Implement EncodeHTML function and tests

microsoft · Mar 8, 2024 · ed43022 · ed43022
1 parent 685aa72
commit ed43022
Show file tree

Hide file tree

Showing 9 changed files with 63 additions and 56 deletions.
diff --git a/src/libraries/Microsoft.PowerFx.Core/Localization/Strings.cs b/src/libraries/Microsoft.PowerFx.Core/Localization/Strings.cs
@@ -518,8 +518,8 @@ internal static class TexlStrings
         public static StringGetter AboutEncodeUrl = (b) => StringResources.Get("AboutEncodeUrl", b);
         public static StringGetter EncodeUrlArg1 = (b) => StringResources.Get("EncodeUrlArg1", b);
 
-        public static StringGetter AboutEscapeHtml = (b) => StringResources.Get("AboutEscapeHtml", b);
-        public static StringGetter EscapeHtmlArg1 = (b) => StringResources.Get("EscapeHtmlArg1", b);
+        public static StringGetter AboutEncodeHTML = (b) => StringResources.Get("AboutEncodeHTML", b);
+        public static StringGetter EncodeHTMLArg1 = (b) => StringResources.Get("EncodeHTMLArg1", b);
 
         public static StringGetter AboutPlainText = (b) => StringResources.Get("AboutPlainText", b);
         public static StringGetter PlainTextArg1 = (b) => StringResources.Get("PlainTextArg1", b);

diff --git a/src/libraries/Microsoft.PowerFx.Core/Texl/BuiltinFunctionsCore.cs b/src/libraries/Microsoft.PowerFx.Core/Texl/BuiltinFunctionsCore.cs
@@ -94,10 +94,10 @@ internal class BuiltinFunctionsCore
         public static readonly TexlFunction DropColumns = _library.Add(new DropColumnsFunction());
         public static readonly TexlFunction EDate = _library.Add(new EDateFunction());
         public static readonly TexlFunction EOMonth = _library.Add(new EOMonthFunction());
+        public static readonly TexlFunction EncodeHTML = _library.Add(new EncodeHTMLFunction());
         public static readonly TexlFunction EncodeUrl = _library.Add(new EncodeUrlFunction());
         public static readonly TexlFunction EndsWith = _library.Add(new EndsWithFunction());
         public static readonly TexlFunction Error = _library.Add(new ErrorFunction());
-        public static readonly TexlFunction EscapeHtml = _library.Add(new EscapeHtmlFunction());
         public static readonly TexlFunction Exp = _library.Add(new ExpFunction());
         public static readonly TexlFunction ExpT = _library.Add(new ExpTableFunction());
         public static readonly TexlFunction Filter = _library.Add(new FilterFunction());

