tests: update coops for new API

searvey/coops.py

@@ -85,6 +85,9 @@ class COOPS_StationMetadataSource(Enum):
 ]
 
 
+# NOTE:
+# DAILY_MEAN is only available for Great Lakes stations and at LST
+# for SALINITY couldn't find a station that provides data!
 class COOPS_Product(Enum):  # noqa: N801
     WATER_LEVEL = (
         "water_level"
@@ -100,10 +103,10 @@ class COOPS_Product(Enum):  # noqa: N801
         "visibility"  # Visibility from the station's visibility sensor. A measure of atmospheric clarity.
     )
     HUMIDITY = "humidity"  # Relative humidity as measured at the station.
-    SALINITY = "salinity"  # Salinity and specific gravity data for the station.
+    # SALINITY = "salinity"  # Salinity and specific gravity data for the station.
     HOURLY_HEIGHT = "hourly_height"  # Verified hourly height water level data for the station.
     HIGH_LOW = "high_low"  # Verified high/low water level data for the station.
-    DAILY_MEAN = "daily_mean"  # Verified daily mean water level data for the station.
+    # DAILY_MEAN = "daily_mean"  # Verified daily mean water level data for the station.
     MONTHLY_MEAN = "monthly_mean"  # Verified monthly mean water level data for the station.
     ONE_MINUTE_WATER_LEVEL = (
         "one_minute_water_level"

tests/coops_test.py

@@ -1,6 +1,13 @@
+import contextlib
 from datetime import datetime
+from datetime import timedelta
+from urllib.parse import quote
 
+import httpx
+import numpy as np
+import pandas as pd
 import pytest
+import pytz
 from shapely.geometry import box
 
 from searvey import fetch_coops_station
@@ -55,6 +62,7 @@ def test_coops_stations_main_api():
     assert len(stations_main_api) > 0
     assert list(stations_main_api.columns) == [
         "nws_id",
+        "station_type",
         "name",
         "state",
         "lon",
@@ -77,7 +85,7 @@ def test_coops_stations_main_api():
 
     with pytest.raises(ValueError) as excinfo:
         get_coops_stations(metadata_source="someothersource")
-    assert "not a valid stationmetadatasource" in str(excinfo.value).lower()
+    assert "not a valid coops_stationmetadatasource" in str(excinfo.value).lower()
 
 
 @pytest.mark.vcr
@@ -89,6 +97,7 @@ def test_coops_stations_within_region_main_api():
     assert len(stations) > 0
     assert list(stations.columns) == [
         "nws_id",
+        "station_type",
         "name",
         "state",
         "lon",
@@ -171,3 +180,174 @@ def test_coops_predictions_product():
     )
 
