feat: ChemoMetec Nucleoview - add support for tab separated files (#1015)

Co-authored-by: Nathan Stender <nathan.stender@benchling.com>

Author: joshua-benchling

src/allotropy/parsers/chemometec_nucleoview/nucleoview_reader.py

@@ -1,3 +1,5 @@
+from io import StringIO
+
 import pandas as pd
 
 from allotropy.exceptions import AllotropeConversionError, AllotropeParsingError
@@ -10,9 +12,29 @@ class NucleoviewReader:
 
     @classmethod
     def read(cls, contents: IOType) -> pd.DataFrame:
-        df = read_csv(
-            contents, sep="[;,]+", engine="python", skipinitialspace=True, index_col=0
-        )
+        content = contents.read()
+        content_str = content.decode("UTF-8") if isinstance(content, bytes) else content
+        separators = [":\t", "[;,]+"]
+        last_error = None
+
+        for sep in separators:
+            try:
+                df = read_csv(
+                    StringIO(content_str),
+                    sep=sep,
+                    engine="python",
+                    skipinitialspace=True,
+                    index_col=0,
+                )
+                if df.shape[1] >= 1:
+                    break
+            except AllotropeParsingError as e:
+                last_error = e
+                continue
+        else:
+            if last_error:
+                raise last_error
+
         try:
             df = df[:-1].dropna(axis="index", how="all").T
         except KeyError as e:

src/allotropy/parsers/chemometec_nucleoview/nucleoview_structure.py

@@ -107,7 +107,8 @@ def _converted_value_or_none(key: str) -> float | None:
             Measurement(
                 measurement_identifier=random_uuid_str(),
                 timestamp=timestamp or DEFAULT_EPOCH_TIMESTAMP,
-                sample_identifier=data[str, "Image"].split("-")[3],
+                sample_identifier=data.get(str, "Sample ID")
+                or data[str, "Image"].split("-")[3],
                 cell_density_dilution_factor=data.get(float, "Multiplication factor"),
                 viability=data[float, "Viability (%)"],
                 # Cell counts are measured in cells/mL, but reported in millions of cells/mL

tests/parsers/chemometec_nucleoview/testdata/chemometec_nucleoview_example08.csv

@@ -0,0 +1,22 @@
+Source ID:	99999999-00009
+Time (UTC):	1999-09-99 09:49:29
+Sample ID:	1111_viability count new 99x_pass 99
+Protocol:	Count & Viability; 1.49
+Software version:	9.9.9.99
+Secure Mode status:	Enable
+Signed by:	
+Sample Media:	Via2-Cassette
+User:	gih
+Operator:	gih
+Instrument:	S/N:99999999999
+Tags:	
+Comments:	
+Total (cells/ml):	1.12E+06
+Live (cells/ml):	9.28E+05
+Dead (cells/ml):	1.90E+05
+Viability (%):	83.0
+Diameter (um):	9.7
+Aggregates (%):	0.0
+Debris Index:	50
+Dilution Factor:	10
+Status:	OK

tests/parsers/chemometec_nucleoview/testdata/chemometec_nucleoview_example08.json

@@ -0,0 +1,76 @@
+{
+    "$asm.manifest": "http://purl.allotrope.org/manifests/cell-counting/REC/2024/09/cell-counting.manifest",
+    "cell counting aggregate document": {
+        "cell counting document": [
+            {
+                "measurement aggregate document": {
+                    "measurement document": [
+                        {
+                            "device control aggregate document": {
+                                "device control document": [
+                                    {
+                                        "device type": "dark field imager (cell counter)",
+                                        "detection type": "dark field"
+                                    }
+                                ]
+                            },
+                            "measurement identifier": "CHEMOMETEC_NUCLEOVIEW_TEST_ID_0",
+                            "measurement time": "1970-01-01T00:00:00+00:00",
+                            "processed data aggregate document": {
+                                "processed data document": [
+                                    {
+                                        "viability (cell counter)": {
+                                            "value": 83.0,
+                                            "unit": "%"
+                                        },
+                                        "viable cell density (cell counter)": {
+                                            "value": 0.928,
+                                            "unit": "10^6 cells/mL"
+                                        },
+                                        "total cell density (cell counter)": {
+                                            "value": 1.12,
+                                            "unit": "10^6 cells/mL"
+                                        },
+                                        "dead cell density (cell counter)": {
+                                            "value": 0.19,
+                                            "unit": "10^6 cells/mL"
+                                        }
+                                    }
+                                ]
+                            },
+                            "sample document": {
+                                "sample identifier": "1111_viability count new 99x_pass 99"
+                            },
+                            "custom information document": {
+                                "User": "gih",
+                                "Secure Mode status": "Enable",
+                                "Sample Media": "Via2-Cassette",
+                                "Software version": "9.9.9.99",
+                                "Time (UTC)": "1999-09-99 09:49:29",
+                                "Debris Index": 50.0,
+                                "Instrument": "S/N:99999999999",
+                                "Protocol": "Count & Viability; 1.49",
+                                "Aggregates (%)": 0.0,
+                                "Dilution Factor": 10.0,
+                                "Diameter (um)": 9.7
+                            }
+                        }
+                    ]
+                },
+                "analyst": "gih"
+            }
+        ],
+        "data system document": {
+            "ASM file identifier": "chemometec_nucleoview_example08.json",
+            "data system instance identifier": "N/A",
+            "file name": "chemometec_nucleoview_example08.csv",
+            "UNC path": "tests/parsers/chemometec_nucleoview/testdata/chemometec_nucleoview_example08.csv",
+            "ASM converter name": "allotropy_chemometec_nucleoview",
+            "ASM converter version": "0.1.93",
+            "software name": "NucleoView"
+        },
+        "device system document": {
+            "model number": "NucleoCounter NC-200"
+        }
+    }
+}