Fix syntax highlighting multiline Swift string literals with backslashes (#723)

Fixes an issue where syntax highlighting looks broken for Swift code listings
with multi-line string literals that have backslash newline delimiters.

Resolves: rdar://102993199

Author: mportiz08

src/utils/custom-highlight-lang/swift.js

@@ -41,5 +41,40 @@ export default function swiftOverride(hljs) {
     };
   }
 
+  // Checks if a given language sub-mode matches the "ESCAPED_NEWLINE" from the
+  // built-in Swift parser from hljs
+  const isEscapedNewlineMode = (mode) => {
+    const { className, match } = mode;
+    if (className !== 'subst' || !match) {
+      return false;
+    }
+
+    const matchStr = match.toString();
+    return matchStr.startsWith('\\') && matchStr.endsWith('[\\t ]*(?:[\\r\\n]|\\r\\n)');
+  };
+
+  // replace the "ESCAPED_NEWLINE" sub-mode in the multiline string literal mode
+  // variants so that it doesn't include the actual newline characters in the
+  // span token that it generates, because this causes issues with our
+  // line-number + multi-line string literal logic when the span for the
+  // backslash token is split across multiple lines
+  const strIndex = language.contains.findIndex(({ className }) => className === 'string');
+  language.contains[strIndex] = {
+    ...language.contains[strIndex],
+    variants: language.contains[strIndex].variants.map(variant => ({
+      ...variant,
+      contains: variant.contains.map(mode => (isEscapedNewlineMode(mode) ? ({
+        className: 'subst',
+        begin: /\\#{0,3}/,
+        end: /[\t ]*(?:[\r\n]|\r\n)/,
+        // same match as the original one but with an explicit start/end match so
+        // that the end one can be excluded from the resulting span
+        excludeEnd: true,
+      }) : (
+        mode
+      ))),
+    })),
+  };
+
   return language;
 }

tests/unit/utils/syntax-highlight.spec.js

@@ -136,6 +136,29 @@ describe("syntax-highlight", () => {
     `);
   });
 
+  it('keeps escaped newline tokens on the same line for multiline string literals', async () => {
+    const content = [
+      'let multiline = """',
+      'a \\',
+      'b',
+      '',
+      'c \\',
+      'd',
+      '"""',
+    ];
+    const { highlightedCode, sanitizedCode } = await prepare(content, 'swift');
+    expect(sanitizedCode).not.toEqual(highlightedCode);
+    expect(sanitizedCode).toMatchInlineSnapshot(`
+      <span class="syntax-keyword">let</span> multiline <span class="syntax-operator">=</span> <span class="syntax-string">"""</span>
+      <span class="syntax-string">a <span class="syntax-subst">\\</span></span>
+      <span class="syntax-string">b</span>
+      <span class="syntax-string"></span>
+      <span class="syntax-string">c <span class="syntax-subst">\\</span></span>
+      <span class="syntax-string">d</span>
+      <span class="syntax-string">"""</span>
+    `);
+  });
+
   it("wraps multiline nested html elements", () => {
     const code = document.createElement("CODE");
     code.innerHTML = `<span class="syntax-function">function <span class="syntax-title function_">someName</span>(<span class="syntax-params">foo,