Merge branch 'Cynser-text-brute-force'

CHANGELOG.md

@@ -2,6 +2,9 @@
 All major and minor version changes will be documented in this file. Details of patch-level version changes can be found in [commit messages](https://github.com/gchq/CyberChef/commits/master).
 
 
+### [8.15.0] - 2018-12-18
+- 'Text Encoding Brute Force' operation added [@Cynser] | [#439]
+
 ### [8.14.0] - 2018-12-18
 - 'To Base62' and 'From Base62' operations added [@tcode2k16] | [#443]
 
@@ -73,6 +76,7 @@ All major and minor version changes will be documented in this file. Details of
 
 
 
+[8.15.0]: https://github.com/gchq/CyberChef/releases/tag/v8.15.0
 [8.14.0]: https://github.com/gchq/CyberChef/releases/tag/v8.14.0
 [8.13.0]: https://github.com/gchq/CyberChef/releases/tag/v8.13.0
 [8.12.0]: https://github.com/gchq/CyberChef/releases/tag/v8.12.0
@@ -107,6 +111,7 @@ All major and minor version changes will be documented in this file. Details of
 [@bwhitn]: https://github.com/bwhitn
 [@jarmovanlenthe]: https://github.com/jarmovanlenthe
 [@tcode2k16]: https://github.com/tcode2k16
+[@Cynser]: https://github.com/Cynser
 
 [#95]: https://github.com/gchq/CyberChef/pull/299
 [#173]: https://github.com/gchq/CyberChef/pull/173
@@ -130,5 +135,6 @@ All major and minor version changes will be documented in this file. Details of
 [#387]: https://github.com/gchq/CyberChef/pull/387
 [#394]: https://github.com/gchq/CyberChef/pull/394
 [#428]: https://github.com/gchq/CyberChef/pull/428
+[#439]: https://github.com/gchq/CyberChef/pull/439
 [#441]: https://github.com/gchq/CyberChef/pull/441
 [#443]: https://github.com/gchq/CyberChef/pull/443

src/core/config/Categories.json

@@ -51,6 +51,7 @@
             "Change IP format",
             "Encode text",
             "Decode text",
+            "Text Encoding Brute Force",
             "Swap endianness",
             "To MessagePack",
             "From MessagePack",

src/core/operations/TextEncodingBruteForce.mjs

@@ -0,0 +1,92 @@
+/**
+ * @author Cynser
+ * @author n1474335 [n1474335@gmail.com]
+ * @copyright Crown Copyright 2018
+ * @license Apache-2.0
+ */
+
+import Operation from "../Operation";
+import Utils from "../Utils";
+import cptable from "../vendor/js-codepage/cptable.js";
+import {IO_FORMAT} from "../lib/ChrEnc";
+
+/**
+ * Text Encoding Brute Force operation
+ */
+class TextEncodingBruteForce extends Operation {
+
+    /**
+     * TextEncodingBruteForce constructor
+     */
+    constructor() {
+        super();
+
+        this.name = "Text Encoding Brute Force";
+        this.module = "CharEnc";
+        this.description = [
+            "Enumerates all supported text encodings for the input, allowing you to quickly spot the correct one.",
+            "<br><br>",
+            "Supported charsets are:",
+            "<ul>",
+            Object.keys(IO_FORMAT).map(e => `<li>${e}</li>`).join("\n"),
+            "</ul>"
+        ].join("\n");
+        this.infoURL = "https://wikipedia.org/wiki/Character_encoding";
+        this.inputType = "string";
+        this.outputType = "json";
+        this.presentType = "html";
+        this.args = [
+            {
+                name: "Mode",
+                type: "option",
+                value: ["Encode", "Decode"]
+            }
+        ];
+    }
+
+    /**
+     * @param {string} input
+     * @param {Object[]} args
+     * @returns {json}
+     */
+    run(input, args) {
+        const output = {},
+            charsets = Object.keys(IO_FORMAT),
+            mode = args[0];
+
+        charsets.forEach(charset => {
+            try {
+                if (mode === "Decode") {
+                    output[charset] = cptable.utils.decode(IO_FORMAT[charset], input);
+                } else {
+                    output[charset] = Utils.arrayBufferToStr(cptable.utils.encode(IO_FORMAT[charset], input));
+                }
+            } catch (err) {
+                output[charset] = "Could not decode.";
+            }
+        });
+
+        return output;
+    }
+
+    /**
+     * Displays the encodings in an HTML table for web apps.
+     *
+     * @param {Object[]} encodings
+     * @returns {html}
+     */
+    present(encodings) {
+        let table = "<table class='table table-hover table-sm table-bordered table-nonfluid'><tr><th>Encoding</th><th>Value</th></tr>";
+
+        for (const enc in encodings) {
+            const value = Utils.printable(encodings[enc], true);
+            table += `<tr><td>${enc}</td><td>${value}</td></tr>`;
+        }
+
+        table += "<table>";
+        return table;
+    }
+
+}
+
+export default TextEncodingBruteForce;

src/core/operations/ToBase62.mjs

@@ -23,7 +23,7 @@ class ToBase62 extends Operation {
         this.name = "To Base62";
         this.module = "Default";
         this.description = "Base62 is a notation for encoding arbitrary byte data using a restricted set of symbols that can be conveniently used by humans and processed by computers. The high number base results in shorter strings than with the decimal or hexadecimal system.";
-        this.infoURL = "https://en.wikipedia.org/wiki/List_of_numeral_systems";
+        this.infoURL = "https://wikipedia.org/wiki/List_of_numeral_systems";
         this.inputType = "byteArray";
         this.outputType = "string";
         this.args = [

test/index.mjs

@@ -75,6 +75,7 @@ import "./tests/operations/SetIntersection";
 import "./tests/operations/SetUnion";
 import "./tests/operations/StrUtils";
 import "./tests/operations/SymmetricDifference";
+import "./tests/operations/TextEncodingBruteForce";
 import "./tests/operations/ToGeohash.mjs";
 import "./tests/operations/TranslateDateTimeFormat";
 import "./tests/operations/Magic";

test/tests/operations/TextEncodingBruteForce.mjs

@@ -0,0 +1,35 @@
+/**
+ * Text Encoding Brute Force tests.
+ *
+ * @author Cynser
+ *
+ * @copyright Crown Copyright 2018
+ * @license Apache-2.0
+ */
+import TestRegister from "../../TestRegister";
+
+TestRegister.addTests([
+    {
+        name: "Text Encoding Brute Force - Encode",
+        input: "Булкі праз ляніва сабаку.",
+        expectedMatch: /Windows-1251 Cyrillic \(1251\).{1,10}Булкі праз ляніва сабаку\./,
+        recipeConfig: [
+            {
+                op: "Text Encoding Brute Force",
+                args: ["Encode"],
+            },
+        ],
+    },
+    {
+        name: "Text Encoding Brute Force - Decode",
+        input: "Áóëê³ ïðàç ëÿí³âà ñàáàêó.",
+        expectedMatch: /Windows-1251 Cyrillic \(1251\).{1,10}Булкі праз ляніва сабаку\./,
+        recipeConfig: [
+            {
+                op: "Text Encoding Brute Force",
+                args: ["Decode"],
+            },
+        ],
+    }
+]);
+