Merge branch 'features/unicode-format' of https://github.com/mattnotm…

…itt/CyberChef into mattnotmitt-features/unicode-format

src/core/config/Categories.json

@@ -200,6 +200,7 @@
         "ops": [
             "Encode text",
             "Decode text",
+            "Unicode Text Format",
             "Remove Diacritics",
             "Unescape Unicode Characters",
             "Convert to NATO alphabet"

src/core/operations/RemoveDiacritics.mjs

@@ -19,7 +19,7 @@ class RemoveDiacritics extends Operation {
 
         this.name = "Remove Diacritics";
         this.module = "Default";
-        this.description = "Replaces accented characters with their latin character equivalent.";
+        this.description = "Replaces accented characters with their latin character equivalent. Accented characters are made up of Unicode combining characters, so unicode text formatting such as strikethroughs and underlines will also be removed.";
         this.infoURL = "https://wikipedia.org/wiki/Diacritic";
         this.inputType = "string";
         this.outputType = "string";

src/core/operations/UnicodeTextFormat.mjs

@@ -0,0 +1,67 @@
+/**
+ * @author Matt C [me@mitt.dev]
+ * @copyright Crown Copyright 2020
+ * @license Apache-2.0
+ */
+
+import Operation from "../Operation.mjs";
+import Utils from "../Utils.mjs";
+
+/**
+ * Unicode Text Format operation
+ */
+class UnicodeTextFormat extends Operation {
+
+    /**
+     * UnicodeTextFormat constructor
+     */
+    constructor() {
+        super();
+
+        this.name = "Unicode Text Format";
+        this.module = "Default";
+        this.description = "Adds Unicode combining characters to change formatting of plaintext.";
+        this.infoURL = "https://en.wikipedia.org/wiki/Combining_character";
+        this.inputType = "byteArray";
+        this.outputType = "byteArray";
+        this.args = [
+            {
+                name: "Underline",
+                type: "boolean",
+                value: "false"
+            },
+            {
+                name: "Strikethrough",
+                type: "boolean",
+                value: "false"
+            }
+        ];
+    }
+
+    /**
+     * @param {byteArray} input
+     * @param {Object[]} args
+     * @returns {byteArray}
+     */
+    run(input, args) {
+        const [underline, strikethrough] = args;
+        let output = input.map(char => [char]);
+        if (strikethrough) {
+            output = output.map(charFormat => {
+                charFormat.push(...Utils.strToUtf8ByteArray("\u0336"));
+                return charFormat;
+            });
+        }
+        if (underline) {
+            output = output.map(charFormat => {
+                charFormat.push(...Utils.strToUtf8ByteArray("\u0332"));
+                return charFormat;
+            });
+        }
+        // return output.flat(); - Not supported in Node 10, polyfilled
+        return [].concat(...output);
+    }
+
+}
+
+export default UnicodeTextFormat;

tests/operations/index.mjs

@@ -69,7 +69,6 @@ import "./tests/ParseQRCode.mjs";
 import "./tests/PowerSet.mjs";
 import "./tests/Regex.mjs";
 import "./tests/Register.mjs";
-import "./tests/RemoveDiacritics.mjs";
 import "./tests/Rotate.mjs";
 import "./tests/SeqUtils.mjs";
 import "./tests/SetDifference.mjs";
@@ -101,6 +100,7 @@ import "./tests/LuhnChecksum.mjs";
 import "./tests/CipherSaber2.mjs";
 import "./tests/Colossus.mjs";
 import "./tests/ParseObjectIDTimestamp.mjs";
+import "./tests/Unicode.mjs";
 
 
 // Cannot test operations that use the File type yet

tests/operations/tests/Unicode.mjs

@@ -0,0 +1,83 @@
+/**
+ * Unicode operation tests.
+ *
+ * @author Matt C [me@mitt.dev]
+ * @author Klaxon [klaxon@veyr.com]
+ *
+ * @copyright Crown Copyright 2020
+ * @license Apache-2.0
+ */
+import TestRegister from "../../lib/TestRegister.mjs";
+
+TestRegister.addTests([
+    {
+        name: "Unicode Text Format: underline",
+        input: "a",
+        expectedOutput: "a\u0332",
+        recipeConfig: [
+            {
+                "op": "Unicode Text Format",
+                "args": [true, false],
+            }
+        ],
+    },
+    {
+        name: "Unicode Text Format: strikethrough",
+        input: "a",
+        expectedOutput: "a\u0336",
+        recipeConfig: [
+            {
+                "op": "Unicode Text Format",
+                "args": [false, true],
+            }
+        ],
+    },
+    {
+        name: "Unicode Text Format: both",
+        input: "a",
+        expectedOutput: "a\u0336\u0332",
+        recipeConfig: [
+            {
+                "op": "Unicode Text Format",
+                "args": [true, true],
+            }
+        ],
+    },
+    {
+        name: "Remove Diacritics: text formatting",
+        input: "a",
+        expectedOutput: "a",
+        recipeConfig: [
+            {
+                "op": "Unicode Text Format",
+                "args": [true, true],
+            },
+            {
+                "op": "Remove Diacritics",
+                "args": []
+            }
+        ],
+    },
+    {
+        name: "Remove Diacritics: all diacritical marks one char",
+        input: "à̴̵̶̷̸̡̢̧̨̛̖̗̘̙̜̝̞̟̠̣̤̥̦̩̪̫̬̭̮̯̰̱̲̳̹̺̻̼́̂̃̄̅̆̇̈̉̊̋̌̍̎̏̐̑̒̓̔̽̾̿̀́͂̓̈́̕̚͠͡ͅ", // sorry about this line lol
+        expectedOutput: "a",
+        recipeConfig: [
+            {
+                "op": "Remove Diacritics",
+                "args": []
+            }
+        ],
+    },
+    {
+        name: "Remove Diacritics: default",
+        input: "\xe0, \xe8, \xec, \xf2, \xf9  \xc0, \xc8, \xcc, \xd2, \xd9\n\xe1, \xe9, \xed, \xf3, \xfa, \xfd \xc1, \xc9, \xcd, \xd3, \xda, \xdd\n\xe2, \xea, \xee, \xf4, \xfb \xc2, \xca, \xce, \xd4, \xdb\n\xe3, \xf1, \xf5 \xc3, \xd1, \xd5\n\xe4, \xeb, \xef, \xf6, \xfc, \xff \xc4, \xcb, \xcf, \xd6, \xdc, \u0178\n\xe5, \xc5",
+        expectedOutput: "a, e, i, o, u  A, E, I, O, U\na, e, i, o, u, y A, E, I, O, U, Y\na, e, i, o, u A, E, I, O, U\na, n, o A, N, O\na, e, i, o, u, y A, E, I, O, U, Y\na, A",
+        recipeConfig: [
+            {
+                "op": "Remove Diacritics",
+                "args": []
+            },
+        ],
+    },
+]);