Copied over most code from another of my projects

src/automlst/engine/local/abif.py

@@ -0,0 +1,104 @@
+import asyncio
+from numbers import Number
+from os import path
+from typing import Sequence, Union
+from automlst.engine.data.genomics import SangerTraceData
+from Bio.SeqRecord import SeqRecord
+from Bio import SeqIO
+
+
+def _biopython_read_abif_sequence(seq_path: str) -> SeqRecord:
+    with open(seq_path, "rb") as seq_handle:
+        return SeqIO.read(seq_handle, "abi")
+
+
+async def read_abif(seq_path: str) -> SangerTraceData:
+    ext = path.splitext(seq_path)[1]
+    if ext.lower() != ".ab1" and ext.lower() != "abi":
+        raise ValueError(
+            'seq_path must have file extension of "ab1", or "abi".')
+    biopython_seq = await asyncio.to_thread(_biopython_read_abif_sequence, seq_path)
+    biopython_annotations = biopython_seq.annotations
+
+    # Lot of type ignoring since Biopython did not define their typing.
+    biopython_abif_raw = biopython_annotations["abif_raw"] # type: ignore
+    trace_data = SangerTraceData(
+        biopython_seq.seq,
+        biopython_abif_raw.get("APFN2"), # type: ignore
+        biopython_abif_raw.get("APrN1"), # type: ignore
+        biopython_abif_raw.get("APrV1"), # type: ignore
+        biopython_abif_raw.get("APrX1"), # type: ignore
+        biopython_abif_raw.get("APXV1"), # type: ignore
+        biopython_abif_raw.get("CMNT1"), # type: ignore
+        biopython_abif_raw.get("CpEP1"), # type: ignore
+        biopython_abif_raw.get("CTID1"), # type: ignore
+        biopython_abif_raw.get("CTNM1"), # type: ignore
+        biopython_abif_raw.get("CTTL1"), # type: ignore
+        biopython_abif_raw.get("DATA1"), # type: ignore
+        biopython_abif_raw.get("DATA2"), # type: ignore
+        biopython_abif_raw.get("DATA3"), # type: ignore
+        biopython_abif_raw.get("DATA4"), # type: ignore
+        biopython_abif_raw.get("DATA5"), # type: ignore
+        biopython_abif_raw.get("DATA6"), # type: ignore
+        biopython_abif_raw.get("DATA7"), # type: ignore
+        biopython_abif_raw.get("DATA8"), # type: ignore
+        biopython_abif_raw.get("DSam1"), # type: ignore
+        biopython_abif_raw.get("DyeN1"), # type: ignore
+        biopython_abif_raw.get("DyeN2"), # type: ignore
+        biopython_abif_raw.get("DyeN3"), # type: ignore
+        biopython_abif_raw.get("DyeN4"), # type: ignore
+        biopython_abif_raw.get("DyeW1"), # type: ignore
+        biopython_abif_raw.get("DyeW2"), # type: ignore
+        biopython_abif_raw.get("DyeW3"), # type: ignore
+        biopython_abif_raw.get("DyeW4"), # type: ignore
+        biopython_abif_raw.get("DySN1"), # type: ignore
+        biopython_abif_raw.get("EPVt1"), # type: ignore
+        biopython_abif_raw.get("EVNT1"), # type: ignore
+        biopython_abif_raw.get("EVNT2"), # type: ignore
+        biopython_abif_raw.get("EVNT3"), # type: ignore
+        biopython_abif_raw.get("EVNT4"), # type: ignore
+        biopython_abif_raw.get("FWO_1"), # type: ignore
+        biopython_abif_raw.get("GTyp1"), # type: ignore
+        biopython_abif_raw.get("InSc1"), # type: ignore
+        biopython_abif_raw.get("InVt1"), # type: ignore
+        biopython_abif_raw.get("LANE1"), # type: ignore
+        biopython_abif_raw.get("LIMS1"), # type: ignore
+        biopython_abif_raw.get("LNTD1"), # type: ignore
+        biopython_abif_raw.get("LsrP1"), # type: ignore
+        biopython_abif_raw.get("MCHN1"), # type: ignore
+        biopython_abif_raw.get("MODF1"), # type: ignore
+        biopython_abif_raw.get("MODL1"), # type: ignore
+        biopython_abif_raw.get("NAVG1"), # type: ignore
+        biopython_abif_raw.get("NLNE1"), # type: ignore
+        biopython_abif_raw.get("OfSc1"), # type: ignore
+        biopython_abif_raw.get("PDMF1"), # type: ignore
+        biopython_abif_raw.get("PXLB1"), # type: ignore
+        biopython_abif_raw.get("RGCm1"), # type: ignore
+        biopython_abif_raw.get("RGNm1"), # type: ignore
+        biopython_abif_raw.get("RMdV1"), # type: ignore
+        biopython_abif_raw.get("RMdX1"), # type: ignore
+        biopython_abif_raw.get("RMXV1"), # type: ignore
+        biopython_abif_raw.get("RPrN1"), # type: ignore
+        biopython_abif_raw.get("RPrV1"), # type: ignore
+        biopython_abif_raw.get("RUND1"), # type: ignore
+        biopython_abif_raw.get("RUND2"), # type: ignore
+        biopython_abif_raw.get("RUND3"), # type: ignore
+        biopython_abif_raw.get("RUND4"), # type: ignore
+        biopython_abif_raw.get("RunN1"), # type: ignore
+        biopython_abif_raw.get("RUNT1"), # type: ignore
+        biopython_abif_raw.get("RUNT2"), # type: ignore
+        biopython_abif_raw.get("RUNT3"), # type: ignore
+        biopython_abif_raw.get("RUNT4"), # type: ignore
+        biopython_abif_raw.get("Satd"), # type: ignore
+        biopython_abif_raw.get("Scal1"), # type: ignore
+        biopython_abif_raw.get("SCAN1"), # type: ignore
+        biopython_abif_raw.get("SMED1"), # type: ignore
+        biopython_abif_raw.get("SMLt"), # type: ignore
+        biopython_abif_raw.get("SMPL1"), # type: ignore
+        biopython_abif_raw.get("SVER1"), # type: ignore
+        biopython_abif_raw.get("SVER3"), # type: ignore
+        biopython_abif_raw.get("Tmpr1"), # type: ignore
+        biopython_abif_raw.get("TUBE"), # type: ignore
+        biopython_abif_raw.get("User") # type: ignore
+    )
+    return trace_data

