@moxn/kb-migrate 0.4.2 → 0.4.6

package/dist/sources/notion-blocks.js

@@ -23,7 +23,7 @@ export async function blocksToSections(blocks, client, pagePathMap, options) {
         if (block.type === 'heading_2') {
             // Flush current section if it has content
             if (currentBlocks.length > 0) {
-                sections.push({ name: currentSectionName, content: currentBlocks });
+                sections.push({ name: currentSectionName, content: mergeConsecutiveListBlocks(currentBlocks) });
             }
             const h2 = block;
             currentSectionName = richTextToPlain(h2.heading_2.rich_text) || 'Untitled';
@@ -36,7 +36,7 @@ export async function blocksToSections(blocks, client, pagePathMap, options) {
     }
     // Flush last section
     if (currentBlocks.length > 0) {
-        sections.push({ name: currentSectionName, content: currentBlocks });
+        sections.push({ name: currentSectionName, content: mergeConsecutiveListBlocks(currentBlocks) });
     }
     // If no sections were created at all (empty page), return empty array
     return sections;
@@ -74,10 +74,10 @@ async function convertBlock(block, client, pagePathMap, visitedSyncedBlocks) {
             results.push(...convertToDo(block));
             break;
         case 'quote':
-            results.push(...convertQuote(block));
+            results.push(...(await convertQuote(block, client, pagePathMap, visitedSyncedBlocks)));
             break;
         case 'callout':
-            results.push(...convertCallout(block));
+            results.push(...(await convertCallout(block, client, pagePathMap, visitedSyncedBlocks)));
             break;
         case 'divider':
             results.push(textBlock('---'));
@@ -140,25 +140,15 @@ async function convertBlock(block, client, pagePathMap, visitedSyncedBlocks) {
             console.warn(`  Skipping unsupported Notion block type: ${block.type}`);
             break;
     }
-    // If block has children (except table, toggle, synced_block, column_list which handle their own)
+    // If block has children (except types that handle their own)
     if (block.has_children &&
-        !['table', 'toggle', 'synced_block', 'column_list', 'column'].includes(block.type)) {
+        !['table', 'toggle', 'synced_block', 'column_list', 'column', 'quote', 'callout'].includes(block.type)) {
+        const indent = ['bulleted_list_item', 'numbered_list_item', 'to_do'].includes(block.type)
+            ? '  '
+            : undefined;
         const children = await client.getBlockChildren(block.id);
-        for (const child of children) {
-            const childBlocks = await convertBlock(child, client, pagePathMap, visitedSyncedBlocks);
-            // Indent child content for list items
-            if (['bulleted_list_item', 'numbered_list_item'].includes(block.type)) {
-                for (const cb of childBlocks) {
-                    if (cb.blockType === 'text' && cb.text) {
-                        cb.text = cb.text
-                            .split('\n')
-                            .map((line) => '  ' + line)
-                            .join('\n');
-                    }
-                }
-            }
-            results.push(...childBlocks);
-        }
+        const childBlocks = await convertAndMergeChildren(children, client, pagePathMap, visitedSyncedBlocks, indent);
+        results.push(...childBlocks);
     }
     return results;
 }
@@ -203,18 +193,26 @@ function convertToDo(block) {
     const text = richTextToMarkdown(td.to_do.rich_text);
     return [textBlock(`${checkbox} ${text}`)];
 }
-function convertQuote(block) {
+async function convertQuote(block, client, pagePathMap, visitedSyncedBlocks) {
     const q = block;
     const text = richTextToMarkdown(q.quote.rich_text);
-    if (!text)
+    if (!text && !block.has_children)
         return [];
     const quoted = text
-        .split('\n')
-        .map((line) => '> ' + line)
-        .join('\n');
-    return [textBlock(quoted)];
+        ? text
+            .split('\n')
+            .map((line) => '> ' + line)
+            .join('\n')
+        : '';
+    const results = quoted ? [textBlock(quoted)] : [];
+    if (block.has_children) {
+        const children = await client.getBlockChildren(block.id);
+        const childBlocks = await convertAndMergeChildren(children, client, pagePathMap, visitedSyncedBlocks, '> ');
+        results.push(...childBlocks);
+    }
+    return results;
 }
-function convertCallout(block) {
+async function convertCallout(block, client, pagePathMap, visitedSyncedBlocks) {
     const c = block;
     const text = richTextToMarkdown(c.callout.rich_text);
     const emoji = c.callout.icon?.emoji ?? '';
@@ -223,7 +221,13 @@ function convertCallout(block) {
         .split('\n')
         .map((line) => '> ' + line)
         .join('\n');
-    return [textBlock(quoted)];
+    const results = [textBlock(quoted)];
+    if (block.has_children) {
+        const children = await client.getBlockChildren(block.id);
+        const childBlocks = await convertAndMergeChildren(children, client, pagePathMap, visitedSyncedBlocks, '> ');
+        results.push(...childBlocks);
+    }
+    return results;
 }
 async function convertTable(block, client) {
     const tableBlock = block;
@@ -255,19 +259,8 @@ async function convertToggle(block, client, pagePathMap, visitedSyncedBlocks) {
     const results = [textBlock(`**${header}**`)];
     if (block.has_children) {
         const children = await client.getBlockChildren(block.id);
-        for (const child of children) {
-            const converted = await convertBlock(child, client, pagePathMap, visitedSyncedBlocks);
-            // Indent toggle content
-            for (const cb of converted) {
-                if (cb.blockType === 'text' && cb.text) {
-                    cb.text = cb.text
-                        .split('\n')
-                        .map((line) => '> ' + line)
