add handling for validation errors to return 400 and message

[dtam]

Pull Request Type

• Breaking Change
• Feature
• Bug Fix
• Non-bug Patch (dependency update, non-production code, etc.)

Link to Notion Task or Github Issue

Summary of Feature(s)

config.py

regexG = Guard().use(RegexMatch(regex="^[A-Z][a-z]*$", on_fail="exception"))
regexG.name = "reg_ex"
regexG.configure(num_reasks=0)

post to guards/reg_ex/validate with any prompt that returns more than 1 word will now return a 400 with the correct string
or use the openai sdk with the code below

import os
os.environ["OPENAI_LOG"] = "debug"

import openai

# OpenAI compatible Guardrails API Guard
openai.base_url = "http://localhost:8000/guards/reg_ex/openai/v1/"

openai.api_key = os.getenv("OPENAI_API_KEY") or 'some key'

messages = [
        {
            "role": "user",
            "content": "return a two words to describe a good day",
        },
    ]

completion = openai.chat.completions.create(
    model="gpt-4o",
    messages=messages,
)


print("===output===", completion.choices[0].message.content)

Summary of Bug Fix(es)

Previous Behaviour

Description of the bug and it's impact

New Behaviour

Description of the bug fix and it's impact

Other details

Dependencies

[zsimjee]

lgtm other than not using .last