Pygooglenews – Python library for advanced Google News data mining

栏目: IT技术 · 发布时间: 4年前

内容简介:If Google News had a Python libraryCreated by Artem from

pygooglenews

If Google News had a Python library

Created by Artem from newscatcherapi.com but you do not need anything from us or from anyone else to get the software going, it just works out of the box.

Demo

Pygooglenews – Python library for advanced Google News data mining

You might also like to check our Google News API or Financial Google News API

Table of Contents

  • Examples of Use Cases
  • Working with Google News in Production
  • Advanced Query Search Examples

About

A python wrapper of the Google News RSS feed.

Top stories, topic related news feeds, geolocation news feed, and an extensive full text search feed.

This work is more of a collection of all things I could find out about how Google News functions.

How is it different from other Pythonic Google News libraries?

  1. URL-escaping user input helper for the search function
  2. Extensive support for the search function that makes it simple to use:
    • exact match
    • in title match, in url match, etc
    • search by date range ( from_ & to_ ), latest published ( when )
  3. Parsing of the sub articles. Almost always, all feeds except the search one contain a subset of similar news for each article in a feed. This package takes care of extracting those sub articles. This feature might be highly useful to ML task when you need to collect a data of similar article headlines

Examples of Use Cases

  1. Integrating a news feed to your platform/application/website
  2. Collecting data by topic to train your own ML model
  3. Search for latest mentions for your new product
  4. Media monitoring of people/organizations — PR

Working with Google News in Production

Before we start, if you want to integrate Google News data to your production then I would advise you to use one of the 3 methods described above. Why? Because you do not want your servers IP address to be locked by Google. Every time you call any function there is an HTTPS request to Google's servers. Don't get me wrong, this Python package still works out of the box.

  1. NewsCatcher's Google News API — all code is written for you, clean & structured JSON output. Low price. You can test it yourself with no credit card. Plus, financial version of API is also available.
  2. ScrapingBee API which handles proxy rotation for you. Each function in this package has scraping_bee parameter where you paste your API key. You can also try it for free, no credit card required. See
  3. Your own proxy — already have a pool of proxies? Each function in this package has proxies parameter (python dictionary) where you just paste your own proxies.

I love working with the news data. I love it so much that I created  my own company that crawls for hundreds of thousands of news articles, and allow you to search it via a news API . But this time, I want to share with the community a Python package that makes it simple to get the news data from the best search engine ever created -  Google .

Most likely, you know already that Google has its own news service . It is different from the usual Google search that we use on a daily basis (sorry  DuckDuckGo , maybe next time).

This package uses the RSS feed of the Google News. The  top stories page , for example.

RSS is an XML page that is already well structured. I heavily rely on Feedparser package to parse the RSS feed.

Google News used to have an API but it was deprecated many years ago. (Unofficial) information about RSS syntax is decentralized over the web. There is no official documentation. So, I tried my best to collect all this informaion in one place.

Installation

$ pip install pygooglenews

Quickstart

from pygooglenews import GoogleNews

gn = GoogleNews()

Top Stories

top = gn.top_news()

Stories by Topic

business = gn.topic_headlines('business')

Geolocation Specific Stories

headquaters = gn.geo_headlines('San Fran')

Stories by a Query Search

# search for the best matching articles that mention MSFT and 
# do not mention AAPL (over the past 6 month
search = gn.search('MSFT -APPL', when = '6m')

Documentation - Functions & Classes

GoogleNews Class

from pygooglenews import GoogleNews
# default GoogleNews instance
gn = GoogleNews(lang = 'en', country = 'US')

To get the access to all the functions, you first have to initiate the GoogleNews class.

It has 2 required variables: lang and  country

You can try any combination of those 2, however, it does not exist for all. Only the combinations that are supported by GoogleNews will work. Check the official Google News page to check what is covered:

On the bottom left side of the Google News page you may find a Language & region section where you can find all of the supported combinations.

For example, for country=UA (Ukraine), there are 2 languages supported:

lang=uk
lang=ru

Top Stories

top = gn.top_news(proxies=None, scraping_bee = None)

top_news() returns the top stories for the selected country and language that are defined in  GoogleNews class. The returned object contains feed (FeedParserDict) and entries list of articles found with all data parsed.

Stories by Topic

business = gn.topic_headlines('BUSINESS', proxies=None, scraping_bee = None)

The returned object contains feed (FeedParserDict) and entries list of articles found with all data parsed.

Accepted topics are:

WORLD
NATION
BUSINESS
TECHNOLOGY
ENTERTAINMENT
SCIENCE
SPORTS
HEALTH

However, you can find some other topics that are also supported by Google News.

For example, if you search for corona in the search tab of  enUS you will find  COVID-19 as a topic.

