OpenAI JSON response + pydantic example

openai_json_pydantic_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pydantic import BaseModel\n",
    "from openai import OpenAI\n",
    "import json\n",
    "from typing import Literal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## OpenAI client set up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "openai_key = \"YOUR_OPENAI_KEY\"\n",
    "openai_client = OpenAI(api_key=openai_key)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create prompt to analyze product review sentiment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "def analyze_review_prompt(review: str) -> str:\n",
    "    return (\n",
    "        f\"Analyse review sentiment (positive, negative or neutral) \"\n",
    "        \"and name of the product (if not available return empty string) \"\n",
    "        f\"for the following review:\\n```\\n{review}\\n```\\n\"\n",
    "        \"Response format:\\n\"\n",
    "        \"json object with the following fields:\\n\"\n",
    "        \"- sentiment: string, sentiment of the review (positive, negative or neutral)\\n\"\n",
    "        \"- product_name: string, name of the product (if not available return empty string)\"\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Pydantic class to validate JSON response"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ReviewAnalysis(BaseModel):\n",
    "    sentiment: Literal[\"positive\", \"negative\", \"neutral\"]\n",
    "    product_name: str"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Invoke OpenAI API"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp = openai_client.chat.completions.create(\n",
    "    response_format={\"type\": \"json_object\"},\n",
    "    messages=[{\"role\": \"user\", \"content\": analyze_review_prompt(\"Samsung Galaxy S21 is the best phone ever!\")}],\n",
    "    model=\"gpt-3.5-turbo\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Parse and validate the response"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sentiment='positive' product_name='Samsung Galaxy S21'\n"
     ]
    }
   ],
   "source": [
    "parsed_json = json.loads(resp.choices[0].message.content)\n",
    "validated_response = ReviewAnalysis.model_validate(parsed_json)\n",
    "print(validated_response)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}