bluera-knowledge 0.11.18 → 0.11.20

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (48) hide show
  1. package/.claude-plugin/plugin.json +1 -1
  2. package/CHANGELOG.md +55 -0
  3. package/dist/{chunk-6FHWC36B.js → chunk-HRQD3MPH.js} +8 -6
  4. package/dist/chunk-HRQD3MPH.js.map +1 -0
  5. package/dist/{chunk-ZZNABJMQ.js → chunk-MQGRQ2EG.js} +99 -34
  6. package/dist/chunk-MQGRQ2EG.js.map +1 -0
  7. package/dist/{chunk-ZDEO4WJT.js → chunk-Q2ZGPJ66.js} +22 -70
  8. package/dist/chunk-Q2ZGPJ66.js.map +1 -0
  9. package/dist/{chunk-5NUI6JL6.js → chunk-ZSKQIMD7.js} +5 -2
  10. package/dist/chunk-ZSKQIMD7.js.map +1 -0
  11. package/dist/index.js +36 -18
  12. package/dist/index.js.map +1 -1
  13. package/dist/mcp/server.js +3 -3
  14. package/dist/watch.service-OPLKIDFQ.js +7 -0
  15. package/dist/workers/background-worker-cli.js +3 -3
  16. package/package.json +1 -1
  17. package/src/cli/commands/crawl.ts +1 -1
  18. package/src/cli/commands/index-cmd.test.ts +14 -4
  19. package/src/cli/commands/index-cmd.ts +11 -4
  20. package/src/cli/commands/store.test.ts +211 -18
  21. package/src/cli/commands/store.ts +26 -8
  22. package/src/crawl/article-converter.test.ts +30 -61
  23. package/src/crawl/article-converter.ts +2 -8
  24. package/src/crawl/bridge.test.ts +14 -0
  25. package/src/crawl/bridge.ts +17 -5
  26. package/src/crawl/intelligent-crawler.test.ts +65 -76
  27. package/src/crawl/intelligent-crawler.ts +33 -69
  28. package/src/db/lance.test.ts +3 -4
  29. package/src/db/lance.ts +14 -19
  30. package/src/mcp/server.test.ts +56 -1
  31. package/src/mcp/server.ts +5 -1
  32. package/src/plugin/git-clone.test.ts +44 -0
  33. package/src/plugin/git-clone.ts +4 -0
  34. package/src/services/code-unit.service.test.ts +59 -6
  35. package/src/services/code-unit.service.ts +47 -2
  36. package/src/services/index.ts +19 -3
  37. package/src/services/job.service.test.ts +10 -7
  38. package/src/services/job.service.ts +12 -6
  39. package/src/services/search.service.ts +15 -9
  40. package/src/services/services.test.ts +19 -6
  41. package/src/services/watch.service.test.ts +80 -56
  42. package/src/services/watch.service.ts +9 -6
  43. package/dist/chunk-5NUI6JL6.js.map +0 -1
  44. package/dist/chunk-6FHWC36B.js.map +0 -1
  45. package/dist/chunk-ZDEO4WJT.js.map +0 -1
  46. package/dist/chunk-ZZNABJMQ.js.map +0 -1
  47. package/dist/watch.service-BJV3TI3F.js +0 -7
  48. /package/dist/{watch.service-BJV3TI3F.js.map → watch.service-OPLKIDFQ.js.map} +0 -0
package/src/crawl/article-converter.test.ts CHANGED
@@ -44,7 +44,7 @@ describe('convertHtmlToMarkdown', () => {
44
44
  html,
45
45
  'https://example.com'
46
46
  );
47
- expect(result.success).toBe(true);
47
+ // Function throws on error, so reaching here means success
48
48
  });
49
49
 
50
50
  it('should include title from extracted article', async () => {
@@ -71,7 +71,7 @@ describe('convertHtmlToMarkdown', () => {
71
71
  const html = '<html><body><h1>Full HTML</h1></body></html>';
72
72
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
73
73
 
