address @charles-cowart comment + some extra tests

Author: antgonza

qiita_client/qiita_client.py

@@ -26,7 +26,7 @@
 MAX_RETRIES = 3
 MIN_TIME_SLEEP = 180
 MAX_TIME_SLEEP = 360
-MIN_FILEPATH_SIZE = 100
+BLANK_FILE_THRESHOLD = 100
 
 
 class ArtifactInfo(object):
@@ -575,6 +575,10 @@ def artifact_and_preparation_files(self, artifact_id,
 
         Returns
         -------
+        dict
+            files available in the artifact
+        pandas.DataFrame
+            the prep information file for that artifact
 
         Raises
         ------
@@ -621,7 +625,7 @@ def _process_files_per_sample_fastq(self, files, prep_info,
         used_prefixes = []
         for i, (fwd, rev) in enumerate(zip_longest(fwds, revs)):
             fwd_fn = basename(fwd['filepath'])
-            file_smaller_than_min = fwd['size'] < MIN_FILEPATH_SIZE
+            file_smaller_than_min = fwd['size'] < BLANK_FILE_THRESHOLD
 
             # iterate over run prefixes and make sure only one matches
             run_prefix = None
@@ -647,7 +651,7 @@ def _process_files_per_sample_fastq(self, files, prep_info,
                 # matches the run prefix:
                 rev_fn = basename(rev['filepath'])
                 if not file_smaller_than_min:
-                    file_smaller_than_min = rev['size'] < MIN_FILEPATH_SIZE
+                    file_smaller_than_min = rev['size'] < BLANK_FILE_THRESHOLD
                 if not rev_fn.startswith(run_prefix):
                     raise ValueError(
                         'Reverse read does not match run prefix. run_prefix: '

qiita_client/tests/test_qiita_client.py

@@ -309,6 +309,35 @@ def test_artifact_and_preparation_files(self):
         self.assertEqual(fobs, fexp)
         self.assertEqual(piobs.shape, (2, 1))
 
+        # just fwd
+        files = {
+            'raw_forward_seqs': [
+                {'filepath': '/X/file_3_R1.fastq.gz', 'size': 101},
+                {'filepath': '/X/file_1_R1.fastq.gz', 'size': 99},
+                {'filepath': '/X/file_2_R1.fastq.gz', 'size': 101}]}
+        prep_info = pd.DataFrame.from_dict({
+            'run_prefix': {"sample.1": 'file_1',
+                           "sample.2": 'file_2',
+                           "sample.3": 'file_3'}}, dtype=str)
+        prep_info.index.name = 'sample_name'
+        fobs, piobs = self.tester._process_files_per_sample_fastq(
+            files, prep_info, False)
+        fexp = {
+            'sample.1': ({'filepath': '/X/file_1_R1.fastq.gz', 'size': 99},
+                         None),
+            'sample.2': ({'filepath': '/X/file_2_R1.fastq.gz', 'size': 101},
+                         None),
+            'sample.3': ({'filepath': '/X/file_3_R1.fastq.gz', 'size': 101},
+                         None)}
+        self.assertEqual(fobs, fexp)
+        self.assertEqual(piobs.shape, (3, 1))
+
+        fobs, piobs = self.tester._process_files_per_sample_fastq(
+            files, prep_info, True)
+        del fexp['sample.1']
+        self.assertEqual(fobs, fexp)
+        self.assertEqual(piobs.shape, (2, 1))
+
 
 if __name__ == '__main__':
     main()