Merge PR #1315: Fix typechecking for top-level samtools commands

pysam-developers · Oct 24, 2024 · effb935 · effb935
2 parents ed699eb + 256adfc
commit effb935
Show file tree

Hide file tree

Showing 6 changed files with 104 additions and 264 deletions.
diff --git a/pysam/bcftools.py b/pysam/bcftools.py
@@ -1,65 +1,32 @@
-try:
-    from typing import Final
-    HAVE_FINAL = True
-except ImportError:
-    HAVE_FINAL = False
+import pysam.utils
 
-from pysam.utils import PysamDispatcher
+annotate = pysam.utils.PysamDispatcher('bcftools', 'annotate')
+call = pysam.utils.PysamDispatcher('bcftools', 'call')
+cnv = pysam.utils.PysamDispatcher('bcftools', 'cnv')
+concat = pysam.utils.PysamDispatcher('bcftools', 'concat')
+consensus = pysam.utils.PysamDispatcher('bcftools', 'consensus')
+convert = pysam.utils.PysamDispatcher('bcftools', 'convert')
+csq = pysam.utils.PysamDispatcher('bcftools', 'csq')
+filter = pysam.utils.PysamDispatcher('bcftools', 'filter')
+gtcheck = pysam.utils.PysamDispatcher('bcftools', 'gtcheck')
+head = pysam.utils.PysamDispatcher('bcftools', 'head')
+index = pysam.utils.PysamDispatcher('bcftools', 'index')
+isec = pysam.utils.PysamDispatcher('bcftools', 'isec')
+merge = pysam.utils.PysamDispatcher('bcftools', 'merge')
+mpileup = pysam.utils.PysamDispatcher('bcftools', 'mpileup')
+norm = pysam.utils.PysamDispatcher('bcftools', 'norm')
+plugin = pysam.utils.PysamDispatcher('bcftools', 'plugin')
+query = pysam.utils.PysamDispatcher('bcftools', 'query')
+reheader = pysam.utils.PysamDispatcher('bcftools', 'reheader')
+roh = pysam.utils.PysamDispatcher('bcftools', 'roh')
+sort = pysam.utils.PysamDispatcher('bcftools', 'sort')
+stats = pysam.utils.PysamDispatcher('bcftools', 'stats')
+view = pysam.utils.PysamDispatcher('bcftools', 'view')
 