74
- expect(result.success).toBe(true);
74
+ // Function throws on error, so reaching here means success
75
75
  // Should have processed the full HTML through markdown conversion
76
76
  expect(vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks)).toHaveBeenCalledWith(html);
77
77
  });
@@ -92,7 +92,7 @@ describe('convertHtmlToMarkdown', () => {
92
92
  const html = '<html><body><h1>Full HTML</h1></body></html>';
93
93
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
94
94
 
95
- expect(result.success).toBe(true);
95
+ // Function throws on error, so reaching here means success
96
96
  expect(vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks)).toHaveBeenCalledWith(html);
97
97
  });
98
98
 
@@ -112,7 +112,7 @@ describe('convertHtmlToMarkdown', () => {
112
112
  const html = '<html><body><h1>Full HTML</h1></body></html>';
113
113
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
114
114
 
115
- expect(result.success).toBe(true);
115
+ // Function throws on error, so reaching here means success
116
116
  expect(vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks)).toHaveBeenCalledWith(html);
117
117
  });
118
118
 
@@ -122,7 +122,7 @@ describe('convertHtmlToMarkdown', () => {
122
122
  const html = '<html><body><h1>Full HTML</h1></body></html>';
123
123
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
124
124
 
125
- expect(result.success).toBe(true);
125
+ // Function throws on error, so reaching here means success
126
126
  expect(vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks)).toHaveBeenCalledWith(html);
127
127
  });
128
128
 
@@ -192,7 +192,7 @@ describe('convertHtmlToMarkdown', () => {
192
192
 
193
193
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
194
194
 
195
- expect(result.success).toBe(true);
195
+ // Function throws on error, so reaching here means success
196
196
  expect(result.markdown).toContain('# Heading 1');
197
197
  expect(result.markdown).toContain('## Heading 2');
198
198
  expect(result.markdown).toContain('### Heading 3');
@@ -213,7 +213,7 @@ describe('convertHtmlToMarkdown', () => {
213
213
 
214
214
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
215
215
 
216
- expect(result.success).toBe(true);
216
+ // Function throws on error, so reaching here means success
217
217
  expect(result.markdown).toContain('```');
218
218
  });
219
219
 
@@ -232,7 +232,7 @@ describe('convertHtmlToMarkdown', () => {
232
232
 
233
233
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
234
234
 
235
- expect(result.success).toBe(true);
235
+ // Function throws on error, so reaching here means success
236
236
  expect(result.markdown).toContain('[Link Text](https://example.com)');
237
237
  });
238
238
 
@@ -251,7 +251,7 @@ describe('convertHtmlToMarkdown', () => {
251
251
 
252
252
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
253
253
 
254
- expect(result.success).toBe(true);
254
+ // Function throws on error, so reaching here means success
255
255
  expect(result.markdown).toContain('|');
256
256
  });
257
257
 
@@ -270,7 +270,7 @@ describe('convertHtmlToMarkdown', () => {
270
270
 
271
271
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
272
272
 
273
- expect(result.success).toBe(true);
273
+ // Function throws on error, so reaching here means success
274
274
  expect(result.markdown).toContain('# Heading with Anchor');
275
275
  expect(result.markdown).not.toContain('[]()');
276
276
  });
@@ -290,7 +290,7 @@ describe('convertHtmlToMarkdown', () => {
290
290
 
291
291
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
292
292
 
293
- expect(result.success).toBe(true);
293
+ // Function throws on error, so reaching here means success
294
294
  expect(result.markdown).toContain('# Heading with spaces');
295
295
  });
296
296
 
@@ -309,62 +309,31 @@ describe('convertHtmlToMarkdown', () => {
309
309
 
310
310
  const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
311
311
 
312
- expect(result.success).toBe(true);
312
+ // Function throws on error, so reaching here means success
313
313
  // Empty heading should not appear in markdown
314
314
  expect(result.markdown).not.toMatch(/^#\s*$/m);
315
315
  });
316
316
  });
317
317
 
