Python client keeps getting "We could not parse the JSON body of your request."

API
, ,
1

I’m trying to use chat completion API using openai python lib, but keep getting the above error. Full error is:

Traceback (most recent call last):
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/Src/SingleUserBot/main.py", line 4, in <module>
    res = simple_textgen_pipeline_obj.run(question=q)
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/Src/SingleUserBot/textgen_pipeline.py", line 90, in run
    model_response = self.client.chat.completions.create(
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/.venv/lib/python3.10/site-packages/openai/_utils/_utils.py", line 301, in wrapper
    return func(*args, **kwargs)
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/.venv/lib/python3.10/site-packages/openai/resources/chat/completions.py", line 598, in create
    return self._post(
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/.venv/lib/python3.10/site-packages/openai/_base_client.py", line 1098, in post
    return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/.venv/lib/python3.10/site-packages/openai/_base_client.py", line 858, in request
    return self._request(
  File "/mnt/c/Users/taran/Documents/work/aras-chatbot/chatbot-for-aras/.venv/lib/python3.10/site-packages/openai/_base_client.py", line 910, in _request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON. If you have trouble figuring out how to fix this, please contact us through our help center at help.openai.com.)", 'type': 'invalid_request_error', 'param': None, 'code': None}}

So I’m making request like below:

model_response = self.client.chat.completions.create(
            messages = [{"role": "system", "content": f"{self.system_prompt}"},
                        {"role" : "user", "content" : self.prompt.gen_template()}], 
            temperature = self.model.temperature,
            model = self.model.model_name
        )

The model I’m using is “gpt-3.5-turbo” and the client looks below:

chat_model_client = openai.OpenAI(api_key=openai_api_key, 
                                  timeout=httpx.Timeout(chat_model_req_res_timeout, 
                                                        connect=chat_model_connection_timeout_sec))
1 Like
2

generated prompt looks like below:

