python-esios 2.4.1__tar.gz → 2.4.2__tar.gz

python_esios-2.4.2/.release-please-manifest.json

@@ -0,0 +1,3 @@
+{
+  ".": "2.4.2"
+}

{python_esios-2.4.1 → python_esios-2.4.2}/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## [2.4.2](https://github.com/datons/python-esios/compare/python-esios-v2.4.1...python-esios-v2.4.2) (2026-06-11)
+
+
+### Bug Fixes
+
+* **i90:** pair MW and price per quarter-hour in multi-measure sheets ([06efdb6](https://github.com/datons/python-esios/commit/06efdb660077cc36c4435f0e36ae5f24e68e9189))
+
 ## [2.4.1](https://github.com/datons/python-esios/compare/python-esios-v2.4.0...python-esios-v2.4.1) (2026-05-19)
 
 

{python_esios-2.4.1 → python_esios-2.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: python-esios
-Version: 2.4.1
+Version: 2.4.2
 Summary: A Python wrapper for the ESIOS API (Spanish electricity market)
 Project-URL: Homepage, https://github.com/datons/python-esios
 Project-URL: Repository, https://github.com/datons/python-esios

{python_esios-2.4.1 → python_esios-2.4.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "python-esios"
-version = "2.4.1"
+version = "2.4.2"
 description = "A Python wrapper for the ESIOS API (Spanish electricity market)"
 readme = "README.md"
 license = "GPL-3.0-only"

{python_esios-2.4.1 → python_esios-2.4.2}/src/esios/processing/i90.py

@@ -200,18 +200,29 @@ class I90Sheet:
         arr[arr == ""] = np.nan
         return arr
 
-    def _normalize_datetime_columns(self, columns: np.ndarray) -> np.ndarray:
+    def _normalize_datetime_columns(
+        self, columns: np.ndarray, n_measures: int = 1
+    ) -> np.ndarray:
         """Normalize time column headers to integer period indices.
 
-        Handles four column formats found in I90 files:
+        Handles five column formats found in I90 files:
 
         1. Sequential integers: 1–24 (hourly) or 1–96 (quarterly)
         2. H-Q format: "1-1", "1-2", "1-3", "1-4", "2-1", …
-        3. NaN-filler format: [1, NaN, NaN, NaN, 2, …]
+        3. NaN-filler format: [1, NaN, NaN, NaN, 2, …] — one period number per
+           hour, three NaN fillers for the remaining quarters (single measure).
         4. Range format (DST days): "00-01", "01-02", "02-03a", "02-03b", …
            where the first number is the start hour and a/b suffix marks
            the repeated hour on fall-back days. Detected by the first
            column starting with "0" (e.g. "00-01").
+        5. Multi-measure format: [1, NaN, 2, NaN, …] — each period number labels
+           a block of `n_measures` sub-columns (e.g. MW + €/MWh paired per
+           quarter-hour in offer sheets), the NaN fillers being the extra
+           measures. The period number is on every n_measures-th column.
+
+        `n_measures` is the count of distinct measure sub-columns per period
+        (1 for single-measure sheets). It disambiguates format 3 from format 5,
+        which share the same NaN-filler shape but mean opposite things.
         """
         if any(pd.isna(columns)):
             self._n_columns_totals = 3
@@ -240,6 +251,10 @@ class I90Sheet:
             )
             return ((hours - 1) * 4 + quarters).values
 
+        # Multi-measure format: each period number labels a block of n_measures sub-columns (e.g. MW + €/MWh). The ffilled value already IS the period index and the duplication is the measures — keep it so every measure of a period shares one datetime (paired), instead of spreading them across sequential slots.
+        if n_measures > 1:
+            return hours.values
+
         # NaN-filler quarterly format: after ffill the same hour number repeats
         # four times (quarters share the hour label).  Assign sequential indices.
         if hours.duplicated().any():
@@ -257,13 +272,18 @@ class I90Sheet:
         if idx_col_start == -1:
             return pd.DataFrame()
 
