How to Scrape Zillow
When you're searching for a house in the US, Zillow is one of the most prominent platforms to explore. It provides comprehensive listings for properties, including detailed descriptions, photos, and even virtual tours. However, Zillow is also known for implementing robust anti-scraping measures, making it challenging to extract data using traditional methods. In this article, we’ll explore how to overcome these challenges using Python and Selenium, which allows us to interact with websites in a way that mimics human behavior, making it more resilient against anti-bot measures. Feel free to choose the language and setup that best fits your development environment.Let’s dive into the step-by-step process and learn how to effectively scrape Zillow using Python.
Feel free to change any of the following from
When we look at a specific home, we get a pop-up about the home information, but no worries, we still get a URL that we can work with.For the house below, our URL isBuilding these URLs from scratch would be pretty difficult, luckily for us, we can scrape them on our initial crawl.
Here is a look at some the HTML we want to parse for an individual property page.
If we want page 3, we would end the url with
Create a New Virtual EnvironmentActivate the EnvironmentInstall Our Dependencies
In our parsing function, we do the following things.
Here is our updated code. We added a
Here is our Here is our code now that it's been fully updated for storage.
Pay close attention to the arguments we pass into
Arguments we pass into the ScrapeOps API:
We'll change 
Our crawler parsed the results at approximately one second per page. This is great!
Things you need to pay attention to here:
Here is our fully updated code.Our code now reads the CSV file generated by the crawler and runs
In our full code below, instead of printing the data, we instantiate a
Quite similar to when we added concurrency earlier,
Here is our finalized code, ready for production.
The full code ran and exited in 42.8 seconds.
In total, there were 45 properties in the CSV file generated from the crawler. If you recall, the crawler took roughly 5 seconds. 42.8 - 5 = 37.8 seconds spent parsing properties. 37.8 seconds / 45 properties = 0.84 seconds per property.
How to Scrape Zillow With Requests and BeautifulSoup
- TLDR How to Scrape Zillow
- How To Architect Our Scraper
- Understanding How To Scrape zillow
- Setting Up Our Zillow Scraper
- Build A Zillow Search Crawler
- Build A Zillow Scraper
- Legal and Ethical Considerations
- Conclusion
- More Cool Articles
Need help scraping the web?
Then check out ScrapeOps, the complete toolkit for web scraping.
TLDR - How to Scrape Zillow
Need to scrape Zillow? We've got you covered.- Create a new project folder.
- Inside that project folder, add a
config.jsonfile with your API key and then add this script. - Run it and you're good to go!
"pr" (Puerto Rico), it spits out a file called pr.csv.It then reads this file and creates an individual report on each house from pr.csv.import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) @dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: scrapeops_proxy_url = get_scrapeops_url(url, location=location) response = requests.get(scrapeops_proxy_url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] != "BreadcrumbList": search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) def process_property(row, location, retries=3): url = row["url"] tries = 0 success = False while tries <= retries and not success: response = requests.get(get_scrapeops_url(url, location=location)) try: if response.status_code == 200: logger.info(f"Status: {response.status_code}") soup = BeautifulSoup(response.text, "html.parser") price_holder = soup.select_one("span[data-testid='price']") price = int(price_holder.text.replace("$", "").replace(",", "")) info_holders = soup.select("dt") time_listed = info_holders[0].text views = int(info_holders[2].text.replace(",", "")) saves = info_holders[4].text property_pipeline = DataPipeline(csv_filename=f"{row['name']}.csv") property_data = PropertyData( name=row["name"], price=price, time_on_zillow=time_listed, views=views, saves=saves ) property_pipeline.add_data(property_data) property_pipeline.close_pipeline() success = True else: logger.warning(f"Failed Response: {response.status_code}") raise Exception(f"Failed Request, status code: {response.status_code}") except Exception as e: logger.error(f"Exception thrown: {e}") logger.warning(f"Failed to process page: {row['url']}") logger.warning(f"Retries left: {retries-tries}") tries += 1 if not success: raise Exception(f"Max Retries exceeded: {retries}") else: logger.info(f"Successfully parsed: {row['url']}") def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( process_property, reader, [location] * len(reader), [retries] * len(reader) ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
main:MAX_THREADS: Determines the maximum number of threads used for concurrent scraping and processing.MAX_RETRIES: Sets the maximum number of retries for each request in case of failure (e.g., network issues, server errors).PAGES: Specifies the number of pages to scrape for each keyword. Each page contains multiple property listings.LOCATION: Defines the geographical location for the scraping. This parameter is used to adjust the proxy location to simulate requests from a specific country.keyword_list: A list of keywords representing different geographical areas or search terms on Zillow. Each keyword triggers a separate scraping job. ("pr"is Puerto Ricto, if you want to do Michigan, add"mi")
How To How To Architect Our Zillow Scraper
Like many other scrapers from our "How To Scrape" series, our Zillow project will consist of two scrapers:- a crawler: Our crawler will perform a search for houses in a specific area and spit out a CSV report on all the properties it finds.
- a parser: The scraper will then read the CSV file, and run an individual scrape on each property from the list.
- Parsing to extract valuable data from webpages.
- Pagination to get our results in batches.
- Data Storage to generate CSV reports from that data.
- Concurrency to handle all the above steps on multiple pages simultaneously.
- Proxy Integration to get past anti-bot systems and anything else that might get in the way of our scraper.
