@moxn/kb-migrate 0.4.34 → 0.4.36

package/dist/sources/local.d.ts

@@ -29,6 +29,22 @@ export interface LocalSourceConfig extends SourceConfig {
     /** Date filter for source files */
     dateFilter?: DateFilter;
 }
+/**
+ * A local file to be imported as GRAMMAR markdown (Phase 9a-3).
+ *
+ * The grammar path reads the file VERBATIM (it is already moxn-grammar
+ * markdown — front-matter + body + `:::…` embeds, e.g. produced by the
+ * export) and hands it to `import_markdown`; it does NOT parse to MCP blocks
+ * like the legacy {@link LocalSource.extract} path.
+ */
+export interface GrammarFile {
+    /** Absolute path to the source file on disk (for reading + media base dir). */
+    fullPath: string;
+    /** Original relative path from the scan root (for logging). */
+    sourcePath: string;
+    /** Sanitized KB path the doc maps to (e.g. "subdir/My-Doc"). */
+    kbPath: string;
+}
 /**
  * Local filesystem migration source
  */
@@ -40,6 +56,14 @@ export declare class LocalSource extends MigrationSource<LocalSourceConfig> {
     validate(): Promise<void>;
     getDocumentCount(): Promise<number | undefined>;
     extract(): AsyncGenerator<ExtractedDocument, void, unknown>;
+    /**
+     * Yield local files as GRAMMAR markdown for the `import_markdown` path
+     * (Phase 9a-3). Reuses the SAME discovery + collision-skip + date-filter
+     * logic as {@link extract}, but emits the file's KB path + disk location
+     * instead of parsing to MCP blocks. The runner reads the file verbatim,
+     * uploads + rewrites its media refs, then calls `import_markdown`.
+     */
+    extractGrammarFiles(): AsyncGenerator<GrammarFile, void, unknown>;
     private discoverFiles;
     private extractDocument;
     private parseMarkdownSections;

package/dist/sources/local.js

@@ -75,6 +75,28 @@ export class LocalSource extends MigrationSource {
             }
         }
     }
+    /**
+     * Yield local files as GRAMMAR markdown for the `import_markdown` path
+     * (Phase 9a-3). Reuses the SAME discovery + collision-skip + date-filter
+     * logic as {@link extract}, but emits the file's KB path + disk location
+     * instead of parsing to MCP blocks. The runner reads the file verbatim,
+     * uploads + rewrites its media refs, then calls `import_markdown`.
+     */
+    async *extractGrammarFiles() {
+        if (!this.files) {
+            await this.discoverFiles();
+        }
+        for (const file of this.files) {
+            if (this.skippedCollisions.has(file)) {
+                continue;
+            }
+            yield {
+                fullPath: path.join(this.config.directory, file),
+                sourcePath: file,
+                kbPath: filePathToKbPath(file),
+            };
+        }
+    }
     async discoverFiles() {
         const patterns = this.config.extensions.map((ext) => `**/*${ext.startsWith('.') ? ext : '.' + ext}`);
         const allFiles = [];

package/dist/sources/local.test.js

@@ -20,7 +20,7 @@ describe('sanitizePathSegment', () => {
         expect(sanitizePathSegment('foo@bar!')).toBe('foo-bar');
     });
     it('handles parentheses, ampersands, apostrophes', () => {
-        expect(sanitizePathSegment("Q3 Plan (final)")).toBe('Q3-Plan-final');
+        expect(sanitizePathSegment('Q3 Plan (final)')).toBe('Q3-Plan-final');
         expect(sanitizePathSegment("Tom's & Jerry's")).toBe('Tom-s-Jerry-s');
     });
     it('strips accents via NFKD normalization', () => {

package/dist/sources/notion.d.ts

@@ -127,6 +127,12 @@ export interface PageTreeResult {
 export declare function buildPageTree(pages: NotionPage[], options?: BuildPageTreeOptions): PageTreeResult;
 /**
  * Slugify a page title for use as a KB path segment.
- * Lowercase, spaces→hyphens, strip special chars, ltree-compatible.
+ *
+ * Server's validatePathForLtree only allows [a-zA-Z0-9-]. We:
+ *   - lowercase
+ *   - normalize NFKD + strip combining marks so "résumé" → "resume"
+ *   - replace anything not in [a-z0-9-] with "-"
+ *   - collapse runs of "-" and trim leading/trailing "-"
+ * Falls back to "untitled" when the result is empty (e.g. emoji-only or non-Latin titles).
  */
 export declare function slugify(title: string): string;

