read_nesh_timelines_skimap_key

openskistats/nesh/tests/test_timelines.py

@@ -0,0 +1,21 @@
+from datetime import date
+
+import polars as pl
+
+from openskistats.nesh.timelines import read_nesh_timelines_skimap_key
+
+
+def test_read_nesh_timelines_skimap_key() -> None:
+    df = read_nesh_timelines_skimap_key()
+    whaleback_2023_timeline = df.row(
+        by_predicate=pl.col.skimap_url.eq(
+            "https://skimap.org/skiareas/view/1078"
+        ).__and__(pl.col.season == 2023),
+        named=True,
+    )
+    assert whaleback_2023_timeline["opening_date"] == date.fromisoformat("2023-12-26")
+    assert whaleback_2023_timeline["closing_date"] == date.fromisoformat("2024-03-10")
+    assert whaleback_2023_timeline["season_duration"] == 75
+    assert whaleback_2023_timeline["nesh_sources"] == [
+        "https://www.newenglandskihistory.com/NewHampshire/whaleback.php"
+    ]

openskistats/nesh/timelines.py

@@ -147,12 +147,30 @@ def read_nesh_timelines() -> pl.DataFrame:
         )
         .drop("opening", "closing")
         .with_columns(
-            season_duration=pl.col("closing_date").sub("opening_date").dt.total_days()
+            season_duration=pl.col("closing_date").sub("opening_date").dt.total_days(),
+            skimap_url=pl.col("ski_area_page").replace_strict(nesh_to_skimap),
         )
     )
     return df
 
 
+def read_nesh_timelines_skimap_key() -> pl.DataFrame:
+    return (
+        read_nesh_timelines()
+        .filter(pl.col("skimap_url").is_not_null())
+        .group_by("skimap_url", "season")
+        .agg(
+            pl.min("opening_date"),
+            pl.max("closing_date"),
+            pl.col("ski_area_page").alias("nesh_sources"),
+        )
+        .with_columns(
+            season_duration=pl.col("closing_date").sub("opening_date").dt.total_days(),
+        )
+        .sort("skimap_url", "season")
+    )
+
+
 nesh_to_skimap = {
     "https://www.newenglandskihistory.com/NewHampshire/abenaki.php": "https://skimap.org/skiareas/view/4091",
     "https://www.newenglandskihistory.com/NewHampshire/arrowhead.php": "https://skimap.org/skiareas/view/2146",