Filtered by JavaScript, Python

Page 27

Reset

Using lazy loading images on Air Mozilla

April 23, 2015
0 comments Mozilla, JavaScript

Starting today, (almost) all the thumbnails below the fold on Air Mozilla are not loaded.

The way it works, is that I use a library called Lazyr.js which notices when you scroll down and when certain pictures are going to be in view, it changes the <img> tag's src.

So it basically looks like this:


<article>
  <h3>Event 1</h3>
  <img src="event1.png">
</article>

<article>
  <h3>Event 2</h3>
  <img src="event2.png">
</article>

<article>
  <h3>Event 3</h3>
  <img src="event3.png">
</article>

<article>
  <h3>Event 4</h3>
  <img src="placeholder.png" data-lazyr="event4.png">
</article>

<article>
  <h3>Event 5</h3>
  <img src="placeholder.png" data-lazyr="event5.png">
</article>

<article>
  <h3>Event 6</h3>
  <img src="placeholder.png" data-lazyr="event6.png">
</article>

That means that to load this page it needs to download, only:

event1.png
event2.png
event3.png
placeholder.png

Only 4 images instead of the otherwise 6 (in this example).

When you scroll down to see the rest of the list, it then also downloads:

event4.png
event5.png
event6.png

The actual numbers on Air Mozilla is that there are 10 events page page and I lazy load 6 of them.

You can see the results when comparing this WebPageTest with this one.

There is more work to do though. At the moment, the thumbnails in the sidebar (Trending and Upcoming events) are above the fold when you're browsing but below the fold when you're viewing an individual event. That's something I have yet to implement.

AngularJS $q notify and resolve as a local GET proxy

April 18, 2015
1 comment AngularJS, JavaScript

Something tells me there are already solutions like this out there that are written by much smarter people who have tests and package.json etc. Perhaps my Friday-brain failed at googling them up.

So, the issue I'm having is an angular app that uses a ui-router to switch between controllers.

In almost every controller it looks something like this:


app.controller('Ctrl', function($scope, $http) {
  /* The form that needs this looks something like this:
      <input name="first_name" ng-model="stuff.first_name">
   */
  $scope.stuff = {};
  $http.get('/stuff/')
  .success(function(response) {
    $scope.stuff = response.stuff;
  })
  .error(function() {
    console.error.apply(console, arguments);
  });
})

(note; ideally you push this stuff into a service, but doing it here in the controller illustrates what matters in this point)

So far so good. But so far so slow.

Every time the controller is activated, the AJAX GET is fired and it might be slow because of network latency.
And I might switch to this controller repeatedly within one request/response session of loading the app.

So I wrote this:


app.service('localProxy',
    ['$q', '$http', '$timeout',
    function($q, $http, $timeout) {
        var service = {};
        var memory = {};

        service.get = function(url, store, once) {
            var deferred = $q.defer();
            var already = memory[url] || null;
            if (already !== null) {
                $timeout(function() {
                    if (once) {
                        deferred.resolve(already);
                    } else {
                        deferred.notify(already);
                    }
                });
            } else if (store) {
                already = sessionStorage.getItem(url);
                if (already !== null) {
                    already = JSON.parse(already);
                    $timeout(function() {
                        if (once) {
                            deferred.resolve(already);
                        } else {
                            deferred.notify(already);
                        }
                    });
                }
            }

            $http.get(url)
            .success(function(r) {
                memory[url] = r;
                deferred.resolve(r);
                if (store) {
                    sessionStorage.setItem(url, JSON.stringify(r));
                }
            })
            .error(function() {
                deferred.reject(arguments);
            });
            return deferred.promise;
        };

        service.remember = function(url, data, store) {
            memory[url] = data;
            if (store) {
                sessionStorage.setItem(url, JSON.stringify(data));
            }
        };

        return service;
    }]
)

And the way you use it is that it basically returns twice. First from the "cache", then from the network request response.

So, after you've used it at least once, when you request data from it, you first get the cached stuff (from memory or from the browser's sessionStorage) then a little bit later you get the latest and greatest response from the server. For example:


app.controller('Ctrl', function($scope, $http, localProxy) {
  $scope.stuff = {};
  localProxy('/stuff/')
  .then(function(response) {
    // network response
    $scope.stuff = response.stuff;
  }, function() {
    // reject/error
    console.error.apply(console, arguments);
  }, function(response) {
    // update
    $scope.stuff = response.stuff;
  });
})

Note how it sets $scope.stuff = response.stuff twice. That means that the page can load first with the cached data and shortly after the latest and greatest from the server.
You get to look at something whilst waiting for the server but you don't have to worry too much about cache invalidation.

