Creating a Simple Bloom Filter (Posted on February 2nd, 2013)

Bloom filters are super efficient data structures that allow us to tell if an object is most likely in a data set or not by checking a few bits. Bloom filters return some false positives but no false negatives. Luckily we can control the amount of false positives we receive with a trade off of time and memory.

You may have never heard of a bloom filter before but you've probably interacted with one at some point. For instance if you use Chrome, Chrome has a bloom filter of malicious URLs. When you visit a website it checks if that domain is in the filter. This prevents you from having to ping Google's servers every time you visit a website to check if it's malicious or not. Large databases such as Cassandra and Hadoop use bloom filters to see if it should do a large query or not.

From Cassandra's Architecture Overview:

Cassandra uses bloom filters to save IO when performing a key lookup: each [Sorted String Table] has a bloom filter associated with it that Cassandra checks before doing any disk seeks, making queries for keys that don't exist almost free.

What You'll Need

  • A bit array (as the name suggests it's just an array of bits)
  • A quick non-cryptographic hash function like murmurhash3 or cityhash

For Python you can do:

sudo pip install bitarray
sudo pip install mmh3

How They Work

A bloom filter is essentially a huge bit array. Bloom filters work by hashing an object several times using either multiple hash functions or the same hash function with a different seed. This insures that when we hash an object we're unlikely to get the same result. For each time an object is hashed the corresponding hash value in the bit array is then marked as 1. This is why we use a bit array. Rather than needing say 4 bytes to store a 1 or a 0 we can simply do it in a bit.

Here's an example to help make it easier to understand. Note we mod by the size of the bit array to prevent index out of bounds:

from bitarray import bitarray
import mmh3

bit_array = bitarray(10)
bit_array.setall(0)

b1 = mmh3.hash("hello", 41) % 10 #Equals 0
bit_array[b1] = 1
b2 = mmh3.hash("hello", 42) % 10 #Equals 4
bit_array[b2] = 1

At this point our bit array looks like this: 1000100000

If we want to check if "hello" is in our data set we do the same opreation back.

b1 = mmh3.hash("hello", 41) % 10 #Equals 0
b2 = mmh3.hash("hello", 42) % 10 #Equals 4
if bit_array[b1] == 1 and bit_array[b2] == 1:
    print "Probably in set"
else:
    print "Definitely not in set"

The reason it is only probably in the set is because a combination of items added to the data set could end up setting the same bits to 1. For instance say we have an empty bloom filter and maybe hashing "world" with seed 41 equalled 0 and hashing "foo" with seed 42 equalled 4. Then when you attempt to search for "hello" you'd get that it is probably in the set even though it was never added. You can prevent this from happening by using a larger bit array and more hash functions. There is some math to choosing these numbers that I'll get in to towards the end of the post.

The Class

Alright let's do this! Bloom filters take in 2 variables: size and number of times to run our hash function(s).

from bitarray import bitarray
import mmh3

class BloomFilter:
    
    def __init__(self, size, hash_count):
        self.size = size
        self.hash_count = hash_count
        self.bit_array = bitarray(size)
        self.bit_array.setall(0)
        
    def add(self, string):
        #Awesome code goes here
        
    def lookup(self, string):
        #Awesome code goes here

Using our knowledge from earlier we can create a function to add items to our filter in just 3 lines of code.

def add(self, string):
    for seed in xrange(self.hash_count):
        result = mmh3.hash(string, seed) % self.size
        self.bit_array[result] = 1

We hash the string with our hash function modded by the size of the bit array and then set that value in our bit array to 1. For each iteration our seed increases by 1 so that we can get different numbers from each hash.

The code for the lookup function is almost the same as the add function. For each hash we're just checking if the resulting value is set to 1 or 0 in our bit array. If we find a 0 then we respond back with "Nope". Otherwise it's probably there so we respond back with "Probably".

def lookup(self, string):
    for seed in xrange(self.hash_count):
        result = mmh3.hash(string, seed) % self.size
        if self.bit_array[result] == 0:
            return "Nope"
    return "Probably"

That's it. That's a bloom filter. Pretty simple right?