Understanding How To Scrape Zillow
Before we dive head first into coding, we need to get a better understanding of Zillow at a high level. In the following sections, we're going to discuss how to request pages, how to extract data, how to control pagination, and how to handle geolocation.Step 1: How To Request Zillow Pages
We can't get very far if we can't request pages. We can request pages on Zillow with a simple GET request. A GET request does exactly what it sounds like, it gets information. When you lookup a site in your browser, you're actually performing a GET request.Take a look at the image below, specifically our address bar. Here is our URL:https://www.zillow.com/pr/2_p/
pr is our location.When we look at a specific home, we get a pop-up about the home information, but no worries, we still get a URL that we can work with.For the house below, our URL ishttps://www.zillow.com/homedetails/459-Carr-Km-7-2-Int-Bo-Arenales-Aguadilla-PR-00603/363559698_zpid/
Step 2: How To Extract Data From Zillow Results and Pages
Extracting data from Zillow can be a bit tricky. For search results, our data is actually embedded in a JSON blob. For individual property pages, it's nested within the HTML elements. Let's take a closer look at exactly which data we'll be extracting.Here is the search page and the JSON blob inside it.Here is a look at some the HTML we want to parse for an individual property page.Step 3: How To Control Pagination
Pagination might seem a bit cryptic, but take a closer look at our URL from before:https://www.zillow.com/pr/2_p/
2_p actually denotes our page number, 2.If we want to search for page 1, our URL ishttps://www.zillow.com/pr/1_p/
3_p.Step 4: Geolocated Data
For geolocated data, we'll be using the ScrapeOps Proxy API and ourkeyword_list. In the keyword_list, we'll hold the locations we'd like to scrape.When interacting with the ScrapeOps API, we'll pass in a country param as well. country will not have any effect on our actual search results, but instead it will route us through a server in whichever country we specify.For instance, if we want to appear in the US, we'd pass us in as our country.Setting Up Our Zillow Scraper Project
Let's get started. You can run the following commands to get setup.Create a New Project Foldermkdir zillow-scraper cd zillow-scraper
python -m venv venvsource venv/bin/activatepip install requestspip install beautifulsoup4Build A Zillow Search Crawler
Step 1: Create Simple Search Data Parser
We'll start off by building a simple parser. The job of the parser is relatively straightforward. Our parser needs to perform a search and extract data from the search results. This is the bedrock of everything else we'll add into our script.While the script below adds some basic structure (logging, error handling, retry logic), what you should pay attention to is the parsing logic inside of ourscrape_search_results() function.Take a look at the script so far.import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) def scrape_search_results(keyword, location, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/" tries = 0 success = False while tries <= retries and not success: try: response = requests.get(url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] != "BreadcrumbList": search_data = { "name": json_data["name"], "property_type": json_data["@type"], "street_address": json_data["address"]["streetAddress"], "locality": json_data["address"]["addressLocality"], "region": json_data["address"]["addressRegion"], "postal_code": json_data["address"]["postalCode"], "url": json_data["url"] } print(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") scrape_search_results(keyword, LOCATION, retries=retries) logger.info(f"Crawl complete.")
- Find all of our JSON blobs with
script_tags = soup.select("script[type='application/ld+json']") - Iterate through the blobs.
- For each blob that does not have a
"@type"of"BreadcrumbList", we parse its data:nameproperty_typestreet_addresslocalityregionpostal_codeurl
Step 2: Add Pagination
As we discussed earlier, we can add pagination with the following string by constructin our URL like so:https://www.zillow.com/{keyword}/{page_number+1}_p/
- As you know,
keywordis the location we'd like to search. {page_number+1}_p, denotes our page number. We usepage_number+1because we'll be using Python'srange()function to create our page list.range()starts counting at zero and Zillow starts our pages at 1. So, we add 1 to our page when we pass it into the URL.
start_scrape() function to support the pagination we just added.def start_scrape(keyword, pages, location, max_threads=5, retries=3): for page in range(pages): scrape_search_results(keyword, location, page, retries=retries)
start_scrape() function to support multiple pages, but all in all, our code isn't all that different.import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) def scrape_search_results(keyword, location, page_number, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: response = requests.get(url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] != "BreadcrumbList": search_data = { "name": json_data["name"], "property_type": json_data["@type"], "street_address": json_data["address"]["streetAddress"], "locality": json_data["address"]["addressLocality"], "region": json_data["address"]["addressRegion"], "postal_code": json_data["address"]["postalCode"], "url": json_data["url"] } print(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, max_threads=5, retries=3): for page in range(pages): scrape_search_results(keyword, location, page, retries=retries) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") start_scrape(keyword, PAGES, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES) logger.info(f"Crawl complete.")
Step 3: Storing the Scraped Data
When we scrape our data, we need to store it. If we didn't, there wouldn't be a point in scraping it to begin with. When we store our data, we can review it and we can also allow other functions or programs to read it.This lays the groundwork for our scraper, which will read the stored CSV file and then lookup individual data about the properties from the CSV file.First, we need aSearchData class. This class simply holds data.@dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip())
DataPipeline. It takes in a dataclass (such as SearchData) and pipes it to a CSV file. This pipeline then filters out our duplicates and then saves the data to a CSV file.Additionally, our pipeline writes the file safely. If the CSV exists, we append it, otherwise the pipeline creates it.class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv()
import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: response = requests.get(url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] != "BreadcrumbList": search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): for page in range(pages): scrape_search_results(keyword, location, page, data_pipeline=data_pipeline, retries=retries) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.")
Step 4: Adding Concurrency
To add concurrency, we'll useThreadPoolExecutor. We'll also add a max_threads argument to start_scrape(). Take a look at the snippet below.def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages )
executor.map():scrape_search_resultsis the function we'd like to run on each thread.- All other arguments are the arguments that we pass into
scrape_search_results() - Each argument after the first gets passed in as an array.
import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: response = requests.get(url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] != "BreadcrumbList": search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.")
Step 5: Bypassing Anti-Bots
When we scrape the web, we tend to run into anti-bots. Anti-bots are software designed to detect malware and block it from accessing a site. Our scraper isn't malware, but it doesn't look human at all and anti-bots tend to flag this.Take a look at the snippet below, this unleashes the power of the ScrapeOps Proxy API.def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url
"api_key": our ScrapeOps API key."url": the url of the site we'd like to scrape."country": the country we'd like to be routed through."residential": a boolean value. If we set this toTrue, we're telling ScrapeOps to give us a residential IP address which decreases our likelihood of getting blocked.
import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: scrapeops_proxy_url = get_scrapeops_url(url, location=location) response = requests.get(scrapeops_proxy_url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") excluded_types = ["BreadcrumbList", "Event"] for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] not in excluded_types: search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 5 LOCATION = "us" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.")
Step 6: Production Run
We're now ready for a production run. Take a look at ourmain.if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 5 LOCATION = "us" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.")