Sure, there is a risk. If your server response is multiple seconds slow, your user might for example, start typing something into a form (once it's loaded from cache) and when the network request finally resolves, what xhe typed in is overwritten or conflicting.

The solution to that problem is that you perhaps put the form in a read-only mode until the network request resolves. At least you get something to look at sooner rather than later.

The default implementation above doesn't store things in sessionStorage. It just stores it in memory as you're flipping between controllers. Alternatively, you might want to use a more persistent approach so then you instead use:


controller( // same as above
  localProxy('/stuff/', true)
  // same as above
)

Sometimes there's data that is very unlikely to change. Perhaps you just need the payload for a big drop-down widget or something. In that case, it's fine if it exists in the cache and you don't need a server response. Then set the third parameter to true, like this:


controller( // same as above
  localProxy('/stuff/', true, true)
  // same as above
)

This way, it won't fire twice. Just once.

Another interesting expansion on this is, if you change the data after it comes back. A good example is if you request data to fill in a form that user updates. After the user has changed some of it, you might want to pre-emptivly cache that too. Here's an example:


app.controller('Ctrl', function($scope, $http, localProxy) {
  $scope.stuff = {};
  var url = '/stuff/';
  localProxy(url)
  .then(function(response) {
    // network response
    $scope.stuff = response.stuff;
  }, function() {
    // reject/error
    console.error.apply(console, arguments);
  }, function(response) {
    // update
    $scope.stuff = response.stuff;
  });

  $scope.save = function() {
      // update the cache
      localProxy.remember(url, $scope.stuff); 
      $http.post(url, $scope.stuff);
  };
})

What do you think? Is it useful? Is it "bonkers"?

I can think of one possible beautification, but I'm not entirely sure how to accomplish it.
Thing is, I like the API of $http.get that it returns a promise with a function called success, error and finally. The ideal API would look something like this:


app.controller('Ctrl', function($scope, $http) {
  $scope.stuff = {};
  // angular's $http service expanded
  $http.getLocalProxy('/stuff/')
  .success(function(cached, response) {
    /* Imagine something like:
        <p class="warning" ng-if="from_cache">Results you see come from caching</p>
     */
    $scope.from_cache = cached;
    $scope.stuff = response.stuff;
  })
  .error(function() {
    console.error.apply(console, arguments);
  });
})

That API looks and feels just like the regular $http.get function but with an additional first argument to the success promise callback.

Match the whole word in auto complete maybe

April 10, 2015
2 comments Web development, JavaScript

(For context, I released Autocompeter.com last week and now I'm thinking about improvements)

I posted a question on Twitter about which highlighting formatting people prefer and got some interesting feedback. More about that later.

The piece of feedback that really got my attention came from my friend Honza Král.
He wondered if not the whole word should be highlighted instead of just the beginning of the word.

I've actually been thinking about that too but never got around to trying it out. Until now.

Before

Before

After

After

What do you think?

I have the code in a branch and I'm still mulling it over. There's sort of a convention to just highlight based on what you've typed so far. I don't want to be too weird because when people don't feel familiar they don't like what they see even if the new actually is better.

gulp-header is the best!

April 9, 2015
1 comment JavaScript

For Autocompeter I develop with gulp. It's like Grunt but better.

One thing I wanted was that when it makes the src/autocompeter.js --> minify() --> dist/autocompeter.min.js step I also wanted to put in a little preample header into the minified file.

First I thought, since UglifyJS supports a --preamble option that that'd be the route to go. I didn't get very far.

Then I thought I had to write my own plugin. So I started reading the documentation about how to write a plugin and partially thinking "Oh I don't have time to do this" and also "Oh finally a chance to sit down and really understand how gulp plugins work". I was wrong. The documentation for writing plugins say:

"Your plugin shouldn't do things that other plugins are responsible for... ...It should not add headers, gulp-header does that"

Oh! So there is already a great plugin for this! Long story short; here's how I used it. The output is that the version number is now on the first line of autocompeter.min.js.

I'm starting to like gulp more and more. There's even a dedicate nice index of all available plugins.

Autocompeter 1.1.8 and smooth typing

April 6, 2015
0 comments Go, JavaScript

One of the most constructive pieces of feedback I got when Autocompeter was on Hacker News was that when you type something with lots of predictable results the results overlay would "flicker".

E.g. you type "javascript" in a nice and steady pace and the overlay would shrink and grow and shrink and grow very rapidly.