318
318
  describe('Error Handling', () => {
319
- it('should return error result when conversion throws error', async () => {
320
- vi.mocked(articleExtractor.extractFromHtml).mockImplementation(() => {
321
- throw new Error('Fatal conversion error');
322
- });
323
- vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks).mockImplementation(() => {
324
- throw new Error('Fatal conversion error');
325
- });
326
-
327
- const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
328
-
329
- expect(result.success).toBe(false);
330
- expect(result.markdown).toBe('');
331
- expect(result.error).toBe('Fatal conversion error');
332
- });
333
-
334
- it('should handle non-Error thrown values', async () => {
335
- vi.mocked(articleExtractor.extractFromHtml).mockImplementation(() => {
336
- throw 'String error';
337
- });
338
- vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks).mockImplementation(() => {
339
- throw 'String error';
340
- });
341
-
342
- const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
343
-
344
- expect(result.success).toBe(false);
345
- expect(result.error).toBe('String error');
346
- });
347
-
348
- it('should return empty markdown on error', async () => {
319
+ it('should throw when conversion fails due to preprocessing error', async () => {
349
320
  vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks).mockImplementation(() => {
350
321
  throw new Error('Preprocessing failed');
351
322
  });
352
323
 
353
- const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
354
-
355
- expect(result.success).toBe(false);
356
- expect(result.markdown).toBe('');
324
+ await expect(convertHtmlToMarkdown('<html></html>', 'https://example.com')).rejects.toThrow(
325
+ 'Preprocessing failed'
326
+ );
357
327
  });
358
328
 
359
- it('should not include title on error', async () => {
329
+ it('should throw with non-Error values wrapped as Error', async () => {
360
330
  vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks).mockImplementation(() => {
361
- throw new Error('Preprocessing failed');
331
+ throw 'String error';
362
332
  });
363
333
 
364
- const result = await convertHtmlToMarkdown('<html></html>', 'https://example.com');
365
-
366
- expect(result.success).toBe(false);
367
- expect(result.title).toBeUndefined();
334
+ await expect(convertHtmlToMarkdown('<html></html>', 'https://example.com')).rejects.toThrow(
335
+ 'String error'
336
+ );
368
337
  });
369
338
  });
370
339
 
@@ -372,27 +341,27 @@ describe('convertHtmlToMarkdown', () => {
372
341
  it('should handle empty HTML string', async () => {
373
342
  const result = await convertHtmlToMarkdown('', 'https://example.com');
374
343
 
375
- expect(result.success).toBe(true);
344
+ // Function throws on error, so reaching here means success
376
345
  });
377
346
 
378
347
  it('should handle whitespace-only HTML', async () => {
379
348
  const result = await convertHtmlToMarkdown(' \n \t ', 'https://example.com');
380
349
 
381
- expect(result.success).toBe(true);
350
+ // Function throws on error, so reaching here means success
382
351
  });
383
352
 
384
353
  it('should handle malformed HTML', async () => {
385
354
  const html = '<html><body><div><p>Unclosed tags';
386
355
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
387
356
 
388
- expect(result.success).toBe(true);
357
+ // Function throws on error, so reaching here means success
389
358
  });
390
359
 
391
360
  it('should handle HTML with no content', async () => {
392
361
  const html = '<html><head><title>Title</title></head><body></body></html>';
393
362
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
394
363
 
395
- expect(result.success).toBe(true);
364
+ // Function throws on error, so reaching here means success
396
365
  });
397
366
 
398
367
  it('should handle HTML with only navigation elements', async () => {
@@ -411,7 +380,7 @@ describe('convertHtmlToMarkdown', () => {
411
380
  const html = '<html><nav><a href="/">Home</a></nav></html>';
412
381
  const result = await convertHtmlToMarkdown(html, 'https://example.com');
413
382
 
414
- expect(result.success).toBe(true);
383
+ // Function throws on error, so reaching here means success
415
384
  });
416
385
  });
417
386
 
