Date processes (#275)

* update openeo-processes

* bump 2023.9.1

* bump 2023.9.0

* add datetime processes

* pre-commit hook

openeo_processes_dask/process_implementations/dates.py

@@ -0,0 +1,126 @@
+from typing import Optional
+
+import numpy as np
+
+__all__ = [
+    "date_between",
+    "date_difference",
+    "date_shift",
+]
+
+
+def datetime_from_str(date: str):
+    daytime = np.datetime64(date)
+    return daytime
+
+
+def date_between(
+    x: str, min: str, max: str, exclude_max: bool = False
+) -> Optional[bool]:
+    x = datetime_from_str(x)
+    min = datetime_from_str(min)
+    max = datetime_from_str(max)
+    if exclude_max:
+        return bool((x >= min) and (x < max))
+    else:
+        return bool((x >= min) and (x <= max))
+
+
+def date_difference(date1: str, date2: str, unit: Optional[str] = "second") -> float:
+    date1 = datetime_from_str(date1)
+    date2 = datetime_from_str(date2)
+    units = {
+        "millisecond": 1,
+        "second": 1000,
+        "minute": 1000 * 60,
+        "hour": 1000 * 60 * 60,
+        "day": 1000 * 60 * 60 * 24,
+        "week": 1000 * 60 * 60 * 24 * 7,
+        "month": "M",
+        "year": "Y",
+    }
+    if unit in units:
+        unit = units[unit]
+    if unit in ["M", "Y"]:
+        return float(
+            (
+                date2.astype(f"datetime64[{unit}]")
+                - date1.astype(f"datetime64[{unit}]")
+            ).astype(float)
+        )
+    else:
+        # we do this, so the examples are fulfilled:
+        # date_difference(date1 = "2020-01-01T00:00:00.0Z", date2 = "2020-01-01T00:00:15.5Z") -> 15.5
+        return (
+            float(
+                (
+                    date2.astype(f"datetime64[ms]") - date1.astype(f"datetime64[ms]")
+                ).astype(float)
+            )
+            / unit
+        )
+
+
+def date_shift(date: str, value: int, unit: str) -> str:
+    if date.endswith("Z"):
+        end = "Z"
+    elif "+" in date:
+        end = "+" + date.split("+")[-1]
+        date = date.split("+")[0]
+    else:
+        end = ""
+    units = {
+        "millisecond": "ms",
+        "second": "s",
+        "minute": "m",
+        "hour": "h",
+        "day": "D",
+        "week": "W",
+        "month": "M",
+        "year": "Y",
+    }
+    if unit in units:
+        unit = units[unit]
+    if unit in ["M", "Y"]:
+        if len(date) > 7:
+            date_M = np.datetime64(date, "M")
+            day = (
+                int(
+                    (np.datetime64(date, "D") - date_M.astype("datetime64[D]")).astype(
+                        int
+                    )
+                )
+                + 1
+            )
+            if " " in date:
+                time = "T" + date.split(" ")[-1]
+            elif "T" in date:
+                time = "T" + date.split("T")[-1]
+            else:
+                time = ""
+            new_date = str(date_M + np.timedelta64(value, unit))
+            if day in [29, 30, 31]:
+                for i in range(3):
+                    try:
+                        new_daytime = f"{new_date}-{day-i}"
+                        new_daytime_numpy = np.datetime64(new_daytime)
+                        result = f"{new_daytime}{time}"
+                        return result
+                    except:
+                        pass
+            elif int(day) < 10:
+                new_daytime = f"{new_date}-0{day}{time}"
+            else:
+                new_daytime = f"{new_date}-{day}T{time}"
+            new_daytime_numpy = np.datetime64(new_daytime)
+            return new_daytime
+
+        date = datetime_from_str(date)
+        return str(date_M + np.timedelta64(value, unit))
+
+    date = datetime_from_str(date)
+    if unit in ["ms"]:
+        result = str((date + np.timedelta64(value, unit)).astype(f"datetime64[{unit}]"))
+    else:
+        result = str((date + np.timedelta64(value, unit)).astype(date.dtype))
+    return result + end

tests/test_dates.py

@@ -0,0 +1,50 @@
+from openeo_processes_dask.process_implementations.dates import (
+    date_between,
+    date_difference,
+    date_shift,
+)
+
+
+def test_date_between():
+    assert not date_between(x="2020-01-01", min="2021-01-01", max="2022-01-01")
+
+
+def test_date_difference():
+    assert (
+        date_difference(date1="2020-01-01T00:00:00.0Z", date2="2020-01-01T00:00:15.5Z")
+        == 15.5
+    )
+    assert (
+        date_difference(date1="2020-01-01T00:00:00Z", date2="2020-01-01T01:00:00+01:00")
+        == 0
+    )
+    assert date_difference(date1="2020-01-02", date2="2020-01-01") == -86400
+    assert date_difference(date1="2020-01-02", date2="2020-01-01", unit="day") == -1
+
+
+def test_date_shift():
+    month_shift = date_shift(date="2020-02-01T17:22:45Z", value=6, unit="month")
+    assert month_shift == "2020-08-01T17:22:45Z"
+
+    day_shift = date_shift(date="2021-03-31T00:00:00+02:00", value=-7, unit="day")
+    assert day_shift == "2021-03-24T00:00:00+02:00"
+
+    year_shift = date_shift(date="2020-02-29T17:22:45Z", value=1, unit="year")
+    assert year_shift == "2021-02-28T17:22:45Z"
+
+    month_shift = date_shift(date="2020-01-31", value=1, unit="month")
+    assert month_shift == "2020-02-29"
+
+    second_shift = date_shift(date="2016-12-31T23:59:59Z", value=1, unit="second")
+    assert second_shift == "2017-01-01T00:00:00Z"
+
+    millisecond_shift = date_shift(
+        date="2018-12-31T17:22:45Z", value=1150, unit="millisecond"
+    )
+    assert millisecond_shift == "2018-12-31T17:22:46.150Z"
+
+    hour_shift = date_shift(date="2018-01-01", value=25, unit="hour")
+    assert hour_shift == "2018-01-02"
+
+    hour_shift = date_shift(date="2018-01-01", value=-1, unit="hour")
+    assert hour_shift == "2017-12-31"