package/dist/sources/notion.js

@@ -596,15 +596,21 @@ function getParentPageId(page) {
 // ============================================
 /**
  * Slugify a page title for use as a KB path segment.
- * Lowercase, spaces→hyphens, strip special chars, ltree-compatible.
+ *
+ * Server's validatePathForLtree only allows [a-zA-Z0-9-]. We:
+ *   - lowercase
+ *   - normalize NFKD + strip combining marks so "résumé" → "resume"
+ *   - replace anything not in [a-z0-9-] with "-"
+ *   - collapse runs of "-" and trim leading/trailing "-"
+ * Falls back to "untitled" when the result is empty (e.g. emoji-only or non-Latin titles).
  */
 export function slugify(title) {
-    return (title
+    const cleaned = title
+        .normalize('NFKD')
+        .replace(/[̀-ͯ]/g, '')
         .toLowerCase()
-        .trim()
-        .replace(/[^\w\s-]/g, '') // Remove special chars (keep word chars, spaces, hyphens)
-        .replace(/[\s_]+/g, '-') // Spaces and underscores → hyphens (ltree forbids _ and .)
-        .replace(/-+/g, '-') // Collapse multiple hyphens
-        .replace(/^-|-$/g, '') || // Trim leading/trailing hyphens
-        'untitled');
+        .replace(/[^a-z0-9-]/g, '-')
+        .replace(/-+/g, '-')
+        .replace(/^-+|-+$/g, '');
+    return cleaned.length > 0 ? cleaned : 'untitled';
 }

package/dist/sources/notion.test.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/sources/notion.test.js

@@ -0,0 +1,45 @@
+import { describe, it, expect } from 'vitest';
+import { slugify } from './notion.js';
+describe('slugify (notion)', () => {
+    it('lowercases', () => {
+        expect(slugify('MyDoc')).toBe('mydoc');
+    });
+    it('replaces spaces with hyphens', () => {
+        expect(slugify('My Doc')).toBe('my-doc');
+    });
+    it('replaces underscores with hyphens (ltree forbids _)', () => {
+        expect(slugify('my_doc')).toBe('my-doc');
+    });
+    it('replaces periods with hyphens (ltree forbids .)', () => {
+        expect(slugify('release.v2')).toBe('release-v2');
+    });
+    it('handles parens, ampersands, apostrophes', () => {
+        expect(slugify('Q3 Plan (final)')).toBe('q3-plan-final');
+        expect(slugify("Tom's & Jerry's")).toBe('tom-s-jerry-s');
+    });
+    it('strips accents via NFKD normalization', () => {
+        expect(slugify('résumé')).toBe('resume');
+        expect(slugify('café')).toBe('cafe');
+        expect(slugify('naïve')).toBe('naive');
+    });
+    it('replaces non-Latin scripts with hyphens', () => {
+        expect(slugify('файл')).toBe('untitled');
+        expect(slugify('日本語')).toBe('untitled');
+    });
+    it('preserves digits', () => {
+        expect(slugify('Q3-2024')).toBe('q3-2024');
+    });
+    it('collapses runs of separators', () => {
+        expect(slugify('foo___bar...baz')).toBe('foo-bar-baz');
+    });
+    it('trims leading/trailing hyphens', () => {
+        expect(slugify('-foo-')).toBe('foo');
+        expect(slugify('  hello  ')).toBe('hello');
+    });
+    it('falls back to "untitled" for empty result', () => {
+        expect(slugify('')).toBe('untitled');
+        expect(slugify('___')).toBe('untitled');
+        expect(slugify('!!!')).toBe('untitled');
+        expect(slugify('🎉')).toBe('untitled');
+    });
+});

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@moxn/kb-migrate",
-  "version": "0.4.34",
+  "version": "0.4.36",
   "description": "Migration tool for importing documents into Moxn Knowledge Base from local files, Notion, Google Docs, and more",
   "type": "module",
   "main": "dist/index.js",
@@ -70,6 +70,14 @@
       "types": "./dist/client.d.ts",
       "default": "./dist/client.js"
     },
+    "./grammar-media": {
+      "types": "./dist/grammar-media.d.ts",
+      "default": "./dist/grammar-media.js"
+    },
+    "./import-local": {
+      "types": "./dist/import-local.d.ts",
+      "default": "./dist/import-local.js"
+    },
     "./date-filter": {
       "types": "./dist/date-filter.d.ts",
       "default": "./dist/date-filter.js"