RSS 2.0
Sign In
# Tuesday, January 26, 2016

Until now we've been aware of 2 excelent artificial intelligence frameworks written in C#. These are AForge.NET and its successor Accord.NET. The both include a lot of algorithms for solving wide range of tasks.

Yesterday we've discovered that Microsoft has published as an open-source project their Computational Network Toolkit in order to speed up advances in artificial intelligence and made it available to a broader group of developers.

The sources written in C++ which scales good and uses GPUs. This gives a competitive advantage to CNTK, see more details here.

Although the main aim of such development was speech recognition, the CNTK contains a Neural Network framework that may be used for other artificial intelligence tasks.

Tuesday, January 26, 2016 9:09:11 AM UTC  #    Comments [0] -
Announce
# Thursday, January 14, 2016

It's very old theme...

Many years ago we have defined a .NET wrapper around Windows Uniscribe API.

Uniscribe API is used to render bidirectional languages like Hebrew, so it's important mainly here in Israel.

Once in a while we get request from people to give that API, so we published it on GitHub at https://github.com/nesterovsky-bros/BidiVisualConverter.

You're welcome to use it!

Thursday, January 14, 2016 2:19:54 PM UTC  #    Comments [2] -
.NET | Announce | Tips and tricks
# Monday, January 4, 2016

Essence of the problem (see Error during transformation in Saxon 9.7, thread on forum):

  1. XPath engine may arbitrary reorder predicates whose expressions do not depend on a context position.
  2. While an XPath expression $N[@x castable as xs:date][xs:date(@x) gt xs:date("2000-01-01")] cannot raise an error if it's evaluated from the left to right, an expression with reordered predicates $N[xs:date(@x) gt xs:date("2000-01-01")][@x castable as xs:date] may generate an error when @x is not a xs:date.

To avoid a potential problem one should rewrite the expression like this: $N[if (@x castable as xs:date) then xs:date(@x) gt xs:date("2000-01-01") else false()].

Please note that the following rewrite will not work: $N[(@x castable as xs:date) and (xs:date(@x) gt xs:date("2000-01-01"))], as arguments of and expression can be evaluated in any order, and error that occurs during evaluation of any argument may be propageted.

With these facts we faced a task to check our code base and to fix possible problems.

A search has brought ~450 instances of XPath expessions that use two or more consequtive predicates. Accurate analysis limited this to ~20 instances that should be rewritten. But then, all of sudden, we have decided to commit an experiment. What if we split XPath expression in two sub expressions. Can error still resurface?

Consider:

<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  xmlns:xs="http://www.w3.org/2001/XMLSchema">

  <xsl:variable name="elements" as="element()+"><a/><b value="c"/></xsl:variable>

  <xsl:template match="/">
    <xsl:variable name="a" as="element()*" select="$elements[self::d or self::e]"/>
    <xsl:variable name="b" as="element()*" select="$a[xs:integer(@value) = 1]"/>

    <xsl:sequence select="$b"/>
  </xsl:template>

</xsl:stylesheet>

As we expected Saxon 9.7 internally assembles a final XPath with two predicates and reorders them. As result we get an error:

Error at char 20 in xsl:variable/@select on line 8 column 81 of Saxon9.7-filter_speculation.xslt:
  FORG0001: Cannot convert string "c" to an integer

This turn of events greately complicates the code review we have to commit.

Michiel Kay's answer to this example:

I think your argument that the reordering is inappropriate when the expression is written using variables is very powerful. I shall raise the question with my WG colleagues.

In fact we think that either: reordering of predicates is inappropriate, or (weaker, to allow reordering) to treat an error during evaluation of predicate expression as false(). This is what is done in XSLT patterns. Other solutions make XPath less intuitive.

In other words we should use XPath (language) to express ideas, and engine should correctly and efficiently implement them. So, we should not be forced to rewrite expression to please implementation.

