prime lazy loaded accessibility tree

[tbtmuse]

WebView accessibility tree populates asynchronously on first access. The second call always seemed to work so after some research it seems this is the case. This is the best solution I could come up with.

First call primes the tree, second call returns complete cached data.

[coderabbitai]

Walkthrough

The change updates AndroidRobot.getElementsOnScreen in src/android.ts to await getUiAutomatorDump() before parsing the UiAutomator XML. This primes the lazy-loaded accessibility tree prior to the existing XML parsing and element collection logic. No other parsing or collection behavior was modified. The change adds three lines of code.

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely summarizes the main change: priming the lazy-loaded accessibility tree before parsing in the AndroidRobot.getElementsOnScreen function.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Description check	✅ Passed	The pull request description clearly describes the problem (WebView accessibility tree populates asynchronously) and the implemented solution (priming the tree with a first call before the second call returns cached data), which aligns with the changeset in src/android.ts.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (1)

src/android.ts (1)

352-353: Avoid parsing XML in the priming call.

Line [353] primes via getUiAutomatorXml() but discards the parsed object. This adds unnecessary parse/allocation work on every call. Prime with dump only, then parse once for the returned tree.

Proposed diff

 public async getElementsOnScreen(): Promise<ScreenElement[]> {
 	// Prime lazy loaded accessibility tree
-	await this.getUiAutomatorXml();
+	await this.getUiAutomatorDump();

 	const parsedXml = await this.getUiAutomatorXml();
 	const hierarchy = parsedXml.hierarchy;
 	const elements = this.collectElements(hierarchy.node);
 	return elements;
 }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@src/android.ts` around lines 352 - 353, The current priming call uses
getUiAutomatorXml() which parses and allocates the XML even though its parsed
result is discarded; change the priming step to use a dump-only path (e.g., call
a raw dump method such as getUiAutomatorDump() or change getUiAutomatorXml to
accept a flag like skipParse/parse=false) so the code only retrieves the XML
string without parsing, and then parse the XML exactly once when you actually
need the accessibility tree (the existing parse logic can remain in the code
that consumes the returned tree).

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@src/android.ts`:
- Around line 352-353: The current priming call uses getUiAutomatorXml() which
parses and allocates the XML even though its parsed result is discarded; change
the priming step to use a dump-only path (e.g., call a raw dump method such as
getUiAutomatorDump() or change getUiAutomatorXml to accept a flag like
skipParse/parse=false) so the code only retrieves the XML string without
parsing, and then parse the XML exactly once when you actually need the
accessibility tree (the existing parse logic can remain in the code that
consumes the returned tree).

---

ℹ️ Review info

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Cache: Disabled due to data retention organization setting

Knowledge base: Disabled due to data retention organization setting

📥 Commits

Reviewing files that changed from the base of the PR and between 93351ad and bc470bd.

📒 Files selected for processing (1)

• src/android.ts

[coderabbitai]

🧹 Nitpick comments (1)

src/android.ts (1)

351-359: Consider expanding the comment to explain the WebView-specific behavior.

The current comment mentions "Prime lazy loaded accessibility tree" but doesn't explain why two calls are needed. Future maintainers may not understand the double-call pattern or might "optimize" it away.

📝 Suggested comment improvement

 public async getElementsOnScreen(): Promise<ScreenElement[]> {
-	// Prime lazy loaded accessibility tree (raw dump, no parsing)
+	// WebView accessibility tree populates asynchronously on first access,
+	// returning incomplete data. A second call consistently returns the
+	// complete cached tree, so we prime here before the actual fetch.
 	await this.getUiAutomatorDump();

 	const parsedXml = await this.getUiAutomatorXml();

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@src/android.ts` around lines 351 - 359, The comment for getElementsOnScreen
should explain the WebView-specific double-call pattern: the first call to
getUiAutomatorDump() is intentional to prime lazy-loaded accessibility trees
(notably WebViews that defer accessibility node population until a raw dump is
requested), and the subsequent call to getUiAutomatorXml() reads the
now-populated parsed XML; update the comment above getElementsOnScreen to
explicitly state this reason and that both getUiAutomatorDump() and
getUiAutomatorXml() are required and must not be removed or merged.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@src/android.ts`:
- Around line 351-359: The comment for getElementsOnScreen should explain the
WebView-specific double-call pattern: the first call to getUiAutomatorDump() is
intentional to prime lazy-loaded accessibility trees (notably WebViews that
defer accessibility node population until a raw dump is requested), and the
subsequent call to getUiAutomatorXml() reads the now-populated parsed XML;
update the comment above getElementsOnScreen to explicitly state this reason and
that both getUiAutomatorDump() and getUiAutomatorXml() are required and must not
be removed or merged.

---

ℹ️ Review info

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Cache: Disabled due to data retention organization setting

Knowledge base: Disabled due to data retention organization setting

📥 Commits

Reviewing files that changed from the base of the PR and between bc470bd and 0796367.

📒 Files selected for processing (1)

• src/android.ts

[gmegidish]

Holy wow! I was not aware of this! Let me try this out to reproduce. Alternatively, we can use cdp to get the Chrome that's on screen. That will also allow you do to other things, like list DOM that is not visible on screen and directly click on such elements (or inject js). Thanks for this PR, I'll take a look later to reproduce this.

[tbtmuse]

@gmegidish awesome, looking forward to hearing how it goes!

[tbtmuse]

@gmegidish did you manage to have a look at this, I'm running this locally and have had no issues since implementing this.