cdp-skill 1.0.7 → 1.0.14

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 (47) hide show
  1. package/README.md +80 -35
  2. package/SKILL.md +198 -1344
  3. package/install.js +1 -0
  4. package/package.json +1 -1
  5. package/src/aria/index.js +8 -0
  6. package/src/aria/output-processor.js +173 -0
  7. package/src/aria/role-query.js +1229 -0
  8. package/src/aria/snapshot.js +459 -0
  9. package/src/aria.js +237 -43
  10. package/src/cdp/browser.js +22 -4
  11. package/src/cdp-skill.js +268 -68
  12. package/src/dom/click-executor.js +240 -76
  13. package/src/dom/element-locator.js +34 -25
  14. package/src/dom/fill-executor.js +55 -27
  15. package/src/page/dialog-handler.js +119 -0
  16. package/src/page/page-controller.js +190 -3
  17. package/src/runner/context-helpers.js +33 -55
  18. package/src/runner/execute-dynamic.js +34 -143
  19. package/src/runner/execute-form.js +11 -11
  20. package/src/runner/execute-input.js +2 -2
  21. package/src/runner/execute-interaction.js +99 -120
  22. package/src/runner/execute-navigation.js +11 -26
  23. package/src/runner/execute-query.js +8 -5
  24. package/src/runner/step-executors.js +256 -95
  25. package/src/runner/step-registry.js +1064 -0
  26. package/src/runner/step-validator.js +16 -740
  27. package/src/tests/Aria.test.js +1025 -0
  28. package/src/tests/ContextHelpers.test.js +39 -28
  29. package/src/tests/ExecuteBrowser.test.js +572 -0
  30. package/src/tests/ExecuteDynamic.test.js +34 -736
  31. package/src/tests/ExecuteForm.test.js +700 -0
  32. package/src/tests/ExecuteInput.test.js +540 -0
  33. package/src/tests/ExecuteInteraction.test.js +319 -0
  34. package/src/tests/ExecuteQuery.test.js +820 -0
  35. package/src/tests/FillExecutor.test.js +2 -2
  36. package/src/tests/StepValidator.test.js +222 -76
  37. package/src/tests/TestRunner.test.js +36 -25
  38. package/src/tests/integration.test.js +2 -1
  39. package/src/types.js +9 -9
  40. package/src/utils/backoff.js +118 -0
  41. package/src/utils/cdp-helpers.js +130 -0
  42. package/src/utils/devices.js +140 -0
  43. package/src/utils/errors.js +242 -0
  44. package/src/utils/index.js +65 -0
  45. package/src/utils/temp.js +75 -0
  46. package/src/utils/validators.js +433 -0
  47. package/src/utils.js +14 -1142
@@ -70,8 +70,9 @@ export async function clickWithVerification(elementLocator, inputEmulator, x, y,
70
70
  * Feature 13: Supports captureResult to detect new visible elements after hover
71
71
  */
72
72
  export async function executeHover(elementLocator, inputEmulator, ariaSnapshot, params) {
73
- const selector = typeof params === 'string' ? params : params.selector;
73
+ const selector = typeof params === 'string' ? params : (params.selector || null);
74
74
  let ref = typeof params === 'object' ? params.ref : null;
75
+ const text = typeof params === 'object' ? params.text : null;
75
76
  const duration = typeof params === 'object' ? (params.duration || 0) : 0;
76
77
 
77
78
  // Detect if string selector looks like a ref (e.g., "s1e1", "s2e12")
@@ -80,9 +81,35 @@ export async function executeHover(elementLocator, inputEmulator, ariaSnapshot,
80
81
  ref = selector;
81
82
  }
82
83
  const force = typeof params === 'object' && params.force === true;
83
- const timeout = typeof params === 'object' ? (params.timeout || 10000) : 10000; // Reduced from 30s to 10s
84
+ const timeout = typeof params === 'object' ? (params.timeout || 10000) : 10000;
84
85
  const captureResult = typeof params === 'object' && params.captureResult === true;
85
86
 
