ESS-4162: Unpinning pandas (#150)

* Not using extremely low numbers that cause underflow and unpinning pandas

* Forgot about the pyproject.toml change

* Safer checking for None in test

Author: branislav-jenco-4ss

datareservoirio/client.py

@@ -49,7 +49,7 @@ def metric() -> logging.Logger:
 
 # Default values to push as start/end dates. (Limited by numpy.datetime64)
 _END_DEFAULT = 9214646400000000000  # 2262-01-01
-_START_DEFAULT = -9214560000000000000  # 1678-01-01
+_START_DEFAULT = 0  # 1970-01-01
 
 _TIMEOUT_DEAULT = (120, 120)
 
@@ -430,22 +430,16 @@ def get(
         else:
             df = pd.DataFrame(columns=("index", "values")).astype({"index": "int64"})
 
-        try:
-            # When we move to pandas 3, the .loc here breaks with None start and end, haven't dug into why yet
-            series = (
-                df.set_index("index").squeeze("columns").loc[start:end].copy(deep=True)
-            )
-        except KeyError as e:
+        s = df.set_index("index").squeeze("columns")
+
+        # Ensure sorted (cheap if already sorted)
+        if not s.index.is_monotonic_increasing:
             logging.warning(
                 "The time series you requested is not properly ordered. The data will be sorted to attempt to resolve the issue. Please note that this operation may take some time."
             )
-            series = (
-                df.set_index("index")
-                .sort_index()
-                .squeeze("columns")
-                .loc[start:end]
-                .copy(deep=True)
-            )
+            s = s.sort_index()
+
+        series = s.loc[start:end]
         series.index.name = None
 
         if series.empty and raise_empty:  # may become empty after slicing

pyproject.toml

@@ -22,7 +22,7 @@ classifiers = [
 dependencies = [
   "numpy",
   "oauthlib",
-  "pandas < 3",
+  "pandas",
   "pyarrow",
   "requests",
   "requests-oauthlib",

tests/response_cases.py

@@ -135,7 +135,7 @@
     # description: TimeSeries API response (empty data)
     (
         "GET",
-        "https://reservoir-api.4subsea.net/api/timeseries/e3d82cda-4737-4af9-8d17-d9dfda8703d0/data/days?start=-9214560000000000000&end=9214646399999999999",
+        "https://reservoir-api.4subsea.net/api/timeseries/e3d82cda-4737-4af9-8d17-d9dfda8703d0/data/days?start=0&end=9214646399999999999",
     ): {
         "_content": (
             TEST_PATH
@@ -489,7 +489,7 @@
     # description: TimeSeries API response
     (
         "GET",
-        "https://reservoir-api.4subsea.net/api/timeseries/2fee7f8a-664a-41c9-9b71-25090517c275/data/days?start=-9214560000000000000&end=9214646399999999999",
+        "https://reservoir-api.4subsea.net/api/timeseries/2fee7f8a-664a-41c9-9b71-25090517c275/data/days?start=0&end=9214646399999999999",
     ): {
         "status_code": 200,
         "reason": "OK",

tests/test_client.py

@@ -140,10 +140,10 @@ def test_get(self, mock_requests, client, start, end, group1_data, response_case
         series_expect = group1_data.as_series()
         pd.testing.assert_series_equal(series_out, series_expect)
         # Check that the correct HTTP request is made
-        if start and end:
+        if start is not None and end is not None:
             request_url_expect = "https://reservoir-api.4subsea.net/api/timeseries/2fee7f8a-664a-41c9-9b71-25090517c275/data/days?start=1672358400000000000&end=1672703939999999999"
         else:
-            request_url_expect = "https://reservoir-api.4subsea.net/api/timeseries/2fee7f8a-664a-41c9-9b71-25090517c275/data/days?start=-9214560000000000000&end=9214646399999999999"
+            request_url_expect = "https://reservoir-api.4subsea.net/api/timeseries/2fee7f8a-664a-41c9-9b71-25090517c275/data/days?start=0&end=9214646399999999999"
         assert mock_requests.call_args_list[0].args[1] == request_url_expect
 
     def test_get_convert_date(self, client, group1_data, response_cases):