Monday, January 4, 2016 10:07:12 AM UTC  #    Comments [0] -
Thinking aloud | xslt
# Saturday, January 2, 2016

On December, 30 we have opened a thread in Saxon help forum that shows a stylesheet generating an error. This is the stylesheet:

<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  xmlns:xs="http://www.w3.org/2001/XMLSchema">

  <xsl:variable name="elements" as="element()+"><a/><b value="c"/></xsl:variable>

  <xsl:template match="/">
    <xsl:sequence select="$elements[self::d or self::e][xs:integer(@value) = 1]"/>
  </xsl:template>

</xsl:stylesheet>

We get an error:

Error at char 47 in xsl:sequence/@select on line 7 column 83 of Saxon9.7-filter_speculation.xslt:
  FORG0001: Cannot convert string "c" to an integer
Exception in thread "main" ; SystemID: .../Saxon9.7-filter_speculation.xslt; Line#: 7; Column#: 47
ValidationException: Cannot convert string "c" to an integer
  at ...

It's interesting that error happens in Saxon 9.7 but not in earlier versions.

The answer we got was expected but disheartening:

The XPath specification (section 2.3.4, Errors and Optimization) explicitly allows the predicates of a filter expression to be reordered by an optimizer. See this example, which is very similar to yours:

The expression in the following example cannot raise a casting error if it is evaluated exactly as written (i.e., left to right). Since neither predicate depends on the context position, an implementation might choose to reorder the predicates to achieve better performance (for example, by taking advantage of an index). This reordering could cause the expression to raise an error.

$N[@x castable as xs:date][xs:date(@x) gt xs:date("2000-01-01")]

Following the spec, Michael Kay advices us to rewrite XPath:

$elements[self::d or self::e][xs:integer(@value) = 1]

like this:

$elements[if (self::d or self::e) then xs:integer(@value) = 1 else false()]

Such subtleties make it hard to reason about and to teach XPath. We doubt many people will spot the difference immediately.

We think that if such optimization was so much important to spec writers, then they had to change filter rules to treat failed predicates as false(). This would avoid any obscure differences in these two, otherwise equal, expressions. In fact something similar already exists with templates where failed evaluation of pattern is treated as un-match.

Saturday, January 2, 2016 9:32:16 PM UTC  #    Comments [0] -
Thinking aloud | xslt
# Wednesday, December 16, 2015

A collegue has approached to us with a question on how Akinator engine may work.

To our shame we have never heard about this amazing game before. To fill the gap we have immediately started to play it, and have identified it as a Troubleshooting solver.

It took us a couple of minutes to come up with a brilliant solution: "We just need to google and find the engine in the internet". :-)

Unfortunately, this led to nowhere, as no Akinator itself is open sourced, and no other good quality open source solutions are available.

After another hour we have got two more ideas:

  1. The task should fit into SQL;
  2. The task is a good candidate for a neural network.

In fact, the first might be required to teach the second, so we have decided to formalize the problem in terms of SQL, while still keeping in mind a neural network.

With this goal we have created a GitHub project. Please see the algorithm and its implementation at github.com/nesterovsky-bros/KB.

Wednesday, December 16, 2015 12:33:41 PM UTC  #    Comments [0] -
Announce | SQL Server puzzle | Thinking aloud
# Sunday, November 8, 2015

Some of our latest projects used file uploading feature. Whether this is an excel, an audio or an image file, the file uploading mechanism remains the same. In a web application an user selects a file and uploads it to the server. Browser sends this file as a multipart-form file attachment, which is then handled on server.

The default HTML way to upload file to server is to use <input type="file"> element on a form. The rendering of such element is different in different browsers and looks rather ugly. Thus, almost all well known javascript libraries like JQuery, Kendo UI etc. provide their own implementations of file upload component. The key statement here is "almost", since in AngularJS bootstrap we didn't find anything like that. It worth to say that we've found several third-party implementations for file upload, but they either have rather complex implementation for such simple task or don't provide file selection feature. This is the reason why we've decided to implement this task by ourselves.

Sources of our solution with upload-link directive and uiUploader service you may find here.

Their usage is rather simple.
E.g. for upload-ink directive:

      <a upload-link
       class="btn btn-primary"
       accept=".*"
       server-url="api/upload"
       on-success="controller.uploadSucceed(data, file)"
       on-error="controller.uploadFailed(e)">Click here to upload an image</a>
    

Where:

accept
is a comma separated list of acceptable file extensions.
server-url
is server URL where to upload the selected file. In case when there is no "server-url" attribute the content of selected file will be passed to success handler as a data URI.
on-success
A "success" handler, which is called when upload is finished successfully.
on-error
An "error" handler, which is called when upload is failed.

