From 19e42c368184013fc30d1e46753b9e9383bb2bdf Mon Sep 17 00:00:00 2001
From: Bjorn Lu <bjornlu.dev@gmail.com>
Date: Wed, 20 Mar 2024 19:15:07 +0800
Subject: [PATCH] Pass meta to shiki transformers (#10494)

---
 .changeset/stale-jeans-yawn.md                |  5 +++
 .../markdoc/src/extensions/shiki.ts           |  3 ++
 packages/markdown/remark/src/highlight.ts     |  5 +--
 packages/markdown/remark/src/shiki.ts         | 13 +++++++-
 packages/markdown/remark/test/shiki.test.js   | 32 +++++++++++++++++++
 5 files changed, 55 insertions(+), 3 deletions(-)
 create mode 100644 .changeset/stale-jeans-yawn.md

diff --git a/.changeset/stale-jeans-yawn.md b/.changeset/stale-jeans-yawn.md
new file mode 100644
index 0000000000..1cf616cae8
--- /dev/null
+++ b/.changeset/stale-jeans-yawn.md
@@ -0,0 +1,5 @@
+---
+"@astrojs/markdown-remark": patch
+---
+
+Fixes support for Shiki transformers that access the `meta` to conditionally perform transformations
diff --git a/packages/integrations/markdoc/src/extensions/shiki.ts b/packages/integrations/markdoc/src/extensions/shiki.ts
index 3026d80806..a39eb69a9f 100644
--- a/packages/integrations/markdoc/src/extensions/shiki.ts
+++ b/packages/integrations/markdoc/src/extensions/shiki.ts
@@ -12,6 +12,9 @@ export default async function shiki(config?: ShikiConfig): Promise<AstroMarkdocC
 			fence: {
 				attributes: Markdoc.nodes.fence.attributes!,
 				transform({ attributes }) {
+					// NOTE: The `meta` from fence code, e.g. ```js {1,3-4}, isn't quite supported by Markdoc.
+					// Only the `js` part is parsed as `attributes.language` and the rest is ignored. This means
+					// some Shiki transformers may not work correctly as it relies on the `meta`.
 					const lang = typeof attributes.language === 'string' ? attributes.language : 'plaintext';
 					const html = highlighter.highlight(attributes.content, lang);
 
diff --git a/packages/markdown/remark/src/highlight.ts b/packages/markdown/remark/src/highlight.ts
index eaf4c9bdf9..31f11119fa 100644
--- a/packages/markdown/remark/src/highlight.ts
+++ b/packages/markdown/remark/src/highlight.ts
@@ -4,7 +4,7 @@ import { toText } from 'hast-util-to-text';
 import { removePosition } from 'unist-util-remove-position';
 import { visitParents } from 'unist-util-visit-parents';
 
-type Highlighter = (code: string, language: string) => string;
+type Highlighter = (code: string, language: string, options?: { meta?: string }) => string;
 
 const languagePattern = /\blanguage-(\S+)\b/;
 
@@ -55,8 +55,9 @@ export function highlightCodeBlocks(tree: Root, highlighter: Highlighter) {
 			return;
 		}
 
+		const meta = (node.data as any)?.meta ?? node.properties.metastring ?? undefined;
 		const code = toText(node, { whitespace: 'pre' });
-		const html = highlighter(code, languageMatch?.[1] || 'plaintext');
+		const html = highlighter(code, languageMatch?.[1] || 'plaintext', { meta });
 		// The replacement returns a root node with 1 child, the `<pr>` element replacement.
 		const replacement = fromHtml(html, { fragment: true }).children[0] as Element;
 		// We just generated this node, so any positional information is invalid.
diff --git a/packages/markdown/remark/src/shiki.ts b/packages/markdown/remark/src/shiki.ts
index bd9280f8c5..ff1589dac1 100644
--- a/packages/markdown/remark/src/shiki.ts
+++ b/packages/markdown/remark/src/shiki.ts
@@ -7,7 +7,14 @@ export interface ShikiHighlighter {
 	highlight(
 		code: string,
 		lang?: string,
-		options?: { inline?: boolean; attributes?: Record<string, string> }
+		options?: {
+			inline?: boolean;
+			attributes?: Record<string, string>;
+			/**
+			 * Raw `meta` information to be used by Shiki transformers
+			 */
+			meta?: string;
+		}
 	): string;
 }
 
@@ -56,6 +63,10 @@ export async function createShikiHighlighter({
 			return highlighter.codeToHtml(code, {
 				...themeOptions,
 				lang,
+				// NOTE: while we can spread `options.attributes` here so that Shiki can auto-serialize this as rendered
+				// attributes on the top-level tag, it's not clear whether it is fine to pass all attributes as meta, as
+				// they're technically not meta, nor parsed from Shiki's `parseMetaString` API.
+				meta: options?.meta ? { __raw: options?.meta } : undefined,
 				transformers: [
 					{
 						pre(node) {
diff --git a/packages/markdown/remark/test/shiki.test.js b/packages/markdown/remark/test/shiki.test.js
index c6044b019d..601b7fabf6 100644
--- a/packages/markdown/remark/test/shiki.test.js
+++ b/packages/markdown/remark/test/shiki.test.js
@@ -53,4 +53,36 @@ describe('shiki syntax highlighting', () => {
 		assert.match(html, />-<\/span>/);
 		assert.match(html, />+<\/span>/);
 	});
+
+	it('renders attributes', async () => {
+		const highlighter = await createShikiHighlighter();
+
+		const html = highlighter.highlight(`foo`, 'js', {
+			attributes: { 'data-foo': 'bar', autofocus: true },
+		});
+
+		assert.match(html, /data-foo="bar"/);
+		assert.match(html, /autofocus(?!=)/);
+	});
+
+	it('supports transformers that reads meta', async () => {
+		const highlighter = await createShikiHighlighter({
+			transformers: [
+				{
+					pre(node) {
+						const meta = this.options.meta?.__raw;
+						if (meta) {
+							node.properties['data-test'] = meta;
+						}
+					},
+				},
+			],
+		});
+
+		const html = highlighter.highlight(`foo`, 'js', {
+			meta: '{1,3-4}',
+		});
+
+		assert.match(html, /data-test="\{1,3-4\}"/);
+	});
 });