webui: Moved constants to lib/constants.

Author: srogmann

tools/server/webui/src/lib/constants/latex-protection.ts

@@ -0,0 +1,32 @@
+/**
+ * Matches common Markdown code blocks to exclude them from further processing (e.g. LaTeX).
+ * - Fenced: ```...```
+ * - Inline: `...` (does NOT support nested backticks or multi-backtick syntax)
+ *
+ * Note: This pattern does not handle advanced cases like:
+ *       `` `code with `backticks` `` or \\``...\\``
+ */
+export const CODE_BLOCK_REGEXP = /(```[\s\S]*?```|`[^`\n]+`)/g;
+
+/**
+ * Matches LaTeX math delimiters \(...\) and \[...\] only when not preceded by a backslash (i.e., not escaped),
+ * while also capturing code blocks (```, `...`) so they can be skipped during processing.
+ *
+ * Uses negative lookbehind `(?<!\\)` to avoid matching \\( or \\[.
+ * Using the look‑behind pattern `(?<!\\)` we skip matches
+ * that are preceded by a backslash, e.g.
+ * `Definitions\\(also called macros)` (title of chapter 20 in The TeXbook)
+ * or `\\[4pt]` (LaTeX line-break).
+ *
+ * group 1: code-block
+ * group 2: square-bracket
+ * group 3: round-bracket
+ */
+export const LATEX_MATH_AND_CODE_PATTERN =
+	/(```[\S\s]*?```|`.*?`)|(?<!\\)\\\[([\S\s]*?[^\\])\\]|(?<!\\)\\\((.*?)\\\)/g;
+
+/** map from mchem-regexp to replacement */
+export const MHCHEM_PATTERN_MAP: readonly [RegExp, string][] = [
+	[/(\s)\$\\ce{/g, '$1$\\\\ce{'],
+	[/(\s)\$\\pu{/g, '$1$\\\\pu{']
+] as const;

tools/server/webui/src/lib/utils/latex-protection.test.ts

@@ -276,8 +276,7 @@ $$`
 		const input = 'Chemical reaction: \\( \\ce{H2O} \\) and $\\ce{CO2}$';
 		const output = preprocessLaTeX(input);
 
-		expect(output).toBe('Chemical reaction: $ \\ce{H2O} $ and $\\\\ce{CO2}$');
-		// Note: \\ce{...} remains, but $\\ce{...} → $\\\\ce{...} via escapeMhchem
+		expect(output).toBe('Chemical reaction: $ \\ce{H2O} $ and $\\ce{CO2}$');
 	});
 
 	test('preserves code blocks', () => {
@@ -286,4 +285,20 @@ $$`
 
 		expect(output).toBe(input); // Code blocks prevent misinterpretation
 	});
+
+	test('escape backslash in mchem ce', () => {
+		const input = 'mchem ce:\n$\\ce{2H2(g) + O2(g) -> 2H2O(l)}$';
+		const output = preprocessLaTeX(input);
+
+		// mhchem-escape would insert a backslash here.
+		expect(output).toBe('mchem ce:\n$\\ce{2H2(g) + O2(g) -> 2H2O(l)}$');
+	});
+
+	test('escape backslash in mchem pu', () => {
+		const input = 'mchem pu:\n$\\pu{-572 kJ mol^{-1}}$';
+		const output = preprocessLaTeX(input);
+
+		// mhchem-escape would insert a backslash here.
+		expect(output).toBe('mchem pu:\n$\\pu{-572 kJ mol^{-1}}$');
+	});
 });

tools/server/webui/src/lib/utils/latex-protection.ts

@@ -1,3 +1,9 @@
+import {
+	CODE_BLOCK_REGEXP,
+	LATEX_MATH_AND_CODE_PATTERN,
+	MHCHEM_PATTERN_MAP
+} from '$lib/constants/latex-protection';
+
 /**
  * Replaces inline LaTeX expressions enclosed in `$...$` with placeholders, avoiding dollar signs
  * that appear to be part of monetary values or identifiers.
@@ -43,7 +49,8 @@ export function maskInlineLaTeX(content: string, latexExpressions: string[]): st
 
