support templated fielddata filenames SimID_SIMULATIONKEY_JOBINDEX_

Author: jcschaff

pyvcell/_internal/simdata/fielddata.py

@@ -1,6 +1,15 @@
 from pathlib import Path
 
-from pyvcell._internal.simdata.simdata_models import DataFileHeader, DataFileMetadata, DataBlockHeader, VariableInfo, VariableType
+from pyvcell._internal.simdata.simdata_models import (
+    DataBlockHeader,
+    DataFileHeader,
+    DataFileMetadata,
+    VariableInfo,
+    VariableType,
+)
+
+JOBINDEX = "JOBINDEX"
+SIMULATIONKEY = "SIMULATIONKEY"
 
 
 class FieldDataFileMetadata:
@@ -32,8 +41,10 @@ def file_header(self) -> DataFileHeader:
         return self.data_file_metadata.file_header
 
 
-def parse_fielddata_filename(file_name: str, fielddata_name: str) -> tuple[int, int, str, str, VariableType, float]:
-    # parse filename like "SimID_286243594_0_DEMO_fieldData_Channel0_5_23_Volume.fda" into (286243594, 0, DEMO_fieldData, 5.23, 'Volume')
+def parse_fielddata_canonical_filename(
+    file_name: str, fielddata_name: str
+) -> tuple[int, int, str, str, VariableType, float]:
+    # parse filename like "SimID_286243594_0_DEMO_fieldData_Channel0_5_23_Volume.fdat" into (286243594, 0, DEMO_fieldData, 5.23, 'Volume')
     parts = file_name.split("_")
     sim_id = int(parts[1])
     job_id = int(parts[2])
@@ -46,14 +57,47 @@ def parse_fielddata_filename(file_name: str, fielddata_name: str) -> tuple[int,
     var_name = var_name.replace(f"SimID_{sim_id}_{job_id}_", "")
     var_name = var_name.replace(f"_{whole_number}_{fraction}_{var_type_name}.fdat", "")
     var_name = var_name.replace(f"{fielddata_name}_", "")
-    expected_fname = f"SimID_{sim_id}_{job_id}_{fielddata_name}_{var_name}_{whole_number}_{fraction}_{var_type_name}.fdat"
+    expected_fname = (
+        f"SimID_{sim_id}_{job_id}_{fielddata_name}_{var_name}_{whole_number}_{fraction}_{var_type_name}.fdat"
+    )
     if file_name != expected_fname:
         raise ValueError(f"filename {file_name} with fielddata_name {fielddata_name} does not match expected format")
     var_type = VariableType.from_field_data_var_type(var_type_name)
     return sim_id, job_id, fielddata_name, var_name, var_type, time
 
 
-def create_fielddata_filename(sim_id: int, job_id: int, fd_name: str, var_name: str, var_type: VariableType, time:float) -> str:
+def parse_fielddata_template_filename(file_name: str, fielddata_name: str) -> tuple[str, str, VariableType, float]:
+    # parse filename like "SimID_SIMULATIONKEY_JOBINDEX_DEMO_fieldData_Channel0_5_23_Volume.fdat" into (286243594, 0, DEMO_fieldData, 5.23, 'Volume')
+    parts = file_name.split("_")
+    simkey_template = parts[1]  # expecting SIMULATIONKEY
+    jobindex_template = parts[2]  # expecting JOBINDEX
+    if simkey_template != SIMULATIONKEY or jobindex_template != JOBINDEX:
+        raise ValueError(f"filename {file_name} does not match expected template format")
+    var_type_name = parts[-1].split(".")[0]
+
+    whole_number = parts[-3]
+    fraction = parts[-2]
+    time = float(f"{whole_number}.{fraction}")
+    var_name = file_name
+    var_name = var_name.replace(f"SimID_{simkey_template}_{jobindex_template}_", "")
+    var_name = var_name.replace(f"_{whole_number}_{fraction}_{var_type_name}.fdat", "")
+    var_name = var_name.replace(f"{fielddata_name}_", "")
+    expected_fname = (
+        f"SimID_{simkey_template}_{jobindex_template}_{fielddata_name}_{var_name}_{whole_number}_{fraction}_{var_type_name}.fdat"
+    )
+    if file_name != expected_fname:
+        raise ValueError(f"filename {file_name} with fielddata_name {fielddata_name} does not match expected format")
+    var_type = VariableType.from_field_data_var_type(var_type_name)
+    return fielddata_name, var_name, var_type, time
+
+
+def create_fielddata_canonical_filename(
+    sim_id: int, job_id: int, fd_name: str, var_name: str, var_type: VariableType, time: float
+) -> str:
     time_str = str(time).replace(".", "_")
     return f"SimID_{sim_id}_{job_id}_{fd_name}_{var_name}_{time_str}_{var_type.field_data_var_type}.fdat"
 
