanAttributes.LLM_REQUEST_MODEL: model or kwargs.get("model"),
+        SpanAttributes.LLM_REQUEST_MODEL: model or kwargs.get("model") or "gpt-3.5-turbo",
         SpanAttributes.LLM_IS_STREAMING: kwargs.get("stream"),
         SpanAttributes.LLM_REQUEST_TEMPERATURE: kwargs.get("temperature"),
         SpanAttributes.LLM_TOP_K: top_k,

src/langtrace_python_sdk/version.py

@@ -1 +1 @@
-__version__ = "2.3.5"
+__version__ = "2.3.7"

src/run_example.py

@@ -2,13 +2,14 @@
 
 ENABLED_EXAMPLES = {
     "anthropic": False,
+    "azureopenai": True,
     "chroma": False,
     "cohere": False,
     "fastapi": False,
-    "langchain": False,
+    "langchain": True,
     "llamaindex": False,
     "hiveagent": False,
-    "openai": True,
+    "openai": False,
     "perplexity": False,
     "pinecone": False,
     "qdrant": False,
@@ -110,3 +111,9 @@
 
     print(Fore.BLUE + "Running Mistral example" + Fore.RESET)
     MistralRunner().run()
+
+if ENABLED_EXAMPLES["azureopenai"]:
+    from examples.azureopenai_example import AzureOpenAIRunner
+
+    print(Fore.BLUE + "Running Azure OpenAI example" + Fore.RESET)
+    AzureOpenAIRunner().run()

src/tests/anthropic/test_anthropic.py

@@ -27,6 +27,7 @@ def test_anthropic(anthropic_client, exporter):
         "stream": False,
         "max_tokens": 1024,
     }
+
     anthropic_client.messages.create(**kwargs)
     spans = exporter.get_finished_spans()
     completion_span = spans[-1]