issue#419

src/core/config/Categories.json

@@ -63,7 +63,8 @@
             "JSON to CSV",
             "Avro to JSON",
             "CBOR Encode",
-            "CBOR Decode"
+            "CBOR Decode",
+            "Escape Smart Characters"
         ]
     },
     {

src/core/operations/EscapeSmartCharacters.mjs

@@ -0,0 +1,146 @@
+/**
+ * @author john19696 [john19696@protonmail.com]
+ * @copyright Crown Copyright 2021
+ * @license Apache-2.0
+ */
+
+import Operation from "../Operation.mjs";
+// import OperationError from "../errors/OperationError.mjs";
+
+/**
+ * Escape Smart Characters operation
+ */
+class EscapeSmartCharacters extends Operation {
+
+    /**
+     * EscapeSmartCharacters constructor
+     */
+    constructor() {
+        super();
+
+        this.name = "Escape Smart Characters";
+        this.module = "Default";
+        this.description = "An operation to convert smart characters (quotes, dashes, apostrophes, \
+            arrows, copyright signs, ellipses etc.) back to plain ASCII.\
+            <br>";
+        this.infoURL = "http://unicode.scarfboy.com/?s=quotation+mark";
+        this.inputType = "string";
+        this.outputType = "string";
+        this.args = [
+            /* Arguments. See the project wiki for full details.*/
+            {
+                name: "Second arg",
+                type: "option",
+                value: ["Escape", "Remove", "Replace with '.'"]
+            }
+        ];
+    }
+
+
+    /**
+     * @param {string} input
+     * @param {Object[]} args
+     * @returns {string}
+     */
+    run(input, args) {
+        const [outArg] = args;
+
+        /* JSON map of characters to escape
+           generated by EscapeSmartCharacters.xls in doc
+           */
+        const ESCAPE_MAP = {
+            // Pasted from git:/doc/SmartEscape.xlsx
+            "‘": "'",
+            "’": "'",
+            "“": "\"",
+            "”": "\"",
+            "©": "(C)",
+            "®": "(R)",
+            "™": "(TM)",
+            "→": "-->",
+            "←": "<--",
+            "↔": "<->",
+            "‐": "-",
+            "‑": "-",
+            "‒": "-",
+            "–": "-",
+            "—": "--",
+            "―": "--",
+            "‖": "||",
+            "‗": "==",
+            "‘": "'",
+            "’": "'",
+            "‚": "'",
+            "‛": "'",
+            "“": "\"",
+            "”": "\"",
+            "„": "\"",
+            "‟": "\"",
+            "•": ".",
+            "‣": ">",
+            "․": ".",
+            "‥": "..",
+            "…": "...",
+            "‧": ".",
+            "‰": "%0",
+            "‱": "%00",
+            "′": "'",
+            "″": "''",
+            "‴": "'''",
+            "‵": "''",
+            "‶": "''",
+            "‷": "'''",
+            "‸": "^",
+            "‹": "<",
+            "›": ">",
+            "‼": "!!",
+            "‽": "?!",
+            "⁃": "-",
+            "⁄": "/",
+            "⁅": "[-",
+            "⁆": "-]",
+            "⁇": "??",
+            "⁈": "?!",
+            "⁉": "!?",
+            "⁌": ".",
+            "⁍": ".",
+            "⁎": "*",
+            "⁏": ";",
+            "⁒": "%",
+            "⁓": "~",
+            "⁕": "*",
+            "⁗": "''''",
+            "⁢": "*",
+            "⁤": "+",
+        };
+
+        let output, result = "";
+        for (const char of input) {
+            if (Object.keys(ESCAPE_MAP).includes(char)) {
+                output = ESCAPE_MAP[char];
+            } else {
+                // Do Something Sensible with the rest of the arrows
+                const charCode = char.charCodeAt(0);
+                if (charCode >= 8592 && charCode <= 8703)  {
+                    output = "->";
+                } else {
+                    output = char;
+                }
+            }
+            switch (outArg) {
+                case "Remove":
+                    break;
+                case "Replace with '.'":
+                    result += ".";
+                    break;
+                default: { // Escape & no change are the same
+                    result += output;
+                }
+            }
+        }
+        return result;
+    }
+
+}
+
+export default EscapeSmartCharacters;

tests/operations/index.mjs

@@ -107,6 +107,7 @@ import "./tests/CBORDecode.mjs";
 import "./tests/JA3Fingerprint.mjs";
 import "./tests/JA3SFingerprint.mjs";
 import "./tests/HASSH.mjs";
+import "./tests/EscapeSmartCharacters.mjs";
 
 
 // Cannot test operations that use the File type yet

tests/operations/tests/EscapeSmartCharacters.mjs

@@ -0,0 +1,22 @@
+/**
+ * @author john19696 [john19696@protonmail.com]
+ * @copyright Crown Copyright 2021
+ * @license Apache-2.0
+ */
+
+ import TestRegister from "../../lib/TestRegister.mjs";
+
+ TestRegister.addTests([
+     {
+        name: "Escape Smart Characters",
+        input: "“”—‘’ →©…",
+        expectedOutput: "\"\"--'' -->(C)...",
+        recipeConfig: [
+            {
+                op: "Escape Smart Characters",
+                args: ["Escape"],
+            },
+        ],
+    },
+]);
+g