[MRG] Fix multigather so that the output CSV contains all matches. (#…

…2322)

This PR fixes #2321 so that more than one output line is placed in the
CSV. Oops!

It also adds a notification of what the CSV output file name is.

Last but not least, it supports `--output-dir` as a way to set the base
path for all output files.

Fixes #2321.

TODO:

- [x] add tests
- [x] make sure filename output behavior is documented
- [x] consider adding an option to have multigather save CSV results in
some other way, like by md5 or ...something. - punted to
#2328

src/sourmash/cli/multigather.py

@@ -83,6 +83,11 @@ def subparser(subparsers):
     )
     subparser.set_defaults(fail_on_empty_database=True)
 
+    subparser.add_argument(
+        '--output-dir', '--outdir',
+        help='output CSV results to this directory',
+    )
+
     add_ksize_arg(subparser)
     add_moltype_args(subparser)
     add_scaled_arg(subparser, 0)

src/sourmash/commands.py

@@ -1064,19 +1064,22 @@ def multigather(args):
                 query_filename = query.md5sum()
 
             output_base = os.path.basename(query_filename)
+            if args.output_dir:
+                output_base = os.path.join(args.output_dir, output_base)
             output_csv = output_base + '.csv'
 
+            notify(f'saving all CSV matches to "{output_csv}"')
             w = None
             with FileOutputCSV(output_csv) as fp:
                 for result in found:
                     if w is None:
                         w = result.init_dictwriter(fp)
-                        result.write(w)
+                    result.write(w)
 
             output_matches = output_base + '.matches.sig'
             with open(output_matches, 'wt') as fp:
                 outname = output_matches
-                notify(f'saving all matches to "{outname}"')
+                notify(f'saving all matching signatures to "{outname}"')
                 sig.save_signatures([ r.match for r in found ], fp)
 
             output_unassigned = output_base + '.unassigned.sig'

tests/test_sourmash.py

@@ -3692,6 +3692,62 @@ def test_multigather_metagenome_query_from_file(runtmp):
                 'NC_011294.1 Salmonella enterica subsp' in out))
 
 
+def test_multigather_metagenome_output(runtmp):
+    # test multigather CSV output has more than one output line
+    c = runtmp
+    testdata_glob = utils.get_test_data('gather/GCF*.sig')
+    testdata_sigs = glob.glob(testdata_glob)
+
+    query_sig = utils.get_test_data('gather/combined.sig')
+
+    cmd = ['index', 'gcf_all']
+    cmd.extend(testdata_sigs)
+    cmd.extend(['-k', '21'])
+    c.run_sourmash(*cmd)
+
+    assert os.path.exists(c.output('gcf_all.sbt.zip'))
+
+    cmd = f'multigather --query {query_sig} --db gcf_all -k 21 --threshold-bp=0'
+    cmd = cmd.split(' ')
+    c.run_sourmash(*cmd)
+
+    output_csv = runtmp.output('-.csv')
+    assert os.path.exists(output_csv)
+    with open(output_csv, newline='') as fp:
+        x = fp.readlines()
+        assert len(x) == 13
+
+
+def test_multigather_metagenome_output_outdir(runtmp):
+    # test multigather CSV output to different location
+    c = runtmp
+    testdata_glob = utils.get_test_data('gather/GCF*.sig')
+    testdata_sigs = glob.glob(testdata_glob)
+
+    query_sig = utils.get_test_data('gather/combined.sig')
+
+    cmd = ['index', 'gcf_all']
+    cmd.extend(testdata_sigs)
+    cmd.extend(['-k', '21'])
+    c.run_sourmash(*cmd)
+
+    assert os.path.exists(c.output('gcf_all.sbt.zip'))
+
+    # create output directory
+    outdir = runtmp.output('savehere')
+    os.mkdir(outdir)
+
+    cmd = f'multigather --query {query_sig} --db gcf_all -k 21 --threshold-bp=0 --output-dir {outdir}'
+    cmd = cmd.split(' ')
+    c.run_sourmash(*cmd)
+
+    output_csv = runtmp.output('savehere/-.csv')
+    assert os.path.exists(output_csv)
+    with open(output_csv, newline='') as fp:
+        x = fp.readlines()
+        assert len(x) == 13
+
+
 @utils.in_tempdir
 def test_multigather_metagenome_query_with_sbt(c):