spectre-core 0.0.22__py3-none-any.whl → 0.0.24__py3-none-any.whl

spectre_core/batches/_base.py

@@ -13,21 +13,40 @@ from os.path import splitext
 from spectre_core._file_io import BaseFileHandler
 from spectre_core.config import get_batches_dir_path, TimeFormat
 from spectre_core.spectrograms import Spectrogram
+from spectre_core.exceptions import deprecated
 
 
-def parse_batch_base_file_name(
-    base_file_name: str
-) -> Tuple[str, str, str]:
+@deprecated(
+    "The terminology (base file name) is inconsistent with the `_file_io` submodule. "
+    f"Please use `parse_batch_file_name` instead."
+)
+def parse_batch_base_file_name(base_file_name: str) -> Tuple[str, str, str]:
+    """Included for backwards compatability - please use `parse_batch_file_name` instead."""
+    return parse_batch_file_name(base_file_name)
+
+
+def parse_batch_file_name(file_name: str) -> Tuple[str, str, str]:
     """Parse the base file name of a batch file into a start time, tag and extension.
+
+    :param file_name: The file name of the batch file, with optional extension.
+    :return: The file name decomposed into its start time, tag and extension. If no extension is present,
+    the final element of the tuple will be an empty string.
     """
-    batch_name, extension = splitext(base_file_name)
+    batch_name, extension = splitext(file_name)
+
+    num_underscores = batch_name.count("_")
+    if num_underscores != 1:
+        raise ValueError(
+            f"Expected exactly one underscore in the batch name '{batch_name}'. Found {num_underscores}"
+        )
+
     # strip the dot from the extension
-    extension = extension.lstrip('.')
+    extension = extension.lstrip(".")
     start_time, tag = batch_name.split("_", 1)
     return start_time, tag, extension
 
 
-T = TypeVar('T')
+T = TypeVar("T")
 
 
 class BatchFile(BaseFileHandler[T]):
@@ -37,14 +56,12 @@ class BatchFile(BaseFileHandler[T]):
 
         `<start time>_<tag>.<extension>`
 
-    The substring `<start time>_<tag>` is referred to as the batch name. Files with the same batch name 
+    The substring `<start time>_<tag>` is referred to as the batch name. Files with the same batch name
     belong to the same batch.
     """
+
     def __init__(
-        self, 
-        batch_parent_dir_path: str, 
-        batch_name: str, 
-        extension: str
+        self, batch_parent_dir_path: str, batch_name: str, extension: str
     ) -> None:
         """Initialise a `BatchFile` instance.
 
@@ -52,52 +69,39 @@ class BatchFile(BaseFileHandler[T]):
         :param batch_name: Base file name, composed of the batch start time and tag.
         :param extension: File extension.
         """
-        super().__init__(batch_parent_dir_path, 
-                         batch_name, 
-                         extension)
+        super().__init__(batch_parent_dir_path, batch_name, extension)
         self._start_time, self._tag = batch_name.split("_")
-   
-   
+
     @property
-    def start_time(
-        self
-    ) -> str:
+    def start_time(self) -> str:
         """The start time of the batch, formatted as a string up to seconds precision."""
         return self._start_time
 
-
     @cached_property
-    def start_datetime(
-        self
-    ) -> datetime:
+    def start_datetime(self) -> datetime:
         """The start time of the batch, parsed as a datetime up to seconds precision."""
         return datetime.strptime(self.start_time, TimeFormat.DATETIME)
 
-    
     @property
-    def tag(
-        self
-    ) -> str:
+    def tag(self) -> str:
         """The batch name tag."""
         return self._tag
-  
-  
+
+
 @dataclass(frozen=True)
 class _BatchExtension:
     """Supported extensions for a `BaseBatch`, inherited by all derived classes.
-    
+
     :ivar PNG: Corresponds to the `.png` file extension.
     """
+
     PNG: str = "png"
-      
-    
+
+
 class _PngFile(BatchFile[str]):
     """Stores an image visualising the data for the batch."""
-    def __init__(
-        self, 
-        batch_parent_dir_path: str, 
-        batch_name: str
-    ) -> None:
+
+    def __init__(self, batch_parent_dir_path: str, batch_name: str) -> None:
         """Initialise a `_PngFile` instance.
 
