using python regex to extract certain URLs from te

问题:

So I have the HTML from an NPR page, and I want to use regex to extract just certain URLs for me (these call the URLs to specific stories nested within the page). The actual links appear in the text (retrieved manually) as:

<a href="http://www.npr.org/blogs/parallels/2014/11/11/363018388/how-the-islamic-state-wages-its-propaganda-war">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363309020/asked-to-stop-praying-alaska-school-won-t-host-state-tournament">
<a href="http://www.npr.org/2014/11/11/362817642/a-marines-parents-story-their-memories-that-you-should-hear">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363288744/comets-rugged-landscape-makes-landing-a-roll-of-the-dice">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363293514/for-dyslexics-a-font-and-a-dictionary-that-are-meant-to-help">

obviously, I cannot to continue to use manual retrieval if I want to be able to use this on a consistent basis. So far, I have this code:

import nltk
import re

f = open("/Users/shannonmcgregor/Desktop/npr.txt")
npr_lines = f.readlines()
f.close()

I have this code to grab everything between (

for line in npr_lines:
re.findall('<a href="?\'?([^"\'>]*)', line)

But that grabs all urls. I tried adding something like:

(parallels|thetwo-way|a-marines)

but that returns nothing. So what am I doing wrong? How I combine the larger URL stripper with these specific words that target the given URLs?

Please and thank you :)

回答1:

You can use re.search function to match the regex in the line and prints the line if it matches as

>>> file  = open('/Users/shannonmcgregor/Desktop/npr.txt', 'r')
>>> for line in file:
...     if re.search('<a href=[^>]*(parallels|thetwo-way|a-marines)', line):
...             print line

will give an output as

<a href="http://www.npr.org/blogs/parallels/2014/11/11/363018388/how-the-islamic-state-wages-its-propaganda-war">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363309020/asked-to-stop-praying-alaska-school-won-t-host-state-tournament">
<a href="http://www.npr.org/2014/11/11/362817642/a-marines-parents-story-their-memories-that-you-should-hear">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363288744/comets-rugged-landscape-makes-landing-a-roll-of-the-dice">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363293514/for-dyslexics-a-font-and-a-dictionary-that-are-meant-to-help">

回答2:

Through a tool which is specially designed for parsing html and xml files [BeautifulSoup],

>>> from bs4 import BeautifulSoup
>>> s = """<a href="http://www.npr.org/blogs/parallels/2014/11/11/363018388/how-the-islamic-state-wages-its-propaganda-war">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363309020/asked-to-stop-praying-alaska-school-won-t-host-state-tournament">
<a href="http://www.npr.org/2014/11/11/362817642/a-marines-parents-story-their-memories-that-you-should-hear">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363288744/comets-rugged-landscape-makes-landing-a-roll-of-the-dice">
<a href="http://www.npr.org/blogs/thetwo-way/2014/11/11/363293514/for-dyslexics-a-font-and-a-dictionary-that-are-meant-to-help">"""
>>> soup = BeautifulSoup(s) # or pass the file directly into BS like >>> soup = BeautifulSoup(open('/Users/shannonmcgregor/Desktop/npr.txt'))
>>> atag = soup.find_all('a')
>>> links = [i['href'] for i in atag]
>>> import re
>>> for i in links:
        if re.match(r'.*(parallels|thetwo-way|a-marines).*', i):
            print(i)


http://www.npr.org/blogs/parallels/2014/11/11/363018388/how-the-islamic-state-wages-its-propaganda-war
http://www.npr.org/blogs/thetwo-way/2014/11/11/363309020/asked-to-stop-praying-alaska-school-won-t-host-state-tournament
http://www.npr.org/2014/11/11/362817642/a-marines-parents-story-their-memories-that-you-should-hear
http://www.npr.org/blogs/thetwo-way/2014/11/11/363288744/comets-rugged-landscape-makes-landing-a-roll-of-the-dice
http://www.npr.org/blogs/thetwo-way/2014/11/11/363293514/for-dyslexics-a-font-and-a-dictionary-that-are-meant-to-help

回答3:

You can do this by using a lookahead:

<a href="?\'?((?=[^"\'>]*(?:thetwo\-way|parallels|a\-marines))[^"\'>]+)

Debuggex Demo