The URL looks like this: https://news.google.com/topics/CAAqIggKIhxDQkFTRHdvSkwyMHZNREZqY0hsNUVnSmxiaWdBUAE?hl=en-US&gl=US&ceid=US%3Aen

We have to copy the text after topics/ and before  ? , then you can use it as an input for the  top_news() function.

from pygooglenews import GoogleNews

gn = GoogleNews()
covid = gn.topic_headlines('CAAqIggKIhxDQkFTRHdvSkwyMHZNREZqY0hsNUVnSmxiaWdBUAE')

However, be aware that this topic will be unique for each language/country combination.

Stories by Geolocation

gn = GoogleNews('uk', 'UA')
kyiv = gn.geo_headlines('kyiv', proxies=None, scraping_bee = None)
# or 
kyiv = gn.geo_headlines('kiev', proxies=None, scraping_bee = None)
# or
kyiv = gn.geo_headlines('киев', proxies=None, scraping_bee = None)
# or
kyiv = gn.geo_headlines('Київ', proxies=None, scraping_bee = None)

The returned object contains feed (FeedParserDict) and entries list of articles found with all data parsed.

All of the above variations will return the same feed of the latest news about Kyiv, Ukraine:

geo['feed'].title

# 'Київ - Останні - Google Новини'

It is language agnostic, however, it does not guarantee that the feed for any specific place will exist. For example, if you want to find the feed on LA or  Los Angeles you can do it with  GoogleNews('en', 'US') .

The main ( enUS ) Google News client will most likely find the feed about the most places.

Stories by a Query

gn.search(query: str, helper = True, when = None, from_ = None, to_ = None, proxies=None, scraping_bee=None)

The returned object contains feed (FeedParserDict) and entries list of articles found with all data parsed.

Google News search itself is a complex function that has inherited some features from the standard Google Search.

The biggest obstacle that you might have is to write the URL-escaping input. To ease this process, helper = True is turned on by default.

helper uses  urllib.parse.quote_plus to automatically convert the input.

For example:

  • 'New York metro opening'  -->  'New+York+metro+opening'
  • 'AAPL -MSFT'  -->  'AAPL+-MSFT'
  • '"Tokyo Olimpics date changes"'  -->  '%22Tokyo+Olimpics+date+changes%22'

You can turn it off and write your own query in case you need it by helper = False

when parameter ( str ) sets the time range for the published datetime. I could not find any documentation regarding this option, but here is what I deducted:

  • h  for hours.(For me, worked for up to  101h ).  when=12h  will search for only the articles matching the  search  criteri and published for the last 12 hours
  • d  for days
  • m  for month (For me, worked for up to  48m )

I did not set any hard limit here. You may try put here anything. Probably, it will work. However, I would like to warn you that wrong inputs will not lead to an error. Instead, the when parameter will be ignored by the Google.

from_ and to_ accept the following format of date: %Y-%m-%d For example, 2020-07-01

Google's Special Query Terms  Cheat Sheet

Many Google's Special Query Terms have been tested one by one. Most of the core ones have been inherited by Google News service. At first, I wanted to integrate all of those as the search() function parameters. But, I realised that it might be a bit confusing and difficult to make them all work correctly.

Instead, I decided to write some kind of a cheat sheet that should give you a decent understanding of what you could do .

  • Boolean OR Search [ OR ]
from pygooglenews import GoogleNews

gn = GoogleNews()

s = gn.search('boeing OR airbus')

print(s['feed'].title)
# "boeing OR airbus" - Google News
  • Exclude Query Term [-]

"The exclude ( - ) query term restricts results for a particular search request to documents that do not contain a particular word or phrase. To use the exclude query term, you would preface the word or phrase to be excluded from the matching documents with "-" (a minus sign)."

  • Include Query Term [+]

"The include ( + ) query term specifies that a word or phrase must occur in all documents included in the search results. To use the include query term, you would preface the word or phrase that must be included in all search results with "+" (a plus sign).

The URL-escaped version of + (a plus sign) is  %2B ."

  • Phrase Search

"The phrase search ( " ) query term allows you to search for complete phrases by enclosing the phrases in quotation marks or by connecting them with hyphens.

The URL-escaped version of " (a quotation mark) is  %22 .

Phrase searches are particularly useful if you are searching for famous quotes or proper names."

  • allintext

"The allintext: query term requires each document in the search results to contain all of the words in the search query in the body of the document. The query should be formatted as  allintext: followed by the words in your search query.

If your search query includes the allintext: query term, Google will only check the body text of documents for the words in your search query, ignoring links in those documents, document titles and document URLs."

  • intitle

"The intitle: query term restricts search results to documents that contain a particular word in the document title. The search query should be formatted as intitle:WORD with no space between the intitle: query term and the following word."

  • allintitle

"The allintitle: query term restricts search results to documents that contain all of the query words in the document title. To use the  allintitle: query term, include "allintitle:" at the start of your search query.