@@ -447,7 +416,7 @@ describe('convertHtmlToMarkdown', () => {
447
416
 
448
417
  const result = await convertHtmlToMarkdown(mkdocsHtml, 'https://example.com/docs');
449
418
 
450
- expect(result.success).toBe(true);
419
+ // Function throws on error, so reaching here means success
451
420
  expect(vi.mocked(markdownUtils.preprocessHtmlForCodeBlocks)).toHaveBeenCalledWith(mkdocsHtml);
452
421
  });
453
422
 
@@ -483,7 +452,7 @@ describe('convertHtmlToMarkdown', () => {
483
452
 
484
453
  const result = await convertHtmlToMarkdown(sphinxHtml, 'https://example.com/docs');
485
454
 
486
- expect(result.success).toBe(true);
455
+ // Function throws on error, so reaching here means success
487
456
  });
488
457
 
489
458
  it('should handle nested code blocks with syntax highlighting', async () => {
@@ -511,7 +480,7 @@ describe('convertHtmlToMarkdown', () => {
511
480
 
512
481
  const result = await convertHtmlToMarkdown(complexHtml, 'https://example.com');
513
482
 
514
- expect(result.success).toBe(true);
483
+ // Function throws on error, so reaching here means success
515
484
  });
516
485
 
517
486
  it('should handle documentation with table of contents', async () => {
@@ -544,7 +513,7 @@ describe('convertHtmlToMarkdown', () => {
544
513
 
545
514
  const result = await convertHtmlToMarkdown(htmlWithToc, 'https://example.com');
546
515
 
547
- expect(result.success).toBe(true);
516
+ // Function throws on error, so reaching here means success
548
517
  });
549
518
  });
550
519
 
package/src/crawl/article-converter.ts CHANGED
@@ -14,8 +14,6 @@ const logger = createLogger('article-converter');
14
14
  export interface ConversionResult {
15
15
  markdown: string;
16
16
  title?: string;
17
- success: boolean;
18
- error?: string;
19
17
  }
20
18
 
21
19
  /**
@@ -128,7 +126,6 @@ export async function convertHtmlToMarkdown(html: string, url: string): Promise<
128
126
  return {
129
127
  markdown,
130
128
  ...(title !== undefined && { title }),
131
- success: true,
132
129
  };
133
130
  } catch (error) {
134
131
  logger.error(
@@ -139,10 +136,7 @@ export async function convertHtmlToMarkdown(html: string, url: string): Promise<
139
136
  'HTML to markdown conversion failed'
140
137
  );
141
138
 
142
- return {
143
- markdown: '',
144
- success: false,
145
- error: error instanceof Error ? error.message : String(error),
146
- };
139
+ // Re-throw errors - do not return graceful degradation
140
+ throw error instanceof Error ? error : new Error(String(error));
147
141
  }
148
142
  }
package/src/crawl/bridge.test.ts CHANGED
@@ -654,6 +654,20 @@ describe('PythonBridge', () => {
654
654
  expect(mockProcess.kill).toHaveBeenCalled();
655
655
  });
656
656
 
657
+ it('should close stderr readline interface on stop', async () => {
658
+ await bridge.start();
659
+ await bridge.stop();
660
+
661
+ expect(mockStderrReadline.close).toHaveBeenCalled();
662
+ });
663
+
664
+ it('should close stdout readline interface on stop', async () => {
665
+ await bridge.start();
666
+ await bridge.stop();
667
+
668
+ expect(mockReadline.close).toHaveBeenCalled();
669
+ });
670
+
657
671
  it('should set process to null on stop', async () => {
658
672
  await bridge.start();
659
673
  await bridge.stop();
package/src/crawl/bridge.ts CHANGED
@@ -1,6 +1,6 @@
1
1
  import { spawn, type ChildProcess } from 'node:child_process';
2
2
  import { randomUUID } from 'node:crypto';
3
- import { createInterface } from 'node:readline';
3
+ import { createInterface, type Interface as ReadlineInterface } from 'node:readline';
4
4
  import { ZodError } from 'zod';
5
5
  import {
6
6
  type CrawlResult,
@@ -31,6 +31,8 @@ export class PythonBridge {
31
31
  private process: ChildProcess | null = null;
32
32
  private readonly pending: Map<string, PendingRequest> = new Map();
33
33
  private stoppingIntentionally = false;
34
+ private stdoutReadline: ReadlineInterface | null = null;
35
+ private stderrReadline: ReadlineInterface | null = null;
34
36
 