-                        .join('\n');
-                }
-            }
-            results.push(...converted);
-        }
+        const childBlocks = await convertAndMergeChildren(children, client, pagePathMap, visitedSyncedBlocks, '> ');
+        results.push(...childBlocks);
     }
     return results;
 }
@@ -372,12 +365,7 @@ async function convertSyncedBlock(block, client, pagePathMap, visitedSyncedBlock
     visitedSyncedBlocks.add(sourceId);
     try {
         const children = await client.getBlockChildren(sourceId);
-        const results = [];
-        for (const child of children) {
-            const converted = await convertBlock(child, client, pagePathMap, visitedSyncedBlocks);
-            results.push(...converted);
-        }
-        return results;
+        return await convertAndMergeChildren(children, client, pagePathMap, visitedSyncedBlocks);
     }
     finally {
         visitedSyncedBlocks.delete(sourceId);
@@ -389,10 +377,8 @@ async function convertColumnList(block, client, pagePathMap, visitedSyncedBlocks
     for (const column of children) {
         if (column.type === 'column' && column.has_children) {
             const columnChildren = await client.getBlockChildren(column.id);
-            for (const child of columnChildren) {
-                const converted = await convertBlock(child, client, pagePathMap, visitedSyncedBlocks);
-                results.push(...converted);
-            }
+            const converted = await convertAndMergeChildren(columnChildren, client, pagePathMap, visitedSyncedBlocks);
+            results.push(...converted);
         }
     }
     return results;
@@ -429,12 +415,9 @@ export function richTextToMarkdown(richText) {
             text = '*' + text + '*';
         if (rt.annotations.strikethrough)
             text = '~~' + text + '~~';
-        // Apply link
-        if (rt.href) {
-            text = `[${text}](${rt.href})`;
-        }
-        else if (rt.type === 'mention' && rt.mention) {
-            // Handle mention types
+        // Handle mentions BEFORE href — Notion sets href on mentions too
+        // (as a web URL), but we want notion:// protocol for resolution
+        if (rt.type === 'mention' && rt.mention) {
             if (rt.mention.type === 'user' && rt.mention.user?.name) {
                 text = `@${rt.mention.user.name}`;
             }
@@ -450,6 +433,9 @@ export function richTextToMarkdown(richText) {
                     text += ` → ${rt.mention.date.end}`;
             }
         }
+        else if (rt.href) {
+            text = `[${text}](${rt.href})`;
+        }
         else if (rt.type === 'equation' && rt.equation) {
             text = `$${rt.equation.expression}$`;
         }
@@ -469,6 +455,96 @@ export function richTextToPlain(richText) {
 function textBlock(text) {
     return { blockType: 'text', text };
 }
+function detectListType(text) {
+    const firstLine = text.split('\n')[0];
+    if (/^\d+\.\s/.test(firstLine))
+        return 'ordered';
+    if (/^- \[[ x]\]\s/.test(firstLine))
+        return 'todo';
+    if (/^- /.test(firstLine))
+        return 'bullet';
+    return null;
+}
+/**
+ * Merge consecutive list-item text blocks into single text blocks.
+ * This ensures the markdown→TipTap parser sees them as one list
+ * instead of creating multiple single-item lists.
+ */
+function mergeConsecutiveListBlocks(blocks) {
+    const result = [];
+    let accumulator = [];
+    let currentListType = null;
+    function flush() {
+        if (accumulator.length === 0)
+            return;
+        let merged = accumulator.join('\n');
+        if (currentListType === 'ordered') {
+            // Fix numbering: replace all leading `1.` with sequential numbers
+            let counter = 0;
+            merged = merged
+                .split('\n')
+                .map((line) => {
+                if (/^\d+\.\s/.test(line)) {
+                    counter++;
+                    return line.replace(/^\d+\./, `${counter}.`);
+                }
+                return line;
+            })
+                .join('\n');
+        }
+        result.push(textBlock(merged));
+        accumulator = [];
+        currentListType = null;
+    }
+    for (const block of blocks) {
+        if (block.blockType !== 'text' || !block.text) {
+            flush();
+            result.push(block);
+            continue;
+        }
+        const type = detectListType(block.text);
+        // Indented child content (starts with spaces) continues current list group
+        if (currentListType &&
+            block.text.split('\n').every((line) => line.startsWith('  '))) {
+            accumulator.push(block.text);
+            continue;
+        }
+        if (type === null) {
+            flush();
+            result.push(block);
+            continue;
+        }
+        if (type !== currentListType) {
+            flush();
+            currentListType = type;
+        }
+        accumulator.push(block.text);
+    }
+    flush();
+    return result;
+}
+/**
+ * Convert children blocks, merge consecutive list items, and optionally indent.
+ */
+async function convertAndMergeChildren(children, client, pagePathMap, visitedSyncedBlocks, indent) {
+    const blocks = [];
+    for (const child of children) {
+        const converted = await convertBlock(child, client, pagePathMap, visitedSyncedBlocks);
+        blocks.push(...converted);
+    }
+    const merged = mergeConsecutiveListBlocks(blocks);
+    if (indent) {
+        for (const block of merged) {
+            if (block.blockType === 'text' && block.text) {
+                block.text = block.text
+                    .split('\n')
+                    .map((line) => indent + line)
+                    .join('\n');
+            }
+        }
+    }
+    return merged;
+}
 function guessImageMediaType(url) {
     const lower = url.toLowerCase();
     if (lower.includes('.jpg') || lower.includes('.jpeg'))