         :param batch_parent_dir_path: The parent directory for the batch.
@@ -105,7 +109,6 @@ class _PngFile(BatchFile[str]):
         """
         super().__init__(batch_parent_dir_path, batch_name, _BatchExtension.PNG)
 
-
     def _read(self) -> str:
         """Reads the PNG file and returns it base64-encoded.
 
@@ -114,25 +117,22 @@ class _PngFile(BatchFile[str]):
         with open(self.file_path, "rb") as f:
             encoded = b64encode(f.read())
             return encoded.decode("ascii")
-   
- 
+
+
 class BaseBatch(ABC):
     """
     An abstract base class representing a group of data files over a common time interval.
 
     All files in a batch share a base file name and differ only by their extension.
-    Subclasses of `BaseBatch` define the expected data for each file extension and 
+    Subclasses of `BaseBatch` define the expected data for each file extension and
     provide an API for accessing their contents using `BatchFile` subclasses.
 
     Subclasses should expose `BatchFile` instances directly as attributes, which
     simplifies static typing. Additionally, they should call `add_file` in the constructor
     to formally register each `BatchFile`.
     """
-    def __init__(
-        self, 
-        start_time: str,
-        tag: str
-    ) -> None:
+
+    def __init__(self, start_time: str, tag: str) -> None:
         """Initialise a `BaseBatch` instance.
 
         :param start_time: Start time of the batch as a string with seconds precision.
@@ -141,113 +141,82 @@ class BaseBatch(ABC):
         self._start_time = start_time
         self._tag: str = tag
         self._start_datetime = datetime.strptime(self.start_time, TimeFormat.DATETIME)
-        self._parent_dir_path = get_batches_dir_path(year  = self.start_datetime.year,
-                                                     month = self.start_datetime.month,
-                                                     day   = self.start_datetime.day)
-        
+        self._parent_dir_path = get_batches_dir_path(
+            year=self.start_datetime.year,
+            month=self.start_datetime.month,
+            day=self.start_datetime.day,
+        )
+
         # internal register of batch files
         self._batch_files: dict[str, BatchFile] = {}
-    
+
         # Add the files shared by all derived classes
-        self._png_file  = _PngFile(self.parent_dir_path, self.name)
-        self.add_file( self._png_file ) 
-    
-    
+        self._png_file = _PngFile(self.parent_dir_path, self.name)
+        self.add_file(self._png_file)
+
     @property
     @abstractmethod
-    def spectrogram_file(
-        self
-    ) -> BatchFile:
+    def spectrogram_file(self) -> BatchFile:
         """The batch file which contains spectrogram data."""
-        
-    
+
     @property
-    def start_time(
-        self
-    ) -> str:
+    def start_time(self) -> str:
         """The start time of the batch, formatted as a string up to seconds precision."""
         return self._start_time
 
-
     @property
-    def start_datetime(
-        self
-    ) -> datetime:
+    def start_datetime(self) -> datetime:
         """The start time of the batch, parsed as a datetime up to seconds precision."""
         return self._start_datetime
-    
-    
+
     @property
-    def tag(
-        self
-    ) -> str:
+    def tag(self) -> str:
         """The batch name tag."""
         return self._tag
-    
 
     @property
-    def parent_dir_path(
-        self
-    ) -> str:
+    def parent_dir_path(self) -> str:
         """The parent directory for the batch."""
         return self._parent_dir_path
 