PAGES to 5, and LOCATION to "us". Feel free to change any of these constants in the main to tweak your results.Here are our results.Our crawler parsed the results at approximately one second per page. This is great!Build A Zillow Scraper
Our crawler is now performing searches, parsing results and saving the data. Now, it's time to build the scraper. Our scraper is going to do the following:- Read a CSV file.
- Parse all properties from the CSV file.
- Save their data.
- Perform the actions above concurrently.
- Integrate with a proxy.
Step 1: Create Simple Business Data Parser
Like before, we'll start with a basic parsing function. The overall structure looks alot like our first parsing function, but there are some key differences.Mainly, instead of finding JSON data nested within the page, we're going to find our data from different elements on the page.Take a look at the function below.def process_property(row, location, retries=3): url = row["url"] tries = 0 success = False while tries <= retries and not success: response = requests.get(url) try: if response.status_code == 200: logger.info(f"Status: {response.status_code}") soup = BeautifulSoup(response.text, "html.parser") price_holder = soup.select_one("span[data-testid='price']") price = int(price_holder.text.replace("$", "").replace(",", "")) info_holders = soup.select("dt") time_listed = info_holders[0].text views = int(info_holders[2].text.replace(",", "")) saves = info_holders[4].text property_data = { "name": row["name"], "price": price, "time_on_zillow": time_listed, "views": views, "saves": saves } print(property_data) success = True else: logger.warning(f"Failed Response: {response.status_code}") raise Exception(f"Failed Request, status code: {response.status_code}") except Exception as e: logger.error(f"Exception thrown: {e}") logger.warning(f"Failed to process page: {row['url']}") logger.warning(f"Retries left: {retries-tries}") tries += 1 if not success: raise Exception(f"Max Retries exceeded: {retries}") else: logger.info(f"Successfully parsed: {row['url']}")
"span[data-testid='price']"is the CSS selector of ourprice_holder.int(price_holder.text.replace("$", "").replace(",", ""))gives us our actual price and converts it to an integer.- We find all of our
info_holderswithsoup.select("dt") - We pull
time_listed,views, andsavesfrom theinfo_holdersarray.
Step 2: Loading URLs To Scrape
In order to use our parsing function, we need to be able to feed urls into it.for each row in the file, we run process_property() on that row. Later, we'll add concurrency to this function just like we did eariler.def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) for row in reader: process_property(row, location, retries=retries)
import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: scrapeops_proxy_url = get_scrapeops_url(url, location=location) response = requests.get(scrapeops_proxy_url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") excluded_types = ["BreadcrumbList", "Event"] for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] not in excluded_types: search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) def process_property(row, location, retries=3): url = row["url"] tries = 0 success = False while tries <= retries and not success: response = requests.get(url) try: if response.status_code == 200: logger.info(f"Status: {response.status_code}") soup = BeautifulSoup(response.text, "html.parser") price_holder = soup.select_one("span[data-testid='price']") price = int(price_holder.text.replace("$", "").replace(",", "")) info_holders = soup.select("dt") time_listed = info_holders[0].text views = int(info_holders[2].text.replace(",", "")) saves = info_holders[4].text property_data = { "name": row["name"], "price": price, "time_on_zillow": time_listed, "views": views, "saves": saves } print(property_data) success = True else: logger.warning(f"Failed Response: {response.status_code}") raise Exception(f"Failed Request, status code: {response.status_code}") except Exception as e: logger.error(f"Exception thrown: {e}") logger.warning(f"Failed to process page: {row['url']}") logger.warning(f"Retries left: {retries-tries}") tries += 1 if not success: raise Exception(f"Max Retries exceeded: {retries}") else: logger.info(f"Successfully parsed: {row['url']}") def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) for row in reader: process_property(row, location, retries=retries) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
process_post() on each row from the file.Step 3: Storing the Scraped Data
As you already know at this point, we need to store the data we scrape. For the most part, we've already got everything in place to do this.We simply need to add aPropertyData class. This class will act much like the SearchData class from before and it will also get passed into a DataPipeline.Here is our SearchData class.@dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip())
PropertyData object and then pass it into a DataPipeline.import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) @dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: scrapeops_proxy_url = get_scrapeops_url(url, location=location) response = requests.get(scrapeops_proxy_url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") excluded_types = ["BreadcrumbList", "Event"] for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] not in excluded_types: search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) def process_property(row, location, retries=3): url = row["url"] tries = 0 success = False while tries <= retries and not success: response = requests.get(url) try: if response.status_code == 200: logger.info(f"Status: {response.status_code}") soup = BeautifulSoup(response.text, "html.parser") price_holder = soup.select_one("span[data-testid='price']") price = int(price_holder.text.replace("$", "").replace(",", "")) info_holders = soup.select("dt") time_listed = info_holders[0].text views = int(info_holders[2].text.replace(",", "")) saves = info_holders[4].text property_pipeline = DataPipeline(csv_filename=f"{row['name']}.csv") property_data = PropertyData( name=row["name"], price=price, time_on_zillow=time_listed, views=views, saves=saves ) property_pipeline.add_data(property_data) property_pipeline.close_pipeline() success = True else: logger.warning(f"Failed Response: {response.status_code}") raise Exception(f"Failed Request, status code: {response.status_code}") except Exception as e: logger.error(f"Exception thrown: {e}") logger.warning(f"Failed to process page: {row['url']}") logger.warning(f"Retries left: {retries-tries}") tries += 1 if not success: raise Exception(f"Max Retries exceeded: {retries}") else: logger.info(f"Successfully parsed: {row['url']}") def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) for row in reader: process_property(row, location, retries=retries) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
Step 4: Adding Concurrency
With a small refactor, we can now add concurrency. Once again, we'll replace ourfor loop with ThreadPoolExecutor. Take a look at the new function.def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( process_property, reader, [location] * len(reader), [retries] * len(reader) )
executor.map() takes the following arguments.process_propertyis the function we want to run on each thread.readeris the array of property from our CSV file.- All other arguments are passed in as arrays, just like before.