+
+def create_fielddata_template_filename(fd_name: str, var_name: str, var_type: VariableType, time: float) -> str:
+    time_str = str(time).replace(".", "_")
+    return f"SimID_{SIMULATIONKEY}_{JOBINDEX}_{fd_name}_{var_name}_{time_str}_{var_type.field_data_var_type}.fdat"

tests/_internal/simdata/fielddata_test.py

@@ -2,62 +2,135 @@
 
 import pytest
 
-from pyvcell._internal.simdata.fielddata import FieldDataFileMetadata, create_fielddata_filename, parse_fielddata_filename
+from pyvcell._internal.simdata.fielddata import (
+    FieldDataFileMetadata,
+    create_fielddata_canonical_filename,
+    parse_fielddata_canonical_filename,
+    create_fielddata_template_filename,
+    parse_fielddata_template_filename,
+)
 from pyvcell._internal.simdata.simdata_models import VariableType
 
 
-def test_parse_fielddata_filename_good() -> None:
+def test_parse_fielddata_canonical_filename_good() -> None:
     file_name = "SimID_286243594_0_DEMO_fieldData_Channel0_5_23_Volume.fdat"
 
     # "DEMO_fieldData_Channel0" is ambiguous, assume "DEMO_fieldData" is the fielddata_name
     fd_name = "DEMO_fieldData"
     expected_var_name = "Channel0"
-    ret = parse_fielddata_filename(file_name=file_name, fielddata_name=fd_name)
+    ret = parse_fielddata_canonical_filename(file_name=file_name, fielddata_name=fd_name)
     assert ret == (286243594, 0, fd_name, expected_var_name, VariableType.VOLUME, 5.23)
     # round trip - create filename from parsed values and compare
-    assert file_name == create_fielddata_filename(sim_id=ret[0], job_id=ret[1], fd_name=ret[2], var_name=ret[3], var_type=ret[4], time=ret[5])
+    assert file_name == create_fielddata_canonical_filename(
+        sim_id=ret[0], job_id=ret[1], fd_name=ret[2], var_name=ret[3], var_type=ret[4], time=ret[5]
+    )
 
     # "DEMO_fieldData_Channel0" is ambiguous, assume "DEMO" is the fielddata_name
     fd_name = "DEMO"
     expected_var_name = "fieldData_Channel0"
-    ret = parse_fielddata_filename(file_name=file_name, fielddata_name=fd_name)
+    ret = parse_fielddata_canonical_filename(file_name=file_name, fielddata_name=fd_name)
     assert ret == (286243594, 0, fd_name, expected_var_name, VariableType.VOLUME, 5.23)
     # round trip - create filename from parsed values and compare
-    assert file_name == create_fielddata_filename(sim_id=ret[0], job_id=ret[1], fd_name=ret[2], var_name=ret[3], var_type=ret[4], time=ret[5])
+    assert file_name == create_fielddata_canonical_filename(
+        sim_id=ret[0], job_id=ret[1], fd_name=ret[2], var_name=ret[3], var_type=ret[4], time=ret[5]
+    )
 
     # make sure integer time is parsed correctly
     file_name = "SimID_286243594_0_DEMO_fieldData_Channel0_5_0_Volume.fdat"
     # DEMO_fieldData_Channel0 is f'{fielddata_name}_{var_name}'
     fd_name = "DEMO_fieldData"
     expected_var_name = "Channel0"
-    ret = parse_fielddata_filename(file_name=file_name, fielddata_name=fd_name)
+    ret = parse_fielddata_canonical_filename(file_name=file_name, fielddata_name=fd_name)
     assert ret == (286243594, 0, fd_name, expected_var_name, VariableType.VOLUME, 5.0)
     # round trip - create filename from parsed values and compare
-    assert file_name == create_fielddata_filename(sim_id=ret[0], job_id=ret[1], fd_name=ret[2], var_name=ret[3], var_type=ret[4], time=ret[5])
+    assert file_name == create_fielddata_canonical_filename(
+        sim_id=ret[0], job_id=ret[1], fd_name=ret[2], var_name=ret[3], var_type=ret[4], time=ret[5]
+    )
 
 
-def test_parse_fielddata_filename_bad() -> None:
+def test_parse_fielddata_canonical_filename_bad() -> None:
     file_name = "SimID_286243594_0_DEMO_fieldData_Channel0_5_23_Volume.fdat"
 
     # bad fielddata_name
     fd_name = "DEMO_fieldData2"
     with pytest.raises(ValueError) as exc:
-        parse_fielddata_filename(file_name=file_name, fielddata_name=fd_name)
-    assert exc.value.args[0] == f'filename {file_name} with fielddata_name {fd_name} does not match expected format'
+        parse_fielddata_canonical_filename(file_name=file_name, fielddata_name=fd_name)
+    assert exc.value.args[0] == f"filename {file_name} with fielddata_name {fd_name} does not match expected format"
 
