Improve RTE performance when using base64 images (#7530)

This commit is contained in:
Aleksandr Šmailov
2020-01-29 09:53:08 +02:00
committed by GitHub
parent 4058ca6109
commit 537674a0b3
2 changed files with 97 additions and 20 deletions

View File

@@ -1,7 +1,6 @@
using Umbraco.Core.Logging;
using Moq;
using NUnit.Framework;
using Umbraco.Core.Services;
using Umbraco.Tests.Testing.Objects.Accessors;
using Umbraco.Web.Templates;
using Umbraco.Web;
@@ -13,12 +12,10 @@ using System;
using System.Linq;
using Umbraco.Core.Models;
using Umbraco.Core;
using Umbraco.Web.PropertyEditors;
using System.Diagnostics;
namespace Umbraco.Tests.Templates
{
[TestFixture]
public class HtmlImageSourceParserTests
{
@@ -31,10 +28,9 @@ namespace Umbraco.Tests.Templates
</div>
</p><p><img src='/media/234234.jpg' data-udi=""umb://media-type/B726D735E4C446D58F703F3FBCFC97A5"" /></p>";
var logger = Mock.Of<ILogger>();
var umbracoContextAccessor = new TestUmbracoContextAccessor();
var imageSourceParser = new HtmlImageSourceParser(umbracoContextAccessor);
var result = imageSourceParser.FindUdisFromDataAttributes(input).ToList();
Assert.AreEqual(2, result.Count);
Assert.AreEqual(Udi.Parse("umb://media/D4B18427A1544721B09AC7692F35C264"), result[0]);
@@ -44,7 +40,6 @@ namespace Umbraco.Tests.Templates
[Test]
public void Remove_Image_Sources()
{
var logger = Mock.Of<ILogger>();
var umbracoContextAccessor = new TestUmbracoContextAccessor();
var imageSourceParser = new HtmlImageSourceParser(umbracoContextAccessor);
@@ -111,10 +106,83 @@ namespace Umbraco.Tests.Templates
<p>
<div><img src=""/media/1001/my-image.jpg?width=100"" data-udi=""umb://media/81BB2036-034F-418B-B61F-C7160D68DCD4"" /></div>
</p>", result);
}
}
[TestCase(
@"<div><img src="""" /></div>",
ExpectedResult = @"<div><img src="""" /></div>",
TestName = "Empty source is not updated with no data-udi set"
)]
[TestCase(
@"<div><img src="""" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
ExpectedResult = @"<div><img src=""/media/1001/image.jpg"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
TestName = "Empty source is updated with data-udi set"
)]
[TestCase(
@"<div><img src=""non empty src"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
ExpectedResult = @"<div><img src=""/media/1001/image.jpg"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
TestName = "Filled source is overwritten with data-udi set"
)]
[TestCase(
@"<div><img src=""some src"" some-attribute data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4"" another-attribute/></div>",
ExpectedResult = @"<div><img src=""/media/1001/image.jpg"" some-attribute data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4"" another-attribute/></div>",
TestName = "Attributes are persisted"
)]
[TestCase(
@"<div><img src=""somesrc?width=100&height=500"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
ExpectedResult = @"<div><img src=""/media/1001/image.jpg?width=100&height=500"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
TestName = "Source is trimmed and parameters are prefixed"
)]
[TestCase(
@"<div><img src=""?width=100&height=500"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
ExpectedResult = @"<div><img src=""/media/1001/image.jpg?width=100&height=500"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/></div>",
TestName = "Parameters are prefixed"
)]
[TestCase(
@"<div>
<img src="""" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/>
<img src=""src"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD5""/>
<img src=""?asd"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD6""/>
</div>",
ExpectedResult =
@"<div>
<img src=""/media/1001/image.jpg"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD4""/>
<img src=""/media/1001/image.jpg"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD5""/>
<img src=""/media/1001/image.jpg?asd"" data-udi=""umb://media/81BB2036034F418BB61FC7160D68DCD6""/>
</div>",
TestName = "Multiple img tags are handled"
)]
[Category("Ensure image sources")]
public string Ensure_ImageSources_Processing(string sourceHtml)
{
var fakeMediaUrl = "/media/1001/image.jpg";
var parser = new HtmlImageSourceParser((guid) => fakeMediaUrl);
var actual = parser.EnsureImageSources(sourceHtml);
return actual;
}
[Category("Ensure image sources")]
[Test]
public void Ensure_Large_Html_Is_Processed_Quickly()
{
int symbolCount = 25000;
int maxMsToRun = 200;
var longText = new string('*', symbolCount);
var text = $@"<img src=""{longText}"" />";
var fakeMediaUrl = "/media/1001/image.jpg";
var parser = new HtmlImageSourceParser((guid) => fakeMediaUrl);
var timer = new Stopwatch();
timer.Start();
var actual = parser.EnsureImageSources(text);
timer.Stop();
Assert.IsTrue(timer.ElapsedMilliseconds <= maxMsToRun);
}
}
}