Step 5: Bypassing Anti-Bots
Once again, we want to get past anti-bots and anything else that might block us. We only need to change one line.response = requests.get(get_scrapeops_url(url, location=location))import osimport csvimport requestsimport jsonimport loggingfrom urllib.parse import urlencodefrom bs4 import BeautifulSoupimport concurrent.futuresfrom dataclasses import dataclass, field, fields, asdict API_KEY = "" with open("config.json", "r") as config_file: config = json.load(config_file) API_KEY = config["api_key"] def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) @dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3): formatted_keyword = keyword.replace(" ", "+") url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" tries = 0 success = False while tries <= retries and not success: try: scrapeops_proxy_url = get_scrapeops_url(url, location=location) response = requests.get(scrapeops_proxy_url) logger.info(f"Recieved [{response.status_code}] from: {url}") if response.status_code != 200: raise Exception(f"Failed request, Status Code {response.status_code}") ## Extract Data soup = BeautifulSoup(response.text, "html.parser") script_tags = soup.select("script[type='application/ld+json']") excluded_types = ["BreadcrumbList", "Events"] for script_tag in script_tags: json_data = json.loads(script_tag.text) if json_data["@type"] not in excluded_types: search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") success = True except Exception as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, retries left {retries-tries}") tries+=1 if not success: raise Exception(f"Max Retries exceeded: {retries}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) def process_property(row, location, retries=3): url = row["url"] tries = 0 success = False while tries <= retries and not success: response = requests.get(get_scrapeops_url(url, location=location)) try: if response.status_code == 200: logger.info(f"Status: {response.status_code}") soup = BeautifulSoup(response.text, "html.parser") price_holder = soup.select_one("span[data-testid='price']") price = int(price_holder.text.replace("$", "").replace(",", "")) info_holders = soup.select("dt") time_listed = info_holders[0].text views = int(info_holders[2].text.replace(",", "")) saves = info_holders[4].text property_pipeline = DataPipeline(csv_filename=f"{row['name']}.csv") property_data = PropertyData( name=row["name"], price=price, time_on_zillow=time_listed, views=views, saves=saves ) property_pipeline.add_data(property_data) property_pipeline.close_pipeline() success = True else: logger.warning(f"Failed Response: {response.status_code}") raise Exception(f"Failed Request, status code: {response.status_code}") except Exception as e: logger.error(f"Exception thrown: {e}") logger.warning(f"Failed to process page: {row['url']}") logger.warning(f"Retries left: {retries-tries}") tries += 1 if not success: raise Exception(f"Max Retries exceeded: {retries}") else: logger.info(f"Successfully parsed: {row['url']}") def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( process_property, reader, [location] * len(reader), [retries] * len(reader) ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
Step 6: Production Run
We're now ready to test this thing out in production. Once again, I've setPAGES to 5 and our LOCATION to "us".Feel free to change any of the constants within main to tweak your results.if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 5 LOCATION = "us" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
In total, there were 45 properties in the CSV file generated from the crawler. If you recall, the crawler took roughly 5 seconds. 42.8 - 5 = 37.8 seconds spent parsing properties. 37.8 seconds / 45 properties = 0.84 seconds per property.Legal and Ethical Considerations
As with any website, when you access Zillow, you are subject to their Terms of Use. You can view those terms here.It's also important to pay attention to theirrobots.txt which you can view here. It's important to note that violation of these Terms could result in your account getting blocked or even permanently removed from the site.When scraping, public data is generally considered legal throughout the world. Private data is any data that is gated behind a login or some other form of authentication.If you're not sure your scraper is legal, it's best to consult with an attorney who handles the jurisdiction of the site you're scraping.Conclusion
You've finished our tutorial and you've now got another skill for your scraping toolbox. You understand parsing, pagination, data storage, concurrency, and proxy integration. Go out and build something!If you're interested in the tech stack used in this article, take a look at the links below.How to Scrape Zillow with Selenium
Let’s dive into the step-by-step process and learn how to effectively scrape Zillow using Selenium.- TLDR: How to Scrape Zillow
- How To Architect Our Scraper
- Understanding How To Scrape zillow
- Setting Up Our Zillow Scraper
- Build A Zillow Search Crawler
- Build A Zillow Scraper
- Legal and Ethical Considerations
- Conclusion
- More Cool Articles
Need help scraping the web?
Then check out ScrapeOps, the complete toolkit for web scraping.
TLDR - How to Scrape Zillow with Selenium
Need to scrape Zillow? We've got you covered.- Create a new project folder.
- Inside the folder, create a
.envfile and add your API key in this format:
SCRAPEOPS_API_KEY=your_api_key_here
- Create file e.g
main.pyand insert this code into it:
import osimport csvimport jsonimport loggingfrom urllib.parse import urlencodeimport concurrent.futuresfrom dataclasses import dataclass, fields, asdictimport timefrom dotenv import load_dotenvfrom selenium import webdriverfrom selenium.webdriver.common.by import Byfrom selenium.webdriver.support.ui import WebDriverWaitfrom selenium.webdriver.support import expected_conditions as ECfrom selenium.common.exceptions import TimeoutException, WebDriverException, NoSuchElementException load_dotenv() API_KEY = os.getenv("SCRAPEOPS_API_KEY") def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) @dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3, timeout=10): url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') # Run in headless mode for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Wait for and find script elements script_elements = WebDriverWait(driver, timeout).until( EC.presence_of_all_elements_located((By.XPATH, "//script[@type='application/ld+json']")) ) for script in script_elements: json_data = json.loads(script.get_attribute('innerHTML')) if json_data["@type"] != "BreadcrumbList": search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") return # Success, exit the function except (TimeoutException, WebDriverException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) def process_property(row, location, retries=3, timeout=10): url = row["url"] scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Extract price price_element = WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.CSS_SELECTOR, "span[data-testid='price']")) ) price = int(price_element.text.replace("$", "").replace(",", "")) # Extract other information info_elements = driver.find_elements(By.TAG_NAME, "dt") time_listed = info_elements[0].text if len(info_elements) > 0 else "No time listed" views = int(info_elements[2].text.replace(",", "")) if len(info_elements) > 2 else 0 saves = info_elements[4].text if len(info_elements) > 4 else "No saves" property_pipeline = DataPipeline(csv_filename=f"{row['name']}.csv") property_data = PropertyData( name=row["name"], price=price, time_on_zillow=time_listed, views=views, saves=saves ) property_pipeline.add_data(property_data) property_pipeline.close_pipeline() logger.info(f"Successfully parsed: {url}") return # Success, exit the function except (TimeoutException, WebDriverException, NoSuchElementException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}") def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( process_property, reader, [location] * len(reader), [retries] * len(reader) ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
- Run the script, and it will generate a CSV file based on your search. For example, searching for properties in "pr" (Puerto Rico) will output a file called
pr.csv.
