Metadata-Version: 2.4
Name: fds.sdk.Conversational
Version: 0.0.1
Summary: Conversational client library for Python
Home-page: https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0
Author: FactSet Research Systems
License: Apache License, Version 2.0
Keywords: FactSet,API,SDK
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: urllib3>=2.7.0
Requires-Dist: python-dateutil>=2.9.0.post0
Requires-Dist: fds.sdk.utils>=1.0.0
Dynamic: author
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: keywords
Dynamic: license
Dynamic: license-file
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

[![FactSet](https://raw.githubusercontent.com/factset/enterprise-sdk/main/docs/images/factset-logo.svg)](https://www.factset.com)

# Conversational client library for Python

[![API Version](https://img.shields.io/badge/api-v0.2.0-blue)](https://developer.factset.com/api-catalog/conversational-api)
[![PyPi](https://img.shields.io/pypi/v/fds.sdk.Conversational/0.0.1)](https://pypi.org/project/fds.sdk.Conversational/v/0.0.1)
[![Apache-2 license](https://img.shields.io/badge/license-Apache2-brightgreen.svg)](https://www.apache.org/licenses/LICENSE-2.0)

This API lets you send a question to FactSet's 
conversational agent and get an answer back, formatted the same
way OpenAI's Chat Completions API formats its responses. That
means tools already built to talk to OpenAI-style chat APIs (chat
UIs, IDE assistants, evaluation harnesses, and so on) can talk to
conversational agent with little or no changes.

A few things work a bit differently from a typical OpenAI
integration:

- Even though you can send a full list of messages (a \"chat
  history\"), only the most recent message from the user is
  actually used as the question being asked. Any earlier
  messages, or a system prompt, are ignored.
- Instead, the API keeps track of the conversation for you. Each
  conversation has an id (`conversation_id`). Send that id back on
  your next request and the assistant will remember what was said
  before and continue the same conversation. Leave it out and a
  new conversation is started, with a fresh id handed back to you
  to reuse next time.
- The answer isn't returned all at once — it's streamed back a
  little at a time as it's generated, using a format called
  Server-Sent Events (`text/event-stream`). This is the same
  streaming format OpenAI's API uses, so most existing client
  libraries already know how to read it. The stream always ends
  with a `data: [DONE]` line so the client knows there's nothing
  more coming.

**A note on FactSet's internal API standards:** this endpoint's
contract is intentionally OpenAI's Chat Completions API contract,
not FactSet's own API design standards — that's the entire point
of the endpoint, since it lets existing OpenAI-compatible clients
work against it unmodified. Where the two disagree, OpenAI's
contract wins by design, and the FactSet standard does not apply.
In particular:

- Member names (`finish_reason`, `reasoning_content`,
  `factset_configuration`, `conversation_id`,
  `return_raw_link_context`, `factset_context`) are snake_case,
  matching OpenAI's own field naming (and, for
  `reasoning_content`, DeepSeek's), rather than FactSet's
  camelCase convention.
- Error responses are plain text, matching this service's shared
  error-handling framework, rather than FactSet's structured
  `{\"errors\": [...]}` JSON envelope.


This Python package is automatically generated by the [OpenAPI Generator](https://openapi-generator.tech) project:

- API version: 0.2.0
- SDK version: 0.0.1
- Build package: org.openapitools.codegen.languages.PythonClientCodegen

For more information, please visit [https://developer.factset.com/contact](https://developer.factset.com/contact)

## Requirements

* Python >= 3.7

## Installation

### Poetry

```shell
poetry add fds.sdk.utils fds.sdk.Conversational==0.0.1
```

### pip

```shell
pip install fds.sdk.utils fds.sdk.Conversational==0.0.1
```

## Usage

1. [Generate authentication credentials](../../../../README.md#authentication).
2. Setup Python environment.
   1. Install and activate python 3.10+. If you're using [pyenv](https://github.com/pyenv/pyenv):

      ```sh
      pyenv install 3.10.0
      pyenv shell 3.10.0
      ```

   2. (optional) [Install poetry](https://python-poetry.org/docs/#installation).
3. [Install dependencies](#installation).
4. Run the following:

> [!IMPORTANT]
> The parameter variables defined below are just examples and may potentially contain non valid values. Please replace them with valid values.