87
+ // Handle coordinate-based hover
88
+ if (typeof params === 'object' && typeof params.x === 'number' && typeof params.y === 'number' && !ref && !selector && !text) {
89
+ await inputEmulator.hover(params.x, params.y, { duration });
90
+ if (captureResult) {
91
+ await sleep(100);
92
+ return await captureHoverResult(elementLocator.session, []);
93
+ }
94
+ return { hovered: true };
95
+ }
96
+
97
+ // Handle text-based hover
98
+ if (text && ariaSnapshot) {
99
+ const refInfo = await ariaSnapshot.findByText(text);
100
+ if (!refInfo) {
101
+ throw elementNotFoundError(`text:${text}`, 0);
102
+ }
103
+ const x = refInfo.box.x + refInfo.box.width / 2;
104
+ const y = refInfo.box.y + refInfo.box.height / 2;
105
+ await inputEmulator.hover(x, y, { duration });
106
+ if (captureResult) {
107
+ await sleep(100);
108
+ return await captureHoverResult(elementLocator.session, []);
109
+ }
110
+ return { hovered: true };
111
+ }
112
+
86
113
  const session = elementLocator.session;
87
114
  let visibleElementsBefore = [];
88
115
 
@@ -285,7 +312,7 @@ export async function captureHoverResult(session, visibleBefore) {
285
312
  * @returns {Promise<Object>} Drag result
286
313
  */
287
314
  export async function executeDrag(elementLocator, inputEmulator, pageController, ariaSnapshot, params) {
288
- const { source, target, steps = 10, delay = 0 } = params;
315
+ const { source, target, steps = 10, delay = 0, method = 'auto' } = params;
289
316
  const session = elementLocator.session;
290
317
 
291
318
  // Helper to get element bounding box by ref
@@ -305,8 +332,6 @@ export async function executeDrag(elementLocator, inputEmulator, pageController,
305
332
 
306
333
  // Helper to get element bounding box in current frame context
307
334
  async function getElementBox(selector) {
308
- // Use page controller's frame context if available
309
- const contextId = pageController.currentExecutionContextId;
310
335
  const evalParams = {
311
336
  expression: `
312
337
  (function() {
@@ -319,10 +344,8 @@ export async function executeDrag(elementLocator, inputEmulator, pageController,
319
344
  returnByValue: true
320
345
  };
321
346
 
322
- // Add context ID if we're in a non-main frame
323
- if (contextId && pageController.currentFrameId !== pageController.mainFrameId) {
324
- evalParams.contextId = contextId;
325
- }
347
+ const contextId = pageController.getFrameContext();
348
+ if (contextId) evalParams.contextId = contextId;
326
349
 
327
350
  const result = await session.send('Runtime.evaluate', evalParams);
328
351
  if (result.exceptionDetails) {
@@ -415,10 +438,10 @@ export async function executeDrag(elementLocator, inputEmulator, pageController,
415
438
  const targetX = ${targetX};
416
439
  const targetY = ${targetY};
417
440
  const steps = ${steps};
441
+ const method = ${JSON.stringify(method)};
418
442
 
419
443
  // Check if source is an input[type=range] (slider)
420
444
  if (sourceEl && sourceEl.tagName === 'INPUT' && sourceEl.type === 'range') {
421
- // For range inputs, calculate the value based on target position
422
445
  const rect = sourceEl.getBoundingClientRect();
423
446
  const percent = Math.max(0, Math.min(1, (targetX - rect.left) / rect.width));
424
447
  const min = parseFloat(sourceEl.min) || 0;
@@ -430,130 +453,86 @@ export async function executeDrag(elementLocator, inputEmulator, pageController,
430
453
  return { success: true, method: 'range-input', value: newValue };
431
454
  }
432
455
 