Usage of uiUploader service is also easy:

      uiUploader.selectAndUploadFile("api/upload", ".jpg,.png,.gif").
        then(
            function(result)
            {
              // TODO: handle the result.
              //       result.data - contains the server response
              //       result.file - contains uploaded File or Blob instance.              
            },
            $log.error);
    

In case when the first parameter is null the result.data contains the selected file content as a data URI.

Sunday, November 8, 2015 4:21:04 PM UTC  #    Comments [0] -
AngularJS | ASP.NET | javascript
# Monday, September 28, 2015

In a web project we needed to provide a region selection tool.

This requirement is resulted in a javascript module selectionTool, and in an angularjs wrappers selection, and clip.

There are samples test.html, and angularjs-test.html.

The module is implemented through SVG manipulation. Selection paths are defined in terms of SVG.

The simplest way to start with this API is through test pages.

From the client perspective API allows to:

  • create a new selection - click and drag path;
  • select/unselect selection - click selection overlay or image area;
  • move selected path - drag selected overlay, or click arrow keys;
  • move selected edge - drag selected edge;
  • move selected vertex - drag selected vertex;
  • delete selected path - Delete button;
  • add selection vertex - double click or ctrl + click a selection edge;
  • remove selection vertex - double click or ctrl + click a selection vertex;
  • scale selection - shift + drag selection, or shift + arrow keys;
  • rotate selection - ctrl + drag selection, or ctrl + arrow keys.

Sources can be found at GitHub: nesterovsky-bros/selection.

Monday, September 28, 2015 7:05:13 PM UTC  #    Comments [0] -
AngularJS | javascript
# Monday, August 24, 2015

It's time to align csharpxom to the latest version of C#. The article New Language Features in C# 6 sums up what's being added.

Sources can be found at nesterovsky-bros/languages-xom, and C# model is at csharp folder.

In general we feel hostile to any new features until they prove they bring an added value. So, here our list of new features from most to least useless:

  1. String interpolation

    var s = $"{p.Name} is {p.Age} year{{s}} old";

    This is useless, as it does not account resource localization.

  2. Null-conditional operators

    int? first = customers?[0].Orders?.Count();

    They claim to reduce cluttering from null checks, but in our opinion it looks opposite. It's better to get NullReferenceException if arguments are wrong.

  3. Exception filters

    private static bool Log(Exception e) { /* log it */ ; return false; }

    try { … } catch (Exception e) when (Log(e)) {}

    "It is also a common and accepted form of “abuse” to use exception filters for side effects; e.g. logging."

    Design a feature for abuse just does not tastes good.

  4. Expression-bodied function and property members.

    public Point Move(int dx, int dy) => new Point(x + dx, y + dy);
    public string Name => First + " " + Last;

    Not sure it's that usefull.

Monday, August 24, 2015 10:52:07 AM UTC  #    Comments [0] -
.NET | Announce | Java | xslt
# Thursday, August 6, 2015

We have solved this problem years ago, but have run into it once again.

So, we shall log the solution here.

The problem: to minify payload of the JAXB serialized beans.

Java beans have many properties most of them contains default values: zero ints, empty strings, and so on.

JAXB never tries to omit default value from marshalled xml, the only thing it can remove from output is null values. So, our approach is to define xml adapter to map default values to nulls.

Here we refer to the StackOverflow question: Prevent writing default attribute values JAXB, and to our answer.

Though it's not as terse as one would wish, one can create XmlAdapters to avoid marshalling the default values.

The use case is like this:

@XmlRootElement(name = "FIELD")
public class TestLayoutNode
{
  @XmlAttribute(name = "num")
  @XmlJavaTypeAdapter(value = IntegerZero.class, type = int.class)
  public int number;

  @XmlAttribute(name = "str")
  @XmlJavaTypeAdapter(StringDefault.class)
  public String str = "default";
}

And here are adapters.