Finding The Right Values

The big thing we want to control when creating our bloom filter is our false positive rate. To do this we'll need to have an idea of the size of our data set. Wikipedia has the math and fancy graphs behind choosing your values. Here's what you need to know though:

The formula to calculate the size of your bit array (m = size of bit array, n = expected number of items, and p = probability percentage represented as a decimal):

m equals negative n times the natural log p all over the natural log of 2 sqaured

The formula to calculate the number of hashes to use (k = number of hashes to use):

k equals negative m over n times the natural log of 2

You can also play around with values using this formula which will give you your false positive probability:

false probability equals ( 1 minus [ 1 minus 1 over m] raised to the power k times n ) raised to the power of k

Testing Our Code

The built-in american english dictionary on Ubuntu has about 100,000 words in it so let's use that with a bloom filter. According to WolframAlpha a bloom filter with 1.1 million bits and 7 hash functions has about a 0.5% chance of a false positive.

from bitarray import bitarray
import mmh3

class BloomFilter:
    
    def __init__(self, size, hash_count):
        self.size = size
        self.hash_count = hash_count
        self.bit_array = bitarray(size)
        self.bit_array.setall(0)
        
    def add(self, string):
        for seed in xrange(self.hash_count):
            result = mmh3.hash(string, seed) % self.size
            self.bit_array[result] = 1
            
    def lookup(self, string):
        for seed in xrange(self.hash_count):
            result = mmh3.hash(string, seed) % self.size
            if self.bit_array[result] == 0:
                return "Nope"
        return "Probably"

bf = BloomFilter(500000, 7)

lines = open("/usr/share/dict/american-english").read().splitlines()
for line in lines:
    bf.add(line)

print bf.lookup("google")
>>> Nope
print bf.lookup("Max")
>>> Probably
print bf.lookup("mice")
>>> Probably
print bf.lookup("3")
>>> Nope

Here's some code we can use to do a time comparison. We'll create a huge list of all our words then iterate through them one by one til we find our target string.

bf = BloomFilter(500000, 7)
huge = []

lines = open("/usr/share/dict/american-english").read().splitlines()
for line in lines:
    bf.add(line)
    huge.append(line)

import datetime

start = datetime.datetime.now()
bf.lookup("google")
finish = datetime.datetime.now()
print (finish-start).microseconds
>>> 57


start = datetime.datetime.now()
for word in huge:
    if word == "google":
        break
finish = datetime.datetime.now()
print (finish-start).microseconds
>>> 9891


start = datetime.datetime.now()
bf.lookup("apple")
finish = datetime.datetime.now()
print (finish-start).microseconds
>>> 10


start = datetime.datetime.now()
for word in huge:
    if word == "apple":
        break
finish = datetime.datetime.now()
print (finish-start).microseconds
>>> 1935

Those are some pretty big time differences. We saw an over 10,000% performance increase which can really make a difference when doing expensive lookups. Not to mention the memory footprint is much smaller than storing each item in a list. 1.1 million bits is only 0.13 megabytes. Not bad at all if you ask me.

You can grab the source from this post on GitHub.

As always if you have any feedback or questions feel free to drop them in the comments below or contact me privately on my contact page. Thanks for reading!

Tags: Python, Data Structures

Comments:

  • Edgeworth - 3 years, 3 months ago

    The second example is kind of silly. Why not use a set?

    reply

  • José Ricardo - 3 years, 3 months ago

    I think that using a set would use more memory.

    reply

  • instagram followers - 9 months, 4 weeks ago

    I recently came across your blog and have been reading along. I thought I would leave my first comment. I don't know what to say except that I have enjoyed reading. Nice blog. I will keep visiting this blog very often.

    reply

  • comments - 9 months, 4 weeks ago

    This blog is so nice to me. I will keep on coming here again and again. Visit my link as well..

    reply

  • Lars - 3 years, 3 months ago

    The benchmark is flawed. Python's lists are not for item lookup; they're for keeping items in a certain order. I just did some profiling using IPython's %timeit on a larger dictionary, and I found 49ms for your linear search, 16.2ms for Python's built-in linear search (the "in" operator), 109ns for Python's set and 1.96μs for the Bloom filter. I.e., set is >20× faster than your BF. (Of course, that's not entirely fair either because of interpreter overhead in your code.) Also, the (false) positives take more time than the negatives, so you're only measuring the easy case here.

