Merge pull request #1163 from dib-lab/fix/silent_fifo_check

No more complaining about FIFO's being empty (addresses #1147)

ChangeLog

@@ -1,3 +1,9 @@
+2015-07-09  Jacob Fenton  <bocajnotnef@gmail.com>
+
+   * khmer/kfile.py: changed check_valid_file_exists to recognize fifos as
+   non-empty
+   * tests/test_normalize_by_median.py: added test
+
 2015-07-10  Jacob Fenton  <bocajnotnef@gmail.com>
 
    * oxli/functions.py: changed estimate functions to use correct letter

khmer/kfile.py

@@ -149,7 +149,8 @@ def check_valid_file_exists(in_files):
     """
     for in_file in in_files:
         if os.path.exists(in_file):
-            if os.stat(in_file).st_size > 0:
+            mode = os.stat(in_file).st_mode
+            if os.stat(in_file).st_size > 0 or S_ISBLK(mode) or S_ISFIFO(mode):
                 return
             else:
                 print('WARNING: Input file %s is empty' %

tests/test_normalize_by_median.py

@@ -32,6 +32,20 @@ def test_normalize_by_median_indent():
     assert os.path.exists(outfile)
 
 
+def test_normalize_by_median_empty_file():
+    infile = utils.get_temp_filename('empty')
+    shutil.copyfile(utils.get_test_data('empty-file'), infile)
+    script = 'normalize-by-median.py'
+    in_dir = os.path.dirname(infile)
+
+    args = [infile]
+    (status, out, err) = utils.runscript(script, args, in_dir)
+
+    assert 'WARNING:' in err, err
+    assert 'is empty' in err, err
+    assert 'SKIPPED' in err, err
+
+
 def test_normalize_by_median():
     CUTOFF = '1'
 
@@ -495,7 +509,7 @@ def write_by_chunks(infile, outfile, CHUNKSIZE=8192):
     ofile.close()
 
 
-def test_normalize_by_median_streaming():
+def test_normalize_by_median_streaming_0():
     CUTOFF = '20'
 
     infile = utils.get_test_data('100-reads.fq.gz')
@@ -522,6 +536,32 @@ def test_normalize_by_median_streaming():
     assert linecount == 400
 
 
+def test_normalize_by_median_streaming_1():
+    CUTOFF = '20'
+
+    infile = utils.get_test_data('test-filter-abund-Ns.fq')
+    in_dir = os.path.dirname(infile)
+    fifo = utils.get_temp_filename('fifo')
+    outfile = utils.get_temp_filename('outfile')
+
+    # Use a fifo to copy stdout to a file for checking
+    os.mkfifo(fifo)
+    thread = threading.Thread(target=write_by_chunks, args=(infile, fifo))
+    thread.start()
+
+    # Execute diginorm
+    script = 'normalize-by-median.py'
+    args = ['-C', CUTOFF, '-k', '17', '-o', outfile, fifo]
+    (status, out, err) = utils.runscript(script, args, in_dir)
+
+    # Merge the thread
+    thread.join()
+
+    assert os.path.exists(outfile), outfile
+    assert 'Total number of unique k-mers: 98' in err, err
+    assert 'fifo is empty' not in err, err
+
+
 def test_diginorm_basic_functionality_1():
     # each of these pairs has both a multicopy sequence ('ACTTCA...') and
     # a random sequence.  With 'C=1' and '-p', all should be kept.