[MRG] refactor argparse.FileType out of sourmash argument handling.

sourmash/cli/categorize.py

@@ -29,7 +29,7 @@ def subparser(subparsers):
     add_moltype_args(subparser)
 
     # TODO: help messages in these
-    subparser.add_argument('--csv', type=argparse.FileType('at'))
+    subparser.add_argument('--csv', help='output summary CSV to this file')
     subparser.add_argument('--load-csv', default=None)
 
 

sourmash/cli/compare.py

@@ -1,7 +1,5 @@
 """compare genomes"""
 
-from argparse import FileType
-
 from sourmash.cli.utils import add_ksize_arg, add_moltype_args
 
 
@@ -29,7 +27,7 @@ def subparser(subparsers):
         help='compare all signatures underneath directories'
     )
     subparser.add_argument(
-        '--csv', metavar='F', type=FileType('w'),
+        '--csv', metavar='F',
         help='write matrix to specified file in CSV format (with column '
         'headers)'
     )

sourmash/cli/gather.py

@@ -1,8 +1,6 @@
 """search a metagenome signature for multiple non-
                                   overlapping matches"""
 
-from argparse import FileType
-
 from sourmash.cli.utils import add_ksize_arg, add_moltype_args
 
 
@@ -25,11 +23,11 @@ def subparser(subparsers):
         help='search all signatures underneath directories'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
+        '-o', '--output', metavar='FILE',
         help='output CSV containing matches to this file'
     )
     subparser.add_argument(
-        '--save-matches', metavar='FILE', type=FileType('wt'),
+        '--save-matches', metavar='FILE',
         help='save the matched signatures from the database to the '
         'specified file'
     )
@@ -38,7 +36,7 @@ def subparser(subparsers):
         help='threshold (in bp) for reporting results (default=50,000)'
     )
     subparser.add_argument(
-        '--output-unassigned', metavar='FILE', type=FileType('wt'),
+        '--output-unassigned', metavar='FILE',
         help='output unassigned portions of the query as a signature to the '
         'specified file'
     )

sourmash/cli/import_csv.py

@@ -1,16 +1,14 @@
 """'sourmash import_csv' description goes here"""
 
-from argparse import FileType
 import sys
 
 
 def subparser(subparsers):
     subparser = subparsers.add_parser('import_csv')
     subparser.add_argument('mash_csvfile', help='CSV file with mash sketches')
     subparser.add_argument(
-        '-o', '--output', type=FileType('wt'),
-        default=sys.stdout,
-        help='save signature generated from data here'
+        '-o', '--output',
+        help='save signature generated from data to this file (default stdout)'
     )
 
 

sourmash/cli/lca/classify.py

@@ -1,7 +1,5 @@
 """classify genomes"""
 
-from argparse import FileType
-
 
 def subparser(subparsers):
     subparser = subparsers.add_parser('classify')
@@ -17,9 +15,8 @@ def subparser(subparsers):
         help='output debugging output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        help='output CSV to the specified file; by default output to terminal '
-        '(standard output)'
+        '-o', '--output', metavar='FILE', default='-',
+        help='output CSV to the specified file; by default output to stdout'
     )
     subparser.add_argument('--scaled', type=float)
     subparser.add_argument(

sourmash/cli/lca/gather.py

@@ -1,7 +1,5 @@
 """classify metagenomes"""
 
-from argparse import FileType
-
 
 def subparser(subparsers):
     subparser = subparsers.add_parser('gather')
@@ -15,11 +13,11 @@ def subparser(subparsers):
         help='output debugging output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
+        '-o', '--output', metavar='FILE',
         help='output CSV containing matches to this file'
     )
     subparser.add_argument(
-        '--output-unassigned', metavar='FILE', type=FileType('wt'),
+        '--output-unassigned', metavar='FILE',
         help='output unassigned portions of the query as a signature to this '
         'file'
     )

sourmash/cli/lca/summarize.py

@@ -1,7 +1,5 @@
 """summarize mixture"""
 
-from argparse import FileType
-
 
 def subparser(subparsers):
     subparser = subparsers.add_parser('summarize')
@@ -13,7 +11,7 @@ def subparser(subparsers):
         help='load all signatures underneath directories'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
+        '-o', '--output', metavar='FILE',
         help='file to which CSV output will be written'
     )
     subparser.add_argument('--scaled', metavar='FLOAT', type=float)

sourmash/cli/search.py

@@ -1,7 +1,5 @@
 """search a signature against a list of signatures"""
 
-from argparse import FileType
-
 from sourmash.cli.utils import add_ksize_arg, add_moltype_args
 
 
@@ -27,7 +25,7 @@ def subparser(subparsers):
         help='minimum threshold for reporting matches; default=0.08'
     )
     subparser.add_argument(
-        '--save-matches', metavar='FILE', type=FileType('wt'),
+        '--save-matches', metavar='FILE',
         help='output matching signatures to the specified file'
     )
     subparser.add_argument(
@@ -52,7 +50,7 @@ def subparser(subparsers):
         help='downsample query to this scaled factor (yields greater speed)'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
+        '-o', '--output', metavar='FILE',
         help='output CSV containing matches to this file'
     )
     add_ksize_arg(subparser, 31)

sourmash/cli/sig/describe.py

@@ -1,6 +1,5 @@
 """show details of signature"""
 
-from argparse import FileType
 import csv
 
 import sourmash
@@ -15,79 +14,11 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '--csv', metavar='FILE', type=FileType('wt'),
+        '--csv', metavar='FILE',
         help='output information to a CSV file'
     )
 