35
37
  start(): Promise<void> {
36
38
  if (this.process) return Promise.resolve();
@@ -63,8 +65,8 @@ export class PythonBridge {
63
65
 
64
66
  // Add stderr logging
65
67
  if (this.process.stderr) {
66
- const stderrRl = createInterface({ input: this.process.stderr });
67
- stderrRl.on('line', (line) => {
68
+ this.stderrReadline = createInterface({ input: this.process.stderr });
69
+ this.stderrReadline.on('line', (line) => {
68
70
  logger.warn({ stderr: line }, 'Python bridge stderr output');
69
71
  });
70
72
  }
@@ -74,8 +76,8 @@ export class PythonBridge {
74
76
  this.process = null; // Clean up reference
75
77
  return Promise.reject(new Error('Python bridge process stdout is null'));
76
78
  }
77
- const rl = createInterface({ input: this.process.stdout });
78
- rl.on('line', (line) => {
79
+ this.stdoutReadline = createInterface({ input: this.process.stdout });
80
+ this.stdoutReadline.on('line', (line) => {
79
81
  // Filter out non-JSON lines (crawl4ai verbose output)
80
82
  if (!line.trim().startsWith('{')) {
81
83
  return;
@@ -266,6 +268,16 @@ export class PythonBridge {
266
268
  this.stoppingIntentionally = true;
267
269
  this.rejectAllPending(new Error('Python bridge stopped'));
268
270
 
271
+ // Close readline interfaces to prevent resource leaks
272
+ if (this.stdoutReadline) {
273
+ this.stdoutReadline.close();
274
+ this.stdoutReadline = null;
275
+ }
276
+ if (this.stderrReadline) {
277
+ this.stderrReadline.close();
278
+ this.stderrReadline = null;
279
+ }
280
+
269
281
  // Wait for process to actually exit before resolving
270
282
  const proc = this.process;
271
283
  if (proc === null) {
package/src/crawl/intelligent-crawler.test.ts CHANGED
@@ -58,7 +58,6 @@ describe('IntelligentCrawler', () => {
58
58
 
59
59
  // Setup convertHtmlToMarkdown mock
60
60
  vi.mocked(articleConverter.convertHtmlToMarkdown).mockResolvedValue({
61
- success: true,
62
61
  markdown: '# Test\n\nContent',
63
62
  title: 'Test Page',
64
63
  });
@@ -581,24 +580,20 @@ describe('IntelligentCrawler', () => {
581
580
  });
582
581
  });
583
582
 
584
- describe('Intelligent Mode Fallback', () => {
585
- it('should fallback to simple mode when Claude strategy fails', async () => {
583
+ describe('Intelligent Mode Error Handling', () => {
584
+ it('should throw when Claude strategy fails', async () => {
586
585
  mockClaudeClient.determineCrawlUrls.mockRejectedValue(new Error('Claude API error'));
587
- mockPythonBridge.crawl.mockResolvedValue({ pages: [{ links: [] }] });
588
586
 
589
587
  const results = [];
590
- for await (const result of crawler.crawl('https://example.com', {
591
- crawlInstruction: 'Find all docs',
592
- })) {
593
- results.push(result);
594
- }
588
+ await expect(async () => {
589
+ for await (const result of crawler.crawl('https://example.com', {
590
+ crawlInstruction: 'Find all docs',
591
+ })) {
592
+ results.push(result);
593
+ }
594
+ }).rejects.toThrow('Claude API error');
595
595
 
