valediction 1.0.0__py3-none-any.whl → 1.1.0__py3-none-any.whl

valediction/datasets/datasets_helpers.py

@@ -1,46 +1,46 @@
-from __future__ import annotations
-
-from datetime import timedelta
-from pathlib import Path
-from typing import Any, Protocol, runtime_checkable
-
-from pandas import DataFrame
-
-DataLike = Path | str | DataFrame
-
-
-@runtime_checkable
-class DatasetItemLike(Protocol):
-    name: str
-    data: DataLike
-    _dictionary_runtimes: dict[str, timedelta]
-    _padding: int
-
-    @property
-    def is_path(self) -> bool: ...
-    @property
-    def is_dataframe(self) -> bool: ...
-    @property
-    def data(self) -> Any: ...
-    @property
-    def column_count(self) -> int: ...
-    @property
-    def primary_keys(self) -> list[str]: ...
-
-
-def as_folder(path_like: str | Path) -> Path:
-    """Coerce to a Path and verify it is an existing directory."""
-    p = Path(path_like)
-    if not p.exists() or not p.is_dir():
-        raise FileNotFoundError(f"Folder not found or not a directory: {p}")
-    return p
-
-
-def list_csvs(folder: str | Path, *, recursive: bool = False) -> list[Path]:
-    """Return CSV files in a folder.
-
-    Non-recursive by default.
-    """
-    p = as_folder(folder)
-    pattern = "**/*.csv" if recursive else "*.csv"
-    return sorted(p.glob(pattern))
+from __future__ import annotations
+
+from datetime import timedelta
+from pathlib import Path
+from typing import Any, Protocol, runtime_checkable
+
+from pandas import DataFrame
+
+DataLike = Path | str | DataFrame
+
+
+@runtime_checkable
+class DatasetItemLike(Protocol):
+    name: str
+    data: DataLike
+    _dictionary_runtimes: dict[str, timedelta]
+    _padding: int
+
+    @property
+    def is_path(self) -> bool: ...
+    @property
+    def is_dataframe(self) -> bool: ...
+    @property
+    def data(self) -> Any: ...
+    @property
+    def column_count(self) -> int: ...
+    @property
+    def primary_keys(self) -> list[str]: ...
+
+
+def as_folder(path_like: str | Path) -> Path:
+    """Coerce to a Path and verify it is an existing directory."""
+    p = Path(path_like)
+    if not p.exists() or not p.is_dir():
+        raise FileNotFoundError(f"Folder not found or not a directory: {p}")
+    return p
+
+
+def list_csvs(folder: str | Path, *, recursive: bool = False) -> list[Path]:
+    """Return CSV files in a folder.
+
+    Non-recursive by default.
+    """
+    p = as_folder(folder)
+    pattern = "**/*.csv" if recursive else "*.csv"
+    return sorted(p.glob(pattern))

valediction/demo/DEMOGRAPHICS.csv

@@ -1,101 +1,101 @@
-PATIENT_HASH,DATE_OF_BIRTH,ETHNICITY,SEX,DATE_OF_DEATH
-PCB1AC7AEFBC,25/03/1937,D,Male,
-P9822C35CFE2,09/09/1992,A,Other,
-P9741499AAA8,28/07/1971,J,Male,
-P740556E27EF,18/05/1940,H,Female,
-PBB852BC4337,07/03/1974,H,Male,28/04/2019
-P44437C83F93,10/07/1999,E,Female,
-P5EB7C66585C,14/01/1997,D,Other,
-PFBA511DC293,21/07/1991,C,Male,
-PE58D0777A22,07/10/1945,P,Female,
-P2E7B1D36528,13/09/1959,B,Male,
-PADC302CC323,14/05/2008,A,Male,
-P3D355419986,29/09/1956,C,Female,
-P6B120D556F7,19/05/1948,G,Female,
-P3CA18D22E75,16/11/1937,H,Female,
-P2BFFFA51178,21/04/1955,Z,Male,
-P71E3E0FA100,30/07/1981,A,Male,
-P398005303C7,27/04/1941,G,Female,