 				const charBeforeOpen = openDollarIndex > 0 ? line[openDollarIndex - 1] : '';
 				const charAfterOpen = line[openDollarIndex + 1];
-				const charBeforeClose = openDollarIndex + 1 < closeDollarIndex ? line[closeDollarIndex - 1] : '';
+				const charBeforeClose =
+					openDollarIndex + 1 < closeDollarIndex ? line[closeDollarIndex - 1] : '';
 				const charAfterClose = closeDollarIndex + 1 < line.length ? line[closeDollarIndex + 1] : '';
 
 				let shouldSkipAsNonLatex = false;
@@ -87,14 +94,8 @@ export function maskInlineLaTeX(content: string, latexExpressions: string[]): st
 }
 
 function escapeBrackets(text: string): string {
-	// Using the look‑behind pattern `(?<!\\)` we skip matches
-	// that are preceded by a backslash, e.g.
-	// `Definitions\\(also called macros)` (title of chapter 20 in The TeXbook)
-	// or `\\[4pt]`.
-	const pattern = /(```[\S\s]*?```|`.*?`)|(?<!\\)\\\[([\S\s]*?[^\\])\\]|(?<!\\)\\\((.*?)\\\)/g;
-
 	return text.replace(
-		pattern,
+		LATEX_MATH_AND_CODE_PATTERN,
 		(
 			match: string,
 			codeBlock: string | undefined,
@@ -116,14 +117,12 @@ function escapeBrackets(text: string): string {
 
 // Escape $\\ce{...} → $\\ce{...} but with proper handling
 function escapeMhchem(text: string): string {
-	return text.replaceAll('$\\ce{', '$\\\\ce{').replaceAll('$\\pu{', '$\\\\pu{');
+	return MHCHEM_PATTERN_MAP.reduce((result, [pattern, replacement]) => {
+		return result.replace(pattern, replacement);
+	}, text);
 }
 
-// See also:
-// https://github.com/danny-avila/LibreChat/blob/main/client/src/utils/latex.ts
-
-// Protect code blocks: ```...``` and `...`
-const codeBlockRegex = /(```[\s\S]*?```|`[^`\n]+`)/g;
+const doEscapeMhchem = false;
 
 /**
  * Preprocesses markdown content to safely handle LaTeX math expressions while protecting
@@ -145,10 +144,13 @@ const codeBlockRegex = /(```[\s\S]*?```|`[^`\n]+`)/g;
  * // → "Price: $10. The equation is $x^2$."
  */
 export function preprocessLaTeX(content: string): string {
+	// See also:
+	// https://github.com/danny-avila/LibreChat/blob/main/client/src/utils/latex.ts
+
 	// Step 1: Protect code blocks
 	const codeBlocks: string[] = [];
 
-	content = content.replace(codeBlockRegex, (match) => {
+	content = content.replace(CODE_BLOCK_REGEXP, (match) => {
 		codeBlocks.push(match);
 
 		return `<<CODE_BLOCK_${codeBlocks.length - 1}>>`;
@@ -207,7 +209,7 @@ export function preprocessLaTeX(content: string): string {
 	// Step 6: Apply additional escaping functions (brackets and mhchem)
 	content = escapeBrackets(content);
 
-	if (content.includes('\\ce{') || content.includes('\\pu{')) {
+	if (doEscapeMhchem && (content.includes('\\ce{') || content.includes('\\pu{'))) {
 		content = escapeMhchem(content);
 	}
 

tools/server/webui/src/stories/fixtures/math-formulas.ts

@@ -199,6 +199,16 @@ x = \frac{-b \pm \sqrt{\,b^{2}-4ac\,}}{2a}
 | | \(\displaystyle \prod_{k=1}^{n}k = n! \) | Factorial definition |
 | **Geometry** | \( \mathbf{a}\cdot \mathbf{b} = \|\mathbf{a}\|\,\|\mathbf{b}\|\,\cos\theta \) | Dot product & angle |
 
+## No math (but chemical)
+
+Balanced chemical reaction with states:
+
+\[
+\ce{2H2(g) + O2(g) -> 2H2O(l)}
+\]
+
+The standard enthalpy change for the reaction is: $\Delta H^\circ = \pu{-572 kJ mol^{-1}}$.
+
 ---
 
 *This document showcases various mathematical notation and formulas that can be rendered in markdown using LaTeX syntax.*