MRG: add some argparse utility wrappers

.gitignore

@@ -1,3 +1,4 @@
 *~
 __pycache__
 sourmash_utils.egg-info
+dist/

pyproject.toml

@@ -3,7 +3,7 @@ name = "sourmash_utils"
 description = "Python utilities for sourmash plugins and scripts"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.1"
+version = "0.2"
 
 dependencies = ["sourmash>=4.8.8,<5"]
 

src/sourmash_utils.py

@@ -1,9 +1,12 @@
 "Python utilities for sourmash plugins and scripts."
 import sourmash
+from sourmash import sourmash_args
+from sourmash.cli import utils as sourmash_cli
 
-__all__ = ['FracMinHash']
+__all__ = ['FracMinHash', 'add_standard_minhash_args']
 
 class FracMinHash(sourmash.MinHash):
+    "An updated MinHash class with nicer constructor arguments."
     def __init__(self, *,
                  ksize=31,
                  scaled=1000,
@@ -16,6 +19,16 @@ def __init__(self, *,
         assert 'dayhoff' not in kwargs
         assert 'hp' not in kwargs
 
+        if scaled is None:
+            raise ValueError(f"must specify scaled")
+        else:
+            scaled = int(scaled)
+
+        if ksize is None:
+            raise ValueError(f"must specify ksize")
+        else:
+            ksize = int(ksize)
+
         if moltype == 'DNA':
             kwargs['is_protein'] = False
             kwargs['dayhoff'] = False
@@ -34,3 +47,38 @@ def __init__(self, *,
                          scaled=scaled,
                          track_abundance=track_abundance,
                          **kwargs)
+
+    def __str__(self):
+        return f"k={self.ksize} scaled={self.scaled} moltype={self.moltype}"
+
+
+def add_standard_minhash_args(parser):
+    sourmash_cli.add_construct_moltype_args(parser)
+    sourmash_cli.add_ksize_arg(parser, default=31)
+    sourmash_cli.add_scaled_arg(parser, default=1000)
+
+
+def create_minhash_from_args(args, *, track_abundance=False, **defaults):
+    default_moltype = defaults.get('moltype')
+    moltype = sourmash_args.calculate_moltype(args, default=default_moltype)
+    ksize = args.ksize or defaults.get('ksize', 31)
+    scaled = args.scaled or defaults.get('scaled', 1000)
+
+    return FracMinHash(moltype=moltype,
+                       ksize=ksize,
+                       scaled=scaled,
+                       track_abundance=track_abundance)
+
+
+def load_index_and_select(filename, minhash_obj, *, raise_on_empty=True):
+    """Load a sourmash Index object from filename, select sketches compatible
+    with minhash_obj.
+    """
+    idx = sourmash.load_file_as_index(filename)
+    idx = idx.select(ksize=minhash_obj.ksize,
+                     moltype=minhash_obj.moltype,
+                     scaled=minhash_obj.scaled,
+                     abund=minhash_obj.track_abundance)
+    if not idx:
+        raise ValueError(f"no matching sketches in '{filename}' for k={minhash_obj.ksize} moltype={minhash_obj.moltype} scaled={minhash_obj.scaled}")
+    return idx