-P6F871D71CB3,12/06/1952,P,Male,
-P7D47EED6306,28/03/1993,H,Male,
-PE713EBAEA26,06/04/1999,R,Female,
-P46C7BEA5C6A,14/10/1975,J,Female,
-P5345753F1C2,22/05/1994,A,Male,
-P913AC9C6C8F,15/08/1954,F,Male,
-PD8EA4E83BD5,22/07/1996,P,Male,
-P2EF8F723DFE,21/09/1934,L,Female,
-P7B9D911F1B4,07/09/2010,J,Male,
-PFCFA255D1C7,23/10/1983,E,Male,
-P46C8AEA8822,09/06/1986,G,Male,
-P297E227D41B,31/05/1967,L,Female,
-PAE1AEC29BE6,19/06/1943,D,Male,
-P377F9684630,11/11/1960,C,Male,
-PF663CEC0C9B,01/09/1959,N,Female,
-P81105805770,06/08/2003,D,Female,
-PC73CFD35AF8,25/02/1951,M,Male,03/02/2016
-P0870F10A5AA,18/03/1966,M,Male,
-P25E336E78C9,10/02/2003,J,Female,
-P5709FBE5A08,09/07/1967,B,Female,
-PF97770A72E5,21/05/1999,R,Male,
-P56B8D39D872,02/09/1937,D,Male,03/11/2008
-P3C3343A044B,15/07/1944,N,Male,
-PE57E477014D,05/01/1970,C,Female,
-P7E9B4C8F2C4,20/04/1954,K,Female,
-P8098F3C00C9,07/12/2004,M,Female,
-P73CAA33922A,10/04/1942,G,Male,05/03/2024
-P5E08567E985,05/05/1952,C,Male,
-PB4CBEDC3CBE,02/06/1969,B,Male,
-P2FFF6BE25F8,22/02/1974,H,Male,
-P79DB8E23123,05/03/1972,K,Female,
-PC27B6F4DA23,23/06/1954,C,Male,
-PFC8AAD27415,17/11/1974,H,Female,
-P46F7AF57A11,15/01/2006,D,Male,
-P388D6E058C9,05/04/1992,N,Male,
-P68C5CF713BA,09/02/1955,J,Female,
-PCD0B49BE79B,19/03/1997,R,Female,
-P104F45AF99B,15/05/1980,M,Male,
-P9CF442453AC,20/09/1971,F,Male,
-PBA5E0ED6EB1,11/04/1942,M,Female,
-P3155F710A6D,24/03/2002,M,Female,
-P5DDD43F647F,13/01/1968,G,Female,
-P410BC042F3D,04/08/1955,J,Male,
-PB699903C48A,12/10/1960,C,Female,
-PC8E6E7FC8E5,05/04/1951,F,Male,
-PB480BA7C139,27/06/1990,H,Female,
-P302842BAA69,09/07/1944,F,Male,
-PCF4FEDF96B4,15/12/1955,R,Male,15/02/2025
-PDD16B650B11,13/02/1944,N,Male,
-PAE31532ED3D,08/08/1948,J,Male,
-PC1A8622975F,13/10/1963,H,Female,
-P2500040F693,08/03/1946,L,Female,
-P37ABABF8BD5,21/02/1965,B,Female,
-PE050156FBBE,10/01/1940,H,Male,
-PE1CADC24657,31/05/1955,B,Male,15/08/2024
-PD560F78D44A,09/05/1978,L,Female,
-P9E13FC6B1CA,21/12/1974,N,Male,
-PD66154398E7,12/06/1943,J,Female,
-PE800609BD66,12/04/1978,C,Male,23/04/2023
-PFE1E7971E5A,04/11/1938,G,Female,
-PAB9A51DD82E,04/07/1967,L,Female,
-P2BB5669531B,04/11/1948,G,Female,
-P189AFEDAE1A,24/06/1953,S,Male,
-P1B59CB857D2,02/06/1963,N,Female,
-PCA9213E32EC,05/01/1973,R,Male,
-P24BE708D4BB,28/01/1950,E,Male,
-P8D65098CD55,27/12/1958,J,Female,
-PA38E6923B0F,03/11/1941,E,Male,
-P6BA4DD82108,17/08/1942,H,Male,12/07/1994
-P239885271FF,07/06/1987,J,Male,
-P503C3452BC2,27/06/1934,P,Female,
-PCADF960402B,31/01/1997,N,Male,
-PF536D59EF2A,04/10/1955,M,Male,
-P495D365E9E0,22/06/1948,H,Female,
-P57F7A471B88,15/03/1950,E,Male,
-P3424F1DB5BF,18/05/2002,Z,Female,
-PE124C0E760F,27/03/1938,S,Female,07/03/2018
-P7439DF795E1,01/05/1998,C,Female,
-P20F97B2DF89,23/12/1955,B,Female,
-PABA2B255009,18/12/1994,D,Female,
-P4DC03FA4017,28/11/1996,E,Female,
-P69044E391C1,03/06/1973,F,Male,
-P00372442559,27/03/1950,P,Male,
+PATIENT_HASH,DATE_OF_BIRTH,ETHNICITY,SEX,DATE_OF_DEATH
+PCB1AC7AEFBC,25/03/1937,D,Male,
+P9822C35CFE2,09/09/1992,A,Other,
+P9741499AAA8,28/07/1971,J,Male,
+P740556E27EF,18/05/1940,H,Female,
+PBB852BC4337,07/03/1974,H,Male,28/04/2019
+P44437C83F93,10/07/1999,E,Female,
+P5EB7C66585C,14/01/1997,D,Other,