-_BCFTOOLS_DISPATCH = [
-    "index",
-    "annotate",
-    "concat",
-    "convert",
-    "isec",
-    "merge",
-    "norm",
-    "plugin",
-    "query",
-    "reheader",
-    "sort",
-    "view",
-    "head",
-    "call",
-    "consensus",
-    "cnv",
-    "csq",
-    "filter",
-    "gtcheck",
-    "mpileup",
-    "roh",
-    "stats"]
-
-
-def _wrap_command(dispatch: str) -> PysamDispatcher:
-    return PysamDispatcher("bcftools", dispatch, ())
-
-
-if not HAVE_FINAL:
-    # instantiate bcftools commands as python functions
-    for cmd in _BCFTOOLS_DISPATCH:
-        globals()[cmd] = PysamDispatcher("bcftools", cmd, None)
-else:
-    # python >=3.8
-    index: Final[PysamDispatcher] = _wrap_command("index")
-    annotate: Final[PysamDispatcher] = _wrap_command("annotate")
-    concat: Final[PysamDispatcher] = _wrap_command("concat")
-    convert: Final[PysamDispatcher] = _wrap_command("convert")
-    isec: Final[PysamDispatcher] = _wrap_command("isec")
-    merge: Final[PysamDispatcher] = _wrap_command("merge")
-    norm: Final[PysamDispatcher] = _wrap_command("norm")
-    plugin: Final[PysamDispatcher] = _wrap_command("plugin")
-    query: Final[PysamDispatcher] = _wrap_command("query")
-    reheader: Final[PysamDispatcher] = _wrap_command("reheader")
-    sort: Final[PysamDispatcher] = _wrap_command("sort")
-    view: Final[PysamDispatcher] = _wrap_command("view")
-    head: Final[PysamDispatcher] = _wrap_command("head")
-    call: Final[PysamDispatcher] = _wrap_command("call")
-    consensus: Final[PysamDispatcher] = _wrap_command("consensus")
-    cnv: Final[PysamDispatcher] = _wrap_command("cnv")
-    csq: Final[PysamDispatcher] = _wrap_command("csq")
-    filter: Final[PysamDispatcher] = _wrap_command("filter")
-    gtcheck: Final[PysamDispatcher] = _wrap_command("gtcheck")
-    mpileup: Final[PysamDispatcher] = _wrap_command("mpileup")
-    roh: Final[PysamDispatcher] = _wrap_command("roh")
-    stats: Final[PysamDispatcher] = _wrap_command("stats")
+__all__ = [
+    'annotate', 'call', 'cnv', 'concat', 'consensus',
+    'convert', 'csq', 'filter', 'gtcheck', 'head',
+    'index', 'isec', 'merge', 'mpileup', 'norm',
+    'plugin', 'query', 'reheader', 'roh', 'sort',
+    'stats', 'view',
+]
diff --git a/pysam/libcutils.pyx b/pysam/libcutils.pyx
@@ -345,7 +345,7 @@ def _pysam_dispatch(collection,
         if collection == "bcftools":
             # in bcftools, most methods accept -o, the exceptions
             # are below:
-            if method not in ("index", "roh", "stats"):
+            if method not in ("head", "index", "roh", "stats"):
                 stdout_option = "-o {}"
         elif method in MAP_STDOUT_OPTIONS[collection]:
             # special case - samtools view -c outputs on stdout

