bluera-knowledge 0.9.32 → 0.9.36

package/src/crawl/bridge.test.ts

@@ -6,6 +6,7 @@
 import { describe, it, expect, beforeEach, afterEach, vi } from 'vitest';
 import { PythonBridge } from './bridge.js';
 import type { ChildProcess } from 'node:child_process';
+import type { Interface as ReadlineInterface, ReadLineOptions } from 'node:readline';
 import { EventEmitter } from 'node:events';
 
 // Mock child_process
@@ -27,17 +28,27 @@ describe('PythonBridge', () => {
   let mockReadline: MockReadline;
   let mockStderrReadline: MockReadline;
 
+  // Mock classes that satisfy the ChildProcess and Interface contracts
   class MockChildProcess extends EventEmitter {
-    stdin = {
-      write: vi.fn(),
-    };
+    stdin = { write: vi.fn() };
     stdout = new EventEmitter();
     stderr = new EventEmitter();
-    kill = vi.fn();
+    kill = vi.fn(() => {
+      // Emit exit event asynchronously to simulate real process behavior
+      setImmediate(() => this.emit('exit', 0, null));
+    });
+    // Type-safe cast helper
+    asChildProcess(): ChildProcess {
+      return this as unknown as ChildProcess;
+    }
   }
 
   class MockReadline extends EventEmitter {
     close = vi.fn();
+    // Type-safe cast helper
+    asInterface(): ReadlineInterface {
+      return this as unknown as ReadlineInterface;
+    }
   }
 
   beforeEach(() => {
@@ -47,14 +58,13 @@ describe('PythonBridge', () => {
     mockReadline = new MockReadline();
     mockStderrReadline = new MockReadline();
 
-    vi.mocked(spawn).mockReturnValue(mockProcess as unknown as ChildProcess);
-    // Use mockImplementation to create new instances each time
-    vi.mocked(createInterface).mockImplementation((config: any) => {
+    vi.mocked(spawn).mockReturnValue(mockProcess.asChildProcess());
+    vi.mocked(createInterface).mockImplementation((config: ReadLineOptions) => {
       // First call is for stderr, second is for stdout
       if (config.input === mockProcess.stderr) {
-        return mockStderrReadline as any;
+        return mockStderrReadline.asInterface();
       }
-      return mockReadline as any;
+      return mockReadline.asInterface();
     });
   });
 
@@ -71,7 +81,7 @@ describe('PythonBridge', () => {
         ['python/crawl_worker.py'],
         expect.objectContaining({
           stdio: ['pipe', 'pipe', 'pipe'],
-        }),
+        })
       );
     });
 
@@ -89,7 +99,7 @@ describe('PythonBridge', () => {
       expect(createInterface).toHaveBeenCalledWith(
         expect.objectContaining({
           input: mockProcess.stdout,
-        }),
+        })
       );
     });
 
@@ -99,7 +109,7 @@ describe('PythonBridge', () => {
       expect(createInterface).toHaveBeenCalledWith(
         expect.objectContaining({
           input: mockProcess.stderr,
-        }),
+        })
       );
     });
 
@@ -107,13 +117,16 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com');
 
       // Wait for process to start and write
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await promise;
       expect(spawn).toHaveBeenCalled();
@@ -127,18 +140,21 @@ describe('PythonBridge', () => {
 
       // Emit response immediately
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await promise;
 
       expect(mockProcess.stdin.write).toHaveBeenCalledWith(
-        expect.stringContaining('"jsonrpc":"2.0"'),
+        expect.stringContaining('"jsonrpc":"2.0"')
       );
       expect(mockProcess.stdin.write).toHaveBeenCalledWith(
-        expect.stringContaining('"method":"crawl"'),
+        expect.stringContaining('"method":"crawl"')
       );
     });
 
@@ -147,14 +163,19 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com/test');
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await promise;
 
-      expect(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0]).toContain('"url":"https://example.com/test"');
+      expect(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0]).toContain(
+        '"url":"https://example.com/test"'
+      );
     });
 
     it('should generate unique request IDs', async () => {
@@ -184,14 +205,28 @@ describe('PythonBridge', () => {
       const req2 = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[1]?.[0] as string);
 