diff --git a/src/libraries/Microsoft.PowerFx.Core/Texl/Builtins/EncodeHTML.cs b/src/libraries/Microsoft.PowerFx.Core/Texl/Builtins/EncodeHTML.cs
@@ -0,0 +1,23 @@
+// Copyright (c) Microsoft Corporation.
+// Licensed under the MIT license.
+
+using System.Collections.Generic;
+using Microsoft.PowerFx.Core.Localization;
+using Microsoft.PowerFx.Core.Texl.Builtins;
+using Microsoft.PowerFx.Core.Types;
+
+namespace Microsoft.PowerFx.Core.Texl
+{
+    internal sealed class EncodeHTMLFunction : StringOneArgFunction
+    {
+        public EncodeHTMLFunction()
+                : base("EncodeHTML", TexlStrings.AboutEncodeHTML, FunctionCategories.Text)
+        {
+        }
+
+        public override IEnumerable<TexlStrings.StringGetter[]> GetSignatures()
+        {
+            yield return new[] { TexlStrings.EncodeHTMLArg1 };
+        }
+    }
+}
diff --git a/src/libraries/Microsoft.PowerFx.Core/Texl/Builtins/EscapeHtml.cs b/src/libraries/Microsoft.PowerFx.Core/Texl/Builtins/EscapeHtml.cs
diff --git a/src/libraries/Microsoft.PowerFx.Interpreter/Functions/Library.cs b/src/libraries/Microsoft.PowerFx.Interpreter/Functions/Library.cs
@@ -641,6 +641,17 @@ static Library()
                     returnBehavior: ReturnBehavior.AlwaysEvaluateAndReturnResult,
                     targetFunction: EOMonth)
             },
+            {
+                BuiltinFunctionsCore.EncodeHTML,
+                StandardErrorHandling<StringValue>(
+                    BuiltinFunctionsCore.EncodeUrl.Name,
+                    expandArguments: NoArgExpansion,
+                    replaceBlankValues: NoOpAlreadyHandledByIR,
+                    checkRuntimeTypes: ExactValueType<StringValue>,
+                    checkRuntimeValues: DeferRuntimeValueChecking,
+                    returnBehavior: ReturnBehavior.AlwaysEvaluateAndReturnResult,
+                    targetFunction: EncodeHTML)
+            },
             {
                 BuiltinFunctionsCore.EncodeUrl,
                 StandardErrorHandling<StringValue>(
@@ -674,17 +685,6 @@ static Library()
                     returnBehavior: ReturnBehavior.ReturnFalseIfAnyArgIsBlank,
                     targetFunction: Error)
             },
-            {
-                BuiltinFunctionsCore.EscapeHtml,
-                StandardErrorHandling<StringValue>(
-                    BuiltinFunctionsCore.EncodeUrl.Name,
-                    expandArguments: NoArgExpansion,
-                    replaceBlankValues: NoOpAlreadyHandledByIR,
-                    checkRuntimeTypes: ExactValueType<StringValue>,
-                    checkRuntimeValues: DeferRuntimeValueChecking,
-                    returnBehavior: ReturnBehavior.AlwaysEvaluateAndReturnResult,
-                    targetFunction: EscapeHtml)
-            },
             {
                 BuiltinFunctionsCore.Exp,
                 StandardErrorHandling<NumberValue>(

diff --git a/src/libraries/Microsoft.PowerFx.Interpreter/Functions/LibraryText.cs b/src/libraries/Microsoft.PowerFx.Interpreter/Functions/LibraryText.cs
@@ -802,7 +802,7 @@ public static FormulaValue EncodeUrl(IRContext irContext, StringValue[] args)
             return new StringValue(irContext, Uri.EscapeDataString(args[0].Value));
         }
 
-        public static FormulaValue EscapeHtml(IRContext irContext, StringValue[] args)
+        public static FormulaValue EncodeHTML(IRContext irContext, StringValue[] args)
         {
             var encoded = HttpUtility.HtmlEncode(args[0].Value);
             return new StringValue(irContext, encoded);

diff --git a/src/strings/PowerFxResources.en-US.resx b/src/strings/PowerFxResources.en-US.resx
@@ -4092,16 +4092,16 @@
   <data name="AboutEncodeUrl_url" xml:space="preserve">
     <value>A url to be encoded.</value>
   </data>
-  <data name="AboutEscapeHtml" xml:space="preserve">
+  <data name="AboutEncodeHTML" xml:space="preserve">
     <value>Converts a text to an HTML-encoded text.</value>
-    <comment>Description of 'EncodeUrl' function.</comment>
+    <comment>Description of 'EncodeHTML' function.</comment>
   </data>
-  <data name="EscapeHtmlArg1" xml:space="preserve">
+  <data name="EncodeHTMLArg1" xml:space="preserve">
     <value>value</value>
-    <comment>Function_parameter - First parameter for the EscapeHtml function - the value (string) to escape.</comment>
+    <comment>Function_parameter - First parameter for the EncodeHTML function - the value (string) to encode.</comment>
   </data>
-  <data name="AboutEscapeHtml_value" xml:space="preserve">
-    <value>A text to be escaped.</value>
+  <data name="AboutEncodeHTML_value" xml:space="preserve">
+    <value>A text to be encoded.</value>
   </data>
   <data name="ErrUnimplementedFunction" xml:space="preserve">
     <value>'{0}' is a recognized but not supported function.</value>

diff --git a/....Tests/ExpressionTestCases/EscapeHtml.txt → ....Tests/ExpressionTestCases/EncodeHTML.txt b/....Tests/ExpressionTestCases/EscapeHtml.txt → ....Tests/ExpressionTestCases/EncodeHTML.txt
@@ -1,42 +1,49 @@
 // Escaping
->> EscapeHtml("<p>A paragraph</p>")
+>> EncodeHTML("<p>A paragraph</p>")
 "&lt;p&gt;A paragraph&lt;/p&gt;"
 