    reply

  • Max Burstein - 3 years, 3 months ago

    Definitely a valid point. As José pointed out, the main win here would be the memory size. I just did a simple test of adding 100,000 integers to a set and the size was just over 4mb which isn't too bad. You'll definitely run out of memory much quicker by storing large groups of data in a set. Bloom Filters really strike a good balance between memory size and lookup speed.

    reply

  • how to get real instagram followers - 10 months, 1 week ago

    I found your this post while searching for information about blog-related research ... It's a good post .. keep posting and updating information.

    reply

  • buy instagram comments fast - 10 months, 1 week ago

    it was a wonderful chance to visit this kind of site and I am happy to know. thank you so much for giving us a chance to have this opportunity..

    reply

  • followers twitter - 9 months, 1 week ago

    Thank you for helping people get the information they need. Great stuff as usual. Keep up the great work!!!

    reply

  • get likes - 9 months, 1 week ago

    Thanks for this great post, i find it very interesting and very well thought out and put together. I look forward to reading your work in the future.

    reply

  • Geoff - 3 years, 3 months ago

    On a side note, you should be using the "if "apple" in huge" syntax to check if an element is in your list for a more fair comparison. You'll notice a timing improvement, though it won't be anywhere close to your bloom filter. On my machine, I go from ~1500µs to ~700µs.

    reply

  • cialis - 1 year, 1 month ago

    Hello!

    reply

  • entertainment news - 1 year ago

    Thanks for the blog loaded with so many information. Stopping by your blog helped me to get what I was looking for.

    reply

  • full - 4 weeks ago

    <a href="http://www.gotepisodeguide.com/"> game of thrones season 6 live stream </a>

    <a href="http://www.gotepisodeguide.com/"> game of thrones season 6 live streaming </a>

    <a href="http://www.gotepisodeguide.com/"> hbo game of thrones season 6 live stream </a>

    <a href="http://www.gotepisodeguide.com/"> HBO game of thrones season 6 live streamING </a>

    reply

  • hay day cheats - 1 year ago

    Superbly written article, if only all bloggers offered the same content as you, the internet would be a far better place..

    reply

  • 8 ball pool hack - 1 year ago

    Admiring the time and effort you put into your blog and detailed information you offer!..

    reply

  • soundcloud plays buy - 1 year ago

    This is just the information I am finding everywhere. Thanks for your blog, I just subscribe your blog. This is a nice blog..

    reply

  • buying soundcloud followers - 1 year ago

    This type of message always inspiring and I prefer to read quality content, so happy to find good place to many here in the post, the writing is just great, thanks for the post.

    reply

  • Assignment Help UK - 11 months, 3 weeks ago

    This is a great article thanks for sharing this informative information.

    reply

  • custom essay uk - 11 months, 3 weeks ago

    Your article is really insightful! Looking at your work has enlightened me. Discovered a lot from it. I will take a note of your site and will pursue to go through your forthcoming blog posts. Wonderful! Thanks a lot

    reply

  • buy dissertations - 10 months, 2 weeks ago

    Thank you for this useful tips.

    reply

  • girlsdoporn - 10 months, 1 week ago

    I have joined the oracle Trained in Training in Bangalore. Oracle Trainer will be teaching in practically Manner. So I can Easily able to understand the any Concept. This is very helpful for the interviews Purpose . I would refer my frnds to joined this institute.....

    reply

  • Schlüsseldienst Berlin - 10 months, 1 week ago

    Stunning! This could be a standout amongst the most helpful web journals we have ever go over on thesubject. Really brilliant information! I'm additionally a specialist in this subject so I can comprehend your exertion.

    reply

  • dissertation online help - 10 months, 1 week ago

    Just discovered your online journal and was immediately flabbergasted with all the helpful data that is on it. Extraordinary post

    reply

  • Software Testing Training in Chennai - 10 months, 1 week ago

    Very Informative article on Creating a Simple Bloom Filter. Thanks

    reply

  • how to get a lot of followers on instagram - 10 months ago

    Thank you very much for this useful article. I like it.

    reply

  • likes on instagram - 10 months ago

    Your website is really cool and this is a great inspiring article.

    reply

  • Writing Economics Research Paper - 10 months ago

    Hi there, I would like to know What does this line of code means : maxint = (2**(0.size * 8 -2) -1) How did you even know what are the values for maxint?

