Potential fix for code scanning alert no. 115: Clear-text logging of sensitive information

[fiftin]

Potential fix for https://github.com/semaphoreui/semaphore/security/code-scanning/115

General fix: ensure all log messages are sanitized before emission so secrets (especially credentials embedded in URLs like https://user:pass@host/...) are redacted.

Best single fix without changing functionality: update services/tasks/LocalJob.go so Log(msg string) passes a sanitized message to t.Logger.Log. Add a helper that redacts userinfo in http:// / https:// URLs from user[:password]@ to ***@. This preserves operational logging context while removing sensitive data. No behavior change outside log content redaction.

Changes needed:

• File: services/tasks/LocalJob.go
  • Add import: regexp
  • Add helper method on LocalJob (or private function) to sanitize log message.
  • Update Log(msg string) to call sanitizer before logging.
• No changes are required in db/Repository.go for this alert, since the sink-side sanitization addresses both variants centrally.

Suggested fixes powered by Copilot Autofix. Review carefully before merging.

[cursor]

Security review (automation)

Scope: services/tasks/LocalJob.go — sanitizeLogMessage + wiring in Log.

Assessment: No medium / high / critical vulnerabilities introduced by this change.

• What changed: Task log lines passed through LocalJob.Log now replace http(s)://…userinfo@ with http(s)://***@, which addresses clear-text logging of embedded HTTP(S) URL credentials for the common lowercase-scheme form.
• Residual risk (low): The pattern is case-sensitive (https? only), so mixed-case schemes (e.g. HTTPS://user:pass@…) would not be redacted; many tools normalize to lowercase, so practical exposure is limited. Non-HTTP URL schemes (e.g. git@, ftp://) are unchanged — pre-existing breadth of logging, not expanded by this PR.
• Out of scope for this diff: Call sites that invoke t.Logger.Log directly (e.g. git client helpers) bypass LocalJob.Log; that behavior predates this PR.

Slack-style summary

• PR: Potential fix for code scanning alert 115 (clear-text logging of sensitive info in LocalJob.Log).
• Outcome: No medium+ security findings on the added/modified code; change is a targeted redaction improvement with minor edge-case coverage limits noted above.

  

_{Sent by Cursor Automation: Find vulnerabilities}