_decoding tests

Author: dsehnal

ciftools/bin/data.py

@@ -1,4 +1,4 @@
-fromtypingimportDict, List, Optional, Union
+fromtypingimportAny, Dict, List, Optional, Union
 
 importnumpyasnp
 fromciftools.bin.decoderimportdecode_cif_data
@@ -11,11 +11,11 @@ def __init__(
 self,
 name: str,
 values: Union[np.ndarray, list[str]],
-value_kinds: Union[np.ndarray, None],
+value_mask: Union[np.ndarray, None],
  ):
 self.name=name
 self._values=values
-self._value_kinds=value_kinds
+self._value_mask=value_mask
 self._row_count=len(values)
 
 defget_string(self, row: int) ->str:
@@ -28,8 +28,8 @@ def get_float(self, row: int) -> float:
 returnfloat(self._values[row])
 
 defget_value_presence(self, row: int) ->CIFValuePresenceEnum:
-ifself._value_kinds:
-returnself._value_kinds[row]
+ifself._value_mask:
+returnself._value_mask[row]
 return0
 
 defare_values_equal(self, row_a: int, row_b: int) ->bool:
@@ -49,19 +49,23 @@ def as_ndarray(
 returnself._values[start:end]
 returnself._values[start:end].astype(dtype)
 
-def__getitem__(self, idx: int) ->Union[str, float, int, None]:
-ifself._value_kindsandself._value_kinds[idx]:
+def__getitem__(self, idx: Any) ->Any:
+ifisinstance(idx, int) andself._value_maskandself._value_mask[idx]:
 returnNone
 returnself._values[idx]
 
 def__len__(self):
 returnself._row_count
 
+@property
+defvalue_presences(self) ->Optional[np.ndarray]:
+returnself._value_mask
+
 
 def_decode_cif_column(column: EncodedCIFColumn) ->CIFColumn:
 values=decode_cif_data(column["data"])
-value_kinds=decode_cif_data(column["mask"]) ifcolumn["mask"] elseNone
-returnBinaryCIFColumn(column["name"], values, value_kinds)
+value_mask=decode_cif_data(column["mask"]) ifcolumn["mask"] elseNone
+returnBinaryCIFColumn(column["name"], values, value_mask)
 
 
 classBinaryCIFCategory(CIFCategory):
@@ -100,7 +104,7 @@ def n_columns(self) -> int:
 returnlen
 
 @property
-defcolumn_names(self) ->list[str]:
+deffield_names(self) ->list[str]:
 returnself._field_names
 
 
@@ -156,8 +160,8 @@ def from_data(data: EncodedCIFFile, *, lazy=True) -> "BinaryCIFFile":
  - False: decode all columns immediately
  """
 
-min_version=[0, 3, 0]
-version=map(int, data["version"].split("."))
+min_version=(0, 3, 0)
+version=tuple(map(int, data["version"].split(".")))
 ifversion<min_version:
 raiseValueError(f"Invalid version {data['version']}, expected >={'.'.join(map(str, min_version))}")
 

ciftools/models/data.py

@@ -1,5 +1,5 @@
 fromenumimportIntEnum
-fromtypingimportDict, List, Optional, Protocol, Union
+fromtypingimportAny, Dict, List, Optional, Protocol, Union
 
 importnumpyasnp
 
@@ -34,7 +34,7 @@ def as_ndarray(
  ) ->np.ndarray:
  ...
 
-def__getitem__(self, idx: int) ->Union[str, float, int, None]:
+def__getitem__(self, idx: Any) ->Any:
  ...
 
 def__len__(self) ->int:
@@ -62,7 +62,7 @@ def n_columns(self) -> int:
  ...
 
 @property
-defcolumn_names(self) ->List[str]:
+deffield_names(self) ->List[str]:
  ...
 
 def__getattr__(self, name: str) ->Optional[CIFColumn]:

pyproject.toml

@@ -1,5 +1,5 @@
 [project]
-name = "CifTools"
+name = "ciftools"
 version = "0.2.0"
 authors = [
 {name="Ravi Jose Tristao Ramos", email="[email protected]" },

tests/_decoding.py

@@ -2,8 +2,8 @@
 importurllib.request
 
 importmsgpack
-fromciftools.cif_format.binary.fileimportBinaryCIFFile
-fromciftools.cif_format.value_presenceimportValuePresenceEnum
+fromciftools.serializationimportloads
+fromciftools.models.dataimportCIFValuePresenceEnum
 
 
 classTestEncodings_Decoding(unittest.TestCase):
@@ -12,7 +12,7 @@ def test(self):
 
 print("mmCIF test")
 data=urllib.request.urlopen("https://models.rcsb.org/1tqn.bcif").read()
-parsed=BinaryCIFFile.loads(data, lazy=False)
+parsed=loads(data, lazy=False)
 
 atom_site=parsed["1TQN"].atom_site
 entity=parsed[0]["entity"]
@@ -24,22 +24,23 @@ def test(self):
 print(entity.field_names)
 print(atom_site.rowCount)
 print(label_comp_id[0])
-print(label_comp_id.values[-1])
-print(len(label_comp_id.values))
+print(label_comp_id[-1])
 print(len(label_comp_id))
-print(cartn_x.values[0:10])
-print(atom_site["label_alt_id"].value_kinds[0] ==ValuePresenceEnum.NotSpecified)
+print(cartn_x[0:10])
+value_slice=cartn_x.as_ndarray(start=1, end=10, dtype='i4')
+print(value_slice)
+print(atom_site["label_alt_id"].value_presences[0] ==CIFValuePresenceEnum.NotSpecified)
 # print([[f"_{c.name}.{f}" for f in c.field_names] for c in parsed[0].categories.values()])
 
 print("Volume Data test")
 data=urllib.request.urlopen(
 "https://ds.litemol.org/x-ray/1tqn/box/-22.367,-33.367,-21.634/-7.106,-10.042,-0.937?detail=1"
  ).read()
-parsed=BinaryCIFFile.loads(msgpack.loads(data))
+parsed=loads(data)
 
 print([b.headerforbinparsed.data_blocks])
-print(parsed[1].categories().keys())
-print(parsed["FO-FC"].categories().keys())
+print(parsed[1].categories.keys())
+print(parsed["FO-FC"].categories.keys())
 print(parsed[1]["volume_data_3d_info"].field_names)
-print(parsed[1]["volume_data_3d"].rowCount)
-print(parsed[1]["volume_data_3d"]["values"].values[0:10])
+print(parsed[1]["volume_data_3d"].n_rows)
+print(parsed[1]["volume_data_3d"]["values"][0:10])

tests/_run_all_tests.py

@@ -8,7 +8,7 @@
 "interval_quantization",
 "run_length",
 "string_array",
-# "_decoding",
+"_decoding",
 # "_encoding",
 ]