Merge pull request #8 from ECRL/dev

Better subprocess handling

padelpy/__init__.py

@@ -1,3 +1,3 @@
 from padelpy.wrapper import padeldescriptor
 from padelpy.functions import from_mdl, from_smiles
-__version__ = '0.1.5'
+__version__ = '0.1.6'

padelpy/functions.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 #
 # padelpy/functions.py
-# v.0.1.5
+# v.0.1.6
 # Developed in 2019 by Travis Kessler <[email protected]>
 #
 # Contains various functions commonly used with PaDEL-Descriptor
@@ -14,13 +14,14 @@
 from datetime import datetime
 from os import remove
 from re import compile, IGNORECASE
+from time import sleep
 
 # PaDELPy imports
 from padelpy import padeldescriptor
 
 
 def from_smiles(smiles: str, output_csv: str=None, descriptors: bool=True,
-                fingerprints: bool=False, timeout: int=4) -> OrderedDict:
+                fingerprints: bool=False, timeout: int=12) -> OrderedDict:
     ''' from_smiles: converts SMILES string to QSPR descriptors/fingerprints
 
     Args:
@@ -55,13 +56,14 @@ def from_smiles(smiles: str, output_csv: str=None, descriptors: bool=True,
                 d_2d=descriptors,
                 d_3d=descriptors,
                 fingerprints=fingerprints,
-                maxruntime=1000*timeout
+                sp_timeout=timeout
             )
             break
         except RuntimeError as exception:
             if attempt == 2:
                 remove('{}.smi'.format(timestamp))
                 if not save_csv:
+                    sleep(0.5)
                     remove(output_csv)
                 raise RuntimeError(exception)
             else:
@@ -81,7 +83,7 @@ def from_smiles(smiles: str, output_csv: str=None, descriptors: bool=True,
 
 
 def from_mdl(mdl_file: str, output_csv: str=None, descriptors: bool=True,
-             fingerprints: bool=False, timeout: int=4) -> list:
+             fingerprints: bool=False, timeout: int=12) -> list:
     ''' from_mdl: converts MDL file into QSPR descriptors/fingerprints;
     multiple molecules may be represented in the MDL file
 
@@ -121,12 +123,13 @@ def from_mdl(mdl_file: str, output_csv: str=None, descriptors: bool=True,
                 d_2d=descriptors,
                 d_3d=descriptors,
                 fingerprints=fingerprints,
-                maxruntime=1000*timeout
+                sp_timeout=timeout
             )
             break
         except RuntimeError as exception:
             if attempt == 2:
                 if not save_csv:
+                    sleep(0.5)
                     remove(output_csv)
                 raise RuntimeError(exception)
             else:

padelpy/wrapper.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 #
 # padelpy/wrapper.py
-# v.0.1.5
+# v.0.1.6
 # Developed in 2019 by Travis Kessler <[email protected]>
 #
 # Contains the `padeldescriptor` function, a wrapper for PaDEL-Descriptor
@@ -12,6 +12,7 @@
 from os.path import abspath, dirname, join
 from shutil import which
 from subprocess import PIPE, Popen
+from time import sleep
 
 # PaDEL-Descriptor is packaged with PaDELPy
 _PADEL_PATH = join(
@@ -21,6 +22,29 @@
 )
 
 
+def _popen_timeout(command: str, timeout: int) -> tuple:
+    ''' Calls PaDEL-Descriptor, with optional subprocess timeout
+
+    Args:
+        command (str): command to execute via subprocess.Popen
+        timeout (int): if not None, times out after this many seconds
+
+    Returns:
+        tuple: (stdout of process, stderr of process)
+    '''
+
+    p = Popen(command.split(), stdout=PIPE, stderr=PIPE)
+    if timeout is not None:
+        for t in range(timeout):
+            sleep(1)
+            if p.poll() is not None:
+                return p.communicate()
+        p.kill()
+        return (-1, b'PaDEL-Descriptor timed out during subprocess call')
+    else:
+        return p.communicate()
+
+
 def padeldescriptor(maxruntime: int=-1, waitingjobs: int=-1, threads: int=-1,
                     d_2d: bool=False, d_3d: bool=False, config: str=None,
                     convert3d: bool=False, descriptortypes: str=None,
@@ -30,7 +54,8 @@ def padeldescriptor(maxruntime: int=-1, waitingjobs: int=-1, threads: int=-1,
                     removesalt: bool=False, retain3d: bool=False,
                     retainorder: bool=False, standardizenitro: bool=False,
                     standardizetautomers: bool=False, tautomerlist: str=None,
-                    usefilenameasmolname: bool=False) -> None:
+                    usefilenameasmolname: bool=False,
+                    sp_timeout: int=None) -> None:
     ''' padeldescriptor: complete wrapper for PaDEL-Descriptor descriptor/
     fingerprint generation software
 
@@ -117,8 +142,7 @@ def padeldescriptor(maxruntime: int=-1, waitingjobs: int=-1, threads: int=-1,
     if usefilenameasmolname is True:
         command += ' -usefilenameasmolname'
 
-    p = Popen(command.split(), stdout=PIPE, stderr=PIPE)
-    _, err = p.communicate()
+    _, err = _popen_timeout(command, sp_timeout)
     if err != b'':
         raise RuntimeError('PaDEL-Descriptor encountered an error: {}'.format(
             err.decode('utf-8')

setup.py

@@ -2,7 +2,7 @@
 
 setup(
     name='padelpy',
-    version='0.1.5',
+    version='0.1.6',
     description='A Python wrapper for PaDEL-Descriptor',
     url='https://github.com/ecrl/padelpy',
     author='Travis Kessler',