dtree-mailing.py

#!/usr/bin/python
from __future__ import print_function


#-----------------------------------------------------------------------        
#  Main
#-----------------------------------------------------------------------        
import os,sys,getopt,pwd


#-----------------------------------------------------------------------        
#  Constants
#-----------------------------------------------------------------------        
MAIL_SERVER = "mta1.uits.uconn.edu"
RETURN_ADDR = "hpc@uconn.edu"

#  Script header
script_header = """#!/bin/bash
cat<<EOM     | 
"""

#  Script footer template (replace values of {{RETURN_ADDR}} and {{MAIL_SERVER}}
#  before writing to script.
#  The strings {{EMAIL}}, {{NAME}}, and {{LIST-FILE}} are used verbatim
script_footer_template = """EOM

while read netid email name
do
   MailViaServer3  -f {{RETURN_ADDR}} -F {{NAME}} -R "$name" -F {{EMAIL}} -R "$email" -F {{LIST-FILE}} -R "$(cat lists/$netid.list)"  -s {{MAIL_SERVER}}  email.template  $email
   echo $(date "+%Y-%m-%d %H:%M:%S") $netid $email $name >> mail.log
done
"""

#-----------------------------------------------------------------------        
#  Functions
#-----------------------------------------------------------------------        

# Get full_name and email address for netid using pwd module
def get_name_email(netid):
	try:
		#  Extract full name and email address from pwnam.
		#   Example:  'Sue User {sue.users@uconn.edu} ' -> ['Sue User', 'sue.user@uconn.edu']
		return [s.rstrip(' }') for s in pwd.getpwnam(netid).pw_gecos.split('{')]
	except KeyError:
		return None, None

def Usage(msg=None):
	print("""
   Usage: dtree-mailing  DTREE-OUTPUT  [OUTPUT-DIR]

   Read output from dtree program  in file DTREE-OUTPUT and write mailing
   script, email template, and subdirectory with lists for each netid of their
   old directories.
 
   INPUT ARGUMENTS

       DTREE-OUTPUT -  Output from dtree program.  This script will read info
                       on the directories on scratchfs1 and produce a script
                       that you can run (send-mailing.sh) to email the owners
                       of the directories.  Typically, the input file is in 
                        /gpfs/scratchfs1/admin/gpfs-scan-files/output/YYYY-MM-DD/dtree-180-200g.out
                       where YYYY-MM-DD is a date.
       OUTPUT-DIR   -  (Optional) Output directory where send-mailing.sh script
                       is written, email-template, and a subdirectory 'lists/'
                       that contains one file for each email recipient; the
                       file contains a list of directories taken from the
                       DTREE-OUTPUT.  The default value is the current
                       directory.

   OUTPUT PRODUCED
     send-mailing.sh -   A bash script that does the mailing.  This script
			 calls the local script MailViaServer3 to do the actual
			 mailing.
     lists            -  A directory containing files for each individual email
     lists/NETID.lst  -  For each netid mailed, a list of directories from
                         DTREE-OUTPUT that they will be asked to remove.

   REQUIRED SCRIPTS AND FILES
     MailViaServer3   -  The dtree-mailing.sh script, when run, will call this script.
                         This is available via github.uconn.edu:
                            git clone https://github.uconn.edu/jar02014/MailViaServer3
     email.template   -  An email template file.  This is used by the dtree-mailing.sh
			 shell-script produced this program.  The values
			 {{MAIL}}, {{EMAIL}}, and {{LIST-FILE}} in the template
			 are replaced with custom input for each email sent.

""")
	if msg:
		print()
		print(msg)
		print()	
	sys.exit()


#  For script input, read output from previous run of dtree.py
def read_dtree_output(dtree_file):
	#  1) Search input file for section that starts with comment '#SCRATCH DIRECTORIES'.
	#  2) From this section, the first two non-blank lines are the two table
	#     header lines.  Save these.
	#  3) Read and store the six field from each line
	#  4) Stop reading when you encounter either 1) end-of-file 2) blank line 3) comment

	headers   = []
	user_data = {}
	with open(dtree_file) as fin:
		#  Look for '# Scratch Directories
		for line in fin:
			line = line.strip().upper()
			if line.startswith("# SCRATCH DIRECTORIES") or line.startswith("#SCRATCH DIRECTORIES"):
				break
		#  Read two header lines
		for lineno, line in enumerate(fin):
			if lineno==2: break
			headers.append(line)
		#  Store input lines by netid
		for line in fin:
			if not line.strip() or line.strip().startswith("#"): break
			parts = line.split(None,6)
			if not len(parts)==6: continue
			netid = parts[4]
			if not netid in user_data: user_data[netid] = []
			user_data[netid].append(line)
	return headers, user_data
	

def print_strings(strings, print_file=sys.stdout, print_end='\n'):
	for s in strings:
		print(s, file=print_file, end=print_end)
	

#-----------------------------------------------------------------------        
#  Main
#-----------------------------------------------------------------------        
def main():

	opts, args = getopt.getopt(sys.argv[1:], '')
	opts = dict(opts)

	if not args: Usage()
	input_file = args[0]
	output_dir = '.' if len(args)<2 else args[1]

	#  Read user data
        headers, user_data = read_dtree_output(args[0])

	#  Write directory lists for each user
	lists_dir = "%s/lists" % output_dir
	if not os.path.isdir(lists_dir):  os.makedirs(lists_dir)
	for netid, lines in user_data.items():
		list_file = "%s/%s.list" % ( lists_dir, netid)
		with open(list_file, "w") as fout: 
			# headers and lines still contain original trailing \n
			print_strings(headers, fout, '')
			print_strings(lines,   fout, '')

	#  Get list of user netids, full names and email addresses
	netid_info = [ [netid] + get_name_email(netid) for netid in user_data.keys() ]

	# Write mailing script
	with open("send-mailing.sh", "w") as fout:
		print(script_header, file=fout, end='')
		for netid, full_name, email in netid_info:
			if full_name==None: continue
			print(netid, email, full_name, file=fout)
		print( script_footer_template.replace("{{RETURN_ADDR}}", RETURN_ADDR).replace("{{MAIL_SERVER}}", MAIL_SERVER), file=fout, end='' )

	print("Success - All Done")
			

main()