433
- // Try HTML5 Drag and Drop API first
434
- if (sourceEl && targetEl) {
456
+ function doMouseDrag() {
457
+ const sourceElAtPoint = sourceEl || document.elementFromPoint(sourceX, sourceY);
458
+ if (!sourceElAtPoint) {
459
+ return { success: false, error: 'No element at source coordinates' };
460
+ }
461
+
462
+ sourceElAtPoint.dispatchEvent(new MouseEvent('mousedown', {
463
+ bubbles: true, cancelable: true,
464
+ clientX: sourceX, clientY: sourceY, button: 0, buttons: 1
465
+ }));
466
+
467
+ const deltaX = (targetX - sourceX) / steps;
468
+ const deltaY = (targetY - sourceY) / steps;
469
+
470
+ for (let i = 1; i <= steps; i++) {
471
+ const currentX = sourceX + deltaX * i;
472
+ const currentY = sourceY + deltaY * i;
473
+ const elAtPoint = document.elementFromPoint(currentX, currentY) || sourceElAtPoint;
474
+ elAtPoint.dispatchEvent(new MouseEvent('mousemove', {
475
+ bubbles: true, cancelable: true,
476
+ clientX: currentX, clientY: currentY, button: 0, buttons: 1
477
+ }));
478
+ }
479
+
480
+ const targetElAtPoint = document.elementFromPoint(targetX, targetY) || sourceElAtPoint;
481
+ targetElAtPoint.dispatchEvent(new MouseEvent('mouseup', {
482
+ bubbles: true, cancelable: true,
483
+ clientX: targetX, clientY: targetY, button: 0, buttons: 0
484
+ }));
485
+
486
+ return { success: true, method: 'mouse-events' };
487
+ }
488
+
489
+ function doHtml5Drag() {
490
+ if (!sourceEl || !targetEl) {
491
+ return { success: false, error: 'HTML5 DnD requires both source and target elements' };
492
+ }
435
493
  try {
436
- // Create DataTransfer object
437
494
  const dataTransfer = new DataTransfer();
438
495
  dataTransfer.effectAllowed = 'all';
439
496
  dataTransfer.dropEffect = 'move';
440
497
 
441
- // Dispatch dragstart on source
442
- const dragStartEvent = new DragEvent('dragstart', {
443
- bubbles: true,
444
- cancelable: true,
445
- dataTransfer: dataTransfer,
446
- clientX: sourceX,
447
- clientY: sourceY
448
- });
449
- sourceEl.dispatchEvent(dragStartEvent);
450
-
451
- // Dispatch drag on source
452
- const dragEvent = new DragEvent('drag', {
453
- bubbles: true,
454
- cancelable: true,
455
- dataTransfer: dataTransfer,
456
- clientX: sourceX,
457
- clientY: sourceY
458
- });
459
- sourceEl.dispatchEvent(dragEvent);
460
-
461
- // Dispatch dragenter on target
462
- const dragEnterEvent = new DragEvent('dragenter', {
463
- bubbles: true,
464
- cancelable: true,
465
- dataTransfer: dataTransfer,
466
- clientX: targetX,
467
- clientY: targetY
468
- });
469
- targetEl.dispatchEvent(dragEnterEvent);
470
-
471
- // Dispatch dragover on target
472
- const dragOverEvent = new DragEvent('dragover', {
473
- bubbles: true,
474
- cancelable: true,
475
- dataTransfer: dataTransfer,
476
- clientX: targetX,
477
- clientY: targetY
478
- });
479
- targetEl.dispatchEvent(dragOverEvent);
480
-
481
- // Dispatch drop on target
482
- const dropEvent = new DragEvent('drop', {
483
- bubbles: true,
484
- cancelable: true,
485
- dataTransfer: dataTransfer,
486
- clientX: targetX,
487
- clientY: targetY
488
- });
489
- targetEl.dispatchEvent(dropEvent);
490
-
491
- // Dispatch dragend on source
492
- const dragEndEvent = new DragEvent('dragend', {
493
- bubbles: true,
494
- cancelable: true,
495
- dataTransfer: dataTransfer,
496
- clientX: targetX,
497
- clientY: targetY
498
- });
499
- sourceEl.dispatchEvent(dragEndEvent);
498
+ sourceEl.dispatchEvent(new DragEvent('dragstart', {
499
+ bubbles: true, cancelable: true, dataTransfer, clientX: sourceX, clientY: sourceY
500
+ }));
501
+ sourceEl.dispatchEvent(new DragEvent('drag', {
502
+ bubbles: true, cancelable: true, dataTransfer, clientX: sourceX, clientY: sourceY
503
+ }));
504
+ targetEl.dispatchEvent(new DragEvent('dragenter', {
505
+ bubbles: true, cancelable: true, dataTransfer, clientX: targetX, clientY: targetY
506
+ }));
507
+ targetEl.dispatchEvent(new DragEvent('dragover', {
508
+ bubbles: true, cancelable: true, dataTransfer, clientX: targetX, clientY: targetY
509
+ }));
510
+ targetEl.dispatchEvent(new DragEvent('drop', {
511
+ bubbles: true, cancelable: true, dataTransfer, clientX: targetX, clientY: targetY
512
+ }));
513
+ sourceEl.dispatchEvent(new DragEvent('dragend', {
514
+ bubbles: true, cancelable: true, dataTransfer, clientX: targetX, clientY: targetY
515
+ }));
500
516
 