IntegerZero:

public class IntegerZero extends DefaultValue<Integer>
{
  public Integer defaultValue() { return 0; }
}

StringDefault:

public class StringDefault extends DefaultValue<String>
{
  public String defaultValue() { return "default"; }
}

DefaultValueAdapter:

public class DefaultValue<T> extends XmlAdapter<T, T>
{
  public T defaultValue() { return null; }

  public T marshal(T value) throws Exception
  {
    return (value == null) || value.equals(defaultValue()) ? null : value;
  }

  public T unmarshal(T value) throws Exception
  {
    return value;
  }
}

With small number of different default values this approach works well.

Thursday, August 6, 2015 8:01:23 PM UTC  #    Comments [0] -
Java | Tips and tricks
# Friday, July 31, 2015

Taking into an account that we use Saxon for many years, it was strange to run into so simple error like the following:

<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
  <xsl:template match="/">
    <xsl:variable name="doc" as="element()+"><a/><b/><c/></xsl:variable>   
    <xsl:sequence select="$doc = 3"/>
  </xsl:template>
</xsl:stylesheet>

This is a simplified case that should produce an dynamic error FORG0001 as per General Comparisions; the real code is more complex, as it uses SFINAE and continues.

This case crushes in Saxon with exception:

Exception in thread "main" java.lang.RuntimeException: 
      Internal error evaluating template  at line 3 in module ICE9.6.xslt
    at net.sf.saxon.expr.instruct.Template.applyLeavingTail()
    at net.sf.saxon.trans.Mode.applyTemplates()
    at net.sf.saxon.Controller.transformDocument()
    at net.sf.saxon.Controller.transform()
    at net.sf.saxon.s9api.XsltTransformer.transform()
    at net.sf.saxon.jaxp.TransformerImpl.transform()
    ...
Caused by: java.lang.NumberFormatException: For input string: "" 
    at java.lang.NumberFormatException.forInputString()
    at java.lang.Long.parseLong()
    at java.lang.Long.parseLong()
    at net.sf.saxon.expr.GeneralComparison.quickCompare()
    at net.sf.saxon.expr.GeneralComparison.compare()
    at net.sf.saxon.expr.GeneralComparison.evaluateManyToOne()
    at net.sf.saxon.expr.GeneralComparison.evaluateItem()
    at net.sf.saxon.expr.GeneralComparison.evaluateItem()
    at net.sf.saxon.expr.Expression.process()
    at net.sf.saxon.expr.instruct.Template.applyLeavingTail()
    ... 8 more

We have reported the problem at Saxon's forum, and as usual the problem was shortly resolved.

Friday, July 31, 2015 8:36:39 AM UTC  #    Comments [0] -
xslt
# Monday, July 27, 2015

Though ADO.NET and other ORM framworks like EntityFramework and Dapper support async pattern, you should remember that database drivers (at least all we know about) do not support concurrent db commands running against a single connection.

To see what we mean consider a bug we have recently identified. Consider a code:

await Task.WhenAll(
  newImages.
    Select(
      async image =>
      {
        // Load data from url.
        image.Content = await HttpUtils.ReadData(image.Url);

        // Insert image into the database.
        image.ImageID = await context.InsertImage(image);
      }));

The code runs multiple tasks to read images, and to write them into a database.

Framework decides to run all these tasks in parallel. HttpUtils.ReadData() has no problem with parallel execution, while context.InsertImage() does not run well in parallel, and is a subject of race conditions.

To workaround the problem we had to use async variant of a critical section. So the fixed code looks like this:

using(var semaphore = new SemaphoreSlim(1))
{
  await Task.WhenAll(
    newImages.
      Select(
        async image =>
        {
          // Load data from url.
          image.Content = await HttpUtils.ReadData(image.Url);

          await semaphore.WaitAsync();

          try
          {
            // Insert image into the database.
            image.ImageID = await context.InsertImage(image);
          }
          finally
          {
            semaphore.Release();
          }
        }));
}

So, in the async world we still should care about race conditions.

Monday, July 27, 2015 6:44:45 AM UTC  #    Comments [0] -
.NET | Thinking aloud | Tips and tricks
# Friday, July 10, 2015

Here we show two snall directives that help to build fixed menu bar in your angularjs application.