### Example Code

```python
from fds.sdk.utils.authentication import ConfidentialClient

import fds.sdk.Conversational
from fds.sdk.Conversational.api import chat_api
from fds.sdk.Conversational.models import *
from dateutil.parser import parse as dateutil_parser
from pprint import pprint

# See configuration.py for a list of all supported configuration parameters.

# Examples for each supported authentication method are below,
# choose one that satisfies your use case.

# (Preferred) OAuth 2.0: FactSetOAuth2
# See https://github.com/FactSet/enterprise-sdk#oauth-20
# for information on how to create the app-config.json file
#
# The confidential client instance should be reused in production environments.
# See https://github.com/FactSet/enterprise-sdk-utils-python#authentication
# for more information on using the ConfidentialClient class
configuration = fds.sdk.Conversational.Configuration(
    fds_oauth_client=ConfidentialClient('/path/to/app-config.json')
)

# Basic authentication: FactSetApiKey
# See https://github.com/FactSet/enterprise-sdk#api-key
# for information how to create an API key
# configuration = fds.sdk.Conversational.Configuration(
#     username='USERNAME-SERIAL',
#     password='API-KEY'
# )

# Enter a context with an instance of the API client
with fds.sdk.Conversational.ApiClient(configuration) as api_client:
    # Create an instance of the API class
    api_instance = chat_api.ChatApi(api_client)
    chat_completion_request = ChatCompletionRequest(
        model="fds-answer-2.0",
        stream=True,
        messages=[
            ChatMessage(
                role="system",
                content="content_example",
            ),
        ],
        factset_configuration=FactSetConfiguration(
            conversation_id="conversation_id_example",
            return_raw_link_context=False,
        ),
    ) # ChatCompletionRequest | 
    consumer = "developer-portal" # str | A short name identifying which application or integration is making this call, used for FactSet's own usage tracking. If omitted, usage is tracked under a generic default value rather than the request being rejected. (optional)

    try:
        # Ask a question and stream back the answer (OpenAI-compatible)
        # example passing only required values which don't have defaults set
        # and optional values
        api_response = api_instance.chat_completion(chat_completion_request, consumer=consumer)

        pprint(api_response)
    except fds.sdk.Conversational.ApiException as e:
        print("Exception when calling ChatApi->chat_completion: %s\n" % e)

    # # Get response, http status code and response headers
    # try:
    #     # Ask a question and stream back the answer (OpenAI-compatible)
    #     api_response, http_status_code, response_headers = api_instance.chat_completion_with_http_info(chat_completion_request, consumer=consumer)


    #     pprint(api_response)
    #     pprint(http_status_code)
    #     pprint(response_headers)
    # except fds.sdk.Conversational.ApiException as e:
    #     print("Exception when calling ChatApi->chat_completion: %s\n" % e)

    # # Get response asynchronous
    # try:
    #     # Ask a question and stream back the answer (OpenAI-compatible)
    #     async_result = api_instance.chat_completion_async(chat_completion_request, consumer=consumer)
    #     api_response = async_result.get()


    #     pprint(api_response)
    # except fds.sdk.Conversational.ApiException as e:
    #     print("Exception when calling ChatApi->chat_completion: %s\n" % e)

    # # Get response, http status code and response headers asynchronous
    # try:
    #     # Ask a question and stream back the answer (OpenAI-compatible)
    #     async_result = api_instance.chat_completion_with_http_info_async(chat_completion_request, consumer=consumer)
    #     api_response, http_status_code, response_headers = async_result.get()


    #     pprint(api_response)
    #     pprint(http_status_code)
    #     pprint(response_headers)
    # except fds.sdk.Conversational.ApiException as e:
    #     print("Exception when calling ChatApi->chat_completion: %s\n" % e)

```

### Using Pandas

To convert an API response to a Pandas DataFrame, it is necessary to transform it first to a dictionary.
```python
import pandas as pd

response_dict = api_response.to_dict()['data']

simple_json_response = pd.DataFrame(response_dict)
nested_json_response = pd.json_normalize(response_dict)
```

### Debugging

The SDK uses the standard library [`logging`](https://docs.python.org/3/library/logging.html#module-logging) module.