diff --git a/pysam/samtools.py b/pysam/samtools.py
@@ -1,197 +1,59 @@
-import platform
-from typing import (
-    Callable,
-    List,
-    Tuple,
-    Iterable,
-    Union,
-)
-try:
-    from typing import Final
-    HAVE_FINAL = True
-except ImportError:
-    HAVE_FINAL = False
-
-from pysam.utils import PysamDispatcher
-
-
-# samtools command line options to export in python
-_SAMTOOLS_DISPATCH = {
-    # samtools 'documented' commands
-    "view": ("view", ()),
-    "head": ("head", ()),
-    "sort": ("sort", ()),
-    "mpileup": ("mpileup", ()),
-    "consensus": ("consensus", ()),
-    "depth": ("depth", ()),
-    "faidx": ("faidx", ()),
-    "fqidx": ("fqidx", ()),
-    "tview": ("tview", ()),
-    "index": ("index", ()),
-    "idxstats": ("idxstats", ()),
-    "fixmate": ("fixmate", ()),
-    "flagstat": ("flagstat", ()),
-    "calmd": ("calmd", ()),
-    "merge": ("merge", ()),
-    "markdup": ("markdup", ()),
-    "rmdup": ("rmdup", ()),
-    "reference": ("reference", ()),
-    "reheader": ("reheader", ()),
-    "reset": ("reset", ()),
-    "cat": ("cat", ()),
-    "targetcut": ("targetcut", ()),
-    "phase": ("phase", ()),
-    "bam2fq": ("bam2fq", ()),
-    "dict": ("dict", ()),
-    "addreplacerg": ("addreplacerg", ()),
-    "pad2unpad": ("pad2unpad", ()),
-    "depad": ("pad2unpad", ()),
-    "bedcov": ("bedcov", ()),
-    "coverage": ("coverage", ()),
-    "bamshuf": ("bamshuf", ()),
-    "collate": ("collate", ()),
-    "stats": ("stats", ()),
-    "fasta": ("fasta", ()),
-    "fastq": ("fastq", ()),
-    "cram_size": ("cram-size", ()),
-    "quickcheck": ("quickcheck", ()),
-    "split": ("split", ()),
-    "flags": ("flags", ()),
-    "ampliconclip": ("ampliconclip", ()),
-    "ampliconstats": ("ampliconstats", ()),
-    "version": ("version", ()),
-    "fqimport": ("import", ()),
-    "import_": ("import", ()),
-    "samples": ("samples", ()),
-}
-
-
-def _wrap_command(
-    dispatch: str,
-    parsers: Iterable[Tuple[str, Callable[[Union[str, List[str]]], Union[str, List[str]]]]],
-) -> PysamDispatcher:
-    return PysamDispatcher("samtools", dispatch, parsers)
-
-
-if not HAVE_FINAL:
-    # python 3.7
-    for key, options in _SAMTOOLS_DISPATCH.items():
-        cmd, parser = options
-        globals()[key] = PysamDispatcher("samtools", cmd, parser)
-
-    __all__ = list(_SAMTOOLS_DISPATCH)
-else:
-    # python >=3.8
-    view: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["view"][0], _SAMTOOLS_DISPATCH["view"][1])
-
-    head: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["head"][0], _SAMTOOLS_DISPATCH["head"][1])
-
-    sort: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["sort"][0], _SAMTOOLS_DISPATCH["sort"][1])
-
-    mpileup: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["mpileup"][0], _SAMTOOLS_DISPATCH["mpileup"][1])
-
-    consensus: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["consensus"][0],
-        _SAMTOOLS_DISPATCH["consensus"][1],
-    )
-
-    depth: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["depth"][0], _SAMTOOLS_DISPATCH["depth"][1])
-
-    faidx: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["faidx"][0], _SAMTOOLS_DISPATCH["faidx"][1])
-
-    fqidx: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["fqidx"][0], _SAMTOOLS_DISPATCH["fqidx"][1])
-
-    tview: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["tview"][0], _SAMTOOLS_DISPATCH["tview"][1])
-
-    index: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["index"][0], _SAMTOOLS_DISPATCH["index"][1])
-
-    idxstats: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["idxstats"][0], _SAMTOOLS_DISPATCH["idxstats"][1])
-
-    fixmate: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["fixmate"][0], _SAMTOOLS_DISPATCH["fixmate"][1])
-
-    flagstat: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["flagstat"][0], _SAMTOOLS_DISPATCH["flagstat"][1])
-
-    calmd: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["calmd"][0], _SAMTOOLS_DISPATCH["calmd"][1])
-