context: Built on Windows Authentication Framework, the Aras Innovator client is accessed directly in a browser. The Aras Innovator login can be configured to accommodate your corporate authentication requirements. The login screen can also be personalized to display your corporate branding and logo. Built on Windows Authentication Framework, the Aras Innovator client is accessed directly in a browser. The Aras Innovator login can be configured to accommodate your corporate authentication requirements. The login screen can also be personalized to display your corporate branding and logo. The login screen requires you to enter your user ID and password and contains a startup graphic which is configurable by a System Administrator to display a custom company logo. The first time you log in, you must also pick an Authentication Framework from the dropdown list which is then remembered for future logins. The Windows Authentication Framework allows for specialized user authentication requirements like single sign-on, authorization control, and auditing. To log on to Aras Innovator follow the following steps. 1. Open a browser and enter the server URL provided. The URL will be in the following format: 2. On the Login screen, note the Aras Innovator Version and Build Number. 3. Select the appropriate database. 4. If the Authentication Framework option is displayed, leave the “Login with” setting as Aras Innovator. 5. Enter the Username and Password (passwords are case sensitive). 6. Then click on the Login Button. You will be logged in and the Aras Innovator main window will be displayed. Closing the browser window does not close your Aras Innovator session. To end your session, click the User Menu and select Logout. Also Closing the browser window does not close your Aras Innovator session. To end your session, click the User Menu. The user menu dropdown will open from there click and select Logout. The user will be logged out of ARAS Innovator. From the user menu you can access preferences for searches and discussions, create and run reports, view help, as well as log out of the system. Investigate User Settings 1. Click the User Menu to see a list of options. 2. Navigate to select About to see the current version and build of the Aras Innovator Server. 3. Select Session to ensure you are logged into the correct server and database number assigned to you for this class. 4. If the database number does not match the database assigned by the instructor, please select Logout. 5. Select the database assigned to you from the pulldown menu and login again using credentials \npreviously specified.\n1. Learn about the Aras Platform \n2. Understand Aras Model-based Technology \n3. Understand Aras Single Platform Architecture \n4. Aras Applications and Features \n5. Review Course Objectives \n6. Review Course AgendaIn this unit, you learn how to navigate in the Aras Innovator browser client interface. You also learn how to search for items using various query types.When the main window opens you can access all Aras Innovator functions. This window is customizable by an administrator. 1. The Aras Innovator client may be customized by an Administrator to display HTML content as the workspace background or home screen. You can also define the start page for users; for example, you can configure Aras Innovator to display the user’s individual In Basket on startup. 2. The Navigation button provides access to the Navigation panel. 3. The Enterprise Search field enables search across items in the database, as configured by the administrator. 4. Enterprise Search can be configured to include any custom ItemTypes, their related items, files, and properties. 5. Users can find items as long as they have the access privileges and can filter by properties to fine tune their search. 6. See the Aras Innovator Enterprise Search User Guide for more information about using Enterprise Search, and the Aras Innovator Enterprise Search Administrators Guide for instructions on configuring Enterprise Search. 7. Split screen enables you to view item tabs and search grids side-by-side 8. The Notifications button displays alerts to users including Aras Innovator updates and system maintenance.9. The User Menu provides access to user preferences, reports, and custom actions. When the main window opens you can access all Aras Innovator functions. In ARAS Innovator, User Menu is a drop-down, which allows users to set user preferences. The user menu is available at the top right corner of the navigation Header of the ARAS main window. It has users' initials on it. On clicking it, a dropdown will open, and user can set preferences here. It also allows users to set Actions, Reports, Admin, Licenses, and Activate features. It also has tabs like Help, Session, About, and Logout. The user menu is available at the top right corner of the navigation Header of the ARAS main window. Navigate to the top right corner and click the rightmost button to open the user menu. The user menu button has the current user’s initials on it. On clicking the button, a dropdown will open which has user preference settings in it. \nARAS Innovator is a product lifecycle management (PLM) platform, and it includes a centralized database to store and manage information related to product development and lifecycle processes. The database in ARAS Innovator serves as a repository for various types of data, including product data, documents, workflows, change management information, and more. The database is a collection of preloaded training (e.g. InnovatorSolution). We can create a database through SSMS by taking an instance of the present database. In the ARAS Innovator Main window, on the top left, you will find the user menu with the initials of the user. On clicking that user menu, a dropdown will open. In this dropdown, click and select session. A new window will open, which will have information like, login name, database, and server that are in use in the current session. Once you have accessed the ARAS Innovator web interface, you will be prompted to enter your login credentials. These credentials usually include a username and password. If your organization has multiple databases or instances of ARAS Innovator, you may need to select the specific database you want to access here on the login page itself. After entering your credentials and selecting the appropriate database, click the “Login” button. The Aras Innovator Main Window is divided into two sections: Header and Workspace Pane. Header Contains top-level controls to navigate and execute commands. A configurable Workspace Pane to display a custom “home screen”. An administrator can specify an HTML page as the background of the workspace. This window is customizable by an administrator. The Aras Innovator client may be customized by an Administrator to display HTML content as the workspace background or home screen. You can also define the start page for users; for example, you can configure Aras Innovator to display the user’s individual In Basket on startup. The Navigation button provides access to the Navigation panel. The Enterprise Search field enables search across items in the database, as configured by the administrator. Enterprise Search can be configured to include any custom ItemTypes, their related items, files, and properties. Users can find items as long as they have access privileges and can filter by properties to fine-tune their search. Split screen enables you to view item tabs and search grids side-by-side. The Notifications button displays alerts to users including Aras Innovator updates and system maintenance. The User Menu provides access to user preferences, reports, and custom actions. Aras The Aras Innovator Header appears at the top of the main window. The Header contains elements like the Navigation button, Enterprise Search, Split Screen, Notification, and User Menu. The navigation button (or “Nav Button”) provides access to Innovator items. The Enterprise Search text box allows for global searches across ItemTypes. The Enterprise Search feature is configurable and conforms to user permissions. Enterprise Search allows you to perform keyword or full-text searches across various ItemTypes such as Parts, Documents, ECOs, as well as related items and files. You only see the results based on your access privileges and you can filter by property values to fine-tune a search. Splitscreen enables you to view item tabs and search grids side-by-side. split screen button can be found beside the Notification button on the top right corner of the navigation bar. Navigate to the top right corner and click the split screen button which can be used to work on two different processes. The Notification button shows messages received from the System Administrator. User Notifications appear as alerts to a user while working in an Innovator session. These notifications alert a user about an upcoming event (e.g. system maintenance, updated feature, etc.). The notification “bell” icon is highlighted when a notification is received from the System Administrator. The User Menu button displays the User Menu and contains the user’s initials (or a photo if configured by the administrator). User Menu allows users to set preferences, run reports, and execute actions. User Menu is a drop-down, which allows users to set user preferences. The user menu is available at the Top right corner of the navigation Header of the ARAS main window. It also allows users to set Actions, Reports, Admin, Licenses, and Activate features. It also has tabs like Help, Session, About, and Logout. From the user menu, you can access preferences for searches and discussions, create and run reports, view help, as well as log out of the system. To Investigate User Settings, you must Click the User Menu to see a list of options. After that Navigate to select About to see the current version and build of the Aras Innovator Server. From there Select Session to ensure you are logged into the correct server and database number assigned to you for this class. If the database number does not match the database assigned by the instructor, please select Logout. From the Login page Select the database assigned to you from the pulldown menu and log in again using the credentials previously specified. Closing the browser window does not close a user session. You must select Logout from the User Menu to log out of Innovator. The User Menu Actions provides menu selections to initiate various activities based on the user’s roles. As a System Administrator, there are several generic operations available: like Hide Configurable InBasket, Re-index Enterprise Search, Choose CM Options, and Configure Visual Collaboration Notifications. An end-user (non-administrator) will see the following generic Actions: like Reset user preferences and Update time zone data. Reports in the user menu, Initiates the Self-Service Reporting tool. Aras Innovator has integrated the Izenda Reporting Engine for you to report on data in the Aras database. User menu Admin allows the user to set the actions based on the role like Delete user preferences, API Reference (.NET), API Reference (JAVASCRIPT), clearing client Metadata Cache, Rebuild Kyed Names, and Debug. User Menu Licenses Provides administrative access to utilities used to license and customize the Innovator system. User Menu Help Displays Just Ask Innovator, a context-sensitive help system to help while working with Aras Innovator. User Menu Session Displays information about your current session including your login name, database, and server URL. User Menu About Provides links to additional resources, product versions, and build information. We can get Additional Information using the User Menu About in Innovator.\n\n