src/automlst/engine/local/csv.py

@@ -0,0 +1,31 @@
+import csv
+from io import TextIOWrapper
+from os import PathLike
+from typing import AsyncIterable, Iterable, Mapping, Sequence, Union
+
+from automlst.engine.data.MLST import Allele, MLSTProfile
+
+
+def loci_alleles_variants_from_loci(alleles_map: Mapping[str, Sequence[Allele]]):
+    result_dict: dict[str, list[str]] = {}
+    for loci, alleles in alleles_map.items():
+        result_dict[loci] = list()
+        for allele in alleles:
+            result_dict[loci].append(allele.allele_variant)
+    return result_dict
+
+
+async def write_mlst_profiles_as_csv(mlst_profiles_iterable: Iterable[MLSTProfile], handle: Union[str, bytes, PathLike[str], PathLike[bytes]]):
+    mlst_profiles = list(mlst_profiles_iterable)
+    header = ["st", "clonal-complex", *mlst_profiles[0].alleles.keys()]
+    with open(handle, "w", newline='') as filehandle:
+        writer = csv.DictWriter(filehandle, fieldnames=header)
+        writer.writeheader()
+        for mlst_profile in mlst_profiles:
+            row_dictionary = {
+                "st": mlst_profile.sequence_type,
+                "clonal-complex": mlst_profile.clonal_complex,
+                **loci_alleles_variants_from_loci(mlst_profile.alleles)
+            }
+
+            writer.writerow(rowdict=row_dictionary)

src/automlst/engine/local/fasta.py

@@ -0,0 +1,11 @@
+import asyncio
+from io import TextIOWrapper
+from typing import Any, AsyncGenerator, Generator, Sequence, Union
+from Bio import SeqIO
+
+from automlst.engine.data.genomics import NamedString
+
+async def read_fasta(handle: Union[str, TextIOWrapper]) -> AsyncGenerator[NamedString, Any]:
+    fasta_sequences = asyncio.to_thread(SeqIO.parse, handle=handle, format="fasta")
+    for fasta_sequence in await fasta_sequences:
+        yield NamedString(fasta_sequence.id, str(fasta_sequence.seq))