596
- // Should still crawl using simple mode
597
- expect(results).toHaveLength(1);
598
- const errorEvents = progressEvents.filter((e) => e.type === 'error');
599
- expect(errorEvents.some((e) => e.message?.includes('falling back to simple mode'))).toBe(
600
- true
601
- );
596
+ expect(results).toHaveLength(0);
602
597
  });
603
598
  });
604
599
 
@@ -623,22 +618,21 @@ describe('IntelligentCrawler', () => {
623
618
  );
624
619
  });
625
620
 
626
- it('should continue without extraction if extraction fails', async () => {
621
+ it('should throw when extraction fails', async () => {
627
622
  mockClaudeClient.extractContent.mockRejectedValue(new Error('Extraction failed'));
628
623
  mockPythonBridge.crawl.mockResolvedValue({ pages: [{ links: [] }] });
629
624
 
630
625
  const results = [];
631
- for await (const result of crawler.crawl('https://example.com', {
632
- simple: true,
633
- extractInstruction: 'Extract pricing',
634
- })) {
635
- results.push(result);
636
- }
626
+ await expect(async () => {
627
+ for await (const result of crawler.crawl('https://example.com', {
628
+ simple: true,
629
+ extractInstruction: 'Extract pricing',
630
+ })) {
631
+ results.push(result);
632
+ }
633
+ }).rejects.toThrow('Extraction failed');
637
634
 
638
- expect(results).toHaveLength(1);
639
- expect(results[0]?.extracted).toBeUndefined();
640
- const errorEvents = progressEvents.filter((e) => e.type === 'error');
641
- expect(errorEvents.some((e) => e.message?.includes('storing raw markdown'))).toBe(true);
635
+ expect(results).toHaveLength(0);
642
636
  });
643
637
 
644
638
  it('should not extract when extractInstruction is empty', async () => {
@@ -824,7 +818,6 @@ describe('IntelligentCrawler', () => {
824
818
  it('should include title when available', async () => {
825
819
  mockPythonBridge.crawl.mockResolvedValue({ pages: [{ links: [] }] });
826
820
  vi.mocked(articleConverter.convertHtmlToMarkdown).mockResolvedValue({
827
- success: true,
828
821
  markdown: '# Test',
829
822
  title: 'Test Page Title',
830
823
  });
@@ -839,11 +832,9 @@ describe('IntelligentCrawler', () => {
839
832
 
840
833
  it('should handle conversion failures', async () => {
841
834
  mockPythonBridge.crawl.mockResolvedValue({ pages: [{ links: [] }] });
842
- vi.mocked(articleConverter.convertHtmlToMarkdown).mockResolvedValue({
843
- success: false,
844
- markdown: '',
845
- error: 'Conversion error',
846
- });
835
+ vi.mocked(articleConverter.convertHtmlToMarkdown).mockRejectedValue(
836
+ new Error('Conversion error')
837
+ );
847
838
 
848
839
  const results = [];
849
840
  for await (const result of crawler.crawl('https://example.com', { simple: true })) {
@@ -905,69 +896,67 @@ describe('IntelligentCrawler', () => {
905
896
  });
906
897
  });
907
898
 
908
- describe('npm Package Mode (Claude CLI Not Installed)', () => {
909
- it('should use simple mode when Claude CLI is not available', async () => {
910
- // Simulate npm package usage without Claude Code installed
911
- vi.mocked(ClaudeClient.isAvailable).mockReturnValue(false);
899
+ describe('Headless Mode Error Handling', () => {
900
+ it('should throw when headless fetch fails', async () => {
901
+ mockPythonBridge.fetchHeadless.mockRejectedValue(new Error('Browser crashed'));
912
902
 
913
- // Setup link extraction for simple mode
914
- mockPythonBridge.crawl.mockResolvedValue({
915
- pages: [{ links: [] }],
916
- });
903
+ const results = [];
904
+ await expect(async () => {
905
+ for await (const result of crawler.crawl('https://example.com', {
906
+ simple: true,
907
+ useHeadless: true,
908
+ })) {
909
+ results.push(result);
910
+ }
911
+ }).rejects.toThrow('Headless fetch failed: Browser crashed');
917
912
 