Setting `debug` to `True` on an instance of the `Configuration` class sets the log-level of related packages to `DEBUG`
and enables additional logging in Pythons [HTTP Client](https://docs.python.org/3/library/http.client.html).

**Note**: This prints out sensitive information (e.g. the full request and response). Use with care.

```python
import logging
import fds.sdk.Conversational

logging.basicConfig(level=logging.DEBUG)

configuration = fds.sdk.Conversational.Configuration(...)
configuration.debug = True
```

### Configure a Proxy

You can pass proxy settings to the Configuration class:

* `proxy`: The URL of the proxy to use.
* `proxy_headers`: a dictionary to pass additional headers to the proxy (e.g. `Proxy-Authorization`).

```python
import fds.sdk.Conversational

configuration = fds.sdk.Conversational.Configuration(
    # ...
    proxy="http://secret:password@localhost:5050",
    proxy_headers={
        "Custom-Proxy-Header": "Custom-Proxy-Header-Value"
    }
)
```

### Custom SSL Certificate

TLS/SSL certificate verification can be configured with the following Configuration parameters:

* `ssl_ca_cert`: a path to the certificate to use for verification in `PEM` format.
* `verify_ssl`: setting this to `False` disables the verification of certificates.
  Disabling the verification is not recommended, but it might be useful during
  local development or testing.

```python
import fds.sdk.Conversational

configuration = fds.sdk.Conversational.Configuration(
    # ...
    ssl_ca_cert='/path/to/ca.pem'
)
```

### Request Retries

In case the request retry behaviour should be customized, it is possible to pass a `urllib3.Retry` object to the `retry` property of the Configuration.

```python
from urllib3 import Retry
import fds.sdk.Conversational

configuration = fds.sdk.Conversational.Configuration(
    # ...
)

configuration.retries = Retry(total=3, status_forcelist=[500, 502, 503, 504])
```


## Documentation for API Endpoints

All URIs are relative to *https://api.factset.com/conversational/v0*

Class | Method | HTTP request | Description
------------ | ------------- | ------------- | -------------
*ChatApi* | [**chat_completion**](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ChatApi.md#chat_completion) | **POST** /chat/completions | Ask a question and stream back the answer (OpenAI-compatible)
*ChatApi* | [**list_models**](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ChatApi.md#list_models) | **GET** /models | List the models you can use (OpenAI-compatible)


## Documentation For Models

 - [ChatCompletionChunk](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ChatCompletionChunk.md)
 - [ChatCompletionRequest](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ChatCompletionRequest.md)
 - [ChatMessage](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ChatMessage.md)
 - [FactSetConfiguration](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/FactSetConfiguration.md)
 - [Model](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/Model.md)
 - [ModelList](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ModelList.md)
 - [ResponseChoices](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ResponseChoices.md)
 - [ResponseChoicesDelta](https://github.com/FactSet/enterprise-sdk/tree/main/code/python/Conversational/v0/docs/ResponseChoicesDelta.md)


## Documentation For Authorization


## FactSetApiKey

- **Type**: HTTP basic authentication


## FactSetOAuth2

- **Type**: OAuth
- **Flow**: application
- **Authorization URL**: 
- **Scopes**: N/A


## Notes for Large OpenAPI documents
If the OpenAPI document is large, imports in fds.sdk.Conversational.apis and fds.sdk.Conversational.models may fail with a
RecursionError indicating the maximum recursion limit has been exceeded. In that case, there are a couple of solutions:

Solution 1:
Use specific imports for apis and models like:
- `from fds.sdk.Conversational.api.default_api import DefaultApi`
- `from fds.sdk.Conversational.model.pet import Pet`

Solution 2:
Before importing the package, adjust the maximum recursion limit as shown below:
```
import sys
sys.setrecursionlimit(1500)
import fds.sdk.Conversational
from fds.sdk.Conversational.apis import *
from fds.sdk.Conversational.models import *
```

## Contributing

Please refer to the [contributing guide](../../../../CONTRIBUTING.md).

## Copyright

Copyright 2026 FactSet Research Systems Inc

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

