feat(tools): add pdf extraction for minio ingestion

src/tools/builtin/minio-ingest.test.ts

@@ -31,10 +31,11 @@ describe('minio ingest internals', () => {
   it('accepts known text-like extensions', () => {
     expect(minioIngestInternals.isLikelyTextObject('notes/today.md')).toBe(true);
     expect(minioIngestInternals.isLikelyTextObject('logs/daemon.log')).toBe(true);
+    expect(minioIngestInternals.isLikelyTextObject('manual.pdf')).toBe(true);
   });
 
   it('rejects likely binary extensions', () => {
-    expect(minioIngestInternals.isLikelyTextObject('manual.pdf')).toBe(false);
+    expect(minioIngestInternals.isLikelyTextObject('manual.exe')).toBe(false);
   });
 });
 
@@ -72,18 +73,54 @@ describe('createMinioIngestTool', () => {
     );
   });
 
-  it('rejects likely binary object unless force=true', async () => {
+  it('rejects unsupported binary object unless force=true', async () => {
     const write = vi.fn();
     const store = { write } as unknown as MemoryStore;
     const execRunner = vi.fn();
     const tool = createMinioIngestTool(makeBackupConfig(), store, { execRunner });
 
-    const result = await tool.execute({ object_key: 'knowledge/diagram.pdf' });
+    const result = await tool.execute({ object_key: 'knowledge/diagram.docx' });
     expect(result.success).toBe(false);
     expect(result.error).toContain('Unsupported object type');
     expect(execRunner).not.toHaveBeenCalled();
   });
 
+  it('extracts PDF text with pdftotext', async () => {
+    const write = vi.fn();
+    const store = { write } as unknown as MemoryStore;
+    const execRunner = vi.fn(async (_file: string, args: string[]) => {
+      if (args[0] === 'cp') {
+        return { stdout: '', stderr: '' };
+      }
+      if (_file === 'pdftotext') {
+        return { stdout: 'Extracted PDF text', stderr: '' };
+      }
+      return { stdout: '', stderr: '' };
+    });
+    const tool = createMinioIngestTool(makeBackupConfig(), store, { execRunner });
+
+    const result = await tool.execute({
+      object_key: 'knowledge/diagram.pdf',
+      mode: 'replace',
+    });
+    expect(result.success).toBe(true);
+    expect(write).toHaveBeenCalledWith(
+      'global/knowledge',
+      expect.stringContaining('Extracted PDF text'),
+      'replace',
+    );
+    expect(execRunner).toHaveBeenCalledWith(
+      'mc',
+      expect.arrayContaining(['cp', 'flynningest/flynn-knowledge/knowledge/diagram.pdf']),
+      expect.objectContaining({ env: expect.any(Object) }),
+    );
+    expect(execRunner).toHaveBeenCalledWith(
+      'pdftotext',
+      expect.arrayContaining(['-q']),
+      expect.any(Object),
+    );
+  });
+
   it('allows non-text extension when force=true', async () => {
     const write = vi.fn();
     const store = { write } as unknown as MemoryStore;