View File

@@ -1,4 +1,5 @@
using System.Collections.Generic;
using System;
using System.Collections.Generic;
using System.Text.RegularExpressions;
using Umbraco.Core;
@@ -7,18 +8,29 @@ namespace Umbraco.Web.Templates
public sealed class HtmlImageSourceParser
{
public HtmlImageSourceParser(IUmbracoContextAccessor umbracoContextAccessor)
public HtmlImageSourceParser(Func<Guid, string> getMediaUrl)
{
_umbracoContextAccessor = umbracoContextAccessor;
this._getMediaUrl = getMediaUrl;
}
private static readonly Regex ResolveImgPattern = new Regex(@"(<img[^>]*src="")([^""\?]*)([^""]*""[^>]*data-udi="")([^""]*)(""[^>]*>)",
public HtmlImageSourceParser(IUmbracoContextAccessor umbracoContextAccessor)
{
if (umbracoContextAccessor?.UmbracoContext?.UrlProvider == null)
{
return;
}
_getMediaUrl = (guid) => umbracoContextAccessor.UmbracoContext.UrlProvider.GetMediaUrl(guid);
}
private static readonly Regex ResolveImgPattern = new Regex(@"(<img[^>]*src="")([^""\?]*)((?:\?[^""]*)?""[^>]*data-udi="")([^""]*)(""[^>]*>)",
RegexOptions.Compiled | RegexOptions.IgnoreCase | RegexOptions.IgnorePatternWhitespace);
private readonly IUmbracoContextAccessor _umbracoContextAccessor;
private static readonly Regex DataUdiAttributeRegex = new Regex(@"data-udi=\\?(?:""|')(?<udi>umb://[A-z0-9\-]+/[A-z0-9]+)\\?(?:""|')",
RegexOptions.Compiled | RegexOptions.CultureInvariant | RegexOptions.IgnoreCase);
private readonly Func<Guid, string> _getMediaUrl;
/// <summary>
/// Parses out media UDIs from an html string based on 'data-udi' html attributes
/// </summary>
@@ -45,14 +57,13 @@ namespace Umbraco.Web.Templates
/// <remarks>Umbraco image tags are identified by their data-udi attributes</remarks>
public string EnsureImageSources(string text)
{
// don't attempt to proceed without a context
if (_umbracoContextAccessor?.UmbracoContext?.UrlProvider == null)
// no point in doing any processing if we don't have
// a function to retrieve Urls
if (_getMediaUrl == null)
{
return text;
}
var urlProvider = _umbracoContextAccessor.UmbracoContext.UrlProvider;
return ResolveImgPattern.Replace(text, match =>
{
// match groups:
@@ -66,7 +77,7 @@ namespace Umbraco.Web.Templates
{
return match.Value;
}
var mediaUrl = urlProvider.GetMediaUrl(guidUdi.Guid);
var mediaUrl = _getMediaUrl(guidUdi.Guid);
if (mediaUrl == null)
{
// image does not exist - we could choose to remove the image entirely here (return empty string),
@@ -86,7 +97,5 @@ namespace Umbraco.Web.Templates
public string RemoveImageSources(string text)
// see comment in ResolveMediaFromTextString for group reference
=> ResolveImgPattern.Replace(text, "$1$3$4$5");
}
}