-
     @property
-    def name(
-        self
-    ) -> str:
-        """Return the base file name shared by all files in the batch, 
+    def name(self) -> str:
+        """Return the base file name shared by all files in the batch,
         composed of the start time and the batch tag."""
         return f"{self._start_time}_{self._tag}"
-    
-    
+
     @property
-    def extensions(
-        self
-    ) -> list[str]:
+    def extensions(self) -> list[str]:
         """All defined file extensions for the batch."""
         return list(self._batch_files.keys())
-    
-    
+
     @property
-    def batch_files(
-        self
-    ) -> dict[str, BatchFile]:
+    def batch_files(self) -> dict[str, BatchFile]:
         """Map each file extension in the batch to the corresponding batch file instance.
-        
+
         Use `add_file` to add a file to the batch.
         """
         return self._batch_files
-    
-    
+
     @property
-    def png_file(
-        self
-    ) -> _PngFile:
+    def png_file(self) -> _PngFile:
         """The batch file corresponding to the `.png` extension."""
         return self._png_file
-    
 
-    def add_file(
-        self, 
-        batch_file: BatchFile
-    ) -> None:
+    def add_file(self, batch_file: BatchFile) -> None:
         """Add an instance of a batch file to the batch.
 
         :param batch_file: The `BatchFile` instance to add to the batch.
         :raises ValueError: If the `BatchFile` instance does not have a defined file extension.
         """
         if batch_file.extension is None:
-            raise ValueError(f"The `BatchFile` must have a defined file extension. "
-                             f"Received '{batch_file.extension}.")
+            raise ValueError(
+                f"The `BatchFile` must have a defined file extension. "
+                f"Received '{batch_file.extension}."
+            )
         self._batch_files[batch_file.extension] = batch_file
-    
 
-    def get_file(
-        self, 
-        extension: str
-    ) -> BatchFile:
+    def get_file(self, extension: str) -> BatchFile:
         """Get a batch file instance from the batch, according to the file extension.
 
         :param extension: The file extension of the batch file.
@@ -257,13 +226,11 @@ class BaseBatch(ABC):
         try:
             return self._batch_files[extension]
         except KeyError:
-            raise NotImplementedError(f"A batch file with extension '{extension}' is not implemented for this batch.")
-
+            raise NotImplementedError(
+                f"A batch file with extension '{extension}' is not implemented for this batch."
+            )
 
-    def delete_file(
-        self, 
-        extension: str
-    ) -> None:
+    def delete_file(self, extension: str) -> None:
         """Delete a file from the batch, according to the file extension.
 
         :param extension: The file extension of the batch file.
@@ -272,11 +239,7 @@ class BaseBatch(ABC):
         batch_file = self.get_file(extension)
         batch_file.delete()
 
-
-    def has_file(
-        self, 
-        extension: str
-    ) -> bool:
+    def has_file(self, extension: str) -> bool:
         """Determine the existance of a batch file in the file system.
 
         :param extension: The file extension of the batch file.
@@ -287,22 +250,10 @@ class BaseBatch(ABC):
             return batch_file.exists
         except FileNotFoundError:
             return False
-        
-        
-    def read_spectrogram(
-        self
-    ) -> Spectrogram:
+
+    def read_spectrogram(self) -> Spectrogram:
         """Read and return the spectrogram data stored in the batch.
 
         :return: The spectrogram stored by the batch `spectrogram_file`.
         """
         return self.spectrogram_file.read()
-    
-
-
-
-
-
-
-
-

spectre_core/batches/_batches.py

@@ -10,22 +10,23 @@ from datetime import datetime
 from spectre_core.config import TimeFormat
 from spectre_core.spectrograms import Spectrogram, time_chop, join_spectrograms
 from spectre_core.config import get_batches_dir_path
-from spectre_core.exceptions import (
-    BatchNotFoundError
-)
-from ._base import BaseBatch, parse_batch_base_file_name
+from spectre_core.exceptions import BatchNotFoundError
+from ._base import BaseBatch, parse_batch_file_name
+
+T = TypeVar("T", bound=BaseBatch)
+
 
-T = TypeVar('T', bound=BaseBatch)
 class Batches(Generic[T]):
     """Managed collection of `Batch` instances for a given tag. Provides a simple
     interface for read operations on batched data files."""
+
     def __init__(
-        self, 
+        self,
         tag: str,
         batch_cls: Type[T],
-        year: Optional[int] = None, 
-        month: Optional[int] = None, 
-        day: Optional[int] = None
+        year: Optional[int] = None,
+        month: Optional[int] = None,
+        day: Optional[int] = None,
     ) -> None:
         """Initialise a `Batches` instance.
 
