Optimize the save performance of page elements

stevencohn#1122

OneMore/Helpers/Extensions/HashAlgorithmExtensions.cs

@@ -0,0 +1,21 @@
+//************************************************************************************************
+// Copyright © 2032 Steven M Cohn.  All rights reserved.
+//************************************************************************************************
+
+namespace River.OneMoreAddIn
+{
+	using System.Numerics;
+	using System.Security.Cryptography;
+	using System.Text;
+
+
+	internal static class HashAlgorithmExtensions
+	{
+		public static string GetHashString(this HashAlgorithm algorithm, string s)
+		{
+			var data = Encoding.UTF8.GetBytes(s);
+			var hash = algorithm.ComputeHash(data);
+			return new BigInteger(hash).ToString();
+		}
+	}
+}

OneMore/Models/Page.cs

@@ -16,7 +16,7 @@ namespace River.OneMoreAddIn.Models
 	using System.Globalization;
 	using System.Linq;
 	using System.Media;
-	using System.Security.Policy;
+	using System.Security.Cryptography;
 	using System.Text;
 	using System.Text.RegularExpressions;
 	using System.Xml;
@@ -30,6 +30,8 @@ public partial class Page
 	{
 		public int TopOutlinePosition = 86;
 
+		private const string HashAttributeName = "omHash";
+
 
 		/// <summary>
 		/// Initialize a new instance with the given page XML root
@@ -39,16 +41,65 @@ public Page(XElement root)
 		{
 			if (root != null)
 			{
-				Root = root;
 				Namespace = root.GetNamespaceOfPrefix(OneNote.Prefix);
-
 				PageId = root.Attribute("ID")?.Value;
+				ComputeHashes(root);
+
+				Root = root;
 			}
 
 			SelectionScope = SelectionScope.Unknown;
 		}
 
 
+		#region Hashing
+		/// <summary>
+		/// Calculates a hash value for each 1st gen child element on the page, which will
+		/// be used to optimize the page just prior to saving.
+		/// </summary>
+		/// <param name="root">The root element of the page</param>
+		private void ComputeHashes(XElement root)
+		{
+			using var algo = MD5.Create();
+
+			// 1st generation child elements of the Page
+			foreach (var child in root.Elements())
+			{
+				child.Add(new XAttribute(
+					HashAttributeName,
+					algo.GetHashString(child.ToString(SaveOptions.DisableFormatting))
+					));
+			}
+		}
+
+
+		/// <summary>
+		/// Keeps only modified 1st gen child elements of the page to optimize save performance
+		/// especially when the page is loaded with many Ink drawings.
+		/// </summary>
+		public void OptimizeForSave()
+		{
+			// MD5 should be sufficient and performs better than any other algorithm
+			using var algo = MD5.Create();
+
+			// 1st generation child elements of the Page
+			foreach (var child in Root.Elements().ToList())
+			{
+				var att = child.Attribute(HashAttributeName);
+				if (att != null)
+				{
+					att.Remove();
+					var hash = algo.GetHashString(child.ToString(SaveOptions.DisableFormatting));
+					if (hash == att.Value)
+					{
+						child.Remove();
+					}
+				}
+			}
+		}
+		#endregion Hashing
+
+
 		public bool IsValid => Root != null;
 
 

OneMore/OneMore.csproj

@@ -241,6 +241,7 @@
     <Compile Include="Helpers\ColorHelper.cs" />
     <Compile Include="Helpers\Extensions\FontFamilyExtensions.cs" />
     <Compile Include="Helpers\Extensions\GraphicsExtensions.cs" />
+    <Compile Include="Helpers\Extensions\HashAlgorithmExtensions.cs" />
     <Compile Include="Helpers\Extensions\ScreenExtensions.cs" />
     <Compile Include="Models\KnownSchemaAttributes.cs" />
     <Compile Include="Models\PageReader.cs" />

OneMore/OneNote.cs

@@ -1047,22 +1047,13 @@ public async Task Update(Page page)
 				return;
 			}
 
-			await Update(page.Root);
-		}
+			// must optimize before we can validate schema...
 
+			page.OptimizeForSave();
 
-		/// <summary>
-		/// Updates the given content, with a unique ID, on the current page.
-		/// </summary>
-		/// <param name="element">A page or element within a page with a unique objectID</param>
-		public async Task Update(XElement element)
-		{
-			if (element.Name.LocalName == "Page")
+			if (!ValidateSchema(page.Root))
 			{
-				if (!ValidateSchema(element))
-				{
-					return;
-				}
+				return;
 			}
 
 			// dateExpectedLastModified is merely a pessimistic-locking safeguard to prevent
@@ -1072,7 +1063,7 @@ public async Task Update(XElement element)
 			//	? DateTime.Parse(att.Value).ToUniversalTime()
 			//	: DateTime.MinValue;
 
-			var xml = element.ToString(SaveOptions.DisableFormatting);
+			var xml = page.Root.ToString(SaveOptions.DisableFormatting);
 
 			await InvokeWithRetry(() =>
 			{