     # bad prefix
     file_name = "Sim_286243594_0_DEMO_fieldData_Channel0_5_23_Volume.fdat"
     fd_name = "DEMO_fieldData"
     with pytest.raises(ValueError) as exc:
-        parse_fielddata_filename(file_name=file_name, fielddata_name=fd_name)
-    assert exc.value.args[0] == f'filename {file_name} with fielddata_name {fd_name} does not match expected format'
+        parse_fielddata_canonical_filename(file_name=file_name, fielddata_name=fd_name)
+    assert exc.value.args[0] == f"filename {file_name} with fielddata_name {fd_name} does not match expected format"
 
     # bad suffix
     file_name = "SimID_286243594_0_DEMO_fieldData_Channel0_5_23_Volume.fda"
     fd_name = "DEMO_fieldData"
     with pytest.raises(ValueError) as exc:
-        parse_fielddata_filename(file_name=file_name, fielddata_name=fd_name)
-    assert exc.value.args[0] == f'filename {file_name} with fielddata_name {fd_name} does not match expected format'
+        parse_fielddata_canonical_filename(file_name=file_name, fielddata_name=fd_name)
+    assert exc.value.args[0] == f"filename {file_name} with fielddata_name {fd_name} does not match expected format"
+
+
+def test_parse_fielddata_template_filename_good() -> None:
+    file_name = "SimID_SIMULATIONKEY_JOBINDEX_DEMO_fieldData_Channel0_5_23_Volume.fdat"
+
+    # "DEMO_fieldData_Channel0" is ambiguous, assume "DEMO_fieldData" is the fielddata_name
+    fd_name = "DEMO_fieldData"
+    expected_var_name = "Channel0"
+    ret = parse_fielddata_template_filename(file_name=file_name, fielddata_name=fd_name)
+    assert ret == (fd_name, expected_var_name, VariableType.VOLUME, 5.23)
+    # round trip - create filename from parsed values and compare
+    assert file_name == create_fielddata_template_filename(
+        fd_name=ret[0], var_name=ret[1], var_type=ret[2], time=ret[3]
+    )
+
+    # "DEMO_fieldData_Channel0" is ambiguous, assume "DEMO" is the fielddata_name
+    fd_name = "DEMO"
+    expected_var_name = "fieldData_Channel0"
+    ret = parse_fielddata_template_filename(file_name=file_name, fielddata_name=fd_name)
+    assert ret == (fd_name, expected_var_name, VariableType.VOLUME, 5.23)
+    # round trip - create filename from parsed values and compare
+    assert file_name == create_fielddata_template_filename(
+        fd_name=ret[0], var_name=ret[1], var_type=ret[2], time=ret[3]
+    )
+
+    # make sure integer time is parsed correctly
+    file_name = "SimID_SIMULATIONKEY_JOBINDEX_DEMO_fieldData_Channel0_5_0_Volume.fdat"
+    # DEMO_fieldData_Channel0 is f'{fielddata_name}_{var_name}'
+    fd_name = "DEMO_fieldData"
+    expected_var_name = "Channel0"
+    ret = parse_fielddata_template_filename(file_name=file_name, fielddata_name=fd_name)
+    assert ret == (fd_name, expected_var_name, VariableType.VOLUME, 5.0)
+    # round trip - create filename from parsed values and compare
+    assert file_name == create_fielddata_template_filename(
+        fd_name=ret[0], var_name=ret[1], var_type=ret[2], time=ret[3]
+    )
+
+
+def test_parse_fielddata_template_filename_bad() -> None:
+    file_name = "SimID_SIMULATIONKEY_JOBINDEX_DEMO_fieldData_Channel0_5_23_Volume.fdat"
+
+    # bad fielddata_name
+    fd_name = "DEMO_fieldData2"
+    with pytest.raises(ValueError) as exc:
+        parse_fielddata_template_filename(file_name=file_name, fielddata_name=fd_name)
+    assert exc.value.args[0] == f"filename {file_name} with fielddata_name {fd_name} does not match expected format"
+
+    # bad prefix
+    file_name = "Sim_SIMULATIONKEY_JOBINDEX_DEMO_fieldData_Channel0_5_23_Volume.fdat"
+    fd_name = "DEMO_fieldData"
+    with pytest.raises(ValueError) as exc:
+        parse_fielddata_template_filename(file_name=file_name, fielddata_name=fd_name)
+    assert exc.value.args[0] == f"filename {file_name} with fielddata_name {fd_name} does not match expected format"
+
+    # bad suffix
+    file_name = "SimID_SIMULATIONKEY_JOBINDEX_DEMO_fieldData_Channel0_5_23_Volume.fda"
+    fd_name = "DEMO_fieldData"
+    with pytest.raises(ValueError) as exc:
+        parse_fielddata_template_filename(file_name=file_name, fielddata_name=fd_name)
+    assert exc.value.args[0] == f"filename {file_name} with fielddata_name {fd_name} does not match expected format"
+
 
 def test_read_fielddata_file_metadata(fielddata_file_path: Path) -> None:
     fd_metadata = FieldDataFileMetadata(field_data_file=fielddata_file_path)