bug(scheduler): _normalize_dow_field does not handle range-step tokens like '1-7/2' — day 7 never normalized to 0

[mrveiss]

Bug

_normalize_dow_field in trigger_service.py (added by #4918) correctly handles scalars (7→0) and pure ranges (1-7→1-6,0) but does not handle range-step tokens like 1-7/2.

The regex r'(\d+)-(\d+)' does not match 1-7/2 (because of the /2 suffix). The token falls through to the scalar branch, which returns 1-7/2 unchanged. When _parse_cron_field processes 1-7/2, it produces {1, 3, 5, 7} — the 7 is never mapped to Sunday (0), so the cron expression fires on an unrecognized day.

Reproducer

_normalize_dow_field('1-7/2')  # returns '1-7/2' unchanged — should return '1-6/2,0' or equivalent

Fix

Extend _replace_token to match the pattern r'(\d+)-(\d+)/(\d+)' and replace hi=7 with 6 in the range portion, then add ,0 to include Sunday.

File

autobot-backend/services/trigger_service.py — _normalize_dow_field / _replace_token

[mrveiss]

✅ Closed — range-step normalization already shipped

Code: autobot-backend/services/trigger_service.py:265-291 (_replace_token inside _normalize_dow_field)

Current implementation handles all four cases from the issue:

• Scalar: '7' → '0'
• Range: '1-7' → '1-6,0', '0-7' → '0-6'
• Range-step: '1-7/2' → '1-6/2,0', '0-7/2' → '0-6/2'
• Step: '*/7' left unchanged

The regex from the issue description is now present: re.fullmatch(r'(\d+)-(\d+)/(\d+)', token) at L267, triggered BEFORE the plain range match so range-steps are handled first.

Spot-test verification:

_replace_token('1-7/2')  # → '1-6/2,0' ✓
_replace_token('1-7')    # → '1-6,0'   ✓
_replace_token('7')      # → '0'       ✓

Docstring on _normalize_dow_field explicitly lists range-steps (1-7/2->1-6/2,0) as supported.

Acceptance Criteria Met:

• ✅ '1-7/2' normalizes correctly (no longer falls through to scalar branch).
• ✅ All other token shapes preserved (scalar, range, step).
• ✅ Docstring lists range-step as supported.