    reply

  • green smart living - 9 months, 2 weeks ago

    Such intelligent work on the subject and ideal way of writing here. I am really impressed! This post is a helpful overview of the particular topic and very actionable. Interesting approach!

    reply

  • how to get more subscribers on youtube - 9 months, 2 weeks ago

    This is definitely the knowledge We are acquiring all over the place. Cheers for ones web site, I merely join your blog. This is the wonderful web site. .

    reply

  • youtube comment bot - 9 months, 2 weeks ago

    This is actually the content Now i'm searching for anywhere. Regards for use on your web page, I simply subscribe your blog. They can be a excellent web page. .

    reply

  • learn about debt negotiation - 9 months, 1 week ago

    Truly! Whatever an eye opener this unique put up happens to be in my circumstances. Substantially relished, saved, I just can’t look for further!

    reply

  • Anonymous - 9 months, 1 week ago

    This was a really great contest and hopefully I can attend the next one. It was alot of fun and I really enjoyed myself.. <a href="http://gamehack.org/Boom-Beach-Hack/">boom beach hack</a>|<a href="http://clashofclanstriche.net/">clash of clans</a>

    reply

  • web hosting murah - 9 months ago

    WebiiHost telah memberikan layanan web hosting dan registrasi domain di indonesia sejak tahun 2004 dengan harga yang murah. Untuk memenuhi kebutuhan client , kami juga menyediakan layanan reseller hosting, VPS dan Dedicated Server Indonesia.

    reply

  • buyessayanyday.com - 9 months ago

    We have every reason to believe that you are right. Thanks for this post.

    reply

  • Anonymous - 9 months ago

    Here's a better Wolfram Alpha link:

    http://www.wolframalpha.com/input/?i=%281-%281-1%2Fm%29^%28k*n%29%29^k++for+k%3D7+%2C+n%3D100000%2C+m%3D1100000

    reply

  • this site - 9 months ago

    This article gives the light in which we can observe the reality. This is very nice one and gives indepth information. Thanks for this nice article.

    reply

  • IB Maths tutor - 8 months, 3 weeks ago

    I would like to thank you for the efforts you have made in writing this article. I am hoping the same best work from you in the future as well.

    reply

  • youtube subscribers - 8 months, 3 weeks ago

    I felt very happy while reading this site. This was really very informative site for me. I really liked it. This was really a cordial post. Thanks a lot!.

    reply

  • free youtube views - 8 months, 3 weeks ago

    I really like your take on the issue. I now have a clear idea on what this matter is all about..

    reply

  • Puja Kumari - 8 months, 2 weeks ago

    Thanks for post this helpful post - Please visit for More information about: http://www.top9th.in/packers-and-movers-pune/ http://www.top9th.in/packers-and-movers-noida/ http://www.top9th.in/packers-and-movers-chennai/ http://www.top9th.in/packers-and-movers-kolkata/ http://www.top9th.in/packers-and-movers-ahmedabad/ http://www.top9th.in/packers-and-movers-surat/

    reply

  • Puja Kumari - 8 months, 2 weeks ago

    Thanks for Nice and Informative Post. This article is really contains lot more information about This Topic http://www.top9th.in/packers-and-movers-bangalore/ http://www.top9th.in/packers-and-movers-hyderabad/ http://www.top9th.in/packers-and-movers-mumbai/ http://www.top9th.in/packers-and-movers-delhi/ http://www.top9th.in/packers-and-movers-gurgaon/ http://www.top9th.in/

    reply

  • buy stilnoct no prescription needed - 8 months, 1 week ago

    I am so much pleased to get this planning. I am looking ahead to read more about it. This is a very lovely post.

    reply

  • game of thrones episode guide - 4 weeks ago

    <a href="http://www.season6gameofthrones.com">season 6 game of thrones</a>

    http://www.season6gameofthrones.com/bran-stark-meets-nights-king-got6/

    http://www.season6gameofthrones.com/game-thrones-season-episodes/

    http://www.season6gameofthrones.com/jaime-returns-kings-landing-dead-myrcella/

    http://www.season6gameofthrones.com/game-thrones-season-6-spoilers/

    http://www.season6gameofthrones.com/jon-snow-might-alive-game-thrones/

    http://www.season6gameofthrones.com/melisandre-try-resurrect-jon-snow/

    http://www.season6gameofthrones.com/faith-militant-vs-crown/

    http://www.season6gameofthrones.com/interesting-things-watch-season-6/

    http://www.season6gameofthrones.com

    reply

  • Spermaspender gesucht - 7 months, 3 weeks ago

    Many thanks for the exciting blog posting! Simply put your blog post to my favorite blog list and will look forward for additional updates.

