feat(pageSnapshot): add header comment with info by Skn0tt · Pull Request #34941 · microsoft/playwright

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

Simon Knott (Skn0tt) · 2025-02-27T07:43:30Z

Adds a header comment to the pageSnapshot file that should help LLMs understand its contents better.

Copilot

PR Overview

This PR enhances the clarity of pageSnapshot artifacts by adding a header comment with contextual test information.

Updates the snapshot file header generation in the Playwright artifacts recorder.
Adjusts the test expectation to accommodate the new header comment.

Reviewed Changes

File	Description
tests/playwright-test/playwright.artifacts.spec.ts	Updates test expectation to match header additions
packages/playwright/src/index.ts	Prepends a header comment to the ARIA snapshot output

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (2)

tests/playwright-test/playwright.artifacts.spec.ts:463

[nitpick] The test expectation is hard-coded with a specific header format; if the header generation logic changes (e.g., dynamic values like test title or status), the test may fail. Consider constructing the expected header dynamically or using a partial match to reduce fragility.

expect(fs.readFileSync(testInfo.outputPath('test-results', 'artifacts-failing', 'test-failed-1.snapshot.yml'), { encoding: 'utf-8' })).toEqual(`

packages/playwright/src/index.ts:634

[nitpick] Ensure that concatenating the header with the aria snapshot preserves the intended formatting in all cases, especially when the ariaSnapshot content might be empty or formatted unexpectedly.

await fs.promises.writeFile(path, header + ariaSnapshot);

github-actions · 2025-02-27T08:30:08Z

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:403:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18
⚠️ [playwright-test] › tests/ui-mode-test-watch.spec.ts:145:5 › should watch all @windows-latest-node18-1

38034 passed, 650 skipped
✔️✔️✔️

Merge workflow run.

Adam Gastineau (agg23) · 2025-02-27T13:08:53Z

 this._pageSnapshotRecorder = new SnapshotRecorder(this, pageSnapshot, 'pageSnapshot', 'text/plain', '.snapshot.yml', async (page, path) => {
 const ariaSnapshot = await page.locator('body').ariaSnapshot({ timeout: 5000 });
- await fs.promises.writeFile(path, ariaSnapshot);
+ const header = `# ${this._testInfo.titlePath.join(' >> ')}\n# ARIA snapshot of the page contents after test ${this._testInfo._isFailure() ? 'failed' : 'finished'}\n`;


I don't think I like this. Why can't we inject the titlePath into the prompt we generate, rather than including it with the page snapshot? This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.

my hope is to help this usecase:
Copilot is already picking up the file. the title path should help it pick the right one, and the aria snapshot mention should help it understand that it's seeing ARIA roles in the tree.
This is making the assumption that the page snapshot is exclusively for AI, which I don't think is necessarily going to remain the case.
Every other consumer is free to delete the leading comment lines. I think that's fine.

Simon Knott (Skn0tt) · 2025-02-28T07:47:03Z

superceded by #34963, all info is in the file path now.

feat(pageSnapshot): add header comment with info
c66c4fd

Simon Knott (Skn0tt) requested review from Adam Gastineau (agg23) and Copilot February 27, 2025 07:43

Simon Knott (Skn0tt) self-assigned this Feb 27, 2025