501
517
  return { success: true, method: 'html5-dnd' };
502
518
  } catch (e) {
503
- // Fall through to mouse events
519
+ return { success: false, error: e.message };
504
520
  }
505
521
  }
506
522
 
507
- // Fallback: Mouse event simulation for non-DnD dragging (e.g., sortable lists, custom drag)
508
- const sourceElAtPoint = sourceEl || document.elementFromPoint(sourceX, sourceY);
509
- if (!sourceElAtPoint) {
510
- return { success: false, error: 'No element at source coordinates' };
511
- }
523
+ if (method === 'mouse') return doMouseDrag();
524
+ if (method === 'html5') return doHtml5Drag();
525
+
526
+ // auto: try mouse first (works for jQuery UI, sortable lists), then HTML5 DnD
527
+ const mouseResult = doMouseDrag();
528
+ if (mouseResult.success) return mouseResult;
512
529
 
513
- // Dispatch mouse events
514
- const mouseDown = new MouseEvent('mousedown', {
515
- bubbles: true,
516
- cancelable: true,
517
- clientX: sourceX,
518
- clientY: sourceY,
519
- button: 0,
520
- buttons: 1
521
- });
522
- sourceElAtPoint.dispatchEvent(mouseDown);
523
-
524
- // Move in steps
525
- const deltaX = (targetX - sourceX) / steps;
526
- const deltaY = (targetY - sourceY) / steps;
527
-
528
- for (let i = 1; i <= steps; i++) {
529
- const currentX = sourceX + deltaX * i;
530
- const currentY = sourceY + deltaY * i;
531
- const elAtPoint = document.elementFromPoint(currentX, currentY) || sourceElAtPoint;
532
-
533
- const mouseMove = new MouseEvent('mousemove', {
534
- bubbles: true,
535
- cancelable: true,
536
- clientX: currentX,
537
- clientY: currentY,
538
- button: 0,
539
- buttons: 1
540
- });
541
- elAtPoint.dispatchEvent(mouseMove);
530
+ if (sourceEl && targetEl) {
531
+ const html5Result = doHtml5Drag();
532
+ if (html5Result.success) return html5Result;
542
533
  }
543
534
 
544
- // Release at target
545
- const targetElAtPoint = document.elementFromPoint(targetX, targetY) || sourceElAtPoint;
546
- const mouseUp = new MouseEvent('mouseup', {
547
- bubbles: true,
548
- cancelable: true,
549
- clientX: targetX,
550
- clientY: targetY,
551
- button: 0,
552
- buttons: 0
553
- });
554
- targetElAtPoint.dispatchEvent(mouseUp);
555
-
556
- return { success: true, method: 'mouse-events' };
535
+ return mouseResult;
557
536
  })()
558
537
  `,
559
538
  returnByValue: true,
@@ -95,40 +95,30 @@ export async function executeScroll(elementLocator, inputEmulator, pageControlle
95
95
  throw new Error(`Element ref:${ref} is no longer attached to the DOM. Run 'snapshot' again to get fresh refs.`);
96
96
  }
97
97
  // Scroll to element using its coordinates
98
- await pageController.session.send('Runtime.evaluate', {
99
- expression: `
98
+ await pageController.evaluateInFrame(`
100
99
  (function() {
101
100
  const el = window.__ariaRefs && window.__ariaRefs.get(${JSON.stringify(ref)});
102
101
  if (el && el.scrollIntoView) {
103
102
  el.scrollIntoView({ block: 'center', behavior: 'smooth' });
104
103
  }
105
104
  })()
106
- `
107
- });
105
+ `);
108
106
  }
109
107
 