The reason it happened was due to a bug in the javascript code that filtered results whilst waiting for the next AJAX request from the latest typed character. E.g. you type "ash" and the results comes back with "ashley", "ashes", "Ashford". Then you add a "l" so now we start a new AJAX query for "ashl" and whilst waiting for that output from the server we can start filtering out "ashes" and "Ashford" because we can pre-emptively know that that won't be in the new result set.

The bug was a bad function that filtered the existing results on a second rendering whilst waiting for the next AJAX. It was easy to fix and this is included in version 1.1.8.

The reason I failed to notice this was because I had inserted some necessary optimizations when the network latency was very very slow but hadn't tested it in a realistic network latency environment. E.g. a decent DSL connection but nevertheless something more advanced that just connecting to localhost.

Autocompeter.com

April 2, 2015
8 comments Go, JavaScript

About a year ago I found a great article about how to use Redis to index every prefix of every word as an index and thus a super fast way of building an autocomplete service. The idea is that you take all your titles and index them like this; if the title is "My Title" you store a key for m, my, t, ti, tit, titl and title. That means you can do very fast lookups as someone is typing unfinished words.

Anyway. I was running this merrily here on my personal blog but I liked it so much and I wanted to use it on aother site for work that I thought it'd be time to extract it into its own little microservice. All I needed was a name and my friend and colleague jezdez suggested I call it "autocompeter". So that it became.

The original implementation was written in Python and at the time I was learning Go and was eager to have something to build in Go. So I built this microservice in Go using the Negroni web framework.

The idea is that you own and run a website. You have a search feature on your website but you don't have a nifty autocomplete (aka. live search) thing on it. So, you send me all your titles, URLs and optionally their "popularity ranking" (basically a score). I'll index them on autocompeter.com under your domain. You have to sign in with GitHub to set up an API Auth Key.

Then, you put this into your HTML:


<script src="//cdn.jsdelivr.net/autocompeter/1/autocompeter.min.js"></script>
<script>
Autocompeter(document.querySelector('input[name="q"]');
</script>

Also, you'll need to download the CSS and put into your site. I don't recommend pointing to a CDN for CSS.

And that's all you have to do. The REST API, options for the Javascript integration and CSS integration in the documentation.

The Javascript is framework-free meaning it's just pure DOM manipulation and works in IE and modern browsers. The minified file is only 4.2Kb minified (2Kb gzipped).

All code is Open Source under a BSD license. Everything is free but there's no SLA as of yet.

I'm going to be blogging more and more about feature development, benchmarks and other curious things I learn developing this further.

Bye bye httpretty. Welcome back good old mock!

March 19, 2015
4 comments Python

After a day of pushing 9 commits to a PR to finally get Travis to build a simple python package on python 2.6, 2.7, 3.3 and 3.4 I finally gave up and ripped out all of httpretty and replaced it with good old mock.patch()

I was getting all sorts of strange warnings in py3.3 and 3.4 got stuck all the time.
This is not the first time httpretty has been causing confusion so from now on I'm giving up on httpretty. Ithink it was too good to be true to work reliably. Honestly, it might be python's fault for not being better made available to cool libs like httpretty.

By the way, here's one of those errors where Python 3.4 just hangs which stopped being the case once I took out httpretty. And here you can see the clear failure to deactivate the monkeypatch even after the test is complete in Python 3.3.

Median size of Javascript libs on jsDelivr

February 24, 2015
0 comments JavaScript

If you haven't heard of jsDelivr then you've missed out on a great project! It's basically a free CDN to use for Open Source projects. I added my own yesterday and it was as easy as making a pull request with the initial file, some metadata and a file that tells them where to pick up new versions from (e.g. GitHub).

Anyway, they now host A LOT of files. 8,941 to be exact (1,927 unique file names), at the time of writing. So I thought I'd check out what the median size is.

The median size is: 7.4Kb

The average is 28.6Kb and the standard deviation is 73Kb! so we can basically ignore that and just focus on the median.

Check out the histogram

That's pretty big! If you exclude those bigger than 100Kb the median shrinks to 6.5Kb. Still pretty big.

I'm proud to say that my own is only 50% the size of the median size.

Best non-cryptographic hashing function in Python (size and speed)

February 21, 2015
11 comments Python

First of all; hashing is hard. But fortunately it gets a little bit easier if it doesn't have to cryptographic. A non-cryptographic hashing function is basically something that takes a string and converts it to another string in a predictable fashion and it tries to do it with as few clashes as possible and as fast as possible.

MD5 is a non-cryptographic hashing function. Unlike things like sha256 or sha512 the MD5 one is a lot more predictable.

