Save a list of available puppies on GoodDog.com in specific breeds to CSV - GoodDog.com's UX is based on navigating from breed -> breeder -> puppies. There are no options to browse available puppies across breeds or even to identify which breeders may soon have another litter. Good Dog expects its users to dig through its listings, page by page,…

good_dog_scraper.py

import csv
import logging
import requests
import time

logging.basicConfig(level=logging.NOTSET)
logger = logging.getLogger(__name__)

# Seconds between requests
RATE_LIMIT = 2

class GooddogScraper:
    """
    GoodDog.com's UX is based on navigating from breed -> breeder -> puppies. There
        are no options to browse available puppies across breeds or even to identify
        which breeders may soon have another litter. Good Dog expects its users to
        dig through its listings, page by page, even though their API returns a far
        more useful and filterable dataset.

    This scraper accepts a list of breeds and a rate limit, then it gathers all of the 
        currently available puppies of those respective breeds on Good Dog and outputs
        to a CSV.

    Also - Hey Good Dog Engineers; you're gonna be tempted to limit requests to your
        API in a way that doesn't let me do this. Consider talking to your UX folx
        instead.
    """

    API_BASE = "https://www.gooddog.com/api"
    DEFAULT_HEADERS = {
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.90 Safari/537.36',
    }

    def __init__(self, postal_code, breeds, rate_limit):
        self.postal_code = postal_code
        self.breeds = breeds
        self.rate_limit = rate_limit

    def scrape(self):
        logger.info(f"Scraping {len(self.breeds)} breeds")
        puppies = []
        for breed in self.breeds:
            time.sleep(self.rate_limit)
            logger.info(f"    Requesting info for {breed}")
            breed_puppies = self._get_puppies(breed)
            logger.info(f"    {len(breed_puppies)} puppies available")
            puppies.extend(breed_puppies)
        with open("puppies.csv", "w") as csvfile:
            logger.info(f"    Writing {len(puppies)} to csv...")
            fieldnames = puppies[0].keys()
            writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
            writer.writeheader()
            writer.writerows(puppies)
            logger.info(f"    {len(puppies)} puppies saved to CSV")


    def _get_puppies(self, breed):
        url = f"{self.API_BASE}/breeds/{breed}/breeder_profiles.json?city=&distance_in_miles=100&non_standard=true&page=1&per_page=500&postal_code={self.postal_code}&state="
        logger.info(f"        Requesting {url}...")
        response = requests.get(url, headers=self.DEFAULT_HEADERS)
        logger.info(f"            Response code: {response.status_code}")
        breeders = response.json().get('data', {})
        puppies = []
        for breeder in breeders:
            for puppy in breeder.get('puppies', []):
                delivery_methods = [m.get('name') for m in breeder.get('delivery_methods', [])]
                delivery_methods_string = ', '.join(delivery_methods)
                url = f"https://www.gooddog.com{puppy.get('url')}"
                puppy = {
                    'breed': breed,
                    'name': puppy.get('name'),
                    'url': url,
                    'distance': breeder.get('distance_in_miles'),
                    'health_tier': breeder.get('health_tier', {}).get('title'),
                    'breeder_name': breeder.get('name'),
                    'breeder_wait': breeder.get('estimated_wait_description'),
                    'breeder_delivery_methods': delivery_methods_string,
                }
                logger.info(f"         Found puppy: {puppy}")
                puppies.append(puppy)
        return puppies

    
breeds = [
    'alaskan-klee-kai',
    'australian-terrier',
    'biewer-terrier',
    'cairn-terrier',
    'cavapoo',
    'glen-of-imaal-terrier',
    'havanese',
    'havapoo',
    'italian-greyhound',
    'jack-russell-terrier',
    'maltipoo',
    'mi-ki',
    'miniature-pinscher',
    'miniature-schnauzer',
    'morkie',
    'norfolk-terrier',
    'norwich-terrier',
    'papillon',
    'parson-russell-terrier',
    'pomeranian',
    'pomsky',
    'rat-terrier',
    'russell-terrier',
    'schipperke',
    'schnoodle',
    'scottish-terrier',
    'shiba-inu',
    'shichon',
    'shihpoo',
    'shorkie',
    'soft-coated-wheaten-terrier',
    'swedish-vallhund',
    'tibetan-terrier',
    'toy-fox-terrier',
    'welsh-terrier',
    'west-highland-white-terrier',
    'wire-fox-terrier',
    'yorkipoo',
    'yorkshire-terrier',
    'silky-terrier'
]
scraper = GooddogScraper("98101", breeds, RATE_LIMIT)
scraper.scrape()