In IndexHtmlFileProvider, preserve original source formatting

src/Microsoft.AspNetCore.Blazor.Build/Core/FileSystem/IndexHtmlFileProvider.cs

@@ -2,14 +2,14 @@
 // Licensed under the Apache License, Version 2.0. See License.txt in the project root for license information.
 
 using Microsoft.AspNetCore.Blazor.Internal.Common.FileProviders;
-using System.IO;
 using System.Collections.Generic;
 using System.Text;
 using Microsoft.Extensions.FileProviders;
 using System.Linq;
 using AngleSharp.Parser.Html;
-using AngleSharp.Dom;
 using AngleSharp;
+using AngleSharp.Html;
+using System;
 
 namespace Microsoft.AspNetCore.Blazor.Build.Core.FileSystem
 {
@@ -50,38 +50,81 @@ namespace Microsoft.AspNetCore.Blazor.Build.Core.FileSystem
         /// </remarks>
         private static string GetIndexHtmlContents(string htmlTemplate, string assemblyName, IEnumerable<IFileInfo> binFiles)
         {
-            var parser = new HtmlParser();
+            var resultBuilder = new StringBuilder();
-            var dom = parser.Parse(htmlTemplate);
 
-            // First see if the user has declared a 'boot' script,
+            // Search for a tag of the form <script type="boot-blazor"></script>, and replace
-            // then it's their responsibility to load blazor.js
+            // it with a fully-configured Blazor boot script tag
-            var bootScript = dom.Body?.QuerySelectorAll("script")
+            var tokenizer = new HtmlTokenizer(
-                   .Where(x => x.Attributes["type"]?.Value == "blazor-boot").FirstOrDefault();
+                new TextSource(htmlTemplate),
-
+                HtmlEntityService.Resolver);
-            // If we find a script tag that is decorated with a type="blazor-boot"
+            var currentRangeStartPos = 0;
-            // this will be the point at which we start the Blazor boot process
+            var isInBlazorBootTag = false;
-            if (bootScript != null)
+            var resumeOnNextToken = false;
+            while (true)
             {
-                // We need to remove the 'type="blazor-boot"' so that
+                var token = tokenizer.Get();
-                // it reverts to being processed as JS by the browser
+                if (resumeOnNextToken)
-                bootScript.RemoveAttribute("type");
+                {
+                    resumeOnNextToken = false;
+                    currentRangeStartPos = token.Position.Position;
+                }
 
-                // Leave any user-specified attributes on the tag as-is
+                switch (token.Type)
-                // and add/overwrite the config data needed to boot Blazor
+                {
-                InjectBootConfig(bootScript, assemblyName, binFiles);
+                    case HtmlTokenType.StartTag:
-            }
+                        {
+                            // Only do anything special if this is a Blazor boot tag
+                            var tag = token.AsTag();
+                            if (IsBlazorBootTag(tag))
+                            {
+                                // First, emit the original source text prior to this special tag, since
+                                // we want that to be unchanged
+                                resultBuilder.Append(htmlTemplate, currentRangeStartPos, token.Position.Position - currentRangeStartPos - 1);
 
-            // If no blazor-boot script tag was found, we skip it and
+                                // Instead of emitting the source text for this special tag, emit a fully-
-            // leave it up to the user to handle kicking off the boot
+                                // configured Blazor boot script tag
+                                AppendScriptTagWithBootConfig(
+                                    resultBuilder,
+                                    assemblyName,
+                                    binFiles,
+                                    tag.Attributes);
 