MAX_THREADS: Determines the maximum number of threads used for concurrent scraping and processing.MAX_RETRIES: Sets the maximum number of retries for each request in case of failure (e.g., network issues, server errors).PAGES: Specifies the number of pages to scrape for each keyword. Each page contains multiple property listings.LOCATION: Defines the geographical location for the scraping. This parameter is used to adjust the proxy location to simulate requests from a specific country.keyword_list: A list of keywords representing different geographical areas or search terms on Zillow. Each keyword triggers a separate scraping job. ("pr"is Puerto Ricto, if you want to do Michigan, add"mi")
pr.csv.How to Architect Our Zillow Scraper with Selenium
The Zillow scraper we are building will consist of two main components:- Crawler: This component will search for properties in a specific location, gathering relevant data and saving it in a CSV file. Each search will yield results like property names, addresses, and URLs, allowing us to compile a comprehensive dataset.
- Parser: Once the crawler completes, the parser will read the CSV file and scrape additional details for each individual property listed, such as price, time on the market, and views.
- Parsing: Extracting valuable data from Zillow's web pages using Selenium.
- Pagination: Handling multiple pages of search results to ensure all available listings are collected.
- Data Storage: Saving the scraped data into CSV reports for further processing and analysis.
- Concurrency: Speeding up the scraping process by executing multiple tasks (page scrapes) simultaneously using multithreading.
- Proxy Integration: Overcoming anti-bot measures by routing requests through a proxy service to reduce detection and blocking.
Understanding How to Scrape Zillow with Selenium
In this section, we'll look at how Selenium can help us mimic real browser interactions to retrieve pages, extract data, navigate through pagination, and handle geolocation restrictions.Step 1: How To Request Zillow Pages
To scrape Zillow effectively, we first need to load the pages. While a traditional GET request (used in the original code withrequests) can fetch a page’s HTML, Zillow's anti-bot protections make that approach unreliable. Instead, we'll use Selenium to simulate a human browsing experience.In the code, the following URL structure represents a search result page for Puerto Rico, with pagination handled by the number at the end:https://www.zillow.com/pr/2_p/
Here:prrefers to the location (Puerto Rico),2_pspecifies that we are on the second page of results.
https://www.zillow.com/homedetails/459-Carr-Km-7-2-Int-Bo-Arenales-Aguadilla-PR-00603/363559698_zpid/
Step 2: How To Extract Data From Zillow Results and Pages
Extracting data from Zillow using Selenium involves interacting with both JSON data on search result pages and HTML elements on individual property pages.Let’s break down how we handle each scenario:Extracting JSON Data from Search Results
On search results pages, Zillow embeds key property data within a JSON structure, which can be extracted using Selenium.Here is the search page and the JSON blob inside it.
We will wait for all script elements of type application/ld+json, which contain the data we need.Here’s how we would approach this in our scraper:# Extract JSON data from search resultsscript_elements = WebDriverWait(driver, timeout).until( EC.presence_of_all_elements_located((By.XPATH, "//script[@type='application/ld+json']"))) for script in script_elements: json_data = json.loads(script.get_attribute('innerHTML')) if json_data["@type"] != "BreadcrumbList": # Extract relevant fields from JSON search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data)
Extracting HTML Data from Individual Property Pages
On individual property pages, the data is usually buried within the HTML, such as the price, time on Zillow, views, and saves.Here is a look at some the HTML we want to parse for an individual property page.Using Selenium, we can wait for these elements to load and then extract the data using their corresponding CSS selectors or tags:# Wait for the price element and extract its valueprice_element = WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.CSS_SELECTOR, "span[data-testid='price']")))price = int(price_element.text.replace("$", "").replace(",", "")) # Extract other details such as time listed, views, and savesinfo_elements = driver.find_elements(By.TAG_NAME, "dt")time_listed = info_elements[0].text if len(info_elements) > 0 else "No time listed"views = int(info_elements[2].text.replace(",", "")) if len(info_elements) > 2 else 0saves = info_elements[4].text if len(info_elements) > 4 else "No saves"
Step 3: How to Control Pagination
Controlling pagination is straightforward, as Zillow URLs follow a predictable pattern. The page number is embedded directly in the URL, for example:- Page 1:
https://www.zillow.com/pr/1_p/ - Page 2:
https://www.zillow.com/pr/2_p/ - Page 3:
https://www.zillow.com/pr/3_p/
scrape_search_results function handles pagination by iterating over pages:def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3, timeout=10): url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" # The rest of the code follows to scrape data from this page
Step 4: Handling Geolocated Data
To ensure our scraper works even when location-based restrictions apply, we use the ScrapeOps API to route requests through servers located in different countries.When interacting with the ScrapeOps API, we'll pass in acountry param as well. country will not have any effect on our actual search results, but instead it will route us through a server in whichever country we specify.For instance, if we want to appear in the US, we'd pass us in as our country.This helps bypass Zillow’s geolocation blocks and improves the chances of successful scraping.The get_scrapeops_url() function integrates this proxy service:def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url
location parameter determines which country we appear to be browsing from (e.g., us for the US).This doesn’t change the search results but helps us avoid being blocked by Zillow’s anti-scraping mechanisms.Setting Up Our Zillow Scraper Project
You can run the following commands to get setup.Create a New Project Foldermkdir <your_directory_name>cd <your_directory_name>
python -m venv venvsource venv/bin/activate # Linux# ORvenv\Scripts\activate # Windows
pip install selenium python-dotenvselenium to automate web browser interactions.We'll use python-dotenv to securely manage sensitive information like login credentials or API keys by storing them in a separate .env file, which helps keep our main code clean and our secrets safe from accidental exposure.Build A Zillow Search Crawler
In this section, we'll build a Zillow search crawler by combining several key components:- parsing search results,
- handling pagination,
- storing data, and
- adding concurrency.
