From 02cc211e9113ba95ce267555adb49098429cedfe Mon Sep 17 00:00:00 2001
From: pablodanswer <pablo@danswer.ai>
Date: Tue, 15 Oct 2024 16:22:40 -0700
Subject: [PATCH] improved code block copying (#2802)

* improved code block copying

* k
---
 web/src/app/chat/message/Messages.tsx |  2 +-
 web/src/app/chat/message/codeUtils.ts | 42 +++++++++++----------------
 2 files changed, 18 insertions(+), 26 deletions(-)

diff --git a/web/src/app/chat/message/Messages.tsx b/web/src/app/chat/message/Messages.tsx
index 56b917164..181e00845 100644
--- a/web/src/app/chat/message/Messages.tsx
+++ b/web/src/app/chat/message/Messages.tsx
@@ -256,7 +256,7 @@ export const AIMessage = ({
     () => ({
       a: MemoizedLink,
       p: MemoizedParagraph,
-      code: ({ node, inline, className, children, ...props }: any) => {
+      code: ({ node, className, children, ...props }: any) => {
         const codeText = extractCodeText(
           node,
           finalContent as string,
diff --git a/web/src/app/chat/message/codeUtils.ts b/web/src/app/chat/message/codeUtils.ts
index 2aaae71bc..d2aad2990 100644
--- a/web/src/app/chat/message/codeUtils.ts
+++ b/web/src/app/chat/message/codeUtils.ts
@@ -4,40 +4,32 @@ export function extractCodeText(
   children: React.ReactNode
 ): string {
   let codeText: string | null = null;
+
   if (
     node?.position?.start?.offset != null &&
     node?.position?.end?.offset != null
   ) {
-    codeText = content.slice(
-      node.position.start.offset,
-      node.position.end.offset
-    );
-    codeText = codeText.trim();
+    codeText = content
+      .slice(node.position.start.offset, node.position.end.offset)
+      .trim();
 
-    // Find the last occurrence of closing backticks
-    const lastBackticksIndex = codeText.lastIndexOf("```");
-    if (lastBackticksIndex !== -1) {
-      codeText = codeText.slice(0, lastBackticksIndex + 3);
+    // Match code block with optional language declaration
+    const codeBlockMatch = codeText.match(/^```[^\n]*\n([\s\S]*?)\n?```$/);
+    if (codeBlockMatch) {
+      codeText = codeBlockMatch[1];
     }
 
-    // Remove the language declaration and trailing backticks
+    // Normalize indentation
     const codeLines = codeText.split("\n");
-    if (codeLines.length > 1 && codeLines[0].trim().startsWith("```")) {
-      codeLines.shift(); // Remove the first line with the language declaration
-      if (codeLines[codeLines.length - 1]?.trim() === "```") {
-        codeLines.pop(); // Remove the last line with the trailing backticks
-      }
+    const minIndent = codeLines
+      .filter((line) => line.trim().length > 0)
+      .reduce((min, line) => {
+        const match = line.match(/^\s*/);
+        return Math.min(min, match ? match[0].length : min);
+      }, Infinity);
 
-      const minIndent = codeLines
-        .filter((line) => line.trim().length > 0)
-        .reduce((min, line) => {
-          const match = line.match(/^\s*/);
-          return Math.min(min, match ? match[0].length : 0);
-        }, Infinity);
-
-      const formattedCodeLines = codeLines.map((line) => line.slice(minIndent));
-      codeText = formattedCodeLines.join("\n");
-    }
+    const formattedCodeLines = codeLines.map((line) => line.slice(minIndent));
+    codeText = formattedCodeLines.join("\n").trim();
   } else {
     // Fallback if position offsets are not available
     codeText = children?.toString() || null;