-            using (var writer = new StringWriter())
+                                // Set a flag so we know not to emit anything else until the special
-            {
+                                // tag is closed
-                dom.ToHtml(writer, new AutoSelectedMarkupFormatter());
+                                isInBlazorBootTag = true;
-                return writer.ToString();
+                            }
+                            break;
+                        }
+                    
+                    case HtmlTokenType.EndTag:
+                        // If this is an end tag corresponding to the Blazor boot script tag, we
+                        // can switch back into the mode of emitting the original source text
+                        if (isInBlazorBootTag)
+                        {
+                            isInBlazorBootTag = false;
+                            resumeOnNextToken = true;
+                        }
+                        break;
+
+                    case HtmlTokenType.EndOfFile:
+                        // Finally, emit any remaining text from the original source file
+                        resultBuilder.Append(htmlTemplate, currentRangeStartPos, htmlTemplate.Length - currentRangeStartPos);
+                        return resultBuilder.ToString();
+                }
             }
         }
 
-        private static void InjectBootConfig(IElement script, string assemblyName, IEnumerable<IFileInfo> binFiles)
+        private static bool IsBlazorBootTag(HtmlTagToken tag)
+            => string.Equals(tag.Name, "script", StringComparison.Ordinal)
+            && tag.Attributes.Any(pair =>
+                string.Equals(pair.Key, "type", StringComparison.Ordinal)
+                && string.Equals(pair.Value, "blazor-boot", StringComparison.Ordinal));
+
+        private static void AppendScriptTagWithBootConfig(
+            StringBuilder resultBuilder,
+            string assemblyName,
+            IEnumerable<IFileInfo> binFiles,
+            List<KeyValuePair<string, string>> attributes)
         {
             var assemblyNameWithExtension = $"{assemblyName}.dll";
             var referenceNames = binFiles
@@ -89,9 +132,28 @@ namespace Microsoft.AspNetCore.Blazor.Build.Core.FileSystem
                 .Select(file => file.Name);
             var referencesAttribute = string.Join(",", referenceNames.ToArray());
 
-            script.SetAttribute("src", "/_framework/blazor.js");
+            var attributesDict = attributes.ToDictionary(x => x.Key, x => x.Value);
-            script.SetAttribute("main", assemblyNameWithExtension);
+            attributesDict.Remove("type");
-            script.SetAttribute("references", referencesAttribute);
+            attributesDict["src"] = "/_framework/blazor.js";
+            attributesDict["main"] = assemblyNameWithExtension;
+            attributesDict["references"] = referencesAttribute;
+
+            resultBuilder.Append("<script");
+            foreach (var attributePair in attributesDict)
+            {
+                if (!string.IsNullOrEmpty(attributePair.Value))
+                {
+                    resultBuilder.AppendFormat(" {0}=\"{1}\"",
+                        attributePair.Key,
+                        attributePair.Value); // TODO: HTML attribute encode
+                }
+                else
+                {
+                    resultBuilder.AppendFormat(" {0}",
+                        attributePair.Key);
+                }
+            }
+            resultBuilder.Append("></script>");
         }
     }
 }

test/Microsoft.AspNetCore.Blazor.Build.Test/IndexHtmlFileProviderTest.cs

@@ -32,7 +32,6 @@ namespace Microsoft.AspNetCore.Blazor.Server.Test
         {
             // Arrange
             var htmlTemplate = "test";
-            var htmlOutput = $"<html><head></head><body>{htmlTemplate}</body></html>";
             var instance = new IndexHtmlFileProvider(
                 htmlTemplate, "fakeassembly", Enumerable.Empty<IFileInfo>());
 
@@ -44,8 +43,7 @@ namespace Microsoft.AspNetCore.Blazor.Server.Test
             Assert.False(file.IsDirectory);
             Assert.Equal("/index.html", file.PhysicalPath);
             Assert.Equal("index.html", file.Name);
-            Assert.Equal(htmlOutput, ReadString(file));
+            Assert.Equal(htmlTemplate, ReadString(file));
-            Assert.Equal(htmlOutput.Length, file.Length);
         }
 