There are two ideas behind:

  1. Expose element's bounds into a scope for a manipulation (ui-bounds directive).
  2. Allow to react to scroll DOM event (ui-scroll directive).

Directive implementation is very simple. See bounds.html at GitHub.

The use cases are also trivial to unerstand and implement. Take a look at two of them.

  1. Fixed menu:
    <div ng-style="{paddingTop: headerBounds.height.toFixed() + 'px' }">
      <div style="position: fixed; z-index: 1; top: 0; width: 100%; background: menu" 
        ui-bounds="headerBounds">My header</div>
      <div>
        long content that produces a scroll bar.
      </div>
    </div>
  2. Synchronized scroll of table header
    <div style="width: 50em; overflow: hidden; background: pink">
      <div style="position: relative" 
        ng-style="{left: bodyBounds.left.toFixed() + 'px'}">header...<div>
    </div>
    <div style="width: 50em; height: 5em; overflow: auto; background: blue"
      ui-scroll>
      <div ui-bounds="bodyBounds">body...</div>
    </div>

You can see the demo at: nesterovsky-bros/angularjs-api/master/angularjs/bounds.html.

Friday, July 10, 2015 9:04:40 PM UTC  #    Comments [0] -
AngularJS | javascript
# Sunday, June 7, 2015

In our angularjs projects we are often dealing with existing models that do not always fit to angularjs expectations.

Here is an example.

There is a model consisting of two arrays: for data, and for associated data. How to create an ng-repeat that displays data from both sources?

Consider a test controller (see a github sources, and a rawgit working sample):

model.controller(
  "Test",
  function()
  {
    this.records =
    [
      { name: "record 1", state: "Draft" },
      { name: "record 2", state: "Public" },
      { name: "record 3", state: "Disabled" },
      { name: "record 4", state: "Public" },
      { name: "record 5", state: "Public" }
    ];

    this.more =
    [
      { value: 1, selected: true, visible: true },
      { value: 2, selected: false, visible: true },
      { value: 3, selected: true, visible: true },
      { value: 4, selected: false, visible: false },
      { value: 5, selected: false, visible: true }
    ];

    this.delete = function(index)
    {
      this.records.splice(index, 1);
      this.more.splice(index, 1);
    };
  });

Basically there are three approaches here:

  1. Change model.
  2. Adapt model to a single collection.
  3. ng-repeat over first array and access the second array using $index scope variable.

We argued like this:

  • It is often not an option to change a model, as it's how business data are described.
  • A model adaptation when we build a single collection from original two collections, and synchronize it back (if required) may unnecessary complicate things.
  • Thus let's get associated items by $index variable.

This is an example of ng-repeat use:

<table border="1">
  <tr>
    <th>[x]</th>
    <th>Name</th>
    <th>Value</th>
    <th>State</th>
    <th>Actions</th>
  </tr>
  <tr ng-repeat="item in test.records track by $index"
    ng-if="test.more[$index].visible">
    <td>    
      <input type="checkbox" ng-model="test.more[$index].selected"/>
    </td>
    <td>{{item.name}}</td>
    <td>{{test.more[$index].value}}</td>
    <td>{{item.state}}</td>
    <td>
      <a href="#" ng-click="test.delete($index)">Delete</a>
    </td>
  </tr>
</table>

Look at how associated data is accessed: test.more[$index]... Our goal was to optimize that repeating parts, so we looked at ng-init directive.

Though docs warn about its use: "the only appropriate use of ngInit is for aliasing special properties of ngRepeat", we thought that our use of ng-init is rather close to what docs state, so we tried the following:

...
<tr ng-repeat="item in test.records track by $index" 
  ng-init="more = test.more[$index]" 
  ng-if="more.visible">
  <td>    
    <input type="checkbox" ng-model="more.selected"/>
  </td>
  <td>{{item.name}}</td>
  <td>{{more.value}}</td>
  <td>{{item.state}}</td>
  <td>
    <a href="#" ng-click="test.delete($index)">Delete</a>
  </td>
</tr>
...

This code just does not work, as it shows empty table, as if ng-if is always evaluated to false. From docs we found the reason:

  • the priority of the directive ng-if is higher than the prirority of the ng-init, and besides ng-if is a terminal directive;
  • as result ng-if directive is bound, and ng-init is not;
  • when ng-if is evaluated no $scope.more is defined, so more.visible is evaluated to false.