Now, how do you make a hashing function that yields a string that is as short as possible? The simple answer is to make the output use as many different characters as possible. If a hashing function only returns integers you only have 10 permutations per character. If you instead use a-z and A-Z and 0-9 you now have 26 + 26 + 10 permutations per character.

A hex on the other hand only uses 0-9 and a-f which is only 10 + 6 permutations. So you need a longer string to be sure it's unique and can't clash with another hash output. Git for example uses a 40 character log hex string to prepresent a git commit. GitHub is using an appreviated version of that in some of the web UI of only 7 characters which they get away with because things are often in a context of a repo name or something like that. For example github.com/peterbe/django-peterbecom/commit/462ae0c

So, what other choices do you have when it comes to returning a hash output that is sufficiently long that it's "almost guaranteed" to be unique but sufficiently short that it becomes practical in terms of storage space? I have an app for example that turns URLs into unique IDs because they're shorter that way and more space efficient to store as values in a big database. One such solution is to use a base64 encoding.

Base64 uses a-zA-Z0-9 but you'll notice it doesn't have the "hashing" nature in that it's just a direct translation character by character. E.g.


>>> base64.encodestring('peterbengtsson')
'cGV0ZXJiZW5ndHNzb24=\n'
>>> base64.encodestring('peterbengtsson2')
'cGV0ZXJiZW5ndHNzb24y\n'

I.e. these two strings are different but suppose you were to take only the first 10 characters these would be the same. Basically, here's a terrible hashing function:


def hasher(s):  # this is not a good hashing function
    return base64.encodestring(s)[:10]

So, what we want is a hashing function that returns output that is short and very rarely clashing and does this as fast as possible.

To test this I wrote a script that tried a bunch of different ad-hoc hashing functions. I generate a list of 130,000+ different words with an average length of 15 characters. Then I loop over these words until a hashed output is repeated for a second time. And for each, I take the time it takes to generate the 130,000+ hashes and I multiply that with the total number of bytes. For example, if the hash output is 9 characters each in length that's (130000 * 9) / 1024 ~= 1142Kb. And if it took 0.25 seconds to generate all of those the combined score is 1142 * 0.24 ~= 286 bytes second.

Anyway, here are the results:

h11 100.00  0.217s     1184.4 Kb   257.52 kbs
h6  100.00  1.015s  789.6 Kb    801.52 kbs
h10 100.00  1.096s  789.6 Kb    865.75 kbs
h1  100.00  0.215s  4211.2 Kb   903.46 kbs
h4  100.00  1.017s  921.2 Kb    936.59 kbs

(kbs means "kilobytes seconds")

These are the functions that returned 0 clashes amongst 134,758 unique words. There were others too that I'm not bothering to include because they had clashes. So let's look at these functions:



def h11(w):
    return hashlib.md5(w).hexdigest()[:9]

def h6(w):
    h = hashlib.md5(w)
    return h.digest().encode('base64')[:6]

def h10(w):
    h = hashlib.sha256(w)
    return h.digest().encode('base64')[:6]

def h1(w):
    return hashlib.md5(w).hexdigest()

def h4(w):
    h = hashlib.md5(w)
    return h.digest().encode('base64')[:7]    

It's kinda arbitrary to say the "best" one is the one that takes the shortest time multipled by size. Perhaps the size matters more to you in that case the h6() function is better because it returns 6 character strings instead of 9 character strings in h11.

I'm apprehensive about publishing this blog post because I bet I'm doing this entirely wrong. Perhaps there are better ways to digest a hashing function that returns strings that don't need to be base64 encoded. I just haven't found any in the standard library yet.

Almost premature optimization

January 2, 2015
0 comments Python, Web development, Django

In airmozilla the tests almost all derive from one base class whose tearDown deletes the automatically generated settings.MEDIA_ROOT directory and everything in it.

Then there's some code that makes sure a certain thing from the fixtures has a picture uploaded to it.

That means it has do that shutil.rmtree(directory) and that shutil.copy(src, dst) on almost every single test. Some might also not need or depend on it but it's conveninent to put it here.

Anyway, I thought this is all a bit excessive and I could probably optimize that by defining a custom test runner that is first responsible for creating a clean settings.MEDIA_ROOT with the necessary file in it and secondly, when the test suite ends, it deletes the directory.

But before I write that, let's measure how many gazillion milliseconds this is chewing up.

Basically, the tearDown was called 361 times and the _upload_media 281 times. In total, this adds to a whopping total of 0.21 seconds! (of the total of 69.133 seconds it takes to run the whole thing).

I think I'll cancel that optimization idea. Doing some light shutil operations are dirt cheap.