    reply

  • Harry - 3 weeks, 5 days ago

    I really thank the author of this article, the article has great significance for me, it just leads me to do better things for this life. Very meaningful, thank you very much, wish you happy.Viral Loop Review best ultimate viral wordpress them ViralLoop Review tool Viral Loop Free - Viral Loop Bonus

    reply

  • Lisa Henry - 2 weeks, 3 days ago

    Here are the simple steps to add an email address to your Hotmail contacts [url=http://hotmailsigninvl.com/]Hotmail Login[/url] [url=http://hotmailsigninvl.com/]>Hotmail Sign In[/url] [url=http://hotmailsigninvl.com/]Sign In To Hotmail[/url] [url=http://hotmailsigninvl.com/how-to-recover-hotmail-password/]Recover Hotmail Password[/url] [url=http://hotmailsigninvl.com/how-to-create-a-hotmail-account/]Create Hotmail Account[/url] Thanks for sharing nice information with us. i like your post and all you share with us is uptodate and quite informative, i would like to bookmark the page so i can come here again to read you, as you have done a wonderful job.

    reply

  • mike91 - 2 weeks, 3 days ago

    <a href="http://www.jeemainresults2016.co.in/">JEE Main Result 2016</a>

    <a href="http://www.keralahseresult2016.co.in/">Kerala DHSE Result 2016</a>

    <a href="http://karnatakapucresult2016.in/">Karnataka PUC Result 2016</a>

    <a href="http://www.jac10thresult2016.co.in/">JAC 10th Result 2016</a>

    <a href="http://cgbse10thresults2016.in/">Chhattisgarh Board 10th Result 2016</a>

    reply

  • Lisa Henry - 2 weeks, 3 days ago

    http://hotmailsigninvl.com/ Thanks for sharing nice information with us. i like your post and all you share with us is uptodate and quite informative, i would like to bookmark the page so i can come here again to read you, as you have done a wonderful job.

    reply

  • Deepak Kumar - 1 week, 4 days ago

    Packers and Movers Mumbai goto http://www.movers5th.in/packers-and-movers-mumbai/ Packers and Movers Pune goto http://www.movers5th.in/packers-and-movers-pune/ Packers and Movers Bangalore goto http://www.movers5th.in/packers-and-movers-bangalore/

    reply

  • Deepak Kumar - 1 week, 4 days ago

    Packers and Movers Gurgaon goto http://www.movers5th.in/packers-and-movers-gurgaon/ Packers and Movers Hyderabad goto http://www.movers5th.in/packers-and-movers-hyderabad/ Packers and Movers Delhi goto http://www.movers5th.in/packers-and-movers-delhi/

    reply

  • Ritesh Kumar - 1 week, 4 days ago

    Packers and Movers Noida @ http://www.movers5th.in/packers-and-movers-noida/ Packers and Movers Chennai @ http://www.movers5th.in/packers-and-movers-chennai/ Packers and Movers Navi Mumbai @ http://www.movers5th.in/packers-and-movers-navimumbai/ Packers and Movers Thane @ http://www.movers5th.in/packers-and-movers-thane/ Packers and Movers Ghaziabad @ http://www.movers5th.in/packers-and-movers-ghaziabad/ Packers and Movers Faridabad @ http://www.movers5th.in/packers-and-movers-faridabad/

    reply

  • Deepak Kumar - 1 week, 4 days ago

    Movers and Packers Hyderabad @ http://www.moveby5th.in/packers-and-movers-hyderabad.html Movers and Packers Pune @ http://www.moveby5th.in/packers-and-movers-pune.html Movers and Packers Bangalore @ http://www.moveby5th.in/packers-and-movers-bangalore.html Movers and Packers Mumbai @ http://www.moveby5th.in/packers-and-movers-mumbai.html Movers and Packers Delhi @ http://www.moveby5th.in/packers-and-movers-delhi.html Movers and Packers Gurgaon @ http://www.moveby5th.in/packers-and-movers-gurgaon.html

