Download source - 3.8 MB

Table of Contents

Introduction
Background
Prerequisites
Get started to implement
Play the video
Performance considerations
History

Introduction

This article focuses on the implementation of HTTP 206 Partial Content in ASP.NET Web API. I would like to describe how I work on it with ApiController and deal with some potential performance issues. Our goal is to create a video file streaming service and an HTML5 page to play them.

Background

In my last article, we discussed the characteristic of HTTP 206 and its related headers. Also, we had a showcase of video streaming in Node.js and HTML5. This time, we will move to ASP.NET Web API and will have some discussions regarding our implementation. If you would like to learn more details of this HTTP status code, the last article could be a good reference for you. And we will not repeat it in this article.

Prerequisites

The Understanding of HTTP 206 Partial Content
Basic ASP.NET MVC Skill
Basic HTML5 & JavaScript Skill

Get Started to Implement

First of all, we expect the URL for video streaming shall be like this:

http://localhost/movie/api/media/play?f=praise-our-lord.mp4

where movie is our application name in IIS, media is the controller name, play is its action name and parameter f represents the video file we would like to play.

Based on this URL, we will start from the MediaController under the namespace Movie.Controllers, a class derived from ApiController. Before we work on its actual action, we need several static fields and methods to help us in the upcoming steps.

using System;
using System.Collections.Generic;
using System.Collections.ObjectModel;
using System.Diagnostics;
using System.IO;
using System.Linq;
using System.Net;
using System.Net.Http;
using System.Net.Http.Headers;
using System.Net.Mime;
using System.Web.Configuration;
using System.Web.Http;

namespace Movie.Controllers
{
    public class MediaController : ApiController
    {
        #region Fields

        // This will be used in copying input stream to output stream.
        public const int ReadStreamBufferSize = 1024 * 1024;
        // We have a read-only dictionary for mapping file extensions and MIME names. 
        public static readonly IReadOnlyDictionary<string, string> MimeNames;
        // We will discuss this later.
        public static readonly IReadOnlyCollection<char> InvalidFileNameChars;
        // Where are your videos located at? Change the value to any folder you want.
        public static readonly string InitialDirectory;

        #endregion

        #region Constructors

        static MediaController()
        {
            var mimeNames = new Dictionary<string, string>();
			
            mimeNames.Add(".mp3", "audio/mpeg");    // List all supported media types; 
            mimeNames.Add(".mp4", "video/mp4");
            mimeNames.Add(".ogg", "application/ogg");
            mimeNames.Add(".ogv", "video/ogg");
            mimeNames.Add(".oga", "audio/ogg");
            mimeNames.Add(".wav", "audio/x-wav");
            mimeNames.Add(".webm", "video/webm");

            MimeNames = new ReadOnlyDictionary<string, string>(mimeNames);

            InvalidFileNameChars = Array.AsReadOnly(Path.GetInvalidFileNameChars());
            InitialDirectory = WebConfigurationManager.AppSettings["InitialDirectory"];
        }

        #endregion

        #region Actions

        // Later we will do something around here.

        #endregion

        #region Others

        private static bool AnyInvalidFileNameChars(string fileName)
        {
            return InvalidFileNameChars.Intersect(fileName).Any(); 
        }

        private static MediaTypeHeaderValue GetMimeNameFromExt(string ext)
        {
            string value;

            if (MimeNames.TryGetValue(ext.ToLowerInvariant(), out value))
                return new MediaTypeHeaderValue(value);
            else
                return new MediaTypeHeaderValue(MediaTypeNames.Application.Octet);
        }

        private static bool TryReadRangeItem(RangeItemHeaderValue range, long contentLength, 
            out long start, out long end)
        {
            if (range.From != null)
            {
                start = range.From.Value;
                if (range.To != null)
                    end = range.To.Value;
                else
                    end = contentLength - 1;
            }
            else
            {
                end = contentLength - 1;
                if (range.To != null)
                    start = contentLength - range.To.Value;
                else
                    start = 0;
            }
            return (start < contentLength && end < contentLength);
        }

