How can I search through stackoverflow questions from a script?

Question

Given a string of keywords, such as "python best practices", I would like to obtain the first 10 stackoverflow questions that contain that keywords, sorted by relevance (?), say from a python script. My goal is to end up with a list of tuples (title, URL).

How can I accomplish this? Would you consider querying google instead? (How would you do it from python?)

Answer 1

You could screen scrape the returned HTML from a valid HTTP request. But that would result in bad karma, and the loss of the ability to enjoy a good night's sleep.

Answer 2

Since Stackoverflow already has this feature you just need to get the contents of the search results page and scrape the information you need. Here is the URL for a search by relevance:

http://stackoverflow.com/search?q=python+best+practices&sort=relevance

If you View Source, you'll see that the information you need for each question is on a line like this:

<h3><a href="/questions/5119/what-are-the-best-rss-feeds-for-programmersdevelopers#5150" class="answer-title">What are the best RSS feeds for programmers/developers?</a></h3>

So you should be able to get the first ten by doing a regex search for a string of that form.

Answer 3

Suggest that a REST API be added to SO. http://stackoverflow.uservoice.com/

Answer 4

I would just use Pycurl to concatenate the search terms onto the query uri.

Answer 5

>>> from urllib import urlencode
>>> params = urlencode({'q': 'python best practices', 'sort': 'relevance'})
>>> params
'q=python+best+practices&sort=relevance'
>>> from urllib2 import urlopen
>>> html = urlopen("http://stackoverflow.com/search?%s" % params).read()
>>> import re
>>> links = re.findall(r'<h3><a href="([^"]*)" class="answer-title">([^<]*)</a></h3>', html)
>>> links
[('/questions/5119/what-are-the-best-rss-feeds-for-programmersdevelopers#5150', 'What are the best RSS feeds for programmers/developers?'), ('/questions/3088/best-ways-to-teach-a-beginner-to-program#13185', 'Best ways to teach a beginner to program?'), ('/questions/13678/textual-versus-graphical-programming-languages#13886', 'Textual versus Graphical Programming Languages'), ('/questions/58968/what-defines-pythonian-or-pythonic#59877', 'What defines &#8220;pythonian&#8221; or &#8220;pythonic&#8221;?'), ('/questions/592/cxoracle-how-do-i-access-oracle-from-python#62392', 'cx_Oracle - How do I access Oracle from Python? '), ('/questions/7170/recommendation-for-straight-forward-python-frameworks#83608', 'Recommendation for straight-forward python frameworks'), ('/questions/100732/why-is-if-not-someobj-better-than-if-someobj-none-in-python#100903', 'Why is if not someobj: better than if someobj == None: in Python?'), ('/questions/132734/presentations-on-switching-from-perl-to-python#134006', 'Presentations on switching from Perl to Python'), ('/questions/136977/after-c-python-or-java#138442', 'After C++ - Python or Java?')]
>>> from urlparse import urljoin
>>> links = [(urljoin('http://stackoverflow.com/', url), title) for url,title in links]
>>> links
[('http://stackoverflow.com/questions/5119/what-are-the-best-rss-feeds-for-programmersdevelopers#5150', 'What are the best RSS feeds for programmers/developers?'), ('http://stackoverflow.com/questions/3088/best-ways-to-teach-a-beginner-to-program#13185', 'Best ways to teach a beginner to program?'), ('http://stackoverflow.com/questions/13678/textual-versus-graphical-programming-languages#13886', 'Textual versus Graphical Programming Languages'), ('http://stackoverflow.com/questions/58968/what-defines-pythonian-or-pythonic#59877', 'What defines &#8220;pythonian&#8221; or &#8220;pythonic&#8221;?'), ('http://stackoverflow.com/questions/592/cxoracle-how-do-i-access-oracle-from-python#62392', 'cx_Oracle - How do I access Oracle from Python? '), ('http://stackoverflow.com/questions/7170/recommendation-for-straight-forward-python-frameworks#83608', 'Recommendation for straight-forward python frameworks'), ('http://stackoverflow.com/questions/100732/why-is-if-not-someobj-better-than-if-someobj-none-in-python#100903', 'Why is if not someobj: better than if someobj == None: in Python?'), ('http://stackoverflow.com/questions/132734/presentations-on-switching-from-perl-to-python#134006', 'Presentations on switching from Perl to Python'), ('http://stackoverflow.com/questions/136977/after-c-python-or-java#138442', 'After C++ - Python or Java?')]

Converting this to a function should be trivial.

EDIT: Heck, I'll do it...

def get_stackoverflow(query):
    import urllib, urllib2, re, urlparse
    params = urllib.urlencode({'q': query, 'sort': 'relevance'})
    html = urllib2.urlopen("http://stackoverflow.com/search?%s" % params).read()
    links = re.findall(r'<h3><a href="([^"]*)" class="answer-title">([^<]*)</a></h3>', html)
    links = [(urlparse.urljoin('http://stackoverflow.com/', url), title) for url,title in links]

    return links