    reply

  • Devika - 1 week, 4 days ago

    Check your tamilnadu 12th results here: http://tn12thresults2016.co.in/

    reply

  • michael-kors - 1 week, 1 day ago

    Miranda Kerr Has Perfectly Simple Supermodel Shoe Style http://www.giuseppezanotti.us.org Women Giuseppe Zanotti Sneakers - giuseppe zanotti shoes My love for handbags started at a young age, and it began with lusting, not owning. I used to go to the mall with my friends in junior high, and while they were in Limited Too, I was in department stores, looking at bags. http://www.coachfactoryoutlet.org Coach Outlet - Hot Deals, Reviews and Guides about Coach Produtcs 2016 michael kors outlet so yes, http://www.michael-korsoutletonline.us.com first http://www.coachfactoryoutletonlineco.us.com order http://www.oakleysunglassesok.us.com of business: http://www.oakleysunglassesoutlets.us.org we've got http://www.raybansunglasses.eu.com a lot http://www.ralphlaurenpolos.us.com of http://www.michaelkorsoutlet.us.com Rihanna http://www.coach-factoryoutletstore.us.com for http://www.louis-vuittonoutletonline.us.com you http://www.michaelkorsoutletonline.us.com this http://www.valentinoshoesoutlet.com week, and http://www.valentinooutlet.us.org in one http://www.sacsportefeuilles.com instance,http://www.louisvuittonhandbag.org she is http://www.oakleysunglassesstore.us.com actually http://www.coachoutletstore.us.com carrying http://www.coachfactoryoutletinc.us.com something http://www.coachfactoryoutletusa.com other http://www.michaelkorsoutletonlines.us.com than Dior.http://www.michaelkorsoutletinc.us.com (Nobody http://www.michael-korsoutletonline.com tell http://www.michaelkorsoutlet.eu.com Dior's PR http://www.michaelkorsfactoryoutletonline.us.com department.http://www.oakleyoutlet.name oakley outlet) Is http://www.oakleystoresonline.us.com there http://www.truereligionoutletonline.in.net any http://www.marcbymarcjacobs.us.org

    coach factory outlet http://www.coachfactoryoutlet-stores.us.com selling http://www.sacportefeuillespascher.com point than Rihanna? http://www.coach-factoryoutletstores.us.com No, http://www.michaelkorsfactorystore.us.com no there is http://www.north-faceoutlet.us.com not. http://www.coachfactoryoutlet-stores.com Second http://www.ray-banpascher.com order http://www.sacblanc.com of http://www.guccisneakers.us.com business: http://www.raybanstoresonline.us.com celebrities http://www.raybansunglassesclearance.com are http://www.coachoutletusaonline.us.com keeping http://www.tomsshoe.us.com it extra http://www.true-religions.us.com casual http://www.oakleygoggles.us.com this http://www.portefeuillepascher.com week. http://www.coachoutletonline.us coach outlet online http://www.valentino.com.co Sweatpants http://www.valentinoshoes.com.co at http://www.coachfactoryoutletstore.us.com the http://www.coachoutletstores.net airport.http://www.oakleyvaults.us.com Pajamas http://www.pradaoutletstore.us.com at http://www.michaelkorssoutlet.us.com award http://www.christianlouboutinsale.us.com show.http://www.valentinoshoeswebsite.com It http://www.christian-louboutinshoes.us.com makes http://www.beatsbydrdreheadphone.us.com Here's a photo of Dakota Fanning http://www.michaelkors-outletonline.us.com carrying http://www.michaelkorshandbagsoutlet.us.com a mini http://www.beatsbydreheadphone.us.com Givenchy http://www.louboutinshoes.us.com bag around http://www.christianlouboutinoutletonline.us.com the http://www.coachstores.us.com East http://www.coachoutletonlinesale.us.com Village.