        private static void CreatePartialContent(Stream inputStream, Stream outputStream,
            long start, long end)
        {
            int count = 0;
            long remainingBytes = end - start + 1;
            long position = start;
            byte[] buffer = new byte[ReadStreamBufferSize];
            
            inputStream.Position = start;
            do
            {
                try
                {
                    if (remainingBytes > ReadStreamBufferSize)
                        count = inputStream.Read(buffer, 0, ReadStreamBufferSize);
                    else
                        count = inputStream.Read(buffer, 0, (int)remainingBytes); 
                    outputStream.Write(buffer, 0, count);
                }
                catch (Exception error)
                {
                    Debug.WriteLine(error);
                    break;
                }
                position = inputStream.Position;
                remainingBytes = end - position + 1;
            } while (position <= end);
        }
         
        #endregion
    } 
}

And we have:

AnyInvalidFileNameChars() helps us to check if there is any invalid file name character in URL parameter f (by the way, this is a good example of using LINQ on string). This can prevent some unnecessary file system accesses because a file with invalid file name won't exist at all.
GetMimeNameFromExt() helps us to get the corresponding Content-Type header value with file extension from the read-only dictionary MimeNames. If value cannot be found, the default one is application/oct-stream.
TryReadRangeItem() helps us to read Range header from current HTTP request. Returned boolean value represents if range is available. If start or end position is greater than file length (parameter contentLength), it returns false.
CreatePartialContent() helps us to copy content from file stream to response stream with indicated range.

With these tools, to implement the action Play() method will be much easier. The prototype is:

[HttpGet]
public HttpResponseMessage Play(string f) { }

where parameter f represents the URL parameter f. HttpGetAttribute declares that GET is the only acceptable HTTP method. The response headers and content are sent in HttpResponseMessage class. The logic flow behind this method can be described with the following chart.

Naturally, our first job is to see if the file exists. If not, it will result in HTTP 404 Not Found status. Next is to check if Range header is present in current request. If not, the request will be treated as normal request and will result in HTTP 200 OK status. The third step is to determine if Range header can be fulfilled according to target file. If range is not present within file length, HTTP 416 Requested Range Not Satisfiable status will be responded to browser. After these steps, last one is to transmit target file with indicated range, and the story ends with HTTP 206 Partial Content status.

Here is the complete code of Play() action.

[HttpGet]
public HttpResponseMessage Play(string f)
{
    // This can prevent some unnecessary accesses. 
    // These kind of file names won't be existing at all. 
    if (string.IsNullOrWhiteSpace(f) || AnyInvalidFileNameChars(f))
        throw new HttpResponseException(HttpStatusCode.NotFound);

    FileInfo fileInfo = new FileInfo(Path.Combine(InitialDirectory, f));
    
    if (!fileInfo.Exists)
        throw new HttpResponseException(HttpStatusCode.NotFound);

    long totalLength = fileInfo.Length;

    RangeHeaderValue rangeHeader = base.Request.Headers.Range;
    HttpResponseMessage response = new HttpResponseMessage();

    response.Headers.AcceptRanges.Add("bytes");

    // The request will be treated as normal request if there is no Range header.
    if (rangeHeader == null || !rangeHeader.Ranges.Any())
    {
        response.StatusCode = HttpStatusCode.OK;
        response.Content = new PushStreamContent((outputStream, httpContent, transpContext)
		=>
            {
                using (outputStream) // Copy the file to output stream straightforward. 
                using (Stream inputStream = fileInfo.OpenRead())
                {
                    try
                    {
                        inputStream.CopyTo(outputStream, ReadStreamBufferSize);
                    }
                    catch (Exception error)
                    {
                        Debug.WriteLine(error);
                    }
                }
            }, GetMimeNameFromExt(fileInfo.Extension));

        response.Content.Headers.ContentLength = totalLength;
        return response;
    }

    long start = 0, end = 0;

