[rb] trim whitespace around NO_PROXY entries

[aguspe]

🔗 Related Issues

#17460

💥 What does this PR do?

This commit fixes the bug #17460 where NO_PROXY / no_proxy entries were parsed without trimming surrounding whitespace, so values like localhost, 127.0.0.1 would not match the target host and Selenium would not bypass the proxy as expected. I also updated the tests.

🔧 Implementation Notes

Two sites were splitting no_proxy differently:

• remote/http/default.rb used split(',') without trimming, so entries with leading spaces never matched the server host.
• common/proxy.rb#as_json used split(', '), which failed for comma-only separated values like proxy_url,localhost.

Both now use split(',').map(&:strip).reject(&:empty?) so the parsing is consistent and tolerant of surrounding whitespace.

🔄 Types of changes

• Bug fix (backwards compatible)

[qodo-code-review]

Review Summary by Qodo

Trim whitespace around NO_PROXY entries in Ruby

🐞 Bug fix

Walkthroughs

Description

• Fix NO_PROXY parsing to trim whitespace around entries
• Standardize no_proxy splitting across two Ruby files
• Handle both comma-separated and comma-space-separated formats
• Add comprehensive test coverage for whitespace handling

Diagram

flowchart LR
  A["NO_PROXY string with whitespace"] -->|split and trim| B["Clean array of hosts"]
  B -->|proxy matching logic| C["Correct proxy bypass decision"]
  D["common/proxy.rb"] -->|updated parsing| B
  E["remote/http/default.rb"] -->|updated parsing| B

Loading

File Changes

1. rb/lib/selenium/webdriver/common/proxy.rb 🐞 Bug fix +1/-1

Standardize no_proxy parsing with whitespace trimming

• Changed no_proxy string splitting from split(', ') to
 split(',').map(&:strip).reject(&:empty?)
• Handles both comma-only and comma-space-separated formats consistently
• Removes empty entries after splitting and trimming

rb/lib/selenium/webdriver/common/proxy.rb

---

2. rb/lib/selenium/webdriver/remote/http/default.rb 🐞 Bug fix +1/-1

Add whitespace trimming to proxy bypass logic

• Updated use_proxy? method to trim whitespace from no_proxy entries
• Changed from split(',') to split(',').map(&:strip).reject(&:empty?)
• Ensures consistent parsing with common/proxy.rb

rb/lib/selenium/webdriver/remote/http/default.rb

---

3. rb/spec/unit/selenium/webdriver/remote/capabilities_spec.rb 🧪 Tests +12/-0

Add no_proxy whitespace handling tests

• Added test for comma-only separated no_proxy values
• Added test for whitespace trimming around entries
• Validates that both formats are correctly converted to arrays

rb/spec/unit/selenium/webdriver/remote/capabilities_spec.rb

---

View more (1)

4. rb/spec/unit/selenium/webdriver/remote/http/default_spec.rb 🧪 Tests +14/-0

Add proxy bypass whitespace trimming tests

• Added test for trimming whitespace around multiple entries
• Added test for trimming leading whitespace on single entry
• Validates proxy bypass works correctly with whitespace variations

rb/spec/unit/selenium/webdriver/remote/http/default_spec.rb

---

[qodo-code-review]

Code Review by Qodo

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

Context used

✅ Tickets: 🎫 [🐛 Bug]: NO_PROXY/no_proxy entries are parsed without trimming whitespace

1. Array no_proxy unsupported 🐞 Bug ≡ Correctness

Description

Remote::Http::Default#use_proxy? unconditionally calls split on proxy.no_proxy, which raises
NoMethodError if no_proxy is provided as an Array. This is inconsistent with Proxy#as_json and
capabilities behavior that treat array no_proxy values as valid.

Code

rb/lib/selenium/webdriver/remote/http/default.rb[153]

Evidence

use_proxy? calls split on proxy.no_proxy without checking its type, while Proxy#as_json
explicitly supports non-String no_proxy values (e.g., arrays), and the capabilities spec asserts
array no_proxy is valid input.

rb/lib/selenium/webdriver/remote/http/default.rb[149-167]
rb/lib/selenium/webdriver/common/proxy.rb[143-159]
rb/spec/unit/selenium/webdriver/remote/capabilities_spec.rb[44-48]

Agent prompt

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

### Issue description
`Selenium::WebDriver::Remote::Http::Default#use_proxy?` assumes `proxy.no_proxy` is a `String` and calls `.split(',')` on it. If a `Proxy` is configured with `no_proxy` as an `Array` (which other parts of the Ruby bindings accept), this will raise at runtime.

### Issue Context
The PR makes `Proxy#as_json` accept both comma-separated strings and arrays for `noProxy`, but the HTTP client bypass logic isn’t type-aligned.

### Fix Focus Areas
- rb/lib/selenium/webdriver/remote/http/default.rb[149-167]

### Implementation sketch
- Normalize `proxy.no_proxy` into a `hosts` array via:
 - `String`: `split(',').map(&:strip).reject(&:empty?)`
 - `Array`: `map(&:to_s).map(&:strip).reject(&:empty?)`
 - else: treat as empty
- Run `.any?` over `hosts` instead of calling `.split` directly.

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

---

Previous review results

Review updated until commit d7c4046

Results up to commit 92ddd47

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

1. Array no_proxy unsupported 🐞 Bug ≡ Correctness

Description

Remote::Http::Default#use_proxy? unconditionally calls split on proxy.no_proxy, which raises
NoMethodError if no_proxy is provided as an Array. This is inconsistent with Proxy#as_json and
capabilities behavior that treat array no_proxy values as valid.

Code

rb/lib/selenium/webdriver/remote/http/default.rb[153]

Evidence

use_proxy? calls split on proxy.no_proxy without checking its type, while Proxy#as_json
explicitly supports non-String no_proxy values (e.g., arrays), and the capabilities spec asserts
array no_proxy is valid input.

rb/lib/selenium/webdriver/remote/http/default.rb[149-167]
rb/lib/selenium/webdriver/common/proxy.rb[143-159]
rb/spec/unit/selenium/webdriver/remote/capabilities_spec.rb[44-48]

Agent prompt

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

### Issue description
`Selenium::WebDriver::Remote::Http::Default#use_proxy?` assumes `proxy.no_proxy` is a `String` and calls `.split(',')` on it. If a `Proxy` is configured with `no_proxy` as an `Array` (which other parts of the Ruby bindings accept), this will raise at runtime.

### Issue Context
The PR makes `Proxy#as_json` accept both comma-separated strings and arrays for `noProxy`, but the HTTP client bypass logic isn’t type-aligned.

### Fix Focus Areas
- rb/lib/selenium/webdriver/remote/http/default.rb[149-167]

### Implementation sketch
- Normalize `proxy.no_proxy` into a `hosts` array via:
 - `String`: `split(',').map(&:strip).reject(&:empty?)`
 - `Array`: `map(&:to_s).map(&:strip).reject(&:empty?)`
 - else: treat as empty
- Run `.any?` over `hosts` instead of calling `.split` directly.

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

---

[qodo-code-review]

Persistent review updated to latest commit ec1c324

[qodo-code-review]

Persistent review updated to latest commit d7c4046