       // Send responses in reverse order
-      mockReadline.emit('line', JSON.stringify({
-        id: req2.id,
-        result: { pages: [{ url: 'url2', title: 't2', content: 'c2', links: [], crawledAt: '2024-01-02' }] },
-      }));
-      mockReadline.emit('line', JSON.stringify({
-        id: req1.id,
-        result: { pages: [{ url: 'url1', title: 't1', content: 'c1', links: [], crawledAt: '2024-01-01' }] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: req2.id,
+          result: {
+            pages: [
+              { url: 'url2', title: 't2', content: 'c2', links: [], crawledAt: '2024-01-02' },
+            ],
+          },
+        })
+      );
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: req1.id,
+          result: {
+            pages: [
+              { url: 'url1', title: 't1', content: 'c1', links: [], crawledAt: '2024-01-01' },
+            ],
+          },
+        })
+      );
 
       const [result1, result2] = await Promise.all([promise1, promise2]);
 
@@ -204,18 +239,23 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com');
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: {
-          pages: [{
-            url: 'https://example.com',
-            title: 'Test',
-            content: 'Content',
-            links: [],
-            crawledAt: '2024-01-01',
-          }],
-        },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: {
+            pages: [
+              {
+                url: 'https://example.com',
+                title: 'Test',
+                content: 'Content',
+                links: [],
+                crawledAt: '2024-01-01',
+              },
+            ],
+          },
+        })
+      );
 
       const result = await promise;
       expect(result.pages).toHaveLength(1);
@@ -227,10 +267,13 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com');
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        error: { message: 'Crawl failed' },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          error: { message: 'Crawl failed' },
+        })
+      );
 
       await expect(promise).rejects.toThrow('Crawl failed');
     });
@@ -251,10 +294,13 @@ describe('PythonBridge', () => {
 
       // Should not timeout immediately - resolve quickly
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await expect(promise).resolves.toBeDefined();
     });
@@ -264,10 +310,13 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com', 5000);
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await promise;
 
@@ -280,10 +329,13 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com', 5000);
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        error: { message: 'Error' },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          error: { message: 'Error' },
+        })
+      );
 
       await expect(promise).rejects.toThrow();
     });
@@ -297,10 +349,13 @@ describe('PythonBridge', () => {
       // Try to send a late response - should be ignored
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
       expect(() => {
-        mockReadline.emit('line', JSON.stringify({
-          id: request.id,
-          result: { pages: [] },
-        }));
+        mockReadline.emit(
+          'line',
+          JSON.stringify({
+            id: request.id,
+            result: { pages: [] },
+          })
+        );
       }).not.toThrow();
     });
   });
@@ -311,10 +366,13 @@ describe('PythonBridge', () => {
       const promise = bridge.crawl('https://example.com');
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await expect(promise).resolves.toBeDefined();
     });
@@ -333,10 +391,13 @@ describe('PythonBridge', () => {
       await bridge.start();
       const promise = bridge.crawl('https://example.com', 100);
 
-      mockReadline.emit('line', JSON.stringify({
-        id: 'unknown-id',
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: 'unknown-id',
+          result: { pages: [] },
+        })
+      );
 
       // Should timeout since response wasn't matched
       await expect(promise).rejects.toThrow('timeout');
@@ -394,20 +455,25 @@ describe('PythonBridge', () => {
       const newMockReadline = new MockReadline();
       const newMockStderrReadline = new MockReadline();
 
-      vi.mocked(spawn).mockReturnValue(newMockProcess as unknown as ChildProcess);
+      vi.mocked(spawn).mockReturnValue(newMockProcess.asChildProcess());
       vi.mocked(createInterface)
-        .mockReturnValueOnce(newMockStderrReadline as any)
-        .mockReturnValueOnce(newMockReadline as any);
+        .mockReturnValueOnce(newMockStderrReadline.asInterface())
+        .mockReturnValueOnce(newMockReadline.asInterface());
 
       const promise = bridge.crawl('https://example.com');
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
-      const request = JSON.parse(vi.mocked(newMockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      newMockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      const request = JSON.parse(
+        vi.mocked(newMockProcess.stdin.write).mock.calls[0]?.[0] as string
+      );
+      newMockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await promise;
       expect(spawn).toHaveBeenCalledTimes(2);
@@ -415,11 +481,25 @@ describe('PythonBridge', () => {
 