 // Multiple encodings
->> EscapeHtml("<h1>Mac & cheese</h1>")
+>> EncodeHTML("<h1>Mac & cheese</h1>")
 "&lt;h1&gt;Mac &amp; cheese&lt;/h1&gt;"
 
 // Quotes
->> EscapeHtml("A value with ""double"" and 'single' quotes")
+>> EncodeHTML("A value with ""double"" and 'single' quotes")
 "A value with &quot;double&quot; and &#39;single&#39; quotes"
 
 // \u00A0 to \u00af - escaped
->> EscapeHtml(Concat(Sequence(16, 160), UniChar(Value)))
+>> EncodeHTML(Concat(Sequence(16, 160), UniChar(Value)))
 "&#160;&#161;&#162;&#163;&#164;&#165;&#166;&#167;&#168;&#169;&#170;&#171;&#172;&#173;&#174;&#175;"
 
 // \u0090 to \u009f - not escaped
->> With({str:Concat(Sequence(16, 144), UniChar(Value))}, str = EscapeHtml(str))
+>> With({str:Concat(Sequence(16, 144), UniChar(Value))}, str = EncodeHTML(str))
 true
 
 // \u00A0 to \u00ff - escaped
->> With({str:EscapeHtml(Concat(Sequence(96, 160), UniChar(Value))), expectedEscaped:Concat(Sequence(96, 160), $"&#{Value};")}, str = expectedEscaped)
+>> With({str:EncodeHTML(Concat(Sequence(96, 160), UniChar(Value))), expectedEscaped:Concat(Sequence(96, 160), $"&#{Value};")}, str = expectedEscaped)
 true
 
 // \u0100... not escaped
->> EscapeHtml("<b>ĀāĂă</b>")
+>> EncodeHTML("<b>ĀāĂă</b>")
 "&lt;b&gt;ĀāĂă&lt;/b&gt;"
 
 // Blanks
->> EscapeHtml(Blank())
+>> EncodeHTML(Blank())
+""
+
+>> EncodeHTML("")
 ""
 
 // Errors
->> EscapeHtml(Char(-1))
+>> EncodeHTML(Char(-1))
 Error({Kind:ErrorKind.InvalidArgument})
 
 // Escaping surrogate pairs
->> EscapeHtml("<ul><li>not a pair: ❤</li><li>a surrogate pair: 💩</li></ul>")
+>> EncodeHTML("<ul><li>not a pair: ❤</li><li>a surrogate pair: 💩</li></ul>")
 "&lt;ul&gt;&lt;li&gt;not a pair: ❤&lt;/li&gt;&lt;li&gt;a surrogate pair: &#128169;&lt;/li&gt;&lt;/ul&gt;"
 
->> EscapeHtml("Osage alphabet (start): 𐒰𐒱𐒲𐒳𐒴𐒵𐒶")
+>> EncodeHTML("Osage alphabet (start): 𐒰𐒱𐒲𐒳𐒴𐒵𐒶")
 "Osage alphabet (start): &#66736;&#66737;&#66738;&#66739;&#66740;&#66741;&#66742;"
+
+// Unpaired surrogate characters become the replacement character U+FFFD
+>> EncodeHTML($"Unpaired: {UniChar(Hex2Dec("df32"))} {UniChar(Hex2Dec("d823"))}"))
+"Unpaired: � �"
diff --git a/src/tests/Microsoft.PowerFx.Interpreter.Tests/Docs/InterpreterBase.json b/src/tests/Microsoft.PowerFx.Interpreter.Tests/Docs/InterpreterBase.json
@@ -46,11 +46,11 @@
     "Distinct",
     "DropColumns",
     "EDate",
+    "EncodeHTML",
     "EncodeUrl",
     "EndsWith",
     "EOMonth",
     "Error",
-    "EscapeHtml",
     "Exp",
     "Filter",
     "Find",