@@ -40,85 +41,54 @@ class Batches(Generic[T]):
         self._batch_map: dict[str, T] = OrderedDict()
         self.set_date(year, month, day)
 
-
     @property
-    def tag(
-        self
-    ) -> str:
+    def tag(self) -> str:
         """The batch name tag."""
         return self._tag
 
-
     @property
-    def batch_cls(
-        self
-    ) -> Type[T]:
+    def batch_cls(self) -> Type[T]:
         """The `Batch` class used to read the batched files."""
         return self._batch_cls
 
-
     @property
-    def year(
-        self
-    ) -> Optional[int]:
+    def year(self) -> Optional[int]:
         """The numeric year, to filter batch files."""
         return self._year
 
-
-    @property 
-    def month(
-        self
-    ) -> Optional[int]:
+    @property
+    def month(self) -> Optional[int]:
         """The numeric month of the year, to filter batch files."""
         return self._month
-    
 
     @property
-    def day(
-        self
-    ) -> Optional[int]:
+    def day(self) -> Optional[int]:
         """The numeric day of the year, to filter batch files."""
         return self._day
-    
 
     @property
-    def batches_dir_path(
-        self
-    ) -> str:
+    def batches_dir_path(self) -> str:
         """The shared ancestral path for all the batches. `Batches` recursively searches
         this directory to find all batches whose batch name contains `tag`."""
         return get_batches_dir_path(self.year, self.month, self.day)
-    
 
     @property
-    def batch_list(
-        self
-    ) -> list[T]:
+    def batch_list(self) -> list[T]:
         """A list of all batches found within `batches_dir_path`."""
-        return  list(self._batch_map.values())
-    
+        return list(self._batch_map.values())
 
     @property
-    def start_times(
-        self
-    ) -> list[str]:
+    def start_times(self) -> list[str]:
         """The start times of each batch found within `batches_dir_path`."""
         return list(self._batch_map.keys())
 
-
     @property
-    def num_batches(
-        self
-    ) -> int:
+    def num_batches(self) -> int:
         """The total number of batches found within `batches_dir_path`."""
         return len(self.batch_list)
 
-
     def set_date(
-        self, 
-        year: Optional[int],
-        month: Optional[int],
-        day: Optional[int]
+        self, year: Optional[int], month: Optional[int], day: Optional[int]
     ) -> None:
         """Reset `batches_dir_path` according to the numeric date, and refresh the list
         of available batches.
@@ -132,72 +102,57 @@ class Batches(Generic[T]):
         self._day = day
         self.update()
 
-
-    def update(
-        self
-    ) -> None:
-        """Perform a fresh search all files in `batches_dir_path` for batches 
+    def update(self) -> None:
+        """Perform a fresh search all files in `batches_dir_path` for batches
         with `tag` in the batch name."""
         # reset cache
-        self._batch_map = OrderedDict() 
-        
+        self._batch_map = OrderedDict()
+
         # get a list of all batch file names in the batches directory path
-        batch_file_names = [f for (_, _, files) in os.walk(self.batches_dir_path) for f in files]
+        batch_file_names = [
+            f for (_, _, files) in os.walk(self.batches_dir_path) for f in files
+        ]
         for batch_file_name in batch_file_names:
-            start_time, tag, _ = parse_batch_base_file_name(batch_file_name) 
+            start_time, tag, _ = parse_batch_file_name(batch_file_name)
             if tag == self._tag:
                 self._batch_map[start_time] = self.batch_cls(start_time, tag)
-        
+
         self._batch_map = OrderedDict(sorted(self._batch_map.items()))
-    
 