     it('should handle process with null stdout', async () => {
       const nullStdoutProcess = new MockChildProcess();
-      (nullStdoutProcess as any).stdout = null;
+      // Override stdout to null for this test case
+      Object.defineProperty(nullStdoutProcess, 'stdout', { value: null });
+
+      vi.mocked(spawn).mockReturnValue(nullStdoutProcess.asChildProcess());
+
+      await expect(bridge.start()).rejects.toThrow('Python bridge process stdout is null');
+    });
+
+    it('should kill process when stdout is null to prevent zombie process', async () => {
+      const nullStdoutProcess = new MockChildProcess();
+      // Override stdout to null for this test case
+      Object.defineProperty(nullStdoutProcess, 'stdout', { value: null });
 
-      vi.mocked(spawn).mockReturnValue(nullStdoutProcess as unknown as ChildProcess);
+      vi.mocked(spawn).mockReturnValue(nullStdoutProcess.asChildProcess());
 
       await expect(bridge.start()).rejects.toThrow('Python bridge process stdout is null');
+
+      // Critical: process must be killed to prevent zombie
+      expect(nullStdoutProcess.kill).toHaveBeenCalled();
     });
   });
 
@@ -429,23 +509,29 @@ describe('PythonBridge', () => {
       let capturedId: string | undefined;
       const promise = bridge.crawl('https://example.com');
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
       capturedId = request.id;
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await promise;
 
       // Send same ID again - should be ignored (pending was cleared)
       expect(() => {
-        mockReadline.emit('line', JSON.stringify({
-          id: capturedId,
-          result: { pages: [] },
-        }));
+        mockReadline.emit(
+          'line',
+          JSON.stringify({
+            id: capturedId,
+            result: { pages: [] },
+          })
+        );
       }).not.toThrow();
     });
 
@@ -454,23 +540,29 @@ describe('PythonBridge', () => {
       let capturedId: string | undefined;
       const promise = bridge.crawl('https://example.com');
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
       capturedId = request.id;
-      mockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        error: { message: 'Error' },
-      }));
+      mockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          error: { message: 'Error' },
+        })
+      );
 
       await expect(promise).rejects.toThrow();
 
       // Pending should be cleared
       expect(() => {
-        mockReadline.emit('line', JSON.stringify({
-          id: capturedId,
-          result: { pages: [] },
-        }));
+        mockReadline.emit(
+          'line',
+          JSON.stringify({
+            id: capturedId,
+            result: { pages: [] },
+          })
+        );
       }).not.toThrow();
     });
 
@@ -483,10 +575,13 @@ describe('PythonBridge', () => {
       // Pending should be cleared, late response should be ignored
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
       expect(() => {
-        mockReadline.emit('line', JSON.stringify({
-          id: request.id,
-          result: { pages: [] },
-        }));
+        mockReadline.emit(
+          'line',
+          JSON.stringify({
+            id: request.id,
+            result: { pages: [] },
+          })
+        );
       }).not.toThrow();
     });
 
@@ -495,10 +590,14 @@ describe('PythonBridge', () => {
       const promise1 = bridge.crawl('https://example.com/1', 1000);
       const promise2 = bridge.crawl('https://example.com/2', 1000);
 
+      // Attach rejection handlers BEFORE stop to avoid unhandled rejection
+      const rejection1 = expect(promise1).rejects.toThrow('stopped');
+      const rejection2 = expect(promise2).rejects.toThrow('stopped');
+
       await bridge.stop();
 
-      await expect(promise1).rejects.toThrow('stopped');
-      await expect(promise2).rejects.toThrow('stopped');
+      await rejection1;
+      await rejection2;
     });
   });
 
@@ -512,7 +611,7 @@ describe('PythonBridge', () => {
         bridge.crawl('https://example.com/3'),
       ];
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
       const req1 = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
       const req2 = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[1]?.[0] as string);
@@ -532,7 +631,7 @@ describe('PythonBridge', () => {
       const promise1 = bridge.crawl('https://example.com/1', 50);
       const promise2 = bridge.crawl('https://example.com/2', 1000);
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
       // Resolve second immediately
       const req2 = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[1]?.[0] as string);
@@ -564,10 +663,10 @@ describe('PythonBridge', () => {
       const newMockReadline = new MockReadline();
       const newMockStderrReadline = new MockReadline();
 