package/dist/sources/notion-blocks.test.js

@@ -45,8 +45,8 @@ describe('richTextToMarkdown', () => {
             const result = richTextToMarkdown([rt]);
             expect(result).toBe('[My Database](notion://abc123de-f456-abc1-23de-f456abc123de)');
         });
-        it('renders database mention with href using the href (href takes precedence)', () => {
-            // When href is set on the rich text, line 641 handles it BEFORE the mention check
+        it('renders database mention with href using notion:// (mention takes precedence over href)', () => {
+            // Mentions should always use notion:// protocol, even when href is set
             const rt = {
                 type: 'mention',
                 plain_text: 'My Database',
@@ -65,8 +65,7 @@ describe('richTextToMarkdown', () => {
                 },
             };
             const result = richTextToMarkdown([rt]);
-            // href takes precedence — produces a notion.so link (resolved in post-processing)
-            expect(result).toBe('[My Database](https://www.notion.so/workspace/abc123de-f456-abc1-23de-f456abc123de)');
+            expect(result).toBe('[My Database](notion://abc123de-f456-abc1-23de-f456abc123de)');
         });
     });
     describe('page mentions', () => {
@@ -91,7 +90,7 @@ describe('richTextToMarkdown', () => {
             const result = richTextToMarkdown([rt]);
             expect(result).toBe('[My Page](notion://abc123de-f456-abc1-23de-f456abc123de)');
         });
-        it('renders page mention with href using the href', () => {
+        it('renders page mention with href using notion:// (mention takes precedence over href)', () => {
             const rt = {
                 type: 'mention',
                 plain_text: 'My Page',
@@ -110,7 +109,7 @@ describe('richTextToMarkdown', () => {
                 },
             };
             const result = richTextToMarkdown([rt]);
-            expect(result).toBe('[My Page](https://www.notion.so/workspace/abc123de-f456-abc1-23de-f456abc123de)');
+            expect(result).toBe('[My Page](notion://abc123de-f456-abc1-23de-f456abc123de)');
         });
     });
     describe('other mention types preserved', () => {