3

Hi! Welcome to the forums!

are you sure self.prompt.gen_template() is a string, and not some other object?

1 Like
4

Ensure that all referenced variables (like self.system_prompt, self.prompt.gen_template(), self.model.temperature, and self.model.model_name) are properly defined and hold the expected values. Also, make sure that your JSON is properly formatted…

5

Run your requests and responses in something like Jupyter or Colab where you can print every object and see what it actually is. IMHO it’s easier to iterate over an issue inspecting each input/output that way.

1 Like
6

It’s a string I double checked!

7

All strings are not valid json (even though all json is strings) - is it valid json?

1 Like
8

it certainly works on my machine :confused:

what you can do is wrap your generated prompt in a formatted string tag like you wrapped your system prompt:

            messages = [
                        {"role": "system", "content": f"{self.system_prompt}"},
                        {"role" : "user", "content" : f"{self.prompt.gen_template()}"}
            ],

although I wouldn’t recommend it. it will likely make your code work, but you may not get the results you’d expect.

9

all the “self” indicates that you have a class object that would need init methods or properties set in order for those many variables to be instantiated correctly. A return value of another function right in the call, that would be better expressed separately for debugging and recording values.

You can ultimately call with just client.chat.completions.create(**parameters_dict) and ensure that the params are complete as a dictionary of python data structures before any call is made.

1 Like
10

When I got this error I solved it by:

  1. removing all quotation marks from the string, and
  2. encoding the text as utf-8 explicitly.