[py] add Selenium Manager integration tests

[titusfortner]

🔗 Related Issues

Implements #16741 for Python
Part of #17539

💥 What does this PR do?

• Adds integration test coverage of Selenium Manager via Driver Finder class

🔧 Implementation Notes

• Because it goes through Driver Finder, the tests pass with and without pin_browsers being set by taking different paths (or skipping the cache location validation)

• Selected by tag (--test_tag_filters=se-manager) rather than per-browser target names, so all applicable browsers run in a single invocation per OS (3 jobs, not browser×OS).

🤖 AI assistance

• No substantial AI assistance used
• AI assisted (complete below)
  • Tool(s): Claude Code
  • What was generated: the test, BUILD targets, and CI job
  • I reviewed all AI output and can explain the change

💡 Additional Considerations

• Will need to be deconflicted with [py] run affected python tests per-browser on PR, full suite on schedule #17643 depending on which merges first

🔄 Types of changes

• New feature (non-breaking change which adds functionality and tests!)

[qodo-code-review]

Review Summary by Qodo

Add Selenium Manager integration tests for Python

🧪 Tests ✨ Enhancement

Walkthroughs

Description

• Adds comprehensive Selenium Manager integration tests via DriverFinder
• Tests cover driver and browser path resolution with caching
• Supports both pinned and unpinned browser configurations
• Integrates tests into CI pipeline with OS-specific matrix jobs

Diagram

flowchart LR
  A["New Test File"] -->|"Tests DriverFinder"| B["Driver Path Resolution"]
  A -->|"Tests DriverFinder"| C["Browser Path Resolution"]
  B -->|"Validates"| D["Executable Paths"]
  C -->|"Validates"| D
  B -->|"Tests Caching"| E["SE_CACHE_PATH"]
  C -->|"Tests Caching"| E
  F["CI Configuration"] -->|"Runs Tests"| G["Multi-OS Jobs"]
  G -->|"Uses Tag Filter"| H["se-manager"]

Loading

File Changes

1. py/test/selenium/webdriver/common/driver_finder_tests.py 🧪 Tests +103/-0

Selenium Manager integration tests via DriverFinder

• New test file with 5 integration tests for Selenium Manager via DriverFinder
• Tests driver path resolution, browser path resolution, and cache behavior
• Includes conditional skips for pinned runs, Safari, and platform-specific cases
• Uses pytest fixtures for browser_name, pinned status, and driver_finder setup

py/test/selenium/webdriver/common/driver_finder_tests.py

---

2. .github/workflows/ci-python.yml ⚙️ Configuration changes +16/-0

Add Selenium Manager CI job matrix

• Adds new selenium-manager-tests job to CI pipeline
• Runs tests across three OS platforms (ubuntu, windows, macos)
• Uses --test_tag_filters=se-manager to select Selenium Manager tests
• Configured with --pin_browsers=false to test Selenium Manager functionality

.github/workflows/ci-python.yml

---

3. py/BUILD.bazel ⚙️ Configuration changes +29/-5

Configure Bazel targets for driver finder tests

• Defines DRIVER_FINDER_TESTS constant containing the new test file
• Excludes driver_finder tests from common test suites to prevent duplication
• Creates new test--driver-finder targets for each browser
• Tags new targets with se-manager for selective test execution

py/BUILD.bazel

---

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (1) 📘 Rule violations (0) 📎 Requirement gaps (0) 🎨 UX issues (0)

1. Driver count not validated 🐞 Bug ≡ Correctness

Description

The browser_name fixture claims it requires a single --driver/--browser but only checks for presence
and then silently uses the first entry. Because the CLI option supports multiple values, running
with more than one driver will execute the suite against an arbitrary first browser and ignore the
rest.

Code

py/test/selenium/webdriver/common/driver_finder_tests.py[R30-34]

+def browser_name(request):
+    drivers = request.config.option.drivers
+    if not drivers:
+        pytest.skip("Selenium Manager tests require a single --driver/--browser")
+    return drivers[0].lower()

Evidence

The test expects exactly one driver, but the global pytest options explicitly allow multiple via
action=append, so the current fixture can silently pick the wrong browser.

py/test/selenium/webdriver/common/driver_finder_tests.py[29-35]
py/conftest.py[118-134]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`browser_name` skips only when no `--driver/--browser` is provided, but the option supports multiple values. The fixture then uses `drivers[0]`, which can silently run the suite against the wrong browser.

### Issue Context
In `py/conftest.py`, `--driver`/`--browser` uses `action="append"`, so users/CI can pass multiple drivers.

### Fix Focus Areas
- py/test/selenium/webdriver/common/driver_finder_tests.py[29-35]

### Suggested change
Update the fixture to require exactly one driver:
- `if not drivers or len(drivers) != 1: pytest.skip(...)` (or `pytest.fail(...)` if you want misconfiguration to hard-fail)
- Include the received value(s) in the message for easier debugging.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Quoted driver path breaks ✓ Resolved 🐞 Bug ☼ Reliability

Description

The driver_finder fixture strips quotes from --browser-binary but does not strip quotes from
--driver-binary before passing it into the Service. If the driver path is single-quoted (a pattern
already handled elsewhere in this repo), DriverFinder will treat it as a literal path including
quotes and fail file validation.

Code

py/test/selenium/webdriver/common/driver_finder_tests.py[R47-52]

+    if request.config.option.binary:
+        options.binary_location = _resolve_bazel_path(request.config.option.binary).strip("'")
+
+    executable = request.config.option.executable
+    service = module.service.Service(executable_path=_resolve_bazel_path(executable) if executable else None)
+

Evidence

DriverFinder explicitly treats any non-empty Service.path as an override and requires it to be a
real file; leaving quotes on the path will cause Path(...).is_file() to fail. The repo already
strips quotes from executable paths elsewhere, indicating quoted paths are expected in some
configurations.

py/test/selenium/webdriver/common/driver_finder_tests.py[46-53]
py/conftest.py[225-240]
py/selenium/webdriver/common/service.py[73-94]
py/selenium/webdriver/common/driver_finder.py[56-75]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`driver_finder` normalizes `--browser-binary` by stripping surrounding single quotes, but it passes `--driver-binary` through without stripping. When `Service.path` contains quotes, `DriverFinder` validates it with `Path(path).is_file()` and will fail.

### Issue Context
The existing test harness already anticipates quoted executable paths and strips them in other code paths.

### Fix Focus Areas
- py/test/selenium/webdriver/common/driver_finder_tests.py[46-53]

### Suggested change
Normalize the executable path the same way as the browser binary:
```py
executable = request.config.option.executable
exe_path = _resolve_bazel_path(executable).strip("'") if executable else None
service = module.service.Service(executable_path=exe_path)
```
(Optionally, consider a small helper to avoid duplicating quote-stripping logic.)

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[AutomatedTester]

Just the one comment which shouldn't block, just want consistency.

[qodo-code-review]

Code review by qodo was updated up to the latest commit b28e3cf