Not able to identify the field in the modal

[rnyadav3]

1. Setup Test Environment

Create virtual environment

python -m venv venv-hercules
source venv-hercules/bin/activate

Install TestZeus-Hercules

pip install testzeus-hercules
2. Create Feature File (features/modal_login_test.feature)
Feature: Modal Login Test
Test login functionality with modal dialog

Background:
  Given I am on a login page with modal dialog authentication

@modal @login
Scenario: Login with modal dialog
  When I enter email "test@example.com"
  And I click the "Next" button

3. Create Python Runner Script (run_test.py)
   import subprocess
   from pathlib import Path

cmd = [
"testzeus-hercules",
"--input-file", "features/modal_login_test.feature",
"--output-path", "reports/output",
"--project-base", ".",
"--llm-model", "gpt-4o",
"--llm-model-api-key", "your-openai-api-key",
"--llm-model-api-type", "openai",
"--browser-channel", "chrome"
]

Disable headless mode for proper rendering

import os
env = os.environ.copy()
env["HEADLESS"] = "false"

subprocess.run(cmd, env=env)
4. Test Website Requirements
- Any website with modal/popup login dialog
- Modal should contain email input field
- Modal should overlay main page content
5. Run the Test
python run_test.py

Expected Behavior

• Framework should detect the email input field in the modal dialog
• Framework should maintain focus on the modal dialog during text input
• Email should be successfully entered into the input field
• Test should proceed to next step

Actual Behavior

• Framework correctly detects email input field ✅
• Framework loses focus during text input operation ❌
• Text input is attempted on main page content instead of modal ❌
• Email field remains empty ❌
• Test fails silently or proceeds with empty field ❌

Evidence

1. Element Detection Success:
   [
   {"md":"2105","tag":"input","role":"textbox","name":"Email"},
   {"md":"2107","tag":"button","role":"button","name":"Next Button"}
   ]

2. Screenshots Pattern:

• entertext_start_*.png: Shows modal dialog visible and focused
• entertext_end_*.png: Shows main page content (focus lost)

3. Console Logs Pattern:
   [INFO] Registered tool: bulk_enter_text
   [INFO] get_input_fields detected modal elements correctly
   [WARNING] Failed to wait for networkidle: Timeout exceeded

Root Cause Analysis

The framework's bulk_enter_text tool appears to lose DOM focus when
interacting with elements inside modal dialogs or popup overlays. The
element detection works correctly, but the actual text input operation
targets the wrong DOM context.

Impact

• Critical: Prevents testing of any login flows using modal dialogs
• Widespread: Affects any test scenario with popup/overlay/modal elements
• Silent Failure: Framework doesn't report the focus loss as an error

Workaround

None found. The issue appears to be in the core element interaction
mechanism.

Additional Notes

• Issue only occurs with modal/popup elements
• Standard page elements work correctly
• Framework successfully detects elements but fails during interaction
• Disabling headless mode doesn't resolve the issue
• Problem persists across different browser channels

Test Case for Verification

This issue can be reproduced with any website that uses modal dialogs for
authentication. Common examples include:

• Login modals that overlay the main page
• Popup authentication forms
• Any overlay dialog with input fields

[krishnaharshap]

Hi @rnyadav3 , Great findings! I've successfully reproduced the issue where the modal field isn't being identified. After investigating, I believe the problem stems from dynamic rendering or timing issues. To address this, I propose:

• Enhanced Element Locator Strategy: Utilize more robust locators like relative XPath or CSS selectors tailored to the modal's structure.
• Increased Wait Times: Implement explicit waits to ensure the modal and its elements are fully loaded before interaction.
• Modal Detection Mechanism: Add checks to confirm the modal's presence before attempting interactions.

May I proceed with implementing this fix? Thank you for providing a clear and reproducible example!

[shriyanshagnihotri]

@krishnaharshap and @rnyadav3 : can you try the same with 0.2.0 release, we had a bug with text field that is been fixed by one contributor.

try it else, feel free to put up a PR if you find a solution. Or we can discuss the same over Slack channel and find out where its breaking.

[krishnaharshap]

Hi @shriyanshagnihotri , thanks for your suggestion to try version 0.2.0!
I’ll install and test with 0.2.0 in the same environment (modal login scenario) you provided.
I’ll capture logs, screenshots and a simple version-matrix (package version + browser + OS) and report back.
Thanks again for providing a clear repro case. I’d like to help move this forward!