Image Extraction

import instructor
from openai import OpenAI
from pydantic import BaseModel, Field
from typing import List, Optional

client = instructor.from_openai(OpenAI())

class Product(BaseModel):
    name: str = Field(description="Product name")
    price: float = Field(description="Product price in USD")
    description: str = Field(description="Brief product description")
    features: List[str] = Field(description="Key product features")
    brand: Optional[str] = Field(None, description="Brand name if visible")

def extract_product_info(image_path_or_url: str) -> Product:
    return client.chat.completions.create(
        model="gpt-4-vision-preview",
        response_model=Product,
        messages=[
            {
                "role": "user",
                "content": [
                    "Extract detailed product information from this image:",
                    image_path_or_url
                ]
            }
        ],
        autodetect_images=True  # Automatically handle the image
    )

product = extract_product_info("path/to/product_image.jpg")
print(f"Product: {product.name} (${product.price:.2f})")
print(f"Description: {product.description}")
print(f"Features: {', '.join(product.features)}")
if product.brand:
    print(f"Brand: {product.brand}")

from typing import List, Optional
from pydantic import BaseModel, Field
import instructor
from openai import OpenAI

client = instructor.from_openai(OpenAI())

class Ingredient(BaseModel):
    name: str = Field(description="Ingredient name")
    quantity: str = Field(description="Amount needed, including units")
    optional: bool = Field(description="Whether this ingredient is optional")

class Step(BaseModel):
    instruction: str = Field(description="Cooking instruction")
    time_minutes: Optional[int] = Field(None, description="Time required for this step in minutes")

class Recipe(BaseModel):
    title: str = Field(description="Recipe title")
    servings: int = Field(description="Number of servings")
    prep_time_minutes: int = Field(description="Preparation time in minutes")
    cook_time_minutes: int = Field(description="Cooking time in minutes")
    ingredients: List[Ingredient] = Field(description="List of ingredients")
    steps: List[Step] = Field(description="Cooking steps in order")
    difficulty: str = Field(description="Recipe difficulty (easy, medium, hard)")

def extract_recipe(image_path: str) -> Recipe:
    # Create an Image object
    image = instructor.Image.from_path(image_path)

    return client.chat.completions.create(
        model="gpt-4-vision-preview",
        response_model=Recipe,
        messages=[
            {
                "role": "system",
                "content": "Extract complete recipe information from the provided image."
            },
            {
                "role": "user",
                "content": [
                    "Please extract the detailed recipe information from this image:",
                    image
                ]
            }
        ]
    )

$ pip install instructor pydantic

$ python image-to-structured-data.py