package/dist/sources/notion-references.js

@@ -18,6 +18,9 @@ const NOTION_PROTOCOL_RE = /\[([^\]]*)\]\(notion:\/\/([a-f0-9-]{32,36})\)/g;
 const NOTION_WEB_URL_RE = /\[([^\]]*)\]\(https?:\/\/(?:www\.)?notion\.so\/(?:[a-zA-Z0-9_-]+\/)*([a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12})(?:[?#][^)]*)?(?:\))/g;
 // Fallback for IDs without dashes embedded in slug URLs (e.g., ...Page-Title-abc123def456...)
 const NOTION_WEB_URL_SLUG_RE = /\[([^\]]*)\]\(https?:\/\/(?:www\.)?notion\.so\/(?:[a-zA-Z0-9_-]+\/)*[a-zA-Z0-9-]+-([a-f0-9]{32})(?:[?#][^)]*)?\)/g;
+// Bare 32-char hex on notion.so (e.g., https://www.notion.so/3004bf42cfeb818ebe8fe2f4befdad52)
+// Catches mention hrefs that leak through as web URLs
+const NOTION_WEB_URL_BARE_RE = /\[([^\]]*)\]\(https?:\/\/(?:www\.)?notion\.so\/(?:[a-zA-Z0-9_-]+\/)*([a-f0-9]{32})(?:[?#][^)]*)?\)/g;
 // 3. Unresolved placeholders from link_to_page conversion
 const LINK_PLACEHOLDER_RE = /\*\(Link to Notion page: ([a-f0-9-]{32,36})\)\*/g;
 // 4. Relation property markers
@@ -87,6 +90,21 @@ export function resolveNotionReferences(sections, mapping) {
                 }
                 return `[${displayText}](notion://${nid})`;
             });
+            // Pass 3b: Bare 32-char hex Notion URLs (safety net for mention hrefs)
+            text = text.replace(NOTION_WEB_URL_BARE_RE, (_match, displayText, rawId) => {
+                const nid = normalizeId(rawId);
+                const kbPath = mapping.notionIdToKbPath.get(nid);
+                if (kbPath) {
+                    references.push({
+                        sectionIndex,
+                        targetNotionId: nid,
+                        targetKbPath: kbPath,
+                        displayText: displayText || kbPath,
+                    });
+                    return `[${displayText || kbPath}](${kbPath})`;
+                }
+                return `[${displayText}](notion://${nid})`;
+            });
             // Pass 4: Unresolved link_to_page placeholders
             text = text.replace(LINK_PLACEHOLDER_RE, (_match, rawId) => {
                 const nid = normalizeId(rawId);

package/dist/sources/notion-references.test.js

@@ -120,6 +120,24 @@ describe('resolveNotionReferences', () => {
             const { sections: resolved } = resolveNotionReferences(sections, m);
             expect(getText(resolved[0])).toBe('See [link](/docs/my-page)');
         });
+        it('resolves bare 32-char hex Notion URLs (no slug prefix)', () => {
+            const sections = [
+                textSection('Intro', 'See [Dev Setup](https://www.notion.so/3004bf42cfeb818ebe8fe2f4befdad52)'),
+            ];
+            const m = mapping({ '3004bf42cfeb818ebe8fe2f4befdad52': '/docs/dev-setup' });
+            const { sections: resolved, references } = resolveNotionReferences(sections, m);
+            expect(getText(resolved[0])).toBe('See [Dev Setup](/docs/dev-setup)');
+            expect(references).toHaveLength(1);
+            expect(references[0].targetNotionId).toBe('3004bf42cfeb818ebe8fe2f4befdad52');
+        });
+        it('normalizes unresolved bare hex Notion URLs to notion://', () => {
+            const sections = [
+                textSection('Intro', 'See [page](https://www.notion.so/aabbccdd11223344aabbccdd11223344)'),
+            ];
+            const m = mapping({});
+            const { sections: resolved } = resolveNotionReferences(sections, m);
+            expect(getText(resolved[0])).toBe('See [page](notion://aabbccdd11223344aabbccdd11223344)');
+        });
     });
     describe('link_to_page placeholders', () => {
         it('resolves placeholders to KB path links', () => {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@moxn/kb-migrate",
-  "version": "0.4.2",
+  "version": "0.4.6",
   "description": "Migration tool for importing documents into Moxn Knowledge Base from local files, Notion, Google Docs, and more",
   "type": "module",
   "main": "dist/index.js",