Step 1: Create a Simple Search Data Parser
Let’s create the core function,scrape_search_results(). This function handles the search result page scraping, extracting necessary details (like property URLs, addresses, and prices) and storing them in a CSV file.Here’s an outline:def scrape_search_results(keyword, location, retries=3, timeout=10): url = f"https://www.zillow.com/{keyword}/" scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Wait for and find script elements script_elements = WebDriverWait(driver, timeout).until( EC.presence_of_all_elements_located((By.XPATH, "//script[@type='application/ld+json']")) ) for script in script_elements: json_data = json.loads(script.get_attribute('innerHTML')) if json_data["@type"] != "BreadcrumbList": search_data = { "name": json_data["name"], "property_type": json_data["@type"], "street_address": json_data["address"]["streetAddress"], "locality": json_data["address"]["addressLocality"], "region": json_data["address"]["addressRegion"], "postal_code": json_data["address"]["postalCode"], "url": json_data["url"] } print(search_data) logger.info(f"Successfully parsed data from: {url}") return # Success, exit the function except (TimeoutException, WebDriverException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}")
keyword) and page_number. It sends a GET request to fetch the page, extracts the data, and prints the data.We've also included a retry mechanism to handle potential errors during scraping.Step 2: Add Pagination
Pagination is critical for collecting data from multiple pages. Zillow uses page numbers in the URL (_p/) to navigate between search results.As discussed earlier, we need to increment the page number in our URL.Here’s how:def start_scrape(keyword, pages, location, retries=3): for page in range(pages): scrape_search_results(keyword, location, page, retries=retries)
- Keyword: Represents the search location (e.g., city or state).
- Page: A variable we increment with each loop iteration to move through the search result pages.
range(): We use Python’srange()to handle multiple pages of results. Since Zillow pages start at 1 andrange()starts at 0, we add+1to the page number in the URL.
Step 3: Storing the Scraped Data
To store the scraped data, we can create aSearchData class to structure the extracted information, like property address, price, and more.This ensures that we store data consistently in our CSV file.@dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip())
- Uses Python's
@ dataclassdecorator for automatic method generation. - Defines fields for property details like name, type, address components, and URL.
- Implements a post-initialization method to check and clean string fields.
- Sets default values for empty fields and strips whitespace from non-empty ones.
DataPipeline class that handles saving data to a csv file:class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv()
DataPipeline class to handle data efficiently and store it in CSV format:- We accumulate data in a storage queue until it hits a set limit.
- The
save_to_csvmethod writes data to a CSV file. It creates new files or appends to existing ones as needed. - We use
DictWriterfor flexible field handling, writing headers for new files. - After saving, we clear the queue to prevent data duplication.
- We implement a flag
csv_file_opento avoid concurrent CSV writes. - The
close_pipelinemethod saves any leftover data before shutdown.
Step 4: Adding Concurrency
To speed up the scraping process, especially when scraping multiple pages or properties, we can introduce concurrency using Python’sThreadPoolExecutor.This allows us to scrape several pages simultaneously, reducing overall runtime.from concurrent.futures import ThreadPoolExecutor def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages )
- It uses Python's
ThreadPoolExecutorto run multiple scraping tasks simultaneously - We map the
scrape_search_resultsfunction across multiple threads - This function takes parameters like keyword, pages, location, and data pipeline
- We control concurrency with the
max_threadsparameter - We include a retry mechanism for resilience against failures
Step 5: Bypassing Anti-Bots
To bypass Zillow’s anti-bot measures, we integrate ScrapeOps Proxy Aggregator by using theget_scrapeops_url() function. This function generates a proxy URL to route our requests through servers in different regions, making it more difficult for Zillow to block our scrapers.import osfrom dotenv import load_dotenvfrom urllib.parse import urlencode load_dotenv()API_KEY = os.getenv("SCRAPEOPS_API_KEY") def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url print(get_scrapeops_url('https://zillow.com'))
url and a location parameter (e.g., us), we ensure our requests are routed through a residential proxy, minimizing the risk of getting blocked by Zillow’s anti-bot system.Step 6: Production Run
Finally, we define themain method to initiate the scraping process.if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") # INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.")
- We define constants for retries, thread count, pages to scrape, and location
- The script accepts a list of keywords to search for
- We iterate through each keyword:
- We create a unique filename for each keyword
- We instantiate a
DataPipelinefor each keyword - We call
start_scrapeto begin the concurrent scraping process - After scraping, we close the pipeline to ensure data is saved
- We collect filenames for potential aggregation later
- We use logging to track the start and completion of the crawl process
import osimport csvimport jsonimport loggingfrom urllib.parse import urlencodeimport concurrent.futuresfrom dataclasses import dataclass, fields, asdictimport timefrom dotenv import load_dotenvfrom selenium import webdriverfrom selenium.webdriver.common.by import Byfrom selenium.webdriver.support.ui import WebDriverWaitfrom selenium.webdriver.support import expected_conditions as ECfrom selenium.common.exceptions import TimeoutException, WebDriverException load_dotenv() API_KEY = os.getenv("SCRAPEOPS_API_KEY") def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3, timeout=10): url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') # Run in headless mode for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Wait for and find script elements script_elements = WebDriverWait(driver, timeout).until( EC.presence_of_all_elements_located((By.XPATH, "//script[@type='application/ld+json']")) ) for script in script_elements: json_data = json.loads(script.get_attribute('innerHTML')) if json_data["@type"] != "BreadcrumbList": search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") return # Success, exit the function except (TimeoutException, WebDriverException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") # INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.")
