linting

Author: kpedro88

get_files_on_disk.py

@@ -1,13 +1,17 @@
 #!/usr/bin/env python3
 
-import os,sys,getpass,warnings,glob,shlex,subprocess,argparse
+"""Returns a list of files from a dataset including only files that are hosted on disk."""
+
+import os,sys,getpass,warnings,glob,shlex,subprocess,argparse # pylint: disable=multiple-imports
 from collections import defaultdict
 
+"""Gets OS version from shell (other methods return host OS when in container)"""
 def getOS():
     cmd = r"sed -nr 's/[^0-9]*([0-9]+).*/\1/p' /etc/redhat-release"
     osv = subprocess.check_output(shlex.split(cmd), encoding="utf-8").rstrip()
     return osv
 
+"""Gets list of files on disk for a dataset, and list of sites along with how many files each site has"""
 def getHosted(dataset):
     osv = getOS()
     rucio_path = f'/cvmfs/cms.cern.ch/rucio/x86_64/rhel{osv}/py3/current'
@@ -17,7 +21,7 @@ def getHosted(dataset):
     sys.path.insert(0,full_rucio_path+'/site-packages/')
 
     warnings.filterwarnings("ignore", message=".*cryptography.*")
-    from rucio.client.client import Client
+    from rucio.client.client import Client # pylint: disable=import-error,import-outside-toplevel
     client = Client()
 
     # loop over blocks to avoid timeout error from too-large response
@@ -27,12 +31,13 @@ def getHosted(dataset):
     nblocks = 10
     block_groups = [all_blocks[i:i+nblocks] for i in range(0, len(all_blocks), nblocks)]
 
-    from rucio.client.replicaclient import ReplicaClient
+    from rucio.client.replicaclient import ReplicaClient # pylint: disable=import-error,import-outside-toplevel
     rep_client = ReplicaClient()
 
     filelist = set()
     sitelist = defaultdict(int)
-    sitecond = lambda site: "_Tape" not in site
+    def sitecond(site):
+        return "_Tape" not in site
     for block_group in block_groups:
         reps = list(rep_client.list_replicas([{'scope': 'cms', 'name': block['name']} for block in block_group]))
         for rep in reps:
@@ -44,16 +49,17 @@ def getHosted(dataset):
     sys.path.pop(0)
     return filelist, sitelist
 
+"""Prints file list and site list"""
 def main(dataset, outfile=None, verbose=False):
     filelist, sitelist = getHosted(dataset)
 
     if verbose:
         print("Site list:")
         print("\n".join(f'{k}: {v}' for k,v in sitelist.items()))
 
-    file = open(outfile,'w') if outfile is not None else sys.stdout
+    file = open(outfile,'w') if outfile is not None else sys.stdout # pylint: disable=consider-using-with,unspecified-encoding
     print("\n".join(filelist), file=file)
-    if outfile is not None: file.close()
+    if outfile is not None: file.close() # pylint: disable=multiple-statements
 
 if __name__=="__main__":
     parser = argparse.ArgumentParser(