918
- const results: { url: string }[] = [];
913
+ expect(results).toHaveLength(0);
914
+ // Should not have fallen back to axios
915
+ expect(axios.get).not.toHaveBeenCalled();
916
+ });
917
+ });
919
918
 
920
- for await (const result of crawler.crawl('https://example.com', {
921
- crawlInstruction: 'Find all documentation pages', // Would use intelligent mode
922
- maxPages: 5,
923
- })) {
924
- results.push(result);
925
- }
919
+ describe('Claude CLI Not Installed', () => {
920
+ it('should throw when intelligent mode requested but Claude CLI not available', async () => {
921
+ // Simulate npm package usage without Claude Code installed
922
+ vi.mocked(ClaudeClient.isAvailable).mockReturnValue(false);
926
923
 
927
- // Should have crawled using simple BFS mode
928
- expect(results.length).toBeGreaterThan(0);
929
- expect(results[0]?.url).toBe('https://example.com');
924
+ const results: { url: string }[] = [];
930
925
 
931
- // Should have emitted progress event about mode switch
932
- const modeEvent = progressEvents.find(
933
- (e) => e.type === 'error' && e.message?.includes('Claude CLI not found')
934
- );
935
- expect(modeEvent).toBeDefined();
936
- expect(modeEvent?.message).toContain('using simple crawl mode');
926
+ await expect(async () => {
927
+ for await (const result of crawler.crawl('https://example.com', {
928
+ crawlInstruction: 'Find all documentation pages', // Requires intelligent mode
929
+ maxPages: 5,
930
+ })) {
931
+ results.push(result);
932
+ }
933
+ }).rejects.toThrow('Claude CLI not available');
937
934
 
938
- // Should NOT have called Claude's determineCrawlUrls
935
+ expect(results).toHaveLength(0);
939
936
  expect(mockClaudeClient.determineCrawlUrls).not.toHaveBeenCalled();
940
937
  });
941
938
 
942
- it('should skip extraction when Claude CLI is not available', async () => {
939
+ it('should throw when extraction requested but Claude CLI not available', async () => {
943
940
  // Simulate npm package usage without Claude Code installed
944
941
  vi.mocked(ClaudeClient.isAvailable).mockReturnValue(false);
945
942
 
946
- // Setup for simple mode
947
943
  mockPythonBridge.crawl.mockResolvedValue({
948
944
  pages: [{ links: [] }],
949
945
  });
950
946
 
951
947
  const results: { url: string; extracted?: string }[] = [];
952
948
 
953
- for await (const result of crawler.crawl('https://example.com', {
954
- simple: true,
955
- extractInstruction: 'Extract pricing info', // Would use Claude
956
- maxPages: 1,
957
- })) {
958
- results.push(result);
959
- }
960
-
961
- expect(results.length).toBe(1);
962
- expect(results[0]?.extracted).toBeUndefined(); // Should not have extracted
963
-
964
- // Should have emitted skip extraction progress event
965
- const skipEvent = progressEvents.find(
966
- (e) => e.type === 'error' && e.message?.includes('Skipping extraction')
967
- );
968
- expect(skipEvent).toBeDefined();
949
+ await expect(async () => {
950
+ for await (const result of crawler.crawl('https://example.com', {
951
+ simple: true,
952
+ extractInstruction: 'Extract pricing info', // Requires Claude
953
+ maxPages: 1,
954
+ })) {
955
+ results.push(result);
956
+ }
957
+ }).rejects.toThrow('Claude CLI not available');
969
958
 
970
- // Should NOT have called Claude's extractContent
959
+ expect(results).toHaveLength(0);
971
960
  expect(mockClaudeClient.extractContent).not.toHaveBeenCalled();
972
961
  });
973
962
  });