110
108
  if (typeof params === 'string') {
111
109
  // Direction-based scroll
112
110
  switch (params) {
113
111
  case 'top':
114
- await pageController.session.send('Runtime.evaluate', {
115
- expression: 'window.scrollTo(0, 0)'
116
- });
112
+ await pageController.evaluateInFrame('window.scrollTo(0, 0)');
117
113
  break;
118
114
  case 'bottom':
119
- await pageController.session.send('Runtime.evaluate', {
120
- expression: 'window.scrollTo(0, document.body.scrollHeight)'
121
- });
115
+ await pageController.evaluateInFrame('window.scrollTo(0, document.body.scrollHeight)');
122
116
  break;
123
117
  case 'up':
124
- await pageController.session.send('Runtime.evaluate', {
125
- expression: 'window.scrollBy(0, -300)'
126
- });
118
+ await pageController.evaluateInFrame('window.scrollBy(0, -300)');
127
119
  break;
128
120
  case 'down':
129
- await pageController.session.send('Runtime.evaluate', {
130
- expression: 'window.scrollBy(0, 300)'
131
- });
121
+ await pageController.evaluateInFrame('window.scrollBy(0, 300)');
132
122
  break;
133
123
  default:
134
124
  // Check if it looks like a ref (e.g., "s1e1", "s2e12")
@@ -159,22 +149,17 @@ export async function executeScroll(elementLocator, inputEmulator, pageControlle
159
149
  await el.dispose();
160
150
  } else if (params.deltaY !== undefined || params.deltaX !== undefined) {
161
151
  // Scroll by delta using JavaScript (more reliable than CDP mouse wheel events)
162
- await pageController.session.send('Runtime.evaluate', {
163
- expression: `window.scrollBy(${params.deltaX || 0}, ${params.deltaY || 0})`
164
- });
152
+ await pageController.evaluateInFrame(`window.scrollBy(${params.deltaX || 0}, ${params.deltaY || 0})`);
165
153
  } else if (params.y !== undefined) {
166
154
  // Scroll to position
167
- await pageController.session.send('Runtime.evaluate', {
168
- expression: `window.scrollTo(${params.x || 0}, ${params.y})`
169
- });
155
+ await pageController.evaluateInFrame(`window.scrollTo(${params.x || 0}, ${params.y})`);
170
156
  }
171
157
  }
172
158
 
173
159
  // Return current scroll position
174
- const posResult = await pageController.session.send('Runtime.evaluate', {
175
- expression: '({ scrollX: window.scrollX, scrollY: window.scrollY })',
176
- returnByValue: true
177
- });
160
+ const posResult = await pageController.evaluateInFrame(
161
+ '({ scrollX: window.scrollX, scrollY: window.scrollY })'
162
+ );
178
163
 
179
164
  return posResult.result.value;
180
165
  }
@@ -133,10 +133,11 @@ export async function executeGetDom(pageController, params) {
133
133
  };
134
134
  })()`;
135
135
 
136
- const result = await session.send('Runtime.evaluate', {
137
- expression,
138
- returnByValue: true
139
- });
136
+ const evalArgs = { expression, returnByValue: true };
137
+ const contextId = pageController.getFrameContext();
138
+ if (contextId) evalArgs.contextId = contextId;
139
+
140
+ const result = await session.send('Runtime.evaluate', evalArgs);
140
141
 
141
142
  if (result.exceptionDetails) {
142
143
  throw new Error(`getDom error: ${result.exceptionDetails.text}`);
@@ -746,7 +747,9 @@ export async function executeQuery(elementLocator, params) {
746
747
  */
747
748
 
748
749
  export async function executeRoleQuery(elementLocator, params) {
749
- const roleQueryExecutor = createRoleQueryExecutor(elementLocator.session, elementLocator);
750
+ const roleQueryExecutor = createRoleQueryExecutor(elementLocator.session, elementLocator, {
751
+ getFrameContext: elementLocator.getFrameContext
752
+ });
750
753
  return roleQueryExecutor.execute(params);
751
754
  }
752
755