@moxn/kb-migrate 0.1.0 → 0.1.1

package/dist/sources/local.d.ts

@@ -16,6 +16,7 @@ export interface LocalSourceConfig extends SourceConfig {
  */
 export declare class LocalSource extends MigrationSource<LocalSourceConfig> {
     private files;
+    private skippedCollisions;
     get sourceType(): string;
     get sourceLocation(): string;
     validate(): Promise<void>;
@@ -25,6 +26,11 @@ export declare class LocalSource extends MigrationSource<LocalSourceConfig> {
     private extractDocument;
     private parseMarkdownSections;
     private nodesToContentBlocks;
+    /**
+     * Extract image nodes from paragraph children, returning image blocks
+     * and whether non-image text content exists.
+     */
+    private extractImagesFromParagraph;
     private imageToBlock;
     private guessImageType;
     private extensionToMediaType;

package/dist/sources/local.js

@@ -14,6 +14,7 @@ import { MigrationSource } from './base.js';
  */
 export class LocalSource extends MigrationSource {
     files = null;
+    skippedCollisions = new Set();
     get sourceType() {
         return 'local';
     }
@@ -30,13 +31,16 @@ export class LocalSource extends MigrationSource {
         if (!this.files) {
             await this.discoverFiles();
         }
-        return this.files.length;
+        return this.files.length - this.skippedCollisions.size;
     }
     async *extract() {
         if (!this.files) {
             await this.discoverFiles();
         }
         for (const file of this.files) {
+            if (this.skippedCollisions.has(file)) {
+                continue;
+            }
             const doc = await this.extractDocument(file);
             if (doc) {
                 yield doc;
@@ -55,7 +59,29 @@ export class LocalSource extends MigrationSource {
             allFiles.push(...matches);
         }
         // Deduplicate and sort
-        this.files = [...new Set(allFiles)].sort();
+        const uniqueFiles = [...new Set(allFiles)].sort();
+        // Detect KB path collisions (e.g., doc.md and doc.mdx both map to /doc)
+        const pathToFiles = new Map();
+        for (const file of uniqueFiles) {
+            const parsed = path.parse(file);
+            const dirParts = parsed.dir ? parsed.dir.split(path.sep) : [];
+            const kbPath = [...dirParts, parsed.name].join('/').replace(/ /g, '-');
+            const existing = pathToFiles.get(kbPath) || [];
+            existing.push(file);
+            pathToFiles.set(kbPath, existing);
+        }
+        this.skippedCollisions = new Set();
+        for (const [kbPath, files] of pathToFiles) {
+            if (files.length > 1) {
+                console.warn(`  ⚠ Path collision: multiple files map to KB path "/${kbPath}": ${files.join(', ')}`);
+                console.warn(`    → Keeping "${files[0]}", skipping ${files.length - 1} duplicate(s)`);
+                // Skip all but the first file (alphabetically first since files are sorted)
+                for (let i = 1; i < files.length; i++) {
+                    this.skippedCollisions.add(files[i]);
+                }
+            }
+        }
+        this.files = uniqueFiles;
     }
     async extractDocument(relativePath) {
         const fullPath = path.join(this.config.directory, relativePath);
@@ -63,7 +89,12 @@ export class LocalSource extends MigrationSource {
         // Parse relative path to create document path
         const parsed = path.parse(relativePath);
         const dirParts = parsed.dir ? parsed.dir.split(path.sep) : [];
-        const docPath = [...dirParts, parsed.name].join('/');
+        const rawPath = [...dirParts, parsed.name].join('/');
+        // Sanitize: replace spaces with hyphens (server rejects paths with spaces)
+        const docPath = rawPath.replace(/ /g, '-');
+        if (docPath !== rawPath) {
+            console.warn(`  ⚠ Path sanitized: "${rawPath}" → "${docPath}" (spaces replaced with hyphens)`);
+        }
         // Derive name from filename
         const name = parsed.name
             .split(/[-_]/)
@@ -158,12 +189,27 @@ export class LocalSource extends MigrationSource {
         };
         for (const node of nodes) {
             if (node.type === 'image') {
+                // Top-level image (rare in standard markdown, but possible)
                 flushTextBuffer();
                 const imageBlock = this.imageToBlock(node, baseDir);
                 if (imageBlock) {
                     blocks.push(imageBlock);
                 }
             }
+            else if (node.type === 'paragraph') {
+                // Walk paragraph children to extract images separately
+                const paragraph = node;
+                const { images, hasText } = this.extractImagesFromParagraph(paragraph.children, baseDir);
+                if (hasText) {
+                    // Render the paragraph text (images in text become markdown syntax via nodeToMarkdown)
+                    textBuffer += this.nodeToMarkdown(node) + '\n\n';
+                }
+                // Emit extracted image blocks
+                if (images.length > 0) {
+                    flushTextBuffer();
+                    blocks.push(...images);
+                }
+            }
             else {
                 textBuffer += this.nodeToMarkdown(node) + '\n\n';
             }
@@ -171,6 +217,30 @@ export class LocalSource extends MigrationSource {
         flushTextBuffer();
         return blocks;
     }
+    /**
+     * Extract image nodes from paragraph children, returning image blocks
+     * and whether non-image text content exists.
+     */
+    extractImagesFromParagraph(children, baseDir) {
+        const images = [];
+        let hasText = false;
+        for (const child of children) {
+            if (child.type === 'image') {
+                const imageBlock = this.imageToBlock(child, baseDir);
+                if (imageBlock) {
+                    images.push(imageBlock);
+                }
+            }
+            else {
+                // Check if the child has any meaningful text
+                const text = this.nodeToMarkdown(child).trim();
+                if (text) {
+                    hasText = true;
+                }
+            }
+        }
+        return { images, hasText };
+    }
     imageToBlock(node, baseDir) {
         const url = node.url;
         // Handle external URLs
@@ -272,8 +342,13 @@ export class LocalSource extends MigrationSource {
             case 'heading':
                 const heading = node;
                 const hashes = '#'.repeat(heading.depth);
-                const headingText = heading.children.map((c) => this.nodeToMarkdown(c)).join('');
+                const headingText = heading.children
+                    .map((c) => this.nodeToMarkdown(c))
+                    .join('');
                 return `${hashes} ${headingText}`;
+            case 'image':
+                const img = node;
+                return img.alt ? `![${img.alt}](${img.url})` : `![](${img.url})`;
             case 'thematicBreak':
                 return '---';
             case 'table':

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@moxn/kb-migrate",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "description": "Migration tool for importing documents into Moxn Knowledge Base from local files, Notion, Google Docs, and more",
   "type": "module",
   "main": "dist/index.js",