[unstable] Add File size Config

cognite/extractorutils/unstable/configuration/models.py

@@ -434,6 +434,128 @@ def _log_handler_default() -> list[LogHandlerConfig]:
     return [LogConsoleHandlerConfig(type="console", level=LogLevel.INFO)]
 
 
+class FileSizeConfig(ConfigModel):
+    """
+    Configuration parameter for setting a file size.
+    """
+
+    expression: str
+
+    @staticmethod
+    def _parse_size_in_bytes(v: str) -> int:
+        sizes = {
+            "kb": 1000,
+            "mb": 1_000_000,
+            "gb": 1_000_000_000,
+            "tb": 1_000_000_000_000,
+            "kib": 1024,
+            "mib": 1_048_576,
+            "gib": 1_073_741_824,
+            "tib": 1_099_511_627_776,
+        }
+        try:
+            return int(v)
+        except ValueError:
+            pass
+        v_lower = v.lower().replace(" ", "")
+        for size in sizes:
+            if v_lower.endswith(size):
+                num = float(v_lower.replace(size, ""))
+                return int(num * sizes[size])
+        raise InvalidConfigError(f"Invalid unit for file size: {v}. Valid units: {sizes.keys()}")
+
+    @property
+    def _bytes(self) -> int:
+        return self._parse_size_in_bytes(self.expression)
+
+    @property
+    def bytes(self) -> int:
+        """
+        File size in bytes.
+        """
+        return self._bytes
+
+    @property
+    def kilobytes(self) -> float:
+        """
+        File size in kilobytes.
+        """
+        return self._bytes / 1000
+
+    @property
+    def megabytes(self) -> float:
+        """
+        File size in megabytes.
+        """
+        return self._bytes / 1_000_000
+
+    @property
+    def gigabytes(self) -> float:
+        """
+        File size in gigabytes.
+        """
+        return self._bytes / 1_000_000_000
+
+    @property
+    def terabytes(self) -> float:
+        """
+        File size in terabytes.
+        """
+        return self._bytes / 1_000_000_000_000
+
+    @property
+    def kibibytes(self) -> float:
+        """
+        File size in kibibytes (1024 bytes).
+        """
+        return self._bytes / 1024
+
+    @property
+    def mebibytes(self) -> float:
+        """
+        File size in mebibytes (1024 kibibytes).
+        """
+        return self._bytes / 1_048_576
+
+    @property
+    def gibibytes(self) -> float:
+        """
+        File size in gibibytes (1024 mebibytes).
+        """
+        return self._bytes / 1_073_741_824
+
+    @property
+    def tebibytes(self) -> float:
+        """
+        File size in tebibytes (1024 gibibytes).
+        """
+        return self._bytes / 1_099_511_627_776
+
+    def __int__(self) -> int:
+        """
+        Returns the file size as bytes.
+        """
+        return int(self._bytes)
+
+    def __float__(self) -> float:
+        """
+        Returns the file size as bytes.
+        """
+        return float(self._bytes)
+
+    def __str__(self) -> str:
+        """
+        Returns the file size as a human readable string.
+        """
+        return self.expression
+
+    def __repr__(self) -> str:
+        """
+        Returns the file size as a human readable string.
+        """
+        return self.expression
+
+
 class RawDestinationConfig(ConfigModel):
     """
     Configuration parameters for using Raw.

tests/test_unstable/test_configuration.py

@@ -7,6 +7,7 @@
 from cognite.extractorutils.unstable.configuration.loaders import ConfigFormat, load_io
 from cognite.extractorutils.unstable.configuration.models import (
     ConnectionConfig,
+    FileSizeConfig,
     TimeIntervalConfig,
     _ClientCredentialsConfig,
 )
@@ -212,3 +213,17 @@ def test_from_env() -> None:
 
     # Check that the produces cogniteclient object is valid
     assert len(client.assets.list(limit=1)) == 1
+
+
+def test_parse_file_size() -> None:
+    assert FileSizeConfig(expression="154584").bytes == 154584
+    assert FileSizeConfig(expression="1kB").bytes == 1000
+    assert FileSizeConfig(expression="25MB").bytes == 25000000
+    assert FileSizeConfig(expression="1kib").bytes == 1024
+    assert FileSizeConfig(expression="2.7MiB").bytes == 2831155
+    assert FileSizeConfig(expression="4 KB").bytes == 4000
+
+    assert FileSizeConfig(expression="4 KB").kilobytes == pytest.approx(4)
+    assert FileSizeConfig(expression="453 kB").megabytes == pytest.approx(0.453)
+    assert FileSizeConfig(expression="1543 kiB").kilobytes == pytest.approx(1580.032)
+    assert FileSizeConfig(expression="14.5 mb").kilobytes == pytest.approx(14500)