To workaround ng-init/ng-if problem we refactored ng-if as ng-if-start/ng-if-end:

...
<tr ng-repeat="item in test.records track by $index" 
  ng-init="more = test.more[$index]">
  <td ng-if-start="more.visible">
    <input type="checkbox" ng-model="more.selected"/>
  </td>
  <td>{{item.name}}</td>
  <td>{{more.value}}</td>
  <td>{{item.state}}</td>
  <td ng-if-end>
    <a href="#" ng-click="test.delete($index)">Delete</a>
  </td>
</tr>
...

This code works much better and shows a correct content. But then click "Delete" for a row with Name "record 2" and you will find that updated table is out of sync for all data that come from test.more array.

So, why the data goes out of sync? The reason is in the way how the ng-init is implemented: its expression is evaluated just once at directive's pre-link phase. So, the value of $scope.more will persist for the whole ng-init's life cycle, and it does not matter that test.mode[$index] may have changed at some point.

At this point we have decided to introduce a small directive named ui-eval that will act in a way similar to ng-init but that:

  • will run before ng-if;
  • will be re-evaluated when it's value is changed.

This is it:

module.directive(
  "uiEval",
  function()
  {
    var directive =
    {
      restrict: 'A',
      priority: 700,
      link:
      {
        pre: function(scope, element, attr)
        {
          scope.$watch(attr["uiEval"]);
        }
      }
    };

    return directive;
  });

The ui-eval version of the markup is:

...
<tr ng-repeat="item in test.records track by $index" 
  ui-eval="more = test.more[$index]" 
  ng-if="more.visible">
  <td>    
    <input type="checkbox" ng-model="more.selected"/>
  </td>
  <td>{{item.name}}</td>
  <td>{{more.value}}</td>
  <td>{{item.state}}</td>
  <td>
    <a href="#" ng-click="test.delete($index)">Delete</a>
  </td>
</tr>
...

It works as expected both during initial rendering and when model is updated.

We consider ui-eval is a "better" ng-init as it solves ng-init's silent limitations. On the other hand it should not try to evaluate any complex logic, as it can be often re-evaluated, so its use case is to alias a sub-expression. It can be used in any context and is not limited to items of ng-repeat.

Source code can be found at github, and a working sample at rawgit.

Sunday, June 7, 2015 11:46:11 AM UTC  #    Comments [0] -
AngularJS | javascript | Tips and tricks
# Sunday, May 31, 2015

In one of our last projects we were dealing with audio: capture audio in browser, store it on server and then return it by a request and replay in browser.

Though an audio capturing is by itself rather interesting and challenging task, it's addressed by HTML5, so for example take a look at this article. Here we share our findings about other problem, namely an audio conversion.

You might thought that if you have already captured an audio in browser then you will be able to play back it. Thus no additional audio conversion is required.

In practice we are limited by support of various audio formats in browsers. Browsers can capture audio in WAV format, but this format is rather heavy for storing and streaming back. Moreover, not all browsers support this format for playback. See wikipedia for details. There are only two audio formats that more or less widely supported by mainstream browsers: MP3 and AAC. So, you have either convert WAV to MP3, or to AAC.

The obvious choice is to select WAV to MP3 conversion, the benefit that there are many libraries and APIs for such conversion. But in this case you risk falling into a trap with MP3 licensing, especially if you deal with iteractive software products.

Eventually, you will come to the only possible solution (at least for moment of writting) - conversion WAV to AAC.

The native solution is to use NAudio library, which behind the scene uses Media Foundation Transforms. You'll shortly get a working example. Actually the core of solution will contain few lines only:

var source = Path.Combine(root, "audio.wav");
var target = Path.Combine(root, "audio.m4a");

using(var reader = new NAudio.Wave.WaveFileReader(source))
{
  MediaFoundationEncoder.EncodeToAac(reader, target);
}

Everything is great. You'll deploy your code on server (by the way server must be Windows Server 2008R2 or higher) and at this point you may find that your code fails. The problem is that Media Foundation API is not preinstalled as a part of Windows Server installation, and must be installed separately. If you own this server then everything is all right, but in case you use a public web hosting server then you won't have ability to install Media Foundation API and your application will never work properly. That's what happened to us...

