fix: guard against empty dict in clean_incrementing_reverse

[Copilot]

On fresh installs, pv_today history is often < 24 hours old (age.days == 0), producing an empty mdata dict that gets passed to clean_incrementing_reverse(), which immediately crashes on max(data).

Change

Added an early-return guard in utils.py:

def clean_incrementing_reverse(data, max_increment=0):
    new_data = {}
    if not data:
        return new_data  # nothing to clean
    length = max(data) + 1

Returning {} is semantically correct — the caller (minute_data) already handles empty mdata gracefully, and there is nothing to smooth when no data exists.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>clean_incrementing_reverse crashes with ValueError: max() arg is empty when history < 24 hours</issue_title>
<issue_description>## Description

clean_incrementing_reverse() in utils.py crashes with ValueError: max() arg is an empty sequence when called with an empty dict. This happens during the SolarAPI / PV calibration flow on a fresh install where pv_today history is less than 24 hours old.

Stack trace

File "/config/solcast.py", line 967, in fetch_pv_forecast
    pv_forecast_minute, pv_forecast_minute10, pv_forecast_data = self.pv_calibration(...)
File "/config/solcast.py", line 657, in pv_calibration
    pv_today_hist = self.base.minute_data_import_export(days, self.now_utc, "pv_today", required_unit="kWh", pad=False)
File "/config/fetch.py", line 578, in minute_data_import_export
    import_today, _ = minute_data(...)
File "/config/utils.py", line 624, in minute_data
    mdata = clean_incrementing_reverse(mdata, max_increment)
ValueError: max() arg is an empty sequence

Root cause

In minute_data_import_export(), when pad=False is passed (as pv_calibration does), the function uses age.days instead of the requested max_days_previous:

age = now_utc - last_updated_time
age_days = min(age_days, age.days)

If the oldest history entry is less than 24 hours old, age.days == 0. This is passed to minute_data() which computes minute_min = 0 and minute_max = 0, so no history entries fall within bounds and mdata remains {}.

mdata is then passed to clean_incrementing_reverse(), which immediately calls max(data) on the empty dict:

def clean_incrementing_reverse(data, max_increment=0):
    new_data = {}
    length = max(data) + 1  # <-- crashes here when data is {}

Steps to reproduce

1. Fresh Predbat install with Solcast enabled
2. pv_today sensor (e.g. sensor.foxess_solar_energy_today) has less than 24 hours of history
3. Predbat runs its first cycle with SolarAPI → crash every cycle until history crosses 24h

Suggested fix

Add an empty-data guard at the top of clean_incrementing_reverse():

def clean_incrementing_reverse(data, max_increment=0):
    new_data = {}
    if not data:
        return new_data
    length = max(data) + 1

Workaround applied

I patched utils.py directly on the addon container by adding an empty-data guard to clean_incrementing_reverse():

def clean_incrementing_reverse(data, max_increment=0):
    new_data = {}
    if not data:
        return new_data  # <-- added guard
    length = max(data) + 1

This works because:

• The function is only called from minute_data() when clean_increment=True and data is backwards (historical). When mdata is empty, there is nothing to clean — returning an empty dict is the correct no-op behaviour.
• The caller (minute_data) already handles empty mdata gracefully downstream — it simply means no historical data was available for that time window.
• After applying this patch and restarting Predbat, the SolarAPI crash stopped immediately. Predbat completed its cycle successfully with status "Demand" and no errors.

Note: This patch will be overwritten on the next Predbat addon update since the addon replaces all source files. However, the crash condition is only triggered during the first ~24 hours after a fresh install (while age.days == 0), so it is unlikely to recur after an update on an established system.

Environment

• Predbat v8.34.5 (standalone addon)
• Inverter: FoxESS KH10-AUX (foxess_modbus v1.13.6)
• Solcast: BJReplay/ha-solcast-solar v4.5.1
• Home Assistant 2026.3.2
  </issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes clean_incrementing_reverse crashes with ValueError: max() arg is empty when history < 24 hours #3601

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

Pull request overview

Fixes a startup crash in Predbat’s historical minute-data smoothing path by making clean_incrementing_reverse() a no-op when called with empty data (which can occur on fresh installs when history age results in days=0).

Changes:

• Add an early return in clean_incrementing_reverse() when data is empty to avoid max(data) on an empty dict.

[Copilot]

This guard fixes the crash, but there’s no regression test covering the specific failure mode (clean_increment=True with days=0 / empty mdata). Please add a unit test (e.g., in apps/predbat/tests/test_minute_data.py) that exercises minute_data() with days=0 and clean_increment=True and asserts it returns an empty dict without raising.