- Bypassing Anti-Bots: We implemented
get_scrapeops_urlto use ScrapeOps proxy, allowing us to bypass anti-bot measures. - Simple Search Data Parser: We created
scrape_search_resultsto extract property data from Zillow pages using Selenium. - Pagination: We added
start_scrapeto handle multiple pages of search results (initially without concurrency). - Data Storage: We developed
SearchDataclass to structure our scraped information andDataPipelineclass to manage data storage and CSV writing. - Concurrency: We enhanced the
start_scrapefunction withThreadPoolExecutorto scrape multiple pages simultaneously, improving efficiency. - Production Setup: We set up the main execution block to handle multiple keywords and manage the overall scraping process.
Build A Zillow Scraper
Our Zillow crawler is already capable of searching for properties, parsing the results, and saving the data in a CSV. Now, we’ll move on to building a scraper that processes this CSV and scrapes individual property details. Our scraper will:- Read the CSV file generated by the crawler.
- Scrape detailed information about each property.
- Save the property details in a structured format.
- Implement concurrency for efficiency.
- Integrate with a proxy to bypass anti-bot protections.
Step 1: Create a Simple Property Data Parser
We’ll begin by creating a function to parse property data from the Zillow property pages. This function will look similar to the initial parsing function we wrote for search results, but now we’re dealing with individual property pages, which have different HTML structures.Once again, we want to get past anti-bots and anything else that might block us. We will use the sameget_scrapeops_url() function:scrapeops_proxy_url = get_scrapeops_url(url, location=location)
def process_property(row, location, retries=3, timeout=10): url = row["url"] scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Extract price price_element = WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.CSS_SELECTOR, "span[data-testid='price']")) ) price = int(price_element.text.replace("$", "").replace(",", "")) # Extract other information info_elements = driver.find_elements(By.TAG_NAME, "dt") time_listed = info_elements[0].text if len(info_elements) > 0 else "No time listed" views = int(info_elements[2].text.replace(",", "")) if len(info_elements) > 2 else 0 saves = info_elements[4].text if len(info_elements) > 4 else "No saves" property_data = { 'name': row["name"], 'price': price, 'time_on_zillow': time_listed, 'views': views, 'saves': saves } print(property_data) logger.info(f"Successfully parsed: {url}") return # Success, exit the function except (TimeoutException, WebDriverException, NoSuchElementException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}")
"span[data-testid='price']"is the CSS selector for the price.int(price_holder.text.replace("$", "").replace(",", ""))cleans up and converts the price into an integer.- We extract
time_listed,views, andsavesfrom theinfo_holderslist using the correct CSS selectors.
Step 2: Loading URLs to Scrape
To feed property URLs into our parsing function, we’ll load the CSV file generated by the crawler. For each row (property) in the CSV, we’ll callprocess_property(). Later, we’ll add concurrency to speed things up.def process_results(csv_file): with open(csv_file, newline='') as file: reader = csv.DictReader(file) for row in reader: process_property(row['url'])
process_property() function.Step 3: Storing the Scraped Data
We’ll store the scraped property details in a structured format using aPropertyData class.This class will be similar to the SearchData class we used earlier but specific to the details scraped from individual property pages.@dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip())
DataPipeline we created in the crawler section as follows:class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() @dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) def process_property(row, location, retries=3, timeout=10): url = row["url"] scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Extract price price_element = WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.CSS_SELECTOR, "span[data-testid='price']")) ) price = int(price_element.text.replace("$", "").replace(",", "")) # Extract other information info_elements = driver.find_elements(By.TAG_NAME, "dt") time_listed = info_elements[0].text if len(info_elements) > 0 else "No time listed" views = int(info_elements[2].text.replace(",", "")) if len(info_elements) > 2 else 0 saves = info_elements[4].text if len(info_elements) > 4 else "No saves" property_pipeline = DataPipeline(csv_filename=f"{row['name']}.csv") property_data = PropertyData( name=row["name"], price=price, time_on_zillow=time_listed, views=views, saves=saves ) property_pipeline.add_data(property_data) property_pipeline.close_pipeline() logger.info(f"Successfully parsed: {url}") return # Success, exit the function except (TimeoutException, WebDriverException, NoSuchElementException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}")
Step 4: Adding Concurrency
To scrape multiple properties in parallel, we’ll use Python’sThreadPoolExecutor. This helps us speed up the process by running multiple process_property() functions concurrently.def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( process_property, reader, [location] * len(reader), [retries] * len(reader) )
executor.map() handles the parallel processing of multiple property URLs. The process_property() function is called on each URL, and the results are saved concurrently.Step 5: Production Run
We're now ready to test this thing out in production.Once again, we've setPAGES to 5 and our LOCATION to "uk". Feel free to change any of the constants within main to tweak your results.if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES)
