Repository files navigation

yagooglesearch is a Python library for executing intelligent, realistic-looking, and tunable Google searches. Itsimulates real human Google search behavior to prevent rate limiting by Google (the dreaded HTTP 429 response), and ifHTTP 429 blocked by Google, logic to back off and continue trying. The library does not use the Google API and isheavily based off thegooglesearch library. The features include:

• Tunable search client attributes mid searching
• Returning a list of URLs instead of a generator
• HTTP 429 / rate-limit detection (Google is blocking your IP for making too many search requests) and recovery
• Randomizing delay times between retrieving paged search results (i.e., clicking on page 2 for more results)
• HTTP(S) and SOCKS5 proxy support
• Leveragingrequests library for HTTP requests and cookie management
• Adds "&filter=0" by default to search URLs to prevent any omission or filtering of search results by Google
• Console and file logging
• Python 3.6+

This code is supplied as-is and you are fully responsible for how it is used. Scraping Google Search results mayviolate theirTerms of Service. Another Python Google search library had someinteresting information/discussion on it:

• Original issue
• A response
• Author created a separateTerms and Conditions
• ...that contained link to thisblog

Google's preferred method is to use theirAPI.

pip install yagooglesearch

git clone https://github.com/opsdisk/yagooglesearchcd yagooglesearchvirtualenv -p python3 .venv# If using a virtual environment.source .venv/bin/activate# If using a virtual environment.pip install.# Reads from pyproject.toml

importyagooglesearchquery="site:github.com"client=yagooglesearch.SearchClient(query,tbs="li:1",max_search_result_urls_to_return=100,http_429_cool_off_time_in_minutes=45,http_429_cool_off_factor=1.5,# proxy="socks5h://127.0.0.1:9050",verbosity=5,verbose_output=True,# False (only URLs) or True (rank, title, description, and URL))client.assign_random_user_agent()urls=client.search()len(urls)forurlinurls:print(url)

Even though searching Google through the GUI will display a message like "About 13,000,000 results", that does not meanyagooglesearch will find anything close to that. Testing shows that at most, about 400 results are returned. If youset 400 <max_search_result_urls_to_return, a warning message will be printed to the logs. See#28 for the discussion.

Low and slow is the strategy when executing Google searches usingyagooglesearch. If you start getting HTTP 429responses, Google has rightfully detected you as a bot and will block your IP for a set period of time.yagooglesearchis not able to bypass CAPTCHA, but you can do this manually by performing a Google search from a browser and proving youare a human.

The criteria and thresholds to getting blocked is unknown, but in general, randomizing the user agent, waiting enoughtime between paged search results (7-17 seconds), and waiting enough time between different Google searches (30-60seconds) should suffice. Your mileage will definitely vary though. Using this library with Tor will likely get youblocked quickly.

Ifyagooglesearch detects an HTTP 429 response from Google, it will sleep forhttp_429_cool_off_time_in_minutesminutes and then try again. Each time an HTTP 429 is detected, it increases the wait time by a factor ofhttp_429_cool_off_factor.

The goal is to haveyagooglesearch worry about HTTP 429 detection and recovery and not put the burden on the scriptusing it.

If you do not wantyagooglesearch to handle HTTP 429s and would rather handle it yourself, passyagooglesearch_manages_http_429s=False when instantiating the yagooglesearch object. If an HTTP 429 is detected, thestring "HTTP_429_DETECTED" is added to a list object that will be returned, and it's up to you on what the next stepshould be. The list object will contain any URLs found before the HTTP 429 was detected.

importyagooglesearchquery="site:twitter.com"client=yagooglesearch.SearchClient(query,tbs="li:1",verbosity=4,num=10,max_search_result_urls_to_return=1000,minimum_delay_between_paged_results_in_seconds=1,yagooglesearch_manages_http_429s=False,# Add to manage HTTP 429s.)client.assign_random_user_agent()urls=client.search()if"HTTP_429_DETECTED"inurls:print("HTTP 429 detected...it's up to you to modify your search.")# Remove HTTP_429_DETECTED from list.urls.remove("HTTP_429_DETECTED")print("URLs found before HTTP 429 detected...")forurlinurls:print(url)

yagooglesearch supports the use of a proxy. The provided proxy is used for the entire life cycle of the search tomake it look more human, instead of rotating through various proxies for different portions of the search. The generalsearch life cycle is:

1. Simulated "browsing" togoogle.com
2. Executing the search and retrieving the first page of results
3. Simulated clicking through the remaining paged (page 2, page 3, etc.) search results

To use a proxy, provide a proxy string when initializing ayagooglesearch.SearchClient object:

client=yagooglesearch.SearchClient("site:github.com",proxy="socks5h://127.0.0.1:9050",)

Supported proxy schemes are based off those supported in the Pythonrequests library(https://docs.python-requests.org/en/master/user/advanced/#proxies):

• http
• https
• socks5 - "causes the DNS resolution to happen on the client, rather than on the proxy server." You likelydonot want this since all DNS lookups would source from whereyagooglesearch is being run instead of the proxy.
• socks5h - "If you want to resolve the domains on the proxy server, use socks5h as the scheme." This is thebestoption if you are using SOCKS because the DNS lookup and Google search is sourced from the proxy IP address.

If you are using a self-signed certificate for an HTTPS proxy, you will likely need to disable SSL/TLS verification wheneither:

1. Instantiating theyagooglesearch.SearchClient object:

importyagooglesearchquery="site:github.com"client=yagooglesearch.SearchClient(query,proxy="http://127.0.0.1:8080",verify_ssl=False,verbosity=5,)

2. or after instantiation:

query="site:github.com"client=yagooglesearch.SearchClient(query,proxy="http://127.0.0.1:8080",verbosity=5,)client.verify_ssl=False

If you want to use multiple proxies, that burden is on the script utilizing theyagooglesearch library to instantiatea newyagooglesearch.SearchClient object with the different proxy. Below is an example of looping through a list ofproxies:

importyagooglesearchproxies= ["socks5h://127.0.0.1:9050","socks5h://127.0.0.1:9051","http://127.0.0.1:9052",# HTTPS proxy with a self-signed SSL/TLS certificate.]search_queries= ["python","site:github.com pagodo","peanut butter toast","are dragons real?","ssh tunneling",]proxy_rotation_index=0forsearch_queryinsearch_queries:# Rotate through the list of proxies using modulus to ensure the index is in the proxies list.proxy_index=proxy_rotation_index%len(proxies)client=yagooglesearch.SearchClient(search_query,proxy=proxies[proxy_index],    )# Only disable SSL/TLS verification for the HTTPS proxy using a self-signed certificate.ifproxies[proxy_index].startswith("http://"):client.verify_ssl=Falseurls_list=client.search()print(urls_list)proxy_rotation_index+=1

If you have aGOOGLE_ABUSE_EXEMPTION cookie value, it can be passed intogoogle_exemption when instantiating theSearchClient object.

The&tbs= parameter is used to specify either verbatim or time-based filters.

&tbs=li:1

Currently, the.filter_search_result_urls() function will remove any url with the word "google" in it. This is toprevent the returned search URLs from being polluted with Google URLs. Note this if you are trying to explicitly searchfor results that may have "google" in the URL, such assite:google.com computer

Distributed under the BSD 3-Clause License. SeeLICENSE for more information.

@opsdisk

Project Link:https://github.com/opsdisk/yagooglesearch

• Mario Vilas for his amazing work on the originalgooglesearch library.

• KennBro -#9
• ArshansGithub -#21
• pguridi -#38
• libenc -#42