 
-def describe(signatures, quiet=True, csvout=None):
-    siglist = []
-    for sigfile in signatures:
-        this_siglist = []
-        try:
-            this_siglist = sourmash.load_signatures(sigfile, quiet=quiet, do_raise=True)
-            for k in this_siglist:
-                siglist.append((k, sigfile))
-        except Exception as exc:
-            error('\nError while reading signatures from {}:'.format(sigfile))
-            error(str(exc))
-            error('(continuing)')
-
-        notify('loaded {} signatures from {}...', len(siglist), sigfile,
-               end='\r')
-
-    notify('loaded {} signatures total.', len(siglist))
-
-    w = None
-    if csvout:
-        w = csv.DictWriter(
-            csvout, [
-                'signature_file', 'md5', 'ksize', 'moltype', 'num', 'scaled',
-                'n_hashes', 'seed', 'with_abundance','name', 'filename',
-                'license'
-            ],
-            extrasaction='ignore'
-        )
-        w.writeheader()
-
-    # extract info, write as appropriate.
-    for (sig, signature_file) in siglist:
-        mh = sig.minhash
-        ksize = mh.ksize
-        moltype = 'DNA'
-        if mh.is_protein:
-            if mh.dayhoff:
-                moltype = 'dayhoff'
-            elif mh.hp:
-                moltype = 'hp'
-            else:
-                moltype = 'protein'
-        scaled = mh.scaled
-        num = mh.num
-        seed = mh.seed
-        n_hashes = len(mh)
-        with_abundance = 0
-        if mh.track_abundance:
-            with_abundance = 1
-        md5 = sig.md5sum()
-        name = sig.name()
-        filename = sig.filename
-        license = sig.license
-
-        if w:
-            w.writerow(locals())
-
-        print_results('''\
----
-signature filename: {signature_file}
-signature: {name}
-source file: {filename}
-md5: {md5}
-k={ksize} molecule={moltype} num={num} scaled={scaled} seed={seed} track_abundance={with_abundance}
-size: {n_hashes}
-signature license: {license}
-''', **locals())
-
-
 def main(args):
-    describe(args.signatures, quiet=args.quiet, csvout=args.csv)
+    import sourmash
+    return sourmash.sig.__main__.describe(args)

sourmash/cli/sig/downsample.py

@@ -1,6 +1,5 @@
 """downsample one or more signatures"""
 
-from argparse import FileType
 import sys
 
 from sourmash.cli.utils import add_moltype_args, add_ksize_arg
@@ -22,9 +21,8 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        default=sys.stdout,
-        help='output signature to this file'
+        '-o', '--output', metavar='FILE',
+        help='output signature to this file (default stdout)'
     )
     add_ksize_arg(subparser, 31)
     add_moltype_args(subparser)

sourmash/cli/sig/export.py

@@ -1,6 +1,5 @@
 """export a signature, e.g. to mash"""
 
-from argparse import FileType
 import sys
 
 from sourmash.cli.utils import add_ksize_arg, add_moltype_args
@@ -14,9 +13,8 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        default=sys.stdout,
-        help='output signature to this file'
+        '-o', '--output', metavar='FILE',
+        help='output signature to this file (default stdout)'
     )
     add_ksize_arg(subparser, 31)
     add_moltype_args(subparser)

sourmash/cli/sig/extract.py

@@ -1,6 +1,5 @@
 """extract one or more signatures"""
 
-from argparse import FileType
 import sys
 
 from sourmash.cli.utils import add_moltype_args, add_ksize_arg
@@ -14,9 +13,8 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        default=sys.stdout,
-        help='output signature to this file'
+        '-o', '--output', metavar='FILE',
+        help='output signature to this file (default stdout)'
     )
     subparser.add_argument(
         '--md5', default=None,

sourmash/cli/sig/filter.py

@@ -1,6 +1,5 @@
 """filter k-mers on abundance"""
 
-from argparse import FileType
 import sys
 
 from sourmash.cli.utils import add_moltype_args, add_ksize_arg
@@ -14,9 +13,8 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        default=sys.stdout,
-        help='output signature to this file'
+        '-o', '--output', metavar='FILE',
+        help='output signature to this file (default stdout)'
     )
     subparser.add_argument(
         '--md5', type=str, default=None,

sourmash/cli/sig/flatten.py

@@ -1,6 +1,5 @@
 """remove abundances"""
 
-from argparse import FileType
 import sys
 
 from sourmash.cli.utils import add_moltype_args, add_ksize_arg
@@ -14,9 +13,8 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        default=sys.stdout,
-        help='output signature to this file'
+        '-o', '--output', metavar='FILE',
+        help='output signature to this file (default stdout)'
     )
     subparser.add_argument(
         '--md5', default=None,

sourmash/cli/sig/ingest.py

@@ -1,6 +1,5 @@
 """import a mash or other signature"""
 
-from argparse import FileType
 import sys
 
 
@@ -15,9 +14,8 @@ def subparser(subparsers):
             help='suppress non-error output'
         )
         subparser.add_argument(
-            '-o', '--output', metavar='FILE', type=FileType('wt'),
-            default=sys.stdout,
-            help='output signature to this file'
+            '-o', '--output', metavar='FILE',
+            help='output signature to this file (default stdout)'
         )
 
 

sourmash/cli/sig/intersect.py

@@ -1,6 +1,5 @@
 """intersect one or more signatures"""
 
-from argparse import FileType
 import sys
 
 from sourmash.cli.utils import add_moltype_args, add_ksize_arg
@@ -14,9 +13,8 @@ def subparser(subparsers):
         help='suppress non-error output'
     )
     subparser.add_argument(
-        '-o', '--output', metavar='FILE', type=FileType('wt'),
-        default=sys.stdout,
-        help='output signature to this file'
+        '-o', '--output', metavar='FILE',
+        help='output signature to this file (default stdout)'
     )
     subparser.add_argument(
         '-A', '--abundances-from', metavar='FILE',