fix(cli): address PR review feedback on LaTeX renderer

- Remove convertLineBreaks from applyProseConversions so `\\` outside
  math delimiters is no longer rewritten to a newline. This protects
  Windows UNC paths (`\\server\share`) and escaped backslashes in
  code-like prose. LaTeX line breaks inside `$...$` / `$$...$$` are
  still handled by applyMathModeConversions.
- Add `?? match` fallback to the MASK_PATTERN restore step so a
  stray PUA sentinel in user input can never cause "undefined" to
  leak into rendered output.
- Add tests for Windows UNC paths and math-mode-only line breaks.

Addresses review comments from @gemini-code-assist and @jacob314
on PR #25802.

Author: dimssu

packages/cli/src/ui/utils/latexToUnicode.test.ts

@@ -148,8 +148,15 @@ describe('convertLatexToUnicode', () => {
       expect(convertLatexToUnicode('word\\ boundary')).toBe('word boundary');
     });
 
-    it('converts \\\\ to a newline', () => {
-      expect(convertLatexToUnicode('line1\\\\line2')).toBe('line1\nline2');
+    it('converts \\\\ to a newline inside math mode', () => {
+      // `\\` is a LaTeX line break in math/tabular contexts. Only convert
+      // inside `$...$` — outside math this would mangle Windows UNC paths
+      // (`\\server\share`) and escaped backslashes in code-like prose.
+      expect(convertLatexToUnicode('$a\\\\b$')).toBe('a\nb');
+    });
+
+    it('leaves \\\\ alone outside math mode', () => {
+      expect(convertLatexToUnicode('line1\\\\line2')).toBe('line1\\\\line2');
     });
   });
 
@@ -249,6 +256,14 @@ describe('convertLatexToUnicode', () => {
       );
     });
 
+    it('leaves Windows UNC paths alone (no line-break rewrite in prose)', () => {
+      // `\\server\share\file` must NOT be rewritten to a newline. Line-break
+      // conversion is restricted to math mode. See PR #25802.
+      expect(convertLatexToUnicode('\\\\server\\share\\file')).toBe(
+        '\\\\server\\share\\file',
+      );
+    });
+
     it('leaves regex backslash escapes alone', () => {
       expect(convertLatexToUnicode('\\d+\\w*')).toBe('\\d+\\w*');
     });

packages/cli/src/ui/utils/latexToUnicode.ts

@@ -559,7 +559,11 @@ function applyProseConversions(text: string): string {
   out = convertTextFormatting(out);
   out = convertFractionsAndRoots(out);
   out = convertEscapedSpecials(out);
-  out = convertLineBreaks(out);
+  // Deliberately NOT running convertLineBreaks here: outside math delimiters
+  // `\\` is far more likely to be a Windows UNC path (`\\server\share`) or an
+  // escaped backslash in code-like prose than a LaTeX line break. Legitimate
+  // LaTeX line breaks belong inside `$...$` or `$$...$$` and are handled by
+  // applyMathModeConversions. See PR #25802 review.
   out = convertNamedCommands(out);
   out = convertPunctuationCommands(out);
   return out;

packages/cli/src/ui/utils/markdownParsingUtils.ts

@@ -96,7 +96,10 @@ const convertLatexPreservingSpans = (text: string): string => {
   const converted = convertLatexToUnicode(masked);
   return converted.replace(
     MASK_PATTERN,
-    (_, i: string) => preserved[Number(i)],
+    // Fallback to the literal match if the index is somehow out of range —
+    // defensive against the unlikely case where the PUA sentinel appears in
+    // user input. Without the fallback, replace would emit "undefined".
+    (match, i: string) => preserved[Number(i)] ?? match,
   );
 };