    michael kors http://www.onlinemichaelkorsoutlet.us.com Someone http://www.truereligionoutlet.us.com clearly took this http://www.coachoutletstorefactory.us.com on http://www.ray-bansunglasses.it their http://www.cheapsunglassesoakley.us.com cell http://www.coach-factorystoreonline.us.com phone and http://www.michaelkorsoutletco.us.com sold to our http://www.tomsshoesoutletonline.in.net photo http://www.truereligionoutletco.us.com service. http://www.truereligionjeans.us.org Dakota sees http://www.rayban-sunglasses.org you http://www.redbottomshoe.us.com trying to take discreet hip http://www.valentinorockstudshoes.us.com shots http://www.valentinoshoesoutlet.us.com with http://www.sacenpaille.com your http://www.beatsbydrdre.us.com smart phone,http://www.ray-bansunglasses.org ray ban sunglasses aspiring http://www.louboutinoutlet.us.com paparazzo,http://www.modeldesac.com and does not http://www.clshoesoutletonline.com approve. http://www.raybanoutletstoreonline.com I http://www.tomsoutlet.us.com wear a http://www.louboutinshoes.us.com lot of http://www.christianlouboutinoutlet.us.com black, http://www.hermesbirkinoutlet.us.com dye http://www.hermesbag.us.com my hair http://www.wwwcoachoutlet.us.com bright http://www.coachfactory.name colors http://www.raybansunglasse.us.com and http://www.coachoutletstoresfactory.com coach factory outlet sale don't own a http://www.bracelethermes.com single http://www.factorystoreonline.com fashion-related http://www.coachpurses.us.com thing http://www.outletgucci.us.com that http://www.coachfactory-storeoutlet.us.com would http://www.valentinoshoes.us.org qualify http://www.louisvuittonoutlet-online.us.com as http://www.louisvuittonlv.us.com even http://www.michaelkorsoutletsite.us.com vaguely http://www.michaelkorshandbags.us.com pastel, http://www.valentinorockstud.us.com but http://www.louisvuittonpurses.us.com somehow

    louis vuitton http://www.bananelouisvuitton.com they're http://www.oakleysunglasse.us.com the only http://www.pascherhermes.com shades that feel http://www.womens-jewelry.com consistently coright on http://www.valentinooutlet.org my http://www.valentinoshoes.us.com nails. http://www.oakleyssunglassesoutlet.us.com Even http://www.ray-bansunglassesclearance.com black, http://www.raybanoutlet.org my http://www.raybanstoreonline.us.com favorite http://www.raybansunglassesoutlets.com color http://www.cheapoakleyssunglasses.us.com for http://www.cheapoakleysunglasses.com.co everything http://www.tomsoutlet.in.net else http://www.abercrombieoutletonline.us.com I put on my http://www.ralphlauren-outlet.us.com body, doesn't http://www.poloralphlaurenoutlet.us.com fit http://www.raybansoutlet.us.com all http://www.coach-factorystoreoutlet.us.com the http://www.wwwcoachfactoryoutlet.com time, http://www.coach-factorystore.us.com but http://www.coach-factoryonline.us.com I've http://www.coach-outletfactoryonline.us.com never had a http://www.outlet-michaelkors.us.com baby http://www.raybanssunglasses.us.com blue http://www.polosralphlauren.us.com or http://www.sacportefeuilleeu.com mint http://www.marcjacobsoutlet.us.org green manicure I've regretted.http://www.marcbymarcjacobs.us Pale nails have been an outlier in my life for years. http://www.marcbymarcjacobsoutlets.com http://www.chiflatironsofficial.com

    Katherine Callaghan 6546 Comments

    Well, we’ve made it. Here’s to another week marked off our calendar and another fun-filled weekend to look forward to. While the sales were scarce this week, we searched the internet high and low to bring you the best-priced bags we could find.

    reply

  • gg - 5 days, 3 hours ago

    <a href="http://www.movies-out.com" rel="dofollow">movies coming out this weekend</a>

    <a href="http://www.movies-out.com" rel="dofollow">romance movies coming out this year</a>

    <a href="http://www.movies-out.com" rel="dofollow">movie released today</a>

    reply

  • Ray-Ban - 1 day, 5 hours ago

    <a href="http://www.kd7shoes.us.com/"><strong>Kevin Durant Shoes</strong></a>

    reply

  • Ray-Ban - 1 day, 5 hours ago

    [url=http://www.kd7shoes.us.com/][b]Kevin Durant Shoes[/b][/url]

    reply

  • Ray-Ban - 1 day, 5 hours ago

    Kevin Durant Shoes, http://www.kd7shoes.us.com

    reply