Note: Putting allintitle: at the beginning of a search query is equivalent to putting  intitle: in front of each word in the search query."

  • inurl

"The inurl: query term restricts search results to documents that contain a particular word in the document URL. The search query should be formatted as inurl:WORD with no space between the inurl: query term and the following word"

  • allinurl

The allinurl: query term restricts search results to documents that contain all of the query words in the document URL. To use the allinurl: query term, include allinurl: at the start of your search query.

List of operators that do not work (for me, at least):

as_*
allinlinks:
related:

Tip. If you want to build a near real-time feed for a specific topic, use  when='1h' . If Google captured less than 100 articles over the past hour, you should be able to retrieve all of them.

Check the Useful Links section if you want to dig into how Google Search works.

Especially, Special Query Terms section of Google XML reference.

Plus, I will provide some more examples under the Full-Text Search Examples section

Output Body

All 4 functions return the dictionary that has 2 sub-objects:

feed
entries

Both are inherited from the Feedparser . The only change is that each dictionary under  entries also contains  sub_articles which are the similar articles found in the description. Usually, it is non-empty for  top_news() and  topic_headlines() feeds.

TipTo check what is the found feed's name just check the  title under the  feed dictionary

How to use pygooglenews with ScrapingBee

Every function has scrapingbee parameter. It accepts your ScrapingBee API key that will be used to get the response from Google's servers.

You can take a look at what exactly is happening in the source code: check for __scaping_bee_request() function under GoogleNews class

Pay attention to the concurrency of each plan at ScrapingBee

How to use example:

gn = GoogleNews()

# it's a fake API key, do not try to use it
gn.top_news(scraping_bee = 'I5SYNPRFZI41WHVQWWUT0GNXFMO104343E7CXFIISR01E2V8ETSMXMJFK1XNKM7FDEEPUPRM0FYAHFF5')

How to use pygooglenews with proxies

So, if you have your own HTTP/HTTPS proxy(s) that you want to use to make requests to Google, that's how you do it:

gn = GoogleNews()

gn.top_news(proxies = {'https':'34.91.135.38:80'})

Advanced Querying Search Examples

Example 1. Search for artiles that mention  boeing  and do not mention  airbus

from pygooglenews import GoogleNews

gn = GoogleNews()

s = gn.search('boeing -airbus')

print(s['feed'].title)
# "boeing -airbus" - Google News

Example 2. Search for articles that mention  boeing  in title

from pygooglenews import GoogleNews

gn = GoogleNews()

s = gn.search('intitle:boeing')

print(s['feed'].title)
# "intitle:boeing" - Google News

Example 3. Search for artiles that mention  boeing  in title and got published over the past hour

from pygooglenews import GoogleNews

gn = GoogleNews()

s = gn.search('intitle:boeing', when = '1h')

print(s['feed'].title)
# "intitle:boeing when:1h" - Google News

Example 4. Search for artiles that mention  boeing  or  airbus

from pygooglenews import GoogleNews

gn = GoogleNews()

s = gn.search('boeing OR airbus', when = '1h')

print(s['feed'].title)
# "boeing AND airbus when:1h" - Google News

Stack Overflow tread from which it all began

Google XML reference for the search query

Google News Search parameters (The Missing Manual)

Feedparser

Beutifulsoup4

My name is Artem. I :heart: working with news data. I am a co-founder of NewsCatcherAPI - Ultra-fast API to find news articles by any topic, country, language, website, or keyword

If you are interested in hiring me, please, contact me by email - bugara.artem@gmail.com or artem@newscatcherapi.com

Follow me on Twitter - I write about data engineering, python, entrepreneurship, and memes.

Want to read about how it all was done? Subscribe to CODARIUM

thx to Kizy


以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持 码农网

查看所有标签

猜你喜欢:

本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们

注意力商人

注意力商人

吳修銘 / 黃庭敏 / 天下雜誌 / 2018-4-2 / NT$650

電子郵件,免費!照片分享,無上限! 你是否想過,隨手可得的免費內容、便利的免費服務,到底都是誰在付費? 如果商品免費,那你就不是消費者,而是商品! 你我可能都不知不覺地把自己賣給了注意力商人! 「『媒體轉型、網路演化與資訊浪潮」此一主題最具洞見的作者。』──黃哲斌(資深媒體人) 「這是少有的關注產業發展的傳播史,對現在或未來的『注意力產業』」中人來說,不可不讀。」──......一起来看看 《注意力商人》 这本书的介绍吧!

HTML 编码/解码
HTML 编码/解码

HTML 编码/解码

Base64 编码/解码
Base64 编码/解码

Base64 编码/解码

RGB CMYK 转换工具
RGB CMYK 转换工具

RGB CMYK 互转工具