fix: support hard line breaks to start a new paragraph (#73)

* refactor: remove flattening of paragraph parsing logic

* feat: enhance paragraph parsing to support hard line breaks

This update modifies the paragraph parsing logic to correctly split paragraphs on hard line breaks, improving the handling of markdown input. Additionally, new tests have been added to ensure the functionality works as expected.

Author: akoso

src/parser/internal.ts

@@ -118,23 +118,34 @@ function parseParagraph(
   // Notion doesn't deal with inline images, so we need to parse them all out
   // of the paragraph into individual blocks
   const images: notion.Block[] = [];
-  const paragraphs: Array<notion.RichText[]> = [];
+  const paragraphs: notion.RichText[][] = [];
+  let currentParagraph: notion.RichText[] = [];
+
+  const pushParagraph = () => {
+    if (currentParagraph.length > 0) {
+      paragraphs.push(currentParagraph);
+      currentParagraph = [];
+    }
+  };
+
   element.children.forEach(item => {
     if (item.type === 'image') {
       images.push(parseImage(item, options));
-    } else {
-      const richText = parseInline(item) as notion.RichText[];
-      if (richText.length) {
-        paragraphs.push(richText);
-      }
+      return;
     }
+
+    if (item.type === 'break') {
+      pushParagraph();
+      return;
+    }
+
+    const richText = parseInline(item) as notion.RichText[];
+    currentParagraph.push(...richText);
   });
 
-  if (paragraphs.length) {
-    return [notion.paragraph(paragraphs.flat()), ...images];
-  } else {
-    return images;
-  }
+  pushParagraph();
+
+  return [...paragraphs.map(notion.paragraph), ...images];
 }
 
 function parseBlockquote(

test/integration.spec.ts

@@ -218,6 +218,30 @@ const hello = "hello";
 
       expect(actual).toStrictEqual(expected);
     });
+
+    it('should split paragraphs on hard line breaks', () => {
+      const text =
+        'You can _italicize_ or **bold** text.  \nThis is the second line of text.\nAnd this is in the same line';
+
+      const actual = markdownToBlocks(text);
+
+      const expected = [
+        notion.paragraph([
+          notion.richText('You can '),
+          notion.richText('italicize', {annotations: {italic: true}}),
+          notion.richText(' or '),
+          notion.richText('bold', {annotations: {bold: true}}),
+          notion.richText(' text.'),
+        ]),
+        notion.paragraph([
+          notion.richText(
+            'This is the second line of text.\nAnd this is in the same line',
+          ),
+        ]),
+      ];
+
+      expect(actual).toStrictEqual(expected);
+    });
   });
 
   describe('markdownToRichText', () => {

test/parser.spec.ts

@@ -271,6 +271,39 @@ describe('gfm parser', () => {
     expect(actual).toStrictEqual(expected);
   });
 
+  it('should split paragraphs on hard line breaks', () => {
+    // Simulate markdown where a line ends with two spaces followed by \n, which renders as a hard line break.
+    // In the mdast AST this is represented as a `break` node within the paragraph.
+    const br: md.Break = {type: 'break'} as md.Break;
+
+    const ast = md.root(
+      md.paragraph(
+        md.text('You can '),
+        md.emphasis(md.text('italicize')),
+        md.text(' or '),
+        md.strong(md.text('bold')),
+        md.text(' text.'),
+        br,
+        md.text('This is the second line of text'),
+      ),
+    );
+
+    const actual = parseBlocks(ast, options);
+
+    const expected = [
+      notion.paragraph([
+        notion.richText('You can '),
+        notion.richText('italicize', {annotations: {italic: true}}),
+        notion.richText(' or '),
+        notion.richText('bold', {annotations: {bold: true}}),
+        notion.richText(' text.'),
+      ]),
+      notion.paragraph([notion.richText('This is the second line of text')]),
+    ];
+
+    expect(actual).toStrictEqual(expected);
+  });
+
   it('should parse github extensions', () => {
     const ast = md.root(
       md.paragraph(