-    def __iter__(
-        self
-    ) -> Iterator[T]:
+    def __iter__(self) -> Iterator[T]:
         """Iterate over the stored batch instances."""
         yield from self.batch_list
-        
-    
-    def __len__(
-        self
-    ):
-        return self.num_batches
 
+    def __len__(self):
+        return self.num_batches
 
-    def _get_from_start_time(
-        self, 
-        start_time: str
-    ) -> T:
+    def _get_from_start_time(self, start_time: str) -> T:
         """Find and return the `Batch` instance based on the string formatted start time."""
         try:
             return self._batch_map[start_time]
         except KeyError:
-            raise BatchNotFoundError(f"Batch with start time {start_time} could not be found within {self.batches_dir_path}")
+            raise BatchNotFoundError(
+                f"Batch with start time {start_time} could not be found within {self.batches_dir_path}"
+            )
 
-
-    def _get_from_index(
-        self, 
-        index: int
-    ) -> T:
+    def _get_from_index(self, index: int) -> T:
         """Find and return the `Batch` instance based on its numeric index.
-        
+
         Batches are ordered sequentially in time, so index `0` corresponds to the oldest
         `Batch` with respect to the start time.
         """
         if self.num_batches == 0:
             raise BatchNotFoundError("No batches are available")
         elif index > self.num_batches:
-            raise IndexError(f"Index '{index}' is greater than the number of batches '{self.num_batches}'")
+            raise IndexError(
+                f"Index '{index}' is greater than the number of batches '{self.num_batches}'"
+            )
         return self.batch_list[index]
 
-
-    def __getitem__(
-        self, 
-        subscript: str | int
-    ) -> T:
+    def __getitem__(self, subscript: str | int) -> T:
         """Get a `Batch` instanced based on either the start time or chronological index.
 
-        :param subscript: If the subscript is a string, interpreted as a formatted start time. 
+        :param subscript: If the subscript is a string, interpreted as a formatted start time.
         If the subscript is an integer, it is interpreted as a chronological index.
         :return: The corresponding `BaseBatch` subclass.
         """
@@ -205,22 +160,19 @@ class Batches(Generic[T]):
             return self._get_from_start_time(subscript)
         elif isinstance(subscript, int):
             return self._get_from_index(subscript)
-   
 
     def get_spectrogram(
-        self,
-        start_datetime: datetime, 
-        end_datetime: datetime
+        self, start_datetime: datetime, end_datetime: datetime
     ) -> Spectrogram:
         """
-        Retrieve a spectrogram spanning the specified time range. 
+        Retrieve a spectrogram spanning the specified time range.
 
         :param start_datetime: The start time of the range (inclusive).
         :param end_datetime: The end time of the range (inclusive).
         :raises FileNotFoundError: If no spectrogram data is available within the specified time range.
         :return: A spectrogram created by stitching together data from all matching batches.
         """
-        
+
         spectrograms = []
         for batch in self:
             # skip batches without spectrogram data
@@ -233,10 +185,14 @@ class Batches(Generic[T]):
 
             # Check if the batch overlaps with the input time range
             if start_datetime <= upper_bound and lower_bound <= end_datetime:
-                spectrograms.append( time_chop(spectrogram, start_datetime, end_datetime) )
+                spectrograms.append(
+                    time_chop(spectrogram, start_datetime, end_datetime)
+                )
 
         if spectrograms:
             return join_spectrograms(spectrograms)
         else:
-            raise FileNotFoundError(f"No spectrogram data found for the time range "
-                                    f"{start_datetime} to {end_datetime}.")
+            raise FileNotFoundError(
+                f"No spectrogram data found for the time range "
+                f"{start_datetime} to {end_datetime}."
+            )