Parsing search strings
We've all had to write a search form at some point. Beyond simple cases, you reach for the big guns, like haystack, et al.
But what about when it's just something simple? What if you want to, for instance, let people search your blog posts?
In django, that can be done simply with:
Blog.objects.filter(Q(title__icontains=word) | Q(body__icontains=word))
Which is fine, until someone wants to look for a post on, for instance "django" and "cookies", as opposed to the phrase "django cookies".
So, we need to split up the search terms, and iteratively filter for them:
qset = Blog.posts.all() for term in terms.split(' '): qset = qset.filter(Q(title__icontains=term) | Q(body__icontains=term))
Which is all great, until you want to search for "apt-get install" and "postgres" as separate terms. Reflexively you're probably thinking you'd do just that - put quotes around the terms to force groupings. But how do you parse that?
Stdlib to the rescue!
Of course, someone's had to do this before, and there's a lib for that. But it's not made for searching. It's the shlex library.
>>> import shlex >>> shlex.split('"apt-get install" django') ['apt-get install', 'django']
And it's that simple? Well, not quite. If someone leaves unbalanced quotes [or is trying to search for a measurement in inches] shlex will not be happy:
>>> shlex.split('macbook 13"') Traceback (most recent call last): ... ValueError: No closing quotation
To solve this we need to go just a little bit deeper. The shlex.split function is really a wrapper for the shlex.shlex class, in a default and [generally] useful configuration.
We need to configure a shlex to our needs.
from shlex import shlex def parse_search_terms(terms): lexer = shlex(terms) lexer.commenters = '' lexer.quotes = '"\'' while True: term = lexer.get_token() if not term: break yield term
Now we have a generator what will yield search terms, and can now easily filter our queryset!
>>> list(parse_search_terms('13" macbook')) ['13"', 'macbook']