From a141873db816d874298288461d8c65d4b47899ad Mon Sep 17 00:00:00 2001
From: n1474335 <n1474335@gmail.com>
Date: Fri, 16 Sep 2022 16:00:39 +0100
Subject: [PATCH] Highlighting now takes account of character set width

---
 src/core/lib/ChrEnc.mjs               | 42 +++++++++++++++++++++++++++
 src/web/waiters/HighlighterWaiter.mjs | 20 +++++++++----
 2 files changed, 57 insertions(+), 5 deletions(-)

diff --git a/src/core/lib/ChrEnc.mjs b/src/core/lib/ChrEnc.mjs
index 8934d137..6879d736 100644
--- a/src/core/lib/ChrEnc.mjs
+++ b/src/core/lib/ChrEnc.mjs
@@ -6,6 +6,8 @@
  * @license Apache-2.0
  */
 
+import cptable from "codepage";
+
 /**
  * Character encoding format mappings.
  */
@@ -175,6 +177,46 @@ for (const name in CHR_ENC_CODE_PAGES) {
     CHR_ENC_SIMPLE_REVERSE_LOOKUP[CHR_ENC_CODE_PAGES[name]] = simpleName;
 }
 
+
+/**
+ * Returns the width of the character set for the given codepage.
+ * For example, UTF-8 is a Single Byte Character Set, whereas
+ * UTF-16 is a Double Byte Character Set.
+ *
+ * @param {number} page - The codepage number
+ * @returns {number}
+ */
+export function chrEncWidth(page) {
+    if (typeof page !== "number") return 0;
+
+    // Raw Bytes have a width of 1
+    if (page === 0) return 1;
+
+    const pageStr = page.toString();
+    // Confirm this page is legitimate
+    if (!Object.prototype.hasOwnProperty.call(CHR_ENC_SIMPLE_REVERSE_LOOKUP, pageStr))
+        return 0;
+
+    // Statically defined code pages
+    if (Object.prototype.hasOwnProperty.call(cptable, pageStr))
+        return cptable[pageStr].dec.length > 256 ? 2 : 1;
+
+    // Cached code pages
+    if (cptable.utils.cache.sbcs.includes(pageStr))
+        return 1;
+    if (cptable.utils.cache.dbcs.includes(pageStr))
+        return 2;
+
+    // Dynamically generated code pages
+    if (Object.prototype.hasOwnProperty.call(cptable.utils.magic, pageStr)) {
+        // Generate a single character and measure it
+        const a = cptable.utils.encode(page, "a");
+        return a.length;
+    }
+
+    return 0;
+}
+
 /**
  * Unicode Normalisation Forms
  *
diff --git a/src/web/waiters/HighlighterWaiter.mjs b/src/web/waiters/HighlighterWaiter.mjs
index 189d3777..aca55e84 100755
--- a/src/web/waiters/HighlighterWaiter.mjs
+++ b/src/web/waiters/HighlighterWaiter.mjs
@@ -5,7 +5,7 @@
  */
 
 import {EditorSelection} from "@codemirror/state";
-
+import {chrEncWidth} from "../../core/lib/ChrEnc.mjs";
 
 /**
  * Waiter to handle events related to highlighting in CyberChef.
@@ -50,12 +50,23 @@ class HighlighterWaiter {
         // Confirm some non-empty ranges are set
         const selectionRanges = e.state.selection.ranges;
 
+        // Adjust offsets based on the width of the character set
+        const inputCharacterWidth = chrEncWidth(this.manager.input.inputChrEnc);
+        const outputCharacterWidth = chrEncWidth(this.manager.output.outputChrEnc);
+        let ratio = 1;
+        if (inputCharacterWidth !== outputCharacterWidth &&
+            inputCharacterWidth !== 0 && outputCharacterWidth !== 0) {
+            ratio = io === "input" ?
+                inputCharacterWidth / outputCharacterWidth :
+                outputCharacterWidth / inputCharacterWidth;
+        }
+
         // Loop through ranges and send request for output offsets for each one
         const direction = io === "input" ? "forward" : "reverse";
         for (const range of selectionRanges) {
             const pos = [{
-                start: range.from,
-                end: range.to
+                start: Math.floor(range.from * ratio),
+                end: Math.floor(range.to * ratio)
             }];
             this.manager.worker.highlight(this.app.getRecipeConfig(), direction, pos);
         }
@@ -80,8 +91,7 @@ class HighlighterWaiter {
 
 
     /**
-     * Adds the relevant HTML to the specified highlight element such that highlighting appears
-     * underneath the correct offset.
+     * Sends selection updates to the relevant EditorView.
      *
      * @param {string} io - The input or output
      * @param {Object[]} ranges - An array of position objects to highlight