+PFBA511DC293,21/07/1991,C,Male,
+PE58D0777A22,07/10/1945,P,Female,
+P2E7B1D36528,13/09/1959,B,Male,
+PADC302CC323,14/05/2008,A,Male,
+P3D355419986,29/09/1956,C,Female,
+P6B120D556F7,19/05/1948,G,Female,
+P3CA18D22E75,16/11/1937,H,Female,
+P2BFFFA51178,21/04/1955,Z,Male,
+P71E3E0FA100,30/07/1981,A,Male,
+P398005303C7,27/04/1941,G,Female,
+P6F871D71CB3,12/06/1952,P,Male,
+P7D47EED6306,28/03/1993,H,Male,
+PE713EBAEA26,06/04/1999,R,Female,
+P46C7BEA5C6A,14/10/1975,J,Female,
+P5345753F1C2,22/05/1994,A,Male,
+P913AC9C6C8F,15/08/1954,F,Male,
+PD8EA4E83BD5,22/07/1996,P,Male,
+P2EF8F723DFE,21/09/1934,L,Female,
+P7B9D911F1B4,07/09/2010,J,Male,
+PFCFA255D1C7,23/10/1983,E,Male,
+P46C8AEA8822,09/06/1986,G,Male,
+P297E227D41B,31/05/1967,L,Female,
+PAE1AEC29BE6,19/06/1943,D,Male,
+P377F9684630,11/11/1960,C,Male,
+PF663CEC0C9B,01/09/1959,N,Female,
+P81105805770,06/08/2003,D,Female,
+PC73CFD35AF8,25/02/1951,M,Male,03/02/2016
+P0870F10A5AA,18/03/1966,M,Male,
+P25E336E78C9,10/02/2003,J,Female,
+P5709FBE5A08,09/07/1967,B,Female,
+PF97770A72E5,21/05/1999,R,Male,
+P56B8D39D872,02/09/1937,D,Male,03/11/2008
+P3C3343A044B,15/07/1944,N,Male,
+PE57E477014D,05/01/1970,C,Female,
+P7E9B4C8F2C4,20/04/1954,K,Female,
+P8098F3C00C9,07/12/2004,M,Female,
+P73CAA33922A,10/04/1942,G,Male,05/03/2024
+P5E08567E985,05/05/1952,C,Male,
+PB4CBEDC3CBE,02/06/1969,B,Male,
+P2FFF6BE25F8,22/02/1974,H,Male,
+P79DB8E23123,05/03/1972,K,Female,
+PC27B6F4DA23,23/06/1954,C,Male,
+PFC8AAD27415,17/11/1974,H,Female,
+P46F7AF57A11,15/01/2006,D,Male,
+P388D6E058C9,05/04/1992,N,Male,
+P68C5CF713BA,09/02/1955,J,Female,
+PCD0B49BE79B,19/03/1997,R,Female,
+P104F45AF99B,15/05/1980,M,Male,
+P9CF442453AC,20/09/1971,F,Male,
+PBA5E0ED6EB1,11/04/1942,M,Female,
+P3155F710A6D,24/03/2002,M,Female,
+P5DDD43F647F,13/01/1968,G,Female,
+P410BC042F3D,04/08/1955,J,Male,
+PB699903C48A,12/10/1960,C,Female,
+PC8E6E7FC8E5,05/04/1951,F,Male,
+PB480BA7C139,27/06/1990,H,Female,
+P302842BAA69,09/07/1944,F,Male,
+PCF4FEDF96B4,15/12/1955,R,Male,15/02/2025
+PDD16B650B11,13/02/1944,N,Male,
+PAE31532ED3D,08/08/1948,J,Male,
+PC1A8622975F,13/10/1963,H,Female,
+P2500040F693,08/03/1946,L,Female,
+P37ABABF8BD5,21/02/1965,B,Female,
+PE050156FBBE,10/01/1940,H,Male,
+PE1CADC24657,31/05/1955,B,Male,15/08/2024
+PD560F78D44A,09/05/1978,L,Female,
+P9E13FC6B1CA,21/12/1974,N,Male,
+PD66154398E7,12/06/1943,J,Female,
+PE800609BD66,12/04/1978,C,Male,23/04/2023
+PFE1E7971E5A,04/11/1938,G,Female,
+PAB9A51DD82E,04/07/1967,L,Female,
+P2BB5669531B,04/11/1948,G,Female,
+P189AFEDAE1A,24/06/1953,S,Male,
+P1B59CB857D2,02/06/1963,N,Female,
+PCA9213E32EC,05/01/1973,R,Male,
+P24BE708D4BB,28/01/1950,E,Male,
+P8D65098CD55,27/12/1958,J,Female,
+PA38E6923B0F,03/11/1941,E,Male,
+P6BA4DD82108,17/08/1942,H,Male,12/07/1994
+P239885271FF,07/06/1987,J,Male,
+P503C3452BC2,27/06/1934,P,Female,
+PCADF960402B,31/01/1997,N,Male,
+PF536D59EF2A,04/10/1955,M,Male,
+P495D365E9E0,22/06/1948,H,Female,
+P57F7A471B88,15/03/1950,E,Male,
+P3424F1DB5BF,18/05/2002,Z,Female,
+PE124C0E760F,27/03/1938,S,Female,07/03/2018
+P7439DF795E1,01/05/1998,C,Female,
+P20F97B2DF89,23/12/1955,B,Female,
+PABA2B255009,18/12/1994,D,Female,
+P4DC03FA4017,28/11/1996,E,Female,
+P69044E391C1,03/06/1973,F,Male,
+P00372442559,27/03/1950,P,Male,