-    merge: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["merge"][0], _SAMTOOLS_DISPATCH["merge"][1])
-
-    markdup: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["markdup"][0], _SAMTOOLS_DISPATCH["markdup"][1])
-
-    rmdup: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["rmdup"][0], _SAMTOOLS_DISPATCH["rmdup"][1])
-
-    reference: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["reference"][0],
-        _SAMTOOLS_DISPATCH["reference"][1],
-    )
-
-    reheader: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["reheader"][0], _SAMTOOLS_DISPATCH["reheader"][1])
-
-    reset: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["reset"][0], _SAMTOOLS_DISPATCH["reset"][1])
-
-    cat: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["cat"][0], _SAMTOOLS_DISPATCH["cat"][1])
-
-    targetcut: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["targetcut"][0],
-        _SAMTOOLS_DISPATCH["targetcut"][1],
-    )
-
-    phase: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["phase"][0], _SAMTOOLS_DISPATCH["phase"][1])
-
-    bam2fq: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["bam2fq"][0], _SAMTOOLS_DISPATCH["bam2fq"][1])
-
-    dict: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["dict"][0], _SAMTOOLS_DISPATCH["dict"][1])
-
-    addreplacerg: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["addreplacerg"][0],
-        _SAMTOOLS_DISPATCH["addreplacerg"][1],
-    )
-
-    pad2unpad: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["pad2unpad"][0],
-        _SAMTOOLS_DISPATCH["pad2unpad"][1],
-    )
-
-    depad: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["depad"][0], _SAMTOOLS_DISPATCH["depad"][1])
-
-    bedcov: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["bedcov"][0], _SAMTOOLS_DISPATCH["bedcov"][1])
-
-    coverage: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["coverage"][0], _SAMTOOLS_DISPATCH["coverage"][1])
-
-    bamshuf: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["bamshuf"][0], _SAMTOOLS_DISPATCH["bamshuf"][1])
-
-    collate: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["collate"][0], _SAMTOOLS_DISPATCH["collate"][1])
-
-    stats: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["stats"][0], _SAMTOOLS_DISPATCH["stats"][1])
-
-    fasta: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["fasta"][0], _SAMTOOLS_DISPATCH["fasta"][1])
-
-    fastq: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["fastq"][0], _SAMTOOLS_DISPATCH["fastq"][1])
-
-    cram_size: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["cram_size"][0], _SAMTOOLS_DISPATCH["cram_size"][1])
-
-    quickcheck: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["quickcheck"][0],
-        _SAMTOOLS_DISPATCH["quickcheck"][1],
-    )
-
-    split: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["split"][0], _SAMTOOLS_DISPATCH["split"][1])
-
-    flags: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["flags"][0], _SAMTOOLS_DISPATCH["flags"][1])
-
-    ampliconclip: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["ampliconclip"][0],
-        _SAMTOOLS_DISPATCH["ampliconclip"][1],
-    )
-
-    ampliconstats: Final[PysamDispatcher] = _wrap_command(
-        _SAMTOOLS_DISPATCH["ampliconstats"][0],
-        _SAMTOOLS_DISPATCH["ampliconstats"][1],
-    )
-
-    version: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["version"][0], _SAMTOOLS_DISPATCH["version"][1])
-
-    fqimport: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["fqimport"][0], _SAMTOOLS_DISPATCH["fqimport"][1])
-
-    import_: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["import_"][0], _SAMTOOLS_DISPATCH["import_"][1])
-
-    samples: Final[PysamDispatcher] = _wrap_command(_SAMTOOLS_DISPATCH["samples"][0], _SAMTOOLS_DISPATCH["samples"][1])
+import pysam.utils
+
+addreplacerg = pysam.utils.PysamDispatcher('samtools', 'addreplacerg')
+ampliconclip = pysam.utils.PysamDispatcher('samtools', 'ampliconclip')
+ampliconstats = pysam.utils.PysamDispatcher('samtools', 'ampliconstats')
