@moxn/kb-migrate 0.4.3 → 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;
@@ -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/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@moxn/kb-migrate",
-  "version": "0.4.3",
+  "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",