After some research we came to conclusion that another possible solution is a wrapper around an open source video/audio converter - FFPEG. There were two issues with this solution:

  • how to execute ffmpeg.exe on server asynchronously;
  • how to limit maximum parallel requests to conversion service.

All these issues were successfully resolved in our prototype conversion service that you may see here, with source published on github. The solution is Web API based REST service with simple client that uploads audio files using AJAX requests to server and plays it back. As a bonus this solution allows us perform not only WAV to AAC conversions, but from many others format to AAC without additional efforts.

Let's take a close look at crucial details of this solution. The core is FFMpegWrapper class that allows to run ffmpeg.exe asynchronously:

/// <summary>
/// A ffmpeg.exe open source utility wrapper.
/// </summary>
public class FFMpegWrapper
{
  /// <summary>
  /// Creates a wrapper for ffmpeg utility.
  /// </summary>
  /// <param name="ffmpegexe">a real path to ffmpeg.exe</param>
  public FFMpegWrapper(string ffmpegexe)
  {
    if (!string.IsNullOrEmpty(ffmpegexe) && File.Exists(ffmpegexe))
    {
      this.ffmpegexe = ffmpegexe;
    }
  }

  /// <summary>
  /// Runs ffmpeg asynchronously.
  /// </summary>
  /// <param name="args">determines command line arguments for ffmpeg.exe</param>
  /// <returns>
  /// asynchronous result with ProcessResults instance that contains 
  /// stdout, stderr and process exit code.
  /// </returns>
  public Task<ProcessResults> Run(string args)
  {
    if (string.IsNullOrEmpty(ffmpegexe))
    {
      throw new InvalidOperationException("Cannot find FFMPEG.exe");
    }

    //create a process info object so we can run our app
    var info = new ProcessStartInfo 
    { 
      FileName = ffmpegexe,
      Arguments = args,
      CreateNoWindow = true
    };

    return ProcessEx.RunAsync(info);
  }

  private string ffmpegexe;
}

It became possible to run  a process asynchronously thanks to James Manning and his ProcessEx class.  

Another useful part is a semaphore declaration in Global.asax.cs:

   public class WebApiApplication : HttpApplication
  {
    protected void Application_Start()
    {
      GlobalConfiguration.Configure(WebApiConfig.Register);
    }

    /// <summary>
    /// Gets application level semaphore that controls number of running 
    /// in parallel FFMPEG utilities.
    /// </summary>
    public static SemaphoreSlim Semaphore
    {
      get { return semaphore; }
    }

    private static SemaphoreSlim semaphore;

    static WebApiApplication()
    {
      var value =
        ConfigurationManager.AppSettings["NumberOfConcurentFFMpegProcesses"];

      int intValue = 10;

      if (!string.IsNullOrEmpty(value))
      {
        try
        {
          intValue = System.Convert.ToInt32(value);
        }
        catch
        {
          // use the default value
        }
      }

      semaphore = new SemaphoreSlim(intValue, intValue);
    }
  }

And the last piece is the entry point, which was implemented as a REST controller:

/// <summary>
/// A controller to convert audio.
/// </summary>
public class AudioConverterController : ApiController
{
  /// <summary>
  /// Gets ffmpeg utility wrapper.
  /// </summary>
  public FFMpegWrapper FFMpeg
  {
    get
    {
      if (ffmpeg == null)
      {
        ffmpeg = new FFMpegWrapper(
          HttpContext.Current.Server.MapPath("~/lib/ffmpeg.exe"));
      }

      return ffmpeg;
    }
  }

  /// <summary>
  /// Converts an audio in WAV, OGG, MP3 or other formats 
  /// to AAC format (MP4 audio).
  /// </summary>
  /// <returns>A data URI as a  string.</returns>
  [HttpPost]
  public async Task<string> ConvertAudio([FromBody]string audio)
  {
    if (string.IsNullOrEmpty(audio))
    {
      throw new ArgumentException(
        "Invalid audio stream (probably the input audio is too big).");
    }

    var tmp = Path.GetTempFileName();
    var root = tmp + ".dir";

