yagooglesearch - Yet another googlesearch

Overview

yagooglesearch is a Python library for executing intelligent, realistic-looking, and tunable Google searches. It simulates real human Google search behavior to prevent rate limiting by Google (the dreaded HTTP 429 response), and if HTTP 429 blocked by Google, logic to back off and continue trying. The library does not use the Google API and is heavily based off the googlesearch 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
Leveraging requests 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+

Terms and Conditions

This code is supplied as-is and you are fully responsible for how it is used. Scraping Google Search results may violate their Terms of Service. Another Python Google search library had some interesting information/discussion on it:

Original issue
A response
Author created a separate Terms and Conditions
...that contained link to this blog

Google's preferred method is to use their API.

Installation

pip

pip install yagooglesearch

pyproject.toml

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

Usage

import yagooglesearch

query = "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)

for url in urls:
    print(url)

Max ~400 results returned

Even though searching Google through the GUI will display a message like "About 13,000,000 results", that does not mean yagooglesearch will find anything close to that. Testing shows that at most, about 400 results are returned. If you set 400 < max_search_result_urls_to_return, a warning message will be printed to the logs. See https://github.com/opsdisk/yagooglesearch/issues/28 for the discussion.

Google is blocking me!

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

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

HTTP 429 detection and recovery (optional)

If yagooglesearch detects an HTTP 429 response from Google, it will sleep for http_429_cool_off_time_in_minutes minutes and then try again. Each time an HTTP 429 is detected, it increases the wait time by a factor of http_429_cool_off_factor.

The goal is to have yagooglesearch worry about HTTP 429 detection and recovery and not put the burden on the script using it.

If you do not want yagooglesearch to handle HTTP 429s and would rather handle it yourself, pass yagooglesearch_manages_http_429s=False when instantiating the yagooglesearch object. If an HTTP 429 is detected, the string "HTTP_429_DETECTED" is added to a list object that will be returned, and it's up to you on what the next step should be. The list object will contain any URLs found before the HTTP 429 was detected.

import yagooglesearch

query = "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" in urls:
    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...")

    for url in urls:
        print(url)

HTTP and SOCKS5 proxy support

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

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

To use a proxy, provide a proxy string when initializing a yagooglesearch.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 Python requests 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 likely do not want this since all DNS lookups would source from where yagooglesearch 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 the best option if you are using SOCKS because the DNS lookup and Google search is sourced from the proxy IP address.

HTTPS proxies and SSL/TLS certificates

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

Instantiating the yagooglesearch.SearchClient object:

import yagooglesearch

query = "site:github.com"

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

or after instantiation:

query = "site:github.com"

client = yagooglesearch.SearchClient(
    query,
    proxy="http://127.0.0.1:8080",
    verbosity=5,
)

client.verify_ssl = False

Multiple proxies

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

import yagooglesearch

proxies = [
    "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 = 0

for search_query in search_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.
    if proxies[proxy_index].startswith("http://"):
        client.verify_ssl = False

    urls_list = client.search()

    print(urls_list)

    proxy_rotation_index += 1

GOOGLE_ABUSE_EXEMPTION cookie

If you have a GOOGLE_ABUSE_EXEMPTION cookie value, it can be passed into google_exemption when instantiating the SearchClient object.

&tbs= URL filter clarification

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

Verbatim search

&tbs=li:1

time-based filters

Time filter	&tbs= URL parameter	Notes
Past hour	qdr:h
Past day	qdr:d	Past 24 hours
Past week	qdr:w
Past month	qdr:m
Past year	qdr:y
Custom	cdr:1,cd_min:1/1/2021,cd_max:6/1/2021	See yagooglesearch.get_tbs() function

Limitations

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