Checking if filenames include blank spaces.

CHEWBBACA/AlleleCall/allele_call.py

@@ -1960,21 +1960,6 @@ def allele_calling(fasta_files, schema_directory, temp_directory,
     full_to_basename = im.mapping_function(fasta_files, fo.file_basename, [False])
     full_to_unique = {k: fo.split_joiner(v, [0], '.') for k, v in full_to_basename.items()}
 
-    # Detect if some inputs share the same unique prefix
-    basename_list = list(full_to_unique.values())
-    if len(set(basename_list)) < len(fasta_files):
-        basename_counts = [[basename, basename_list.count(basename)]
-                           for basename in set(basename_list)]
-        repeated_basenames = ['{0}: {1}'.format(*l)
-                              for l in basename_counts if l[1] > 1]
-        # Only delete temp directory created for each run
-        # Do not delete output directory because it might include other files
-        fo.delete_directory(temp_directory)
-        sys.exit('\nSome input files share the same filename prefix '
-                 '(substring before the first "." in the filename). '
-                 'Please make sure that every input file has a unique '
-                 'filename prefix.\n{0}'.format('\n'.join(repeated_basenames)))
-
     # Create directory to store files with Pyrodigal results
     pyrodigal_path = fo.join_paths(temp_directory, ['1_cds_prediction'])
     fo.create_directory(pyrodigal_path)

CHEWBBACA/CreateSchema/create_schema.py

@@ -187,18 +187,6 @@ def create_schema_seed(fasta_files, output_directory, schema_name, ptf_path,
     full_to_unique = {k: fo.split_joiner(v, [0], '.')
                         for k, v in full_to_basename.items()}
 
-    # Detect if some inputs share the same unique prefix
-    basename_list = list(full_to_unique.values())
-    if len(set(basename_list)) < len(fasta_files):
-        basename_counts = [[basename, basename_list.count(basename)]
-                           for basename in set(basename_list)]
-        repeated_basenames = ['{0}: {1}'.format(*l)
-                              for l in basename_counts if l[1] > 1]
-        sys.exit('\nSome input files share the same filename prefix '
-                 '(substring before the first "." in the filename). '
-                 'Please make sure that every input file has a unique '
-                 'filename prefix.\n{0}'.format('\n'.join(repeated_basenames)))
-
     # Create directory to store temporary files
     temp_directory = fo.join_paths(output_directory, ['temp'])
     fo.create_directory(temp_directory)

CHEWBBACA/chewBBACA.py

@@ -199,6 +199,10 @@ def msg(name=None):
 
     genome_list = fo.join_paths(args.output_directory, [ct.GENOME_LIST])
     args.input_files = pv.check_input_type(args.input_files, genome_list)
+	# Detect if some inputs share the same unique prefix
+	repeated_prefixes = pv.check_unique_prefixes(args.input_files)
+	# Detect if filenames include blank spaces
+	blank_spaces = pv.check_blanks(args.input_files)
 
     # Add clustering parameters
     args.word_size = ct.WORD_SIZE_DEFAULT
@@ -490,6 +494,10 @@ def msg(name=None):
 
     genome_list = fo.join_paths(args.output_directory, [ct.GENOME_LIST])
     genome_list = pv.check_input_type(args.input_files, genome_list)
+	# Detect if some inputs share the same unique prefix
+	repeated_prefixes = pv.check_unique_prefixes(genome_list)
+	# Detect if filenames include blank spaces
+	blank_spaces = pv.check_blanks(genome_list)
 
     # Determine if schema was downloaded from Chewie-NS
     ns_config = fo.join_paths(args.schema_directory, ['.ns_config'])

CHEWBBACA/utils/parameters_validation.py

@@ -1295,3 +1295,68 @@ def validate_loci_list(input_path, output_file, parent_dir=None):
         fo.write_lines(files, output_file)
 
     return output_file
+
+
+def check_unique_prefixes(input_list):
+	"""Check if all input files have an unique identifier.
+
+	Parameters
+	----------
+	input_list : str
+		Path to file that contains the list of paths to input files.
+
+	Returns
+	-------
+	False if there are no input files sharing the same identifier.
+
+	Raises
+    ------
+    SystemExit
+        - If there are multiple files sharing the same prefix.
+	"""
+	input_paths = fo.read_lines(input_list)
+	basenames = [fo.file_basename(file) for file in input_paths]
+	unique_ids = [fo.split_joiner(name, [0], '.') for name in basenames]
+
+	# Detect if some inputs share the same unique prefix
+	if len(set(unique_ids)) < len(input_paths):
+		basename_counts = [[name, basenames.count(name)]
+							for name in set(basenames)]
+		repeated_basenames = ['{0}: {1}'.format(*l)
+								for l in basename_counts if l[1] > 1]
+		sys.exit('\nSome input files share the same filename prefix '
+				'(substring before the first "." in the filename). '
+				'Please make sure that every input file has a unique '
+				'filename prefix.\n{0}'.format('\n'.join(repeated_basenames)))
+
+	return False
+
+
+def check_blank(input_list):
+	"""Check if input files do not include blank spaces in the filename.
+
+	Parameters
+	----------
+	input_list : str
+		Path to file that contains the list of paths to input files.
+
+	Returns
+	-------
+	False if there are no blank spaces in the filenames.
+
+	Raises
+    ------
+    SystemExit
+        - If there are blank spaces in any of the filenames.
+	"""
+	input_paths = fo.read_lines(input_list)
+	basenames = [fo.file_basename(file) for file in input_paths]
+	include_blanks = [name for name in basenames if ' ' in name]
+
+	if len(include_blanks) > 0:
+		sys.exit('\nThe following input files include blank spaces '
+				 'in the filename:\n{0}\nPlease ensure that filenames '
+				 'do not include blank spaces or special characters '
+				 '(e.g. !@#?$^*()+)'.format('\n'.join(include_blanks)))
+
+	return False