Merge pull request #819 from effigies/enh/padded_runs

ENH: Retain zero-padding in run entities while preserving integer queries and comparisons

Author: effigies

bids/layout/config/bids.json

@@ -34,7 +34,7 @@
         },
         {
             "name": "run",
-            "pattern": "[_/\\\\]+run-0*(\\d+)",
+            "pattern": "[_/\\\\]+run-(\\d+)",
             "dtype": "int"
         },
         {

bids/layout/layout.py

@@ -12,6 +12,7 @@
 
 import sqlalchemy as sa
 from sqlalchemy.orm import aliased
+from sqlalchemy.sql.expression import cast
 from bids_validator import BIDSValidator
 
 from ..utils import listify, natural_sort
@@ -772,10 +773,11 @@ def _build_file_query(self, **kwargs):
                     else:
                         val_clause = tag_alias._value.op('REGEXP')(str(val))
                 else:
-                    if isinstance(val, (list, tuple)):
-                        val_clause = tag_alias._value.in_(val)
+                    vals = listify(val)
+                    if isinstance(vals[0], int):
+                        val_clause = cast(tag_alias._value, sa.Integer).in_(vals)
                     else:
-                        val_clause = tag_alias._value == val
+                        val_clause = tag_alias._value.in_(vals)
 
                 query = join_method(
                     tag_alias,

bids/layout/models.py

@@ -18,7 +18,7 @@
 from ..utils import listify
 from .writing import build_path, write_to_file
 from ..config import get_option
-from .utils import BIDSMetadata
+from .utils import BIDSMetadata, PaddedInt
 
 Base = declarative_base()
 
@@ -71,6 +71,9 @@ def _sanitize_init_args(self, kwargs):
 
         return kwargs
 
+    def __repr__(self):
+        return f"<LayoutInfo {self.root}>"
+
 
 class Config(Base):
     """Container for BIDS configuration information.
@@ -162,6 +165,9 @@ def load(self, config, session=None):
 
         return Config(session=session, **config)
 
+    def __repr__(self):
+        return f"<Config {self.name}>"
+
 
 class BIDSFile(Base):
     """Represents a single file or directory in a BIDS dataset.
@@ -531,12 +537,18 @@ def __init__(self, name, pattern=None, mandatory=False, directory=None,
 
         self._init_on_load()
 
+    def __repr__(self):
+        return f"<Entity {self.name} (pattern={self.pattern}, dtype={self.dtype})>"
+
     @reconstructor
     def _init_on_load(self):
         if self._dtype not in ('str', 'float', 'int', 'bool'):
             raise ValueError("Invalid dtype '{}'. Must be one of 'int', "
                              "'float', 'bool', or 'str'.".format(self._dtype))
-        self.dtype = eval(self._dtype)
+        if self._dtype == "int":
+            self.dtype = PaddedInt
+        else:
+            self.dtype = eval(self._dtype)
         self.regex = re.compile(self.pattern) if self.pattern is not None else None
 
     def __iter__(self):
@@ -684,6 +696,9 @@ def _init_on_load(self):
         if self._dtype == 'json':
             self.value = json.loads(self._value)
             self.dtype = 'json'
+        elif self._dtype == 'int':
+            self.dtype = PaddedInt
+            self.value = self.dtype(self._value)
         else:
             self.dtype = eval(self._dtype)
             self.value = self.dtype(self._value)

bids/layout/tests/test_layout.py

@@ -14,6 +14,7 @@
 from bids.layout import BIDSLayout, Query
 from bids.layout.models import Config
 from bids.layout.index import BIDSLayoutIndexer
+from bids.layout.utils import PaddedInt
 from bids.tests import get_test_data_path
 from bids.utils import natural_sort
 
@@ -711,7 +712,7 @@ def test_get_with_wrong_dtypes(layout_7t_trt):
     ''' Test automatic dtype sanitization. '''
     l = layout_7t_trt
     assert (l.get(run=1) == l.get(run='1') == l.get(run=np.int64(1)) ==
-            l.get(run=[1, '15']))
+            l.get(run=[1, '15']) == l.get(run='01'))
     assert not l.get(run='not_numeric')
     assert l.get(session=1) == l.get(session='1')
 
@@ -807,4 +808,17 @@ def test_load_layout_config_not_overwritten(layout_synthetic_nodb, tmpdir):
     for attr in ['root', 'absolute_paths', 'derivatives']:
         assert getattr(cm1.layout_info, attr) == getattr(cm2.layout_info, attr)
 
-    assert cm1.layout_info.config != cm2.layout_info.config
+    assert cm1.layout_info.config != cm2.layout_info.config
+
+
+def test_padded_run_roundtrip(layout_ds005):
+    for run in (1, "1", "01"):
+        res = layout_ds005.get(subject="01", task="mixedgamblestask",
+                               run=run, extension=".nii.gz")
+        assert len(res) == 1
+    boldfile = res[0]
+    ents = boldfile.get_entities()
+    assert isinstance(ents["run"], PaddedInt)
+    assert ents["run"] == 1
+    newpath = layout_ds005.build_path(ents, absolute_paths=False)
+    assert newpath == "sub-01/func/sub-01_task-mixedgamblestask_run-01_bold.nii.gz"

bids/layout/tests/test_models.py

@@ -13,6 +13,7 @@
 
 from bids.layout.models import (BIDSFile, Entity, Tag, Base, Config,
                                 FileAssociation, BIDSImageFile, LayoutInfo)
+from bids.layout.utils import PaddedInt
 from bids.tests import get_test_data_path
 
 
@@ -129,9 +130,11 @@ def test_tag_dtype(sample_bidsfile, subject_entity):
         Tag(f, e, '4', 'int'),
         Tag(f, e, '4', int),
         Tag(f, e, 4),
-        Tag(file=f, entity=e, dtype=int, value='4')
+        Tag(file=f, entity=e, dtype=int, value='4'),
+        Tag(file=f, entity=e, dtype=int, value='04'),
     ]
-    assert all([t.dtype == int for t in tags])
+    assert all(t.dtype == PaddedInt for t in tags)
+    assert all(t.value == 4 for t in tags)
 
 
 def test_entity_add_file(sample_bidsfile):

bids/layout/utils.py

@@ -20,6 +20,74 @@ def __getitem__(self, key):
                 "Metadata term {!r} unavailable for file {}.".format(key, self._source_file))
 
 
+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
+
+    >>> 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):
+        self.sval = str(val)
+
+    def __eq__(self, val):
+        return val == self.sval or super().__eq__(val)
+
+    def __str__(self):
+        return self.sval
+
+    def __repr__(self):
+        return self.sval
+
+    def __format__(self, format_spec):
+        """ 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:
+            return super().__format__(format_spec)
+
+    def __hash__(self):
+        return super().__hash__()
+
+
 def parse_file_entities(filename, entities=None, config=None,
                         include_unmatched=False):
     """Parse the passed filename for entity/value pairs.

bids/tests/data/ds005/sub-01/func/sub-01_task-mixedgamblestask_run-01_events.json

@@ -1,3 +1,3 @@
 {
-    "run": 1
-}
+    "run": "01"
+}