ENH: Parallel specification of next API

Author: effigies

bids_ng/__init__.py

@@ -0,0 +1,18 @@
+from .api1 import BIDSDataset, BIDSFile, File, Index, Label
+from .enums import Query
+from .utils import PaddedInt
+
+NONE, REQUIRED, OPTIONAL = tuple(Query)
+
+__all__ = (
+    "BIDSDataset",
+    "BIDSFile",
+    "File",
+    "Index",
+    "Label",
+    "NONE",
+    "OPTIONAL",
+    "REQUIRED",
+    "Query",
+    "PaddedInt",
+)

bids_ng/types/__init__.py

@@ -0,0 +1,110 @@
+"""PyBIDS 1.0 API specification"""
+
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Protocol, TypeVar, Union
+
+from .utils import PaddedInt
+
+try:
+    from typing import TypeAlias
+except ImportError:
+    from typing_extensions import TypeAlias
+
+
+# Datasets should be parameterizable on some kind of schema object.
+# External API users should not depend on it, so this is bound to Any,
+# but once a Schema type is defined for an API implementation, type checkers
+# should be able to introspect it.
+SchemaT = TypeVar("SchemaT")
+
+
+Index: TypeAlias = PaddedInt
+Label: TypeAlias = str
+
+
+class File(Protocol[SchemaT]):
+    """Generic file holder
+
+    This serves as a base class for :class:`BIDSFile` and can represent
+    non-BIDS files.
+    """
+
+    path: Path
+    relative_path: Path
+    dataset: Optional["BIDSDataset[SchemaT]"]
+
+    def __fspath__(self) -> str:
+        ...
+
+
+class BIDSFile(File[SchemaT], Protocol):
+    """BIDS file
+
+    This provides access to BIDS concepts such as path components
+    and sidecar metadata.
+
+    BIDS paths take the form::
+
+        [sub-<label>/[ses-<label>/]<datatype>/]<entities>_<suffix><extension>
+    """
+
+    entities: Dict[str, Union[Label, Index]]
+    datatype: Optional[str]
+    suffix: Optional[str]
+    extension: Optional[str]
+
+    metadata: Dict[str, Any]
+    """Sidecar metadata aggregated according to inheritance principle"""
+
+
+class BIDSDataset(Protocol[SchemaT]):
+    """Interface to a single BIDS dataset.
+
+    This structure does not consider the contents of sub-datasets
+    such as `sourcedata/` or `derivatives/`.
+    """
+    root: Path
+    schema: SchemaT
+
+    dataset_description: Dict[str, Any]
+    """Contents of dataset_description.json"""
+
+    ignored: List[File[SchemaT]]
+    """Invalid files found in dataset"""
+
+    files: List[BIDSFile[SchemaT]]
+    """Valid files found in dataset"""
+
+    datatypes: List[str]
+    """Datatype directories found in dataset"""
+
+    modalities: List[str]
+    """BIDS "modalities" found in dataset"""
+
+    subjects: List[str]
+    """Subject/participant identifiers found in the dataset"""
+
+    entities: List[str]
+    """Entities (long names) found in any filename in the dataset"""
+
+    def get(self, **filters) -> List[BIDSFile[SchemaT]]:
+        """Query dataset for files"""
+
+    def get_entities(self, entity: str, **filters) -> List[Label | Index]:
+        """Query dataset for entity values"""
+
+    def get_metadata(self, term: str, **filters) -> List[Any]:
+        """Query dataset for metadata values"""
+
+
+class DatasetCollection(BIDSDataset[SchemaT], Protocol):
+    """Interface to a collection of BIDS dataset.
+
+    This structure allows the user to construct a single view of
+    multiple datasets, such as including source or derivative datasets.
+    """
+    primary: BIDSDataset[SchemaT]
+    datasets: List[BIDSDataset[SchemaT]]
+
+    def add_dataset(self, dataset: BIDSDataset[SchemaT]) -> None:
+        ...

bids_ng/types/api1.py

@@ -0,0 +1,17 @@
+from enum import Enum
+
+
+class Query(Enum):
+    """Special arguments for dataset querying
+
+    * `Query.NONE` - The field MUST NOT be present
+    * `Query.REQUIRED` - The field MUST be present, but may take any value
+    * `Query.OPTIONAL` - The field MAY be present, and may take any value
+
+    `Query.ANY` is a synonym for `Query.REQUIRED`. Its use is discouraged
+    and may be removed in the future.
+    """
+
+    NONE = 1
+    REQUIRED = ANY = 2
+    OPTIONAL = 3

bids_ng/types/enums.py

@@ -0,0 +1,76 @@
+import typing as ty
+
+
+class PaddedInt(int):
+    """Integer type that preserves zero-padding
+
+    Acts like an int in almost all ways except that string formatting
+    will keep the original zero-padding. Numeric format specifiers will
+    work with the integer value.
+
+    >>> PaddedInt(1)
+    1
+    >>> p2 = PaddedInt("02")
+    >>> p2
+    02
+    >>> str(p2)
+    '02'
+    >>> p2 == 2
+    True
+    >>> p2 in range(3)
+    True
+    >>> f"{p2}"
+    '02'
+    >>> f"{p2:s}"
+    '02'
+    >>> f"{p2!s}"
+    '02'
+    >>> f"{p2!r}"
+    '02'
+    >>> f"{p2:d}"
+    '2'
+    >>> f"{p2:03d}"
+    '002'
+    >>> f"{p2:f}"
+    '2.000000'
+    >>> {2: "val"}.get(p2)
+    'val'
+    >>> {p2: "val"}.get(2)
+    'val'
+
+    Note that arithmetic will break the padding.
+
+    >>> str(p2 + 1)
+    '3'
+    """
+
+    def __init__(self, val: ty.Union[str, int]) -> None:
+        self.sval = str(val)
+        if not self.sval.isdigit():
+            raise TypeError(
+                f"{self.__class__.__name__}() argument must be a string of digits "
+                f"or int, not {val.__class__.__name__!r}"
+            )
+
+    def __eq__(self, val: object) -> bool:
+        return val == self.sval or super().__eq__(val)
+
+    def __str__(self) -> str:
+        return self.sval
+
+    def __repr__(self) -> str:
+        return self.sval
+
+    def __format__(self, format_spec: str) -> str:
+        """Format a padded integer
+
+        If a format spec can be used on a string, apply it to the zero-padded string.
+        Otherwise format as an integer.
+        """
+        try:
+            return format(self.sval, format_spec)
+        except ValueError:
+            return super().__format__(format_spec)
+
+    def __hash__(self) -> int:
+        return super().__hash__()