from dotenv import load_dotenvimport osfrom urllib.parse import urlencodefrom selenium import webdriverfrom selenium.webdriver.support.ui import WebDriverWaitfrom selenium.webdriver.support import expected_conditions as ECfrom selenium.webdriver.common.by import Byfrom selenium.common.exceptions import TimeoutException, WebDriverException, NoSuchElementExceptionfrom dataclasses import fields, asdict, dataclassimport csv import loggingimport timeimport concurrent.futuresimport json ## Logginglogging.basicConfig(level=logging.INFO)logger = logging.getLogger(__name__) # Load environment variablesload_dotenv()API_KEY = os.getenv("SCRAPEOPS_API_KEY") def get_scrapeops_url(url, location="us"): payload = { "api_key": API_KEY, "url": url, "country": location, "residential": True } proxy_url = "https://proxy.scrapeops.io/v1/?" + urlencode(payload) return proxy_url @dataclassclass SearchData: name: str = "" property_type: str = "" street_address: str = "" locality: str = "" region: str = "" postal_code: str = "" url: str = "" def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) class DataPipeline: def __init__(self, csv_filename="", storage_queue_limit=50): self.names_seen = [] self.storage_queue = [] self.storage_queue_limit = storage_queue_limit self.csv_filename = csv_filename self.csv_file_open = False def save_to_csv(self): self.csv_file_open = True data_to_save = [] data_to_save.extend(self.storage_queue) self.storage_queue.clear() if not data_to_save: return keys = [field.name for field in fields(data_to_save[0])] file_exists = os.path.isfile(self.csv_filename) and os.path.getsize(self.csv_filename) > 0 with open(self.csv_filename, mode="a", newline="", encoding="utf-8") as output_file: writer = csv.DictWriter(output_file, fieldnames=keys) if not file_exists: writer.writeheader() for item in data_to_save: writer.writerow(asdict(item)) self.csv_file_open = False def is_duplicate(self, input_data): if input_data.name in self.names_seen: logger.warning(f"Duplicate item found: {input_data.name}. Item dropped.") return True self.names_seen.append(input_data.name) return False def add_data(self, scraped_data): if self.is_duplicate(scraped_data) == False: self.storage_queue.append(scraped_data) if len(self.storage_queue) >= self.storage_queue_limit and self.csv_file_open == False: self.save_to_csv() def close_pipeline(self): if self.csv_file_open: time.sleep(3) if len(self.storage_queue) > 0: self.save_to_csv() def scrape_search_results(keyword, location, page_number, data_pipeline=None, retries=3, timeout=10): url = f"https://www.zillow.com/{keyword}/{page_number+1}_p/" scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') # Run in headless mode for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Wait for and find script elements script_elements = WebDriverWait(driver, timeout).until( EC.presence_of_all_elements_located((By.XPATH, "//script[@type='application/ld+json']")) ) for script in script_elements: json_data = json.loads(script.get_attribute('innerHTML')) if json_data["@type"] != "BreadcrumbList": search_data = SearchData( name=json_data["name"], property_type=json_data["@type"], street_address=json_data["address"]["streetAddress"], locality=json_data["address"]["addressLocality"], region=json_data["address"]["addressRegion"], postal_code=json_data["address"]["postalCode"], url=json_data["url"] ) data_pipeline.add_data(search_data) logger.info(f"Successfully parsed data from: {url}") return # Success, exit the function except (TimeoutException, WebDriverException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}") def start_scrape(keyword, pages, location, data_pipeline=None, max_threads=5, retries=3): with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( scrape_search_results, [keyword] * pages, [location] * pages, range(pages), [data_pipeline] * pages, [retries] * pages ) @dataclassclass PropertyData: name: str = "" price: int = 0 time_on_zillow: str = "" views: int = 0 saves: int = 0 def __post_init__(self): self.check_string_fields() def check_string_fields(self): for field in fields(self): # Check string fields if isinstance(getattr(self, field.name), str): # If empty set default text if getattr(self, field.name) == "": setattr(self, field.name, f"No {field.name}") continue # Strip any trailing spaces, etc. value = getattr(self, field.name) setattr(self, field.name, value.strip()) def process_property(row, location, retries=3, timeout=10): url = row["url"] scrapeops_proxy_url = get_scrapeops_url(url, location=location) options = webdriver.ChromeOptions() options.add_argument('--headless') for attempt in range(retries): try: with webdriver.Chrome(options=options) as driver: driver.get(scrapeops_proxy_url) # Wait for the body to ensure page has started loading WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.TAG_NAME, "body")) ) # Extract price price_element = WebDriverWait(driver, timeout).until( EC.presence_of_element_located((By.CSS_SELECTOR, "span[data-testid='price']")) ) price = int(price_element.text.replace("$", "").replace(",", "")) # Extract other information info_elements = driver.find_elements(By.TAG_NAME, "dt") time_listed = info_elements[0].text if len(info_elements) > 0 else "No time listed" views = int(info_elements[2].text.replace(",", "")) if len(info_elements) > 2 else 0 saves = info_elements[4].text if len(info_elements) > 4 else "No saves" property_pipeline = DataPipeline(csv_filename=f"{row['name']}.csv") property_data = PropertyData( name=row["name"], price=price, time_on_zillow=time_listed, views=views, saves=saves ) property_pipeline.add_data(property_data) property_pipeline.close_pipeline() logger.info(f"Successfully parsed: {url}") return # Success, exit the function except (TimeoutException, WebDriverException, NoSuchElementException) as e: logger.error(f"An error occurred while processing page {url}: {e}") logger.info(f"Retrying request for page: {url}, attempts left: {retries-attempt-1}") raise Exception(f"Max retries ({retries}) exceeded for URL: {url}") def process_results(csv_file, location, max_threads=5, retries=3): logger.info(f"processing {csv_file}") with open(csv_file, newline="") as file: reader = list(csv.DictReader(file)) with concurrent.futures.ThreadPoolExecutor(max_workers=max_threads) as executor: executor.map( process_property, reader, [location] * len(reader), [retries] * len(reader) ) if __name__ == "__main__": MAX_RETRIES = 3 MAX_THREADS = 5 PAGES = 1 LOCATION = "uk" logger.info(f"Crawl starting...") ## INPUT ---> List of keywords to scrape keyword_list = ["pr"] aggregate_files = [] ## Job Processes for keyword in keyword_list: filename = keyword.replace(" ", "-") crawl_pipeline = DataPipeline(csv_filename=f"{filename}.csv") start_scrape(keyword, PAGES, LOCATION, data_pipeline=crawl_pipeline, max_threads=MAX_THREADS, retries=MAX_RETRIES) crawl_pipeline.close_pipeline() aggregate_files.append(f"{filename}.csv") logger.info(f"Crawl complete.") logger.info(f"Scrape starting...") for file in aggregate_files: process_results(file, LOCATION, max_threads=MAX_THREADS, retries=MAX_RETRIES) logger.info(f"Scrape complete.")
python <your_script_name>.pypr.csv. It then reads this file and creates an individual report on each house.Legal and Ethical Considerations
When utilizing Zillow, it’s important to follow their Terms of Use. In addition, be sure to review theirrobots.txt file here, which outlines rules for automated access. Not following these guidelines could lead to account suspension or a permanent ban.Generally, scraping publicly available data is legal in many regions, but accessing private data—such as that which requires login or authentication—requires permission.If you’re unsure about the legal aspects of your scraping activities, it’s advisable to seek legal advice from an attorney who is familiar with the laws in your area.