Script PythonFormat uses yapf to check the formatting of Python files

Code/GraphMol/Descriptors/test_data/pmi.py

@@ -51,5 +51,5 @@ if __name__ == '__main__':
   for m in suppl:
     i1, i2, i3 = GetMoments(m, True)
     mi1, mi2, mi3 = GetMoments(m, False)
-    print("%s %.4f %.4f %.4f %.4f %.4f %.4f" %
-          (m.GetProp("_Name"), i1, i2, i3, mi1, mi2, mi3), file=output)
+    print("%s %.4f %.4f %.4f %.4f %.4f %.4f" % (m.GetProp("_Name"), i1, i2, i3, mi1, mi2, mi3),
+          file=output)

Code/GraphMol/MolDraw2D/test_dir/test_rdkit_draw.py

@@ -1,5 +1,5 @@
 #!/usr/bin/env python
-
+from __future__ import print_function
 from rdkit import RDConfig
 import os, sys
 from rdkit import Chem
@@ -12,11 +12,11 @@ if sys.argv[1].endswith('.sdf'):
 elif sys.argv[1].endswith('.smi'):
   suppl = Chem.SmilesMolSupplier(sys.argv[1])
 else:
-  print 'Need a file ending in .sdf or .smi'
+  print('Need a file ending in .sdf or .smi')
   exit(1)
 
 for mol in suppl:
-  print mol.GetProp('_Name')
+  print(mol.GetProp('_Name'))
   fn = mol.GetProp('_Name') + '.png'
   AllChem.Compute2DCoords(mol)
   Draw.MolToFile(mol, fn)

Code/cmake/Modules/fixup_coverage.py

@@ -3,10 +3,10 @@ the coverage tool mistakenly finds in the build tree.
 It replaces the paths with the ones from the source tree
 n.b. if a file with the same name (i.e. sln.yy) is found twice
  in the source tree, this will break"""
-
+from __future__ import print_function
 import os, sys
 source_dir, info_file = sys.argv[1:3]
-print source_dir, info_file
+print(source_dir, info_file)
 
 paths = {}
 for root, dir, files in os.walk(source_dir):
@@ -20,13 +20,13 @@ for line in lines:
   if "SF:" in line:
     fn = line.split("SF:")[-1].strip()
     if not os.path.exists(fn):
-      print "Does not exist:", fn.strip()
+      print("Does not exist:", fn.strip())
       head, rest = os.path.split(fn)
       potential = paths[rest]
       if len(potential) == 1:
         line = "SF:" + potential[0]
       else:
-        asdf
+        raise NotImplementedError('asdf')
   newlines.append(line)
 
 open(info_file, 'w').write("\n".join(newlines))

Scripts/PythonFormat.py

@@ -0,0 +1,69 @@
+'''
+
+Script will test the RDkit python code for conformance with the agreed format using
+yapf.
+
+For each Python file that is found in $RDBASE (excluding the build and External
+directories), yapf is used with the style configuration in $RDBASE/setup.cfg.
+If a change is required, the difference is printed. At the end of the process,
+all non-conformant files are listed and the required yapf command(s) printed.
+
+If changes are found, the script will exit with error code 1, otherwise 0.
+
+'''
+from __future__ import print_function
+import os
+from yapf.yapflib.yapf_api import FormatCode
+import sys
+
+rdbase = os.environ.get('RDBASE', '')
+styleConfig = os.path.join(rdbase, 'setup.cfg')
+
+excludeDirs = [os.path.join(rdbase, 'build'),
+               os.path.join(rdbase, 'External'), ]
+
+
+def pythonFiles(dirname=rdbase):
+  """ Find all python files below directory dirname """
+  for root, _, files in os.walk(dirname):
+    if any(root.startswith(d) for d in excludeDirs):
+      continue
+    for file in files:
+      if file.endswith(".py"):
+        yield os.path.join(root, file)
+
+
+def yapfChanges(filename):
+  """ Use yapf with the default settings to format file filename """
+  try:
+    with open(filename) as f:
+      codeBefore = f.read()
+  except UnicodeError:
+    with open(filename, encoding='latin-1') as f:
+      codeBefore = f.read()
+  try:
+    changes, changed = FormatCode(codeBefore, style_config=styleConfig, print_diff=True,
+                                  filename=filename)
+  except Exception:
+    print(filename)
+    raise
+  if changed:
+    print(changes)
+  return changed
+
+
+if __name__ == "__main__":
+  changedFiles = []
+  for s in pythonFiles():
+    if yapfChanges(s):
+      changedFiles.append(s)
+  print()
+  if changedFiles:
+    print('yapf will make changes to the following files:')
+    print('\n'.join(sorted(changedFiles)))
+    print('To apply the required changes to your code use the following command(s)')
+    for s in sorted(set(s.replace(rdbase, '').split(os.sep)[1] for s in changedFiles)):
+      print('yapf --style $RDBASE/setup.cfg --in-place --recursive $RDBASE/{0}'.format(s))
+    sys.exit(1)
+  print('Code complies with the agreed formatting rules.')
+  sys.exit(0)

rdkit/TestRunner.py

@@ -107,8 +107,8 @@ def RunScript(script, doLongTests, verbose):
   nTests = len(tests) + len(longTests)
   del sys.modules[script]
   if verbose and failed:
-    for exeName,args,extras in failed:
-        print("!!! TEST FAILURE: ",exeName,args,extras,file=sys.stderr)
+    for exeName, args, extras in failed:
+      print("!!! TEST FAILURE: ", exeName, args, extras, file=sys.stderr)
   return failed, nTests