     assert len(data["t"]) > 0
+
+
+def test_generate_urls():
+    station_id = "AAA"
+    start_date = pd.Timestamp("2023-01-01")
+    end_date = pd.Timestamp("2023-06-01")
+    urls = _generate_urls(
+        station_id=station_id,
+        start_date=start_date,
+        end_date=end_date,
+    )
+    assert len(urls) == 6
+    assert all(isinstance(url, httpx.URL) for url in urls)
+    assert all(station_id in str(url) for url in urls)
+    assert quote(_coops_date(start_date)) in str(urls[0])
+    assert quote(_coops_date(end_date)) in str(urls[-1])
+
+
+def test_generate_urls_raises_common_start_date_and_end_date():
+    station_id = "AAA"
+    date = pd.Timestamp("2023-06-01")
+    urls = _generate_urls(
+        station_id=station_id,
+        start_date=date,
+        end_date=date,
+    )
+    assert len(urls) == 0
+    assert not urls
+    assert urls == []
+
+
+def test_generate_urls_raises_when_end_date_before_start_date():
+    start_date = pd.Timestamp("2023-01-01")
+    end_date = pd.Timestamp("2022-01-01")
+    with pytest.raises(ValueError) as exc:
+        _generate_urls(
+            station_id="aaaa",
+            start_date=start_date,
+            end_date=end_date,
+        )
+    assert str(exc.value) == f"'end_date' must be after 'start_date': {end_date} vs {start_date}"
+
+
+@pytest.mark.parametrize(
+    "station_id, product",
+    [
+        (8654467, "water_level"),
+        (8636580, "one_minute_water_level"),
+        (8654467, "predictions"),
+        (8575431, "air_gap"),
+        (8654467, "wind"),
+        (8654467, "air_pressure"),
+        (8654467, "air_temperature"),
+        (8575437, "visibility"),
+        (8720233, "humidity"),
+        (8654467, "water_temperature"),
+        (8737048, "conductivity"),
+        ("cb1101", "currents"),
+        ("cb1101", "currents_predictions"),
+        (8654467, "datums"),
+        # (8636580, "hourly_height"),  # Default dates won't work!
+        # (8636580, "high_low"),  # Default dates won't work!
+        # (8518750, "monthly_mean"),  # Default dates won't work!
+        # (8720233, "salinity"),  # Can't find stations with this data!
+        # (8636580, "daily_mean"),  # Not supported by searvey
+    ],
+)
+def test_coops_data_products_default_args(station_id, product):
+    df = fetch_coops_station(station_id, product=product)
+    assert all(col in COOPS_ProductFieldsNameMap[COOPS_Product(product)].values() for col in df.columns)
+    assert all(
+        df.dtypes[col] == COOPS_ProductFieldTypes[col]
+        for col in df.columns
+        if df.dtypes[col] != np.dtype("O")
+    )
+
+
+@pytest.fixture
+def now_utc():
+    # Seconds are truncated for COOPS query url
+    return pd.Timestamp.now("utc").floor("min")
+
+
+@pytest.mark.parametrize(
+    "days_before_now, station_id, product",
+    [
+        (7, 8654467, "water_level"),
+        (60, 8636580, "hourly_height"),
+        (60, 8636580, "high_low"),
+        (60, 8518750, "monthly_mean"),
+        (7, 8636580, "one_minute_water_level"),
+        (7, 8654467, "predictions"),
+        (7, 8575431, "air_gap"),
+        (7, 8654467, "wind"),
+        (7, 8654467, "air_pressure"),
+        (7, 8654467, "air_temperature"),
+        (7, 8575437, "visibility"),
+        (7, 8720233, "humidity"),
+        (7, 8654467, "water_temperature"),
+        (7, 8737048, "conductivity"),
+        (7, "cb1101", "currents"),
+        (7, "cb1101", "currents_predictions"),
+        (7, 8654467, "datums"),
+        # (7, 8720233, "salinity"),
+        # (30, 8636580, "daily_mean"),
+    ],
+)
+def test_coops_data_products_w_date_input(now_utc, days_before_now, station_id, product):
+    start_date = now_utc - timedelta(days=days_before_now)
+    end_date = now_utc
+
+    df = fetch_coops_station(
+        station_id,
+        product=product,
+        start_date=start_date,
+        end_date=end_date,
+    )
+
+    assert all(col in COOPS_ProductFieldsNameMap[COOPS_Product(product)].values() for col in df.columns)
+    assert all(
+        df.dtypes[col] == COOPS_ProductFieldTypes[col]
+        for col in df.columns
+        if df.dtypes[col] != np.dtype("O")
+    )
+
+    if df.index.name == "time":
+        # When selecting between intervals one time prior to the first
+        # is also included (at most 1 out of range)
+        assert (~((start_date <= df.index) & (df.index <= end_date))).sum() <= 1
+        assert df.index.tz is not None
+        assert df.index.tz.utcoffset(df.index[0]) == timedelta()
+
+
+@pytest.mark.parametrize(
+    "msg_idx, date",
+    [
+        (None, pd.Timestamp.now("utc")),
+        (None, pd.Timestamp.now("gmt")),
+        (None, pd.Timestamp.now(pytz.FixedOffset(0))),
+        (0, pd.Timestamp.now("est")),
+        (0, pd.Timestamp.now(pytz.FixedOffset(4))),
+        (0, pd.Timestamp.now(pytz.FixedOffset(-4))),
+        (1, pd.Timestamp.now()),
+        (1, datetime.now()),
+    ],
+)
+def test_coops_warn_utc(msg_idx, date):
+    msgs = ["Converting to UTC", "Assuming UTC"]
+
+    warn_type = None
+    warn_count = 0
+    warn_msg = ""
+    ctx = contextlib.nullcontext()
+    if msg_idx is not None:
+        warn_type = UserWarning
+        warn_count = 2  # One for start one for end!
+        warn_msg = msgs[msg_idx]
+        ctx = pytest.warns(warn_type, match=warn_msg)
+
+    with ctx as record:
+        fetch_coops_station(
+            8654467,
+            product="water_level",
+            start_date=date - timedelta(days=7),
+            end_date=date,
+        )
+
+    if record is not None:
+        assert len(record) == warn_count
+        if warn_count:
+            assert [warn_msg in r.message.args[0] for r in record]