#! /usr/bin/python

#
# A script to check for large and/or old files in the temp directories
# and remove them.
#
# The script looks for files or directories within the temp directories
#  owned by users who are not logged in. If the files are older than the
#  delete age they are deleted.
#
# 23-feb-06 - davidbbs - first version
# 28-feb-06 - davidbbs - modified to have a list of regexp to ignore
# 01-jun-06 - davidbbs - modified to deal with numeric user IDs on files
#             files by name.
#

import sys, re, os, time, datetime

DEBUG = False

# Directories to check
tempDirectories = ["/tmp/", "/state/partition1/"]

# Age of the file before it is deleted
deleteAgeHours = 24*8

# only delete files belonging to users in these groups
groupsToCheck = ["students", "faculty", "staff", "regression"]

# If this regexp matches the file then it is skipped
namesToSkip = [".X[0-9]+-lock"]

# Commands
lsCommand = "/bin/ls -Alc --full-time -GQ --time-style=long-iso"
wCommand = "/usr/bin/w -h"
groupsCommand = "/usr/bin/groups"
rmCommand = "/bin/rm -rf"


# Get the users who are currently logged in
usersLoggedIn = []

child = os.popen(wCommand, "r")
for line in child.readlines():
    match = re.search("(\S+).*", line)
    if (match):
        user = match.group(1)
        if user not in usersLoggedIn:
            usersLoggedIn.append(match.group(1))

if (DEBUG):
    print "Users logged in:"
    for user in usersLoggedIn:
        print user

users = {}
dates = {}
files = []

# Go through the files in the directories
for tempDir in tempDirectories:
    if (DEBUG):
        print "** Checking directory: " + tempDir
    # Get the files
    lsChild = os.popen(lsCommand + " " + tempDir, "r")
    for line in lsChild:
        match = re.search("(\S+)\s+(\S+)\s+(\S+)\s+(\S+)\s+([0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2})\s+\"(\S+)\"", line)
        if (match):
            validUser = False

            # Extract the data
            user = match.group(3)
            fileName = match.group(6)
            date = match.group(5)
            file = tempDir + fileName

            # If it's just a numeric ID then it's fine to delete. (I.e., the user has been changed/removed.)
            matchNumericID = re.search("[0-9]+", user)
            if (matchNumericID):
                validUser = True
            else:
                # Check if the user is a real user
                groupsChild = os.popen(groupsCommand + " " + user, "r")
                result = groupsChild.readline()
                groupsChild.close()
                for group in groupsToCheck:
                    groupMatch = re.search(group, result)
                    if (groupMatch):
                        validUser = True
                    
            
            # Check if the user is logged in
            if (user in usersLoggedIn):
                validUser = False

            # Check if the file matches an ignore pattern
            validFileName = True
            for name in namesToSkip:
                match = re.search(name, file)
                if (match):
                    validFileName = False
            
            if (validUser & validFileName):
                files.append(file)
                users[file] = user
                epochSeconds = time.mktime(time.strptime(date, "%Y-%m-%d %H:%M"))
                dates[file] = datetime.datetime.fromtimestamp(epochSeconds)
                if (DEBUG):
                    print "Found:   " + fileName.ljust(40) + " for user " + user.ljust(20) + " last accessed at " + date
            else:
                if (DEBUG):
                    print "Ignored: " + fileName.ljust(40) + " for user " + user.ljust(20) + " last accessed at " + date


# For each file check the date
now = datetime.datetime.now()

for file in files:
    user = users[file]
    date = dates[file]
    ageDifference = now - date
    ageInHours = ageDifference.days*24.0+ageDifference.seconds/3600.0

    # If the file is too old delete it.
    if (ageInHours > deleteAgeHours):
        if (DEBUG):
            print "Remove File:  (" + user + " age: %0.1f" % ageInHours  + "h)\t\"" + rmCommand + " " + file + "\""
        else:
            print "Remove File:  (" + user + " age: %0.1f" % ageInHours  + "h)\t\"" + rmCommand + " " + file + "\""
            os.system(rmCommand + " " + file)