         [Fact]
@@ -69,7 +67,19 @@ namespace Microsoft.AspNetCore.Blazor.Server.Test
         public void InjectsScriptTagReferencingAssemblyAndDependencies()
         {
             // Arrange
-            var htmlTemplate = "<html><body><h1>Hello</h1>Some text</body><script type='blazor-boot'></script></html>";
+            var htmlTemplatePrefix = @"
+                <html>
+                <body>
+                    <h1>Hello</h1>
+                    Some text
+                    <script>alert(1)</script>";
+            var htmlTemplateSuffix = @"
+                </body>
+                </html>";
+            var htmlTemplate =
+                $@"{htmlTemplatePrefix}
+                    <script type='blazor-boot' custom1 custom2=""value"">some text that should be removed</script>
+                {htmlTemplateSuffix}";
             var dependencies = new IFileInfo[]
             {
                 new TestFileInfo("System.Abc.dll"),
@@ -80,25 +90,30 @@ namespace Microsoft.AspNetCore.Blazor.Server.Test
 
             // Act
             var file = instance.GetFileInfo("/index.html");
-            var parsedHtml = new HtmlParser().Parse(ReadString(file));
+            var fileContents = ReadString(file);
-            var firstElem = parsedHtml.Body.FirstElementChild;
-            var scriptElem = parsedHtml.Body.QuerySelector("script");
 
-            // Assert
+            // Assert: Start and end is not modified (including formatting)
-            Assert.Equal("h1", firstElem.TagName.ToLowerInvariant());
+            Assert.StartsWith(htmlTemplatePrefix, fileContents);
-            Assert.Equal("script", scriptElem.TagName.ToLowerInvariant());
+            Assert.EndsWith(htmlTemplateSuffix, fileContents);
+
+            // Assert: Boot tag is correct
+            var scriptTagText = fileContents.Substring(htmlTemplatePrefix.Length, fileContents.Length - htmlTemplatePrefix.Length - htmlTemplateSuffix.Length);
+            var parsedHtml = new HtmlParser().Parse("<html><body>" + scriptTagText + "</body></html>");
+            var scriptElem = parsedHtml.Body.QuerySelector("script");
             Assert.False(scriptElem.HasChildNodes);
             Assert.Equal("/_framework/blazor.js", scriptElem.GetAttribute("src"));
             Assert.Equal("MyApp.Entrypoint.dll", scriptElem.GetAttribute("main"));
             Assert.Equal("System.Abc.dll,MyApp.ClassLib.dll", scriptElem.GetAttribute("references"));
             Assert.False(scriptElem.HasAttribute("type"));
+            Assert.Equal(string.Empty, scriptElem.Attributes["custom1"].Value);
+            Assert.Equal("value", scriptElem.Attributes["custom2"].Value);
         }
 
         [Fact]
-        public void MissingBootScriptTagReferencingAssemblyAndDependencies()
+        public void SuppliesHtmlTemplateUnchangedIfNoBootScriptPresent()
         {
             // Arrange
-            var htmlTemplate = "<html><body><h1>Hello</h1>Some text</body></html>";
+            var htmlTemplate = "<!DOCTYPE html><html><body><h1 style='color:red'>Hello</h1>Some text<script type='irrelevant'>blah</script></body></html>";
             var dependencies = new IFileInfo[]
             {
                 new TestFileInfo("System.Abc.dll"),
@@ -109,14 +124,9 @@ namespace Microsoft.AspNetCore.Blazor.Server.Test
 
             // Act
             var file = instance.GetFileInfo("/index.html");
-            var parsedHtml = new HtmlParser().Parse(ReadString(file));
-            var firstElem = parsedHtml.Body.FirstElementChild;
-            var scriptElem = parsedHtml.Body.QuerySelector("script");
-
 
             // Assert
-            Assert.Equal("h1", firstElem.TagName.ToLowerInvariant());
+            Assert.Equal(htmlTemplate, ReadString(file));
-            Assert.Null(scriptElem);
         }
 
         private static string ReadString(IFileInfo file)