+bam2fq = pysam.utils.PysamDispatcher('samtools', 'bam2fq')
+bamshuf = pysam.utils.PysamDispatcher('samtools', 'bamshuf')
+bedcov = pysam.utils.PysamDispatcher('samtools', 'bedcov')
+calmd = pysam.utils.PysamDispatcher('samtools', 'calmd')
+cat = pysam.utils.PysamDispatcher('samtools', 'cat')
+collate = pysam.utils.PysamDispatcher('samtools', 'collate')
+consensus = pysam.utils.PysamDispatcher('samtools', 'consensus')
+coverage = pysam.utils.PysamDispatcher('samtools', 'coverage')
+cram_size = pysam.utils.PysamDispatcher('samtools', 'cram-size')
+depad = pysam.utils.PysamDispatcher('samtools', 'depad')
+depth = pysam.utils.PysamDispatcher('samtools', 'depth')
+dict = pysam.utils.PysamDispatcher('samtools', 'dict')
+faidx = pysam.utils.PysamDispatcher('samtools', 'faidx')
+fasta = pysam.utils.PysamDispatcher('samtools', 'fasta')
+fastq = pysam.utils.PysamDispatcher('samtools', 'fastq')
+fixmate = pysam.utils.PysamDispatcher('samtools', 'fixmate')
+flags = pysam.utils.PysamDispatcher('samtools', 'flags')
+flagstat = pysam.utils.PysamDispatcher('samtools', 'flagstat')
+fqidx = pysam.utils.PysamDispatcher('samtools', 'fqidx')
+fqimport = pysam.utils.PysamDispatcher('samtools', 'import')
+head = pysam.utils.PysamDispatcher('samtools', 'head')
+idxstats = pysam.utils.PysamDispatcher('samtools', 'idxstats')
+index = pysam.utils.PysamDispatcher('samtools', 'index')
+markdup = pysam.utils.PysamDispatcher('samtools', 'markdup')
+merge = pysam.utils.PysamDispatcher('samtools', 'merge')
+mpileup = pysam.utils.PysamDispatcher('samtools', 'mpileup')
+pad2unpad = pysam.utils.PysamDispatcher('samtools', 'pad2unpad')
+phase = pysam.utils.PysamDispatcher('samtools', 'phase')
+quickcheck = pysam.utils.PysamDispatcher('samtools', 'quickcheck')
+reference = pysam.utils.PysamDispatcher('samtools', 'reference')
+reheader = pysam.utils.PysamDispatcher('samtools', 'reheader')
+reset = pysam.utils.PysamDispatcher('samtools', 'reset')
+rmdup = pysam.utils.PysamDispatcher('samtools', 'rmdup')
+samples = pysam.utils.PysamDispatcher('samtools', 'samples')
+sort = pysam.utils.PysamDispatcher('samtools', 'sort')
+split = pysam.utils.PysamDispatcher('samtools', 'split')
+stats = pysam.utils.PysamDispatcher('samtools', 'stats')
+targetcut = pysam.utils.PysamDispatcher('samtools', 'targetcut')
+tview = pysam.utils.PysamDispatcher('samtools', 'tview')
+version = pysam.utils.PysamDispatcher('samtools', 'version')
+view = pysam.utils.PysamDispatcher('samtools', 'view')
+
+__all__ = [
+    'addreplacerg', 'ampliconclip', 'ampliconstats',
+    'bam2fq', 'bamshuf', 'bedcov', 'calmd', 'cat',
+    'collate', 'consensus', 'coverage', 'cram_size',
+    'depad', 'depth', 'dict', 'faidx', 'fasta',
+    'fastq', 'fixmate', 'flags', 'flagstat', 'fqidx',
+    'fqimport', 'head', 'idxstats', 'index',
+    'markdup', 'merge', 'mpileup', 'pad2unpad',
+    'phase', 'quickcheck', 'reference', 'reheader',
+    'reset', 'rmdup', 'samples', 'sort', 'split',
+    'stats', 'targetcut', 'tview', 'version', 'view',
+]
diff --git a/pysam/utils.py b/pysam/utils.py
@@ -1,6 +1,7 @@
 from typing import (
     Callable,
     List,
+    Optional,
     Tuple,
     Iterable,
     Union,
@@ -48,7 +49,7 @@ def __init__(
         self,
         collection: str,
         dispatch: str,
-        parsers: Iterable[Tuple[str, Callable[[Union[str, List[str]]], Union[str, List[str]]]]],
+        parsers: Optional[Iterable[Tuple[str, Callable[[Union[str, List[str]]], Union[str, List[str]]]]]] = None,
     ):
         self.collection = collection
         self.dispatch = dispatch

diff --git a/tests/samtools_test.py b/tests/samtools_test.py
@@ -179,8 +179,6 @@ def check_statement(self, statement):
 
         command = self.get_command(statement)
 
-        # self.assertTrue(command in pysam.SAMTOOLS_DISPATCH)
-
         targets = [x for x in parts if "%(out)s" in x]
         samtools_targets = [x % r_samtools for x in targets]
         pysam_targets = [x % r_pysam for x in targets]