-        columns_date = self._normalize_datetime_columns(columns_prior[idx_col_start:])
+        columns_variable = columns[idx_col_start:]
+        # Distinct measure labels per period (e.g. {MW, €/MWh} -> 2). Disambiguates the NaN-filler time header (single measure, expand to sequential quarters) from the multi-measure paired layout (keep the period index so measures pair).
+        measure_labels = pd.Series(columns_variable).dropna()
+        n_measures = int(measure_labels.nunique()) if len(measure_labels) else 1
+        columns_date = self._normalize_datetime_columns(
+            columns_prior[idx_col_start:], n_measures=n_measures
+        )
         # _normalize_datetime_columns sets _n_columns_totals from the time-block
         # content (NaN-filler vs sequential). Override with a header-label count
         # so the index slice survives REE format revisions that add or drop a
         # "Total" column without touching the time-axis encoding.
         self._n_columns_totals = _count_header_separators(columns_prior, idx_col_start)
-        columns_variable = columns[idx_col_start:]
         columns_index = columns[: idx_col_start - self._n_columns_totals]
 
         return columns, columns_index, columns_date, columns_variable

{python_esios-2.4.1 → python_esios-2.4.2}/tests/test_i90.py

@@ -55,6 +55,17 @@ def _mixed_hq_columns(n_hours: int = 24) -> np.ndarray:
     return np.array(cols, dtype=object)
 
 
+def _multi_measure_columns(n_periods: int = 96, n_measures: int = 2) -> np.ndarray:
+    """Multi-measure paired columns: the period number labels a block of
+    n_measures sub-columns (e.g. MW + €/MWh), the rest NaN: [1, NaN, 2, NaN, …].
+    """
+    cols: list = []
+    for p in range(1, n_periods + 1):
+        cols.append(p)
+        cols.extend([np.nan] * (n_measures - 1))
+    return np.array(cols, dtype=object)
+
+
 # ---------------------------------------------------------------------------
 # _any_value_greater_than_30
 # ---------------------------------------------------------------------------
@@ -317,3 +328,38 @@ class TestPreprocessOverrideIntegration:
         _, columns_index, _, _ = result
         assert list(columns_index) == ["Redespacho", "Tipo"]
         assert s._n_columns_totals == 2
+
+
+class TestNormalizeMultiMeasure:
+    """Multi-measure paired columns ([1, NaN, 2, NaN, …]) share the same
+    NaN-filler shape as the single-measure format but mean the opposite: the
+    duplication is the measures (MW + €/MWh), so the period index must be kept
+    (paired), not expanded to sequential slots. Regression for offer sheets
+    (OFFER_BT / OFFER_BS_*) whose MW and price were otherwise scattered across
+    8 slots/hour instead of paired into 4 quarter-hours.
+    """
+
+    def test_two_measures_keep_period_index(self):
+        s = _make_sheet()
+        result = s._normalize_datetime_columns(
+            _multi_measure_columns(4, n_measures=2), n_measures=2
+        )
+        # 4 quarters x 2 measures -> each quarter index repeated for its measures
+        assert list(result) == [1, 1, 2, 2, 3, 3, 4, 4]
+
+    def test_full_day_paired_96_periods(self):
+        s = _make_sheet()
+        result = s._normalize_datetime_columns(
+            _multi_measure_columns(96, n_measures=2), n_measures=2
+        )
+        assert len(result) == 192
+        assert result.max() == 96
+        _, counts = np.unique(result, return_counts=True)
+        assert set(counts.tolist()) == {2}  # every period appears once per measure
+        assert _any_value_greater_than_30(result)
+
+    def test_single_measure_default_still_sequential(self):
+        # n_measures defaults to 1 -> old NaN-filler behaviour is preserved
+        s = _make_sheet()
+        result = s._normalize_datetime_columns(_full_nan_filler_columns())
+        assert list(result) == list(range(1, 97))

python_esios-2.4.1/.release-please-manifest.json

@@ -1,3 +0,0 @@
-{
-  ".": "2.4.1"
-}