fix blank line processing with markdown

Author: YousefED

packages/core/src/api/exporters/markdown/markdownExporter.ts

@@ -16,7 +16,12 @@ import { addSpacesToCheckboxes } from "./util/addSpacesToCheckboxesRehypePlugin.
 
 // Needs to be sync because it's used in drag handler event (SideMenuPlugin)
 // Ideally, call `await initializeESMDependencies()` before calling this function
-export function cleanHTMLToMarkdown(cleanHTMLString: string) {
+export function cleanHTMLToMarkdown(
+  cleanHTMLString: string,
+  options?: {
+    keepEmptyParagraphs: boolean;
+  }
+) {
   const deps = esmDependencies;
 
   if (!deps) {
@@ -25,6 +30,14 @@ export function cleanHTMLToMarkdown(cleanHTMLString: string) {
     );
   }
 
+  if (options?.keepEmptyParagraphs) {
+    // replace empty paragraphs with [EMPTY-LINE]
+    // otherwise the unified pipeline will get rid of these
+    cleanHTMLString = cleanHTMLString.replace(
+      /<p><\/p>/g,
+      "<p>[EMPTY-LINE]</p>"
+    );
+  }
   const markdownString = deps.unified
     .unified()
     .use(deps.rehypeParse.default, { fragment: true })
@@ -37,7 +50,12 @@ export function cleanHTMLToMarkdown(cleanHTMLString: string) {
     })
     .processSync(cleanHTMLString);
 
-  return markdownString.value as string;
+  let ret = markdownString.value as string;
+  if (options?.keepEmptyParagraphs) {
+    // remove [EMPTY-LINE] hacks we added earlier
+    ret = ret.replace(/\n\[EMPTY-LINE\]\n/g, "\n\n");
+  }
+  return ret;
 }
 
 export async function blocksToMarkdown<
@@ -54,5 +72,9 @@ export async function blocksToMarkdown<
   const exporter = createExternalHTMLExporter(schema, editor);
   const externalHTML = exporter.exportBlocks(blocks, options);
 
-  return cleanHTMLToMarkdown(externalHTML);
+  const ret = cleanHTMLToMarkdown(externalHTML, {
+    keepEmptyParagraphs: true,
+  });
+
+  return ret;
 }