    Directory.CreateDirectory(root);
    File.Delete(tmp);

    try
    {
      var start = audio.IndexOf(':');
      var end = audio.IndexOf(';');
      var mimeType = audio.Substring(start + 1, end - start - 1);
      var ext = mimeType.Substring(mimeType.IndexOf('/') + 1);
      var source = Path.Combine(root, "audio." + ext);
      var target = Path.Combine(root, "audio.m4a");

      await WriteToFileAsync(audio, source);

      switch (ext)
      {
        case "mpeg":
        case "mp3":
        case "wav":
        case "wma":
        case "ogg":
        case "3gp":
        case "amr":
        case "aif":
        case "mid":
        case "au":
        {
          await WebApiApplication.Semaphore.WaitAsync();

          var result = await FFMpeg.Run(
            string.Format(
              "-i {0} -c:a libvo_aacenc -b:a 96k {1}",
              source,
              target));

          WebApiApplication.Semaphore.Release();

          if (result.Process.ExitCode != 0)
          {
            throw new InvalidDataException(
              "Cannot convert this audio file to audio/mp4.");
          }

          break;
        }
        default:
        {
          throw new InvalidDataException(
            "Mime type: '" + mimeType + "' is not supported.");
        }
      }

      var buffer = await ReadAllBytes(target);
      var response = "data:audio/mp4;base64," + System.Convert.ToBase64String(buffer);

      return response;
    }
    finally
    {
      Directory.Delete(root, true);
    }
  }

For those who'd like to read more about audio conversion, we may suggest to read this article.

Sunday, May 31, 2015 10:32:21 AM UTC  #    Comments [0] -
.NET | ASP.NET
# Thursday, May 7, 2015

Stackoverfow shows that people are searching How to intercept $resource requests.

Recently we have written about the way to cancel angularjs $resource requests (see Cancel angularjs resource request). Here we apply the same technique to intercept resource request.

Consider a sample (nesterovsky-bros/angularjs-api/master/angularjs/transform-request.html):

<!DOCTYPE html>
<html>
<head>
  <meta charset="utf-8" />
  <title>Intercept resource request</title>
  <style type="text/css">.ng-cloak { display: none; }</style>
  <script src="angular.js"></script>
  <script src="angular-resource.js"></script>
  <script>
angular.module("app", ["ngResource"]).
  factory(
    "services",
    ["$resource", function ($resource)
    {
      return $resource(
        "http://md5.jsontest.com/",
        {},
        {
          MD5:
          {
            method: "GET",
            params: { text: null },
            then: function (resolve)
            {
              this.params.text = "***" + this.params.text + "***";
              this.then = null;
              resolve(this);
            }
          },
        });
      }]).
  controller(
    "Test",
    ["services", function (services)
    {
      this.value = "Sample text";

      this.call = function()
      {
        this.result = services.MD5({ text: this.value });
      }
    }]);
  </script>
</head>
<body ng-app="app" ng-controller="Test as test">
  <label>Text: <input type="text" ng-model="test.value" /></label>
  <input type="button" value="call" ng-click="test.call()"/>
  <div ng-bind="test.result.md5"></div>
</body>
</html>

How it works.

  1. $resource merges action definition, request params and data to build a config parameter for an $http request.
  2. a config parameter passed into an $http request is treated as a promise like object, so it may contain then function to initialize config.
  3. action's then function may transform request as it wishes.

The demo can be found at transform-request.html

Thursday, May 7, 2015 10:53:34 AM UTC  #    Comments [0] -
AngularJS | javascript | Tips and tricks
Archive
<January 2016>
SunMonTueWedThuFriSat
272829303112
3456789
10111213141516
17181920212223
24252627282930
31123456
Statistics
Total Posts: 387
This Year: 3
This Month: 0
This Week: 0
Comments: 1984
Locations of visitors to this page
Disclaimer
The opinions expressed herein are our own personal opinions and do not represent our employer's view in anyway.

© 2024, Nesterovsky bros
All Content © 2024, Nesterovsky bros
DasBlog theme 'Business' created by Christoph De Baene (delarou)