-      vi.mocked(spawn).mockReturnValue(newMockProcess as unknown as ChildProcess);
+      vi.mocked(spawn).mockReturnValue(newMockProcess.asChildProcess());
       vi.mocked(createInterface)
-        .mockReturnValueOnce(newMockStderrReadline as any)
-        .mockReturnValueOnce(newMockReadline as any);
+        .mockReturnValueOnce(newMockStderrReadline.asInterface())
+        .mockReturnValueOnce(newMockReadline.asInterface());
 
       await bridge.start();
 
@@ -583,9 +682,12 @@ describe('PythonBridge', () => {
       await bridge.start();
       const promise = bridge.crawl('https://example.com');
 
+      // Attach rejection handler BEFORE stop to avoid unhandled rejection
+      const rejection = expect(promise).rejects.toThrow('Python bridge stopped');
+
       await bridge.stop();
 
-      await expect(promise).rejects.toThrow('Python bridge stopped');
+      await rejection;
     });
   });
 
@@ -595,7 +697,7 @@ describe('PythonBridge', () => {
       mockProcess.emit('exit', 0, null);
 
       // Should be able to start again
-      vi.mocked(spawn).mockReturnValue(mockProcess as unknown as ChildProcess);
+      vi.mocked(spawn).mockReturnValue(mockProcess.asChildProcess());
       await bridge.start();
 
       expect(spawn).toHaveBeenCalledTimes(2);
@@ -609,10 +711,10 @@ describe('PythonBridge', () => {
       const newMockReadline = new MockReadline();
       const newMockStderrReadline = new MockReadline();
 
-      vi.mocked(spawn).mockReturnValue(newMockProcess as unknown as ChildProcess);
+      vi.mocked(spawn).mockReturnValue(newMockProcess.asChildProcess());
       vi.mocked(createInterface)
-        .mockReturnValueOnce(newMockStderrReadline as any)
-        .mockReturnValueOnce(newMockReadline as any);
+        .mockReturnValueOnce(newMockStderrReadline.asInterface())
+        .mockReturnValueOnce(newMockReadline.asInterface());
 
       await bridge.start();
 
@@ -631,20 +733,25 @@ describe('PythonBridge', () => {
       const newMockReadline = new MockReadline();
       const newMockStderrReadline = new MockReadline();
 
-      vi.mocked(spawn).mockReturnValue(newMockProcess as unknown as ChildProcess);
+      vi.mocked(spawn).mockReturnValue(newMockProcess.asChildProcess());
       vi.mocked(createInterface)
-        .mockReturnValueOnce(newMockStderrReadline as any)
-        .mockReturnValueOnce(newMockReadline as any);
+        .mockReturnValueOnce(newMockStderrReadline.asInterface())
+        .mockReturnValueOnce(newMockReadline.asInterface());
 
       const promise = bridge.crawl('https://example.com');
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
-      const request = JSON.parse(vi.mocked(newMockProcess.stdin.write).mock.calls[0]?.[0] as string);
-      newMockReadline.emit('line', JSON.stringify({
-        id: request.id,
-        result: { pages: [] },
-      }));
+      const request = JSON.parse(
+        vi.mocked(newMockProcess.stdin.write).mock.calls[0]?.[0] as string
+      );
+      newMockReadline.emit(
+        'line',
+        JSON.stringify({
+          id: request.id,
+          result: { pages: [] },
+        })
+      );
 
       await expect(promise).resolves.toBeDefined();
     });
@@ -653,7 +760,8 @@ describe('PythonBridge', () => {
   describe('Error Edge Cases', () => {
     it('should handle crawl when process stdin is null', async () => {
       await bridge.start();
-      (mockProcess as any).stdin = null;
+      // Override stdin to null for this test case
+      Object.defineProperty(mockProcess, 'stdin', { value: null });
 
       await expect(bridge.crawl('https://example.com')).rejects.toThrow('process not available');
     });
@@ -662,7 +770,7 @@ describe('PythonBridge', () => {
       await bridge.start();
       const promise = bridge.crawl('https://example.com', 100);
 
-      await new Promise(resolve => setImmediate(resolve));
+      await new Promise((resolve) => setImmediate(resolve));
 
       const request = JSON.parse(vi.mocked(mockProcess.stdin.write).mock.calls[0]?.[0] as string);
       mockReadline.emit('line', JSON.stringify({ id: request.id }));