Fix year-round distribution merge for D1.11 and add regression test

Author: javaDevJT

custom_components/dte_rates/pdf_parser.py

@@ -271,13 +271,14 @@ def _parse_rate_section(section: str) -> tuple[list[SeasonalPeriodRate], list[st
         component_label = _normalize_key(value_match.group("label"))
         value = _to_decimal(value_match.group("value"))
         unit = value_match.group("unit").lower().replace(" ", "")
+        target_season = "year_round" if "year_round" in component_label else season
 
         if "¢perkwh" in unit:
             period_name = _period_from_component_label(component_label) or current_period
-            key = (season, period_name)
+            key = (target_season, period_name)
             per_kwh[key][component_label] = value
         else:
-            monthly_by_season[season][component_label] = value
+            monthly_by_season[target_season][component_label] = value
 
     periods: list[SeasonalPeriodRate] = []
     for (period_season, period_name), kwh_components in per_kwh.items():
@@ -292,6 +293,7 @@ def _parse_rate_section(section: str) -> tuple[list[SeasonalPeriodRate], list[st
             )
         )
 
+    periods = _merge_base_components(periods, monthly_by_season)
     return periods, notes
 
 
@@ -308,3 +310,56 @@ def _window_for_period(period_name: str, season: str, specs: dict[str, _WindowSp
         hour_ranges=list(spec.hour_ranges),
         months=_months_for(season),
     )
+
+
+def _merge_base_components(
+    periods: list[SeasonalPeriodRate],
+    monthly_by_season: dict[str, dict[str, Decimal]],
+) -> list[SeasonalPeriodRate]:
+    """Apply base all_kwh/year_round components to specific TOU periods.
+
+    Example: D1.11 has Distribution kWh (Year-round). That value should be part
+    of both peak and off-peak totals instead of existing in a separate all_kwh-only
+    period that is not the active TOU period.
+    """
+    by_key: dict[tuple[str, str], SeasonalPeriodRate] = {
+        (p.season_name, p.period_name): p for p in periods
+    }
+    has_specific_periods = any(p.period_name != "all_kwh" for p in periods)
+    if not has_specific_periods:
+        return periods
+
+    year_round_base = by_key.get(("year_round", "all_kwh"))
+
+    merged: list[SeasonalPeriodRate] = []
+    for period in periods:
+        if period.period_name == "all_kwh":
+            continue
+
+        base_for_season = by_key.get((period.season_name, "all_kwh"))
+        merged_kwh = dict(period.components.per_kwh)
+
+        for source in (year_round_base, base_for_season):
+            if source is None:
+                continue
+            for key, value in source.components.per_kwh.items():
+                merged_kwh.setdefault(key, value)
+
+        merged_monthly: dict[str, Decimal] = {}
+        for source_monthly in (
+            monthly_by_season.get("year_round", {}),
+            monthly_by_season.get(period.season_name, {}),
+        ):
+            for key, value in source_monthly.items():
+                merged_monthly.setdefault(key, value)
+
+        merged.append(
+            SeasonalPeriodRate(
+                season_name=period.season_name,
+                period_name=period.period_name,
+                components=PriceComponents(per_kwh=merged_kwh, monthly=merged_monthly),
+                window=period.window,
+            )
+        )
+
+    return merged

tests/test_pdf_parser.py

@@ -78,6 +78,41 @@ def __init__(self, _bytes):
         self.pages = [_FakePage()]
 
 
+REAL_D111_EXCERPT = """
+Time of Day 3 p.m. - 7 p.m. Standard Base Rate (D1.11)
+Peak hours: Monday-Friday 3 p.m. to 7 p.m.
+Power Supply Charges:
+June through September
+ All peak kWh
+  Capacity Energy .................. 5.360¢ per kWh
+  Non-Capacity Energy ............... 9.047¢ per kWh
+ All off-peak kWh
+  Capacity Energy ................... 3.240¢ per kWh
+  Non-Capacity Energy ............... 5.469¢ per kWh
+October through May
+ All peak kWh
+  Capacity Energy ................... 3.839¢ per kWh
+  Non-Capacity Energy .............. 6.480¢ per kWh
+ All off-peak kWh
+  Capacity Energy ................... 3.240¢ per kWh
+  Non-Capacity Energy ............... 5.469¢ per kWh
+Delivery Charges:
+ Service Charge ....................... $8.50 per month
+ RIA Credit* ......................... ($8.50) per month
+ Distribution kWh (Year-round) ......... 9.726¢ per kWh
+"""
+
+
+class _RealD111Page:
+    def extract_text(self):
+        return REAL_D111_EXCERPT
+
+
+class _RealD111Reader:
+    def __init__(self, _bytes):
+        self.pages = [_RealD111Page()]
+
+
 def test_parse_rate_card_extracts_rates_and_effective_date(monkeypatch):
     monkeypatch.setattr("custom_components.dte_rates.pdf_parser.PdfReader", _FakeReader)
 
@@ -102,6 +137,12 @@ def test_parse_rate_card_tracks_seasonal_and_period_components(monkeypatch):
     assert peak.components.per_kwh["capacity_energy"] == Decimal("5.949")
     assert super_off_peak.components.per_kwh["distribution_super_off_peak_kwh"] == Decimal("4.892")
 
+    d111 = parsed.rates["D1.11"]
+    d111_peak = next(p for p in d111.periods if p.period_name == "peak")
+    d111_off_peak = next(p for p in d111.periods if p.period_name == "off_peak")
+    assert d111_peak.components.per_kwh["distribution_kwh_year_round"] == Decimal("9.726")
+    assert d111_off_peak.components.per_kwh["distribution_kwh_year_round"] == Decimal("9.726")
+
 
 def test_parser_infers_time_windows_from_text(monkeypatch):
     monkeypatch.setattr("custom_components.dte_rates.pdf_parser.PdfReader", _FakeReader)
@@ -127,3 +168,15 @@ def test_parser_supports_new_period_labels_without_code_changes(monkeypatch):
     assert ultra.window.weekdays_only is True
     assert ultra.window.hour_ranges == [(0, 5)]
     assert ultra.components.per_kwh["capacity_energy"] == Decimal("1.111")
+
+
+def test_real_d111_excerpt_applies_year_round_distribution_to_all_tou_periods(monkeypatch):
+    monkeypatch.setattr("custom_components.dte_rates.pdf_parser.PdfReader", _RealD111Reader)
+    parsed = parse_rate_card_pdf(b"fake", "https://example.test/card.pdf")
+
+    d111 = parsed.rates["D1.11"]
+    for season in ("june_through_september", "october_through_may"):
+        peak = next(p for p in d111.periods if p.season_name == season and p.period_name == "peak")
+        off_peak = next(p for p in d111.periods if p.season_name == season and p.period_name == "off_peak")
+        assert peak.components.per_kwh["distribution_kwh_year_round"] == Decimal("9.726")
+        assert off_peak.components.per_kwh["distribution_kwh_year_round"] == Decimal("9.726")