changes to tool as per review advice

[Leky1738]

Re-PRing the translate_tool with changes made from first PR

[kristapratico]

Looking good! There are a few linting errors we should fix up from the github actions tool as well.

[kristapratico]

Let's move these imports under the validate_environment and wrap them in a try/except so we can alert users if they need to install the library, for example:

langchain/libs/community/langchain_community/tools/azure_ai_services/content_safety.py

Lines 55 to 68 in e5fb363

	try:
	import azure.ai.contentsafety as sdk
	from azure.core.credentials import AzureKeyCredential
	values["content_safety_client"] = sdk.ContentSafetyClient(
	endpoint=content_safety_endpoint,
	credential=AzureKeyCredential(content_safety_key),
	)
	except ImportError:
	raise ImportError(
	"azure-ai-contentsafety is not installed. "
	"Run `pip install azure-ai-contentsafety` to install."
	)

[kristapratico]

This isn't related to OpenAI so we can drop that from the env var name

Suggested change

	translate_key = os.getenv("AZURE_OPENAI_TRANSLATE_API_KEY")
	translate_endpoint = os.getenv("AZURE_OPENAI_TRANSLATE_ENDPOINT")
	translate_key = os.getenv("AZURE_TRANSLATE_API_KEY")
	translate_endpoint = os.getenv("AZURE_TRANSLATE_ENDPOINT")

[kristapratico]

I think this should accept the values dict like shown here:

langchain/libs/community/langchain_community/tools/azure_ai_services/image_analysis.py

Line 39 in e5fb363

def validate_environment(cls, values: Dict) -> Any:

[kristapratico]

I think we're expected to create the client here and set it in the values dictionary:

values["translate_client"] = TextTranslationClient(
    endpoint=translate_endpoint, credential=AzureKeyCredential(translate_key)
)

[kristapratico]

can we double-check if we need this init?

[kristapratico]

This gets called internally, there won't be a chance for the user to pass a value to to_language.

[kristapratico]

It looks like this got renamed from azure_translate to translate_tool?

[kristapratico]

let's remove the ModuleNotFoundError output here

[kristapratico]

Suggested change

	"translate_key = os.getenv('AZURE_OPENAI_TRANSLATE_API_KEY')\n",
	"translate_endpoint = os.getenv('AZURE_OPENAI_TRANSLATE_ENDPOINT')\n",
	"\n",
	"assert translate_key is not None, \"AZURE_OPENAI_TRANSLATE_API_KEY is not set.\"\n",
	"assert translate_endpoint is not None, \"AZURE_OPENAI_TRANSLATE_ENDPOINT is not set.\""
	"translate_key = os.getenv('AZURE_TRANSLATE_API_KEY')\n",
	"translate_endpoint = os.getenv('AZURE_TRANSLATE_ENDPOINT')\n",
	"\n",
	"assert translate_key is not None, \"AZURE_TRANSLATE_API_KEY is not set.\"\n",
	"assert translate_endpoint is not None, \"AZURE_TRANSLATE_ENDPOINT is not set.\""

[kristapratico]

Looks like we need to clean up the notebook to show the working tool

[Leky1738]

I'm not sure what is causing these errors!

[Sheepsta300]

You can run the following commands if you have poetry and ruff installed to lint your files how LangChain wants them, just replace them with your file paths.

!poetry run ruff format "path/to/file"
!poetry run ruff check --fix "path/to/file"
!poetry run ruff check --select I --fix "path/to/file"

[kristapratico]

I think to be consistent between translate tools (especially if we decide to merge them), lets add the following:

text_translation_key: str = ""
text_translation_endpoint: str = ""
region: str = ""

[kristapratico]

Missing model_validator (from pydantic import model_validator)

Suggested change

	@classmethod
	def validate_environment(cls, values: Dict) -> Any:
	@model_validator(mode="before")
	@classmethod
	def validate_environment(cls, values: Dict) -> Any:

[kristapratico]

Let's leverage the helper function for env vars that the other tools use:

# Get environment variables
azure_translate_key = get_from_dict_or_env(
    values, "text_translation_key", "AZURE_TRANSLATE_API_KEY"
)
azure_translate_endpoint = get_from_dict_or_env(
    values, "text_translation_endpoint", "AZURE_TRANSLATE_ENDPOINT"
)

region = get_from_dict_or_env(
    values, "region", "AZURE_REGION"
)

[kristapratico]

I needed region to call the client successfully

Suggested change

	values["translate_client"] = TextTranslationClient(
	endpoint=translate_endpoint, credential=AzureKeyCredential(translate_key)
	)
	values["translate_client"] = TextTranslationClient(
	endpoint=translate_endpoint, credential=AzureKeyCredential(translate_key), region=region
	)

[kristapratico]

We shouldn't call the internal methods. I'd expect a user to invoke the tool directly or use an agent.

translate_tool = AzureTranslateTool(
    text_translation_key=os.environ["AZURE_TRANSLATE_API_KEY"],
    text_translation_endpoint=os.environ["AZURE_TRANSLATE_ENDPOINT"],
    region="westus2",
)
answer = translate_tool.invoke({"query": "hola mundo"})

print(answer)

translate_tool = AzureTranslateTool(
    text_translation_key=os.environ["AZURE_TRANSLATE_API_KEY"],
    text_translation_endpoint=os.environ["AZURE_TRANSLATE_ENDPOINT"],
    region="westus2",
)

tools = [translate_tool]

model = AzureChatOpenAI(
    openai_api_version=os.environ["OPENAI_API_VERSION"],
    azure_deployment=os.environ["COMPLETIONS_MODEL"],
    azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"],
    api_key=os.environ["AZURE_OPENAI_API_KEY"],
)

agent = create_structured_chat_agent(model, tools, prompt)

agent_executor = AgentExecutor(
    agent=agent, tools=tools, verbose=True, handle_parsing_errors=True
)

text = "hola mundo"

answer = agent_executor.invoke(
    {"input": f"What does the text say in english? : {text}"}
)
print(answer)