    // 1. If the unit is not 'bytes'.
    // 2. If there are multiple ranges in header value.
    // 3. If start or end position is greater than file length.
    if (rangeHeader.Unit != "bytes" || rangeHeader.Ranges.Count > 1 ||
        !TryReadRangeItem(rangeHeader.Ranges.First(), totalLength, out start, out end))
    {
        response.StatusCode = HttpStatusCode.RequestedRangeNotSatisfiable;
        response.Content = new StreamContent(Stream.Null);  // No content for this status.
        response.Content.Headers.ContentRange = new ContentRangeHeaderValue(totalLength);
        response.Content.Headers.ContentType = GetMimeNameFromExt(fileInfo.Extension);

        return response;
    } 

    var contentRange = new ContentRangeHeaderValue(start, end, totalLength);

    // We are now ready to produce partial content.
    response.StatusCode = HttpStatusCode.PartialContent;
    response.Content = new PushStreamContent((outputStream, httpContent, transpContext)
	=>
        {
            using (outputStream) // Copy the file to output stream in indicated range.
            using (Stream inputStream = fileInfo.OpenRead())
                CreatePartialContent(inputStream, outputStream, start, end);

        }, GetMimeNameFromExt(fileInfo.Extension));

    response.Content.Headers.ContentLength = end - start + 1;
    response.Content.Headers.ContentRange = contentRange;

    return response;
}

Play the Video

Now it is time to play the video. We have a simple HTML5 page with a <video /> element and a <source /> element referring to the URL we have mentioned before.

<!DOCTYPE html>
<html>
    <head>
        <script type="text/javascript">

            function onLoad() {
                var sec = parseInt(document.location.search.substr(1));
                
                if (!isNaN(sec))
                    mainPlayer.currentTime = sec;
            }
        
        </script>
        <title>Partial Content Demonstration</title>
    </head>
    <body>
        <h3>Partial Content Demonstration</h3>
        <hr />
        <video id="mainPlayer" width="640" height="360" 
            autoplay="autoplay" controls="controls" onloadeddata="onLoad()">
            <source src="api/media/play?f=praise-our-lord.mp4" />
        </video>
    </body>
</html>

As you can see, the onLoad() function allows us skipping to indicated second by adding parameter. If parameter is omitted, the <video /> element plays the video from zero. For example, if we want to watch the video starting from 120^th second, then we have:

http://localhost/movie/index.html?120

Let us try this URL in Chrome.

Then we press F12 to open the development tool, switch to Network tab to see what happened behind the scenes.

These headers explain almost everything. Once onLoad() function gets triggered, the player sends a request including a Range header and the start position is exactly equal to the byte position of 120^th second in this video. And the response header Content-Range describes start and end position, and total available bytes. This example shows the biggest benefit of Partial Content mechanism: when a video or audio file is too long, viewers can skip to any second they want.

Performance Considerations

You have probably noticed that we are using PushStreamContent instead of StreamContent in Play() action (excepting empty content) to transfer file stream. Both of them are under the namespace System.Net.Http and derived from HttpContent class. The differences between them could be generally summarized as following points.

PushStreamContent vs. StreamContent

Sequence - For StreamContent, you have to generate content stream before action ends. For PushStreamContent, you will generate it after exit from the action.
File Access - For StreamContent, you generate content stream from file before browser starts receiving. For PushStreamContent, you will do it after browser has received all HTTP headers and is ready to render content, which means if browser receives headers only but cancels rendering content, the file will not be opened.
Memory Usage - For StreamContent, you have to generate partial content stream from file before action ends, which means it will be kept in memory temporarily until browser has received its all bytes. For PushStreamContent, you can directly copy content from file to outgoing stream with specified range and without keeping content temporarily in memory.

Therefore, we choose PushStreamContent for video file streaming. It could reduce memory usage and work more efficiently.

History

2014-11-25

Fixed a potential issue in CreatePartialContent() method when file size is greater than Int32.MaxValue.
Added try-catch section in Play() method to prevent unnecessary exception message in debug mode.
Increased ReadStreamBufferSize value to improve performance.

HTTP 206 Partial Content In ASP.NET Web API - Video File Streaming