rhel-baseline-diff.py

#!/usr/bin/env python
# rhel-