[MRG] autodetect FASTA/FASTQ files if given as signatures

sourmash/sig/__main__.py

@@ -73,7 +73,6 @@ def cat(args):
         try:
             this_siglist = sourmash_args.load_file_as_signatures(sigfile, traverse=True)
         except Exception as exc:
-            error('\nError while reading signatures from {}:'.format(sigfile))
             error(str(exc))
             error('(continuing)')
 
@@ -177,7 +176,6 @@ def describe(args):
             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)')
 

sourmash/sourmash_args.py

@@ -7,6 +7,8 @@
 import itertools
 from enum import Enum
 
+import screed
+
 from sourmash import load_sbt_index
 from sourmash.lca.lca_db import load_single_database
 import sourmash.exceptions
@@ -65,6 +67,11 @@ def calculate_moltype(args, default=None):
 
 
 def load_query_signature(filename, ksize, select_moltype, select_md5=None):
+    """Load a single signature to use as a query.
+
+    Uses load_file_as_signatures underneath, so can load from collections
+    and indexed databases.
+    """
     try:
         sl = load_file_as_signatures(filename, ksize=ksize,
                                      select_moltype=select_moltype)
@@ -404,6 +411,21 @@ def _load_database(filename, traverse, traverse_yield_all):
         except:
             pass
 
+    if not loaded:
+        successful_screed_load = False
+        it = None
+        try:
+            # CTB: could be kind of time consuming for big record, but at the
+            # moment screed doesn't expose format detection cleanly.
+            with screed.open(filename) as it:
+                record = next(iter(it))
+            successful_screed_load = True
+        except:
+            pass
+
+        if successful_screed_load:
+            raise OSError("Error while reading signatures from '{}' - got sequences instead! Is this a FASTA/FASTQ file?".format(filename))
+
     if not loaded:
         raise OSError("Error while reading signatures from '{}'.".format(filename))
 

tests/test_api.py

@@ -46,3 +46,13 @@ def test_load_index_3():
 
     sigs = list(idx.signatures())
     assert len(sigs) == 2
+
+
+def test_load_fasta_as_signature():
+    # try loading a fasta file - should fail with informative exception
+    testfile = utils.get_test_data('short.fa')
+
+    with pytest.raises(OSError) as e:
+        idx = sourmash.load_file_as_index(testfile)
+
+    assert "Error while reading signatures from '{}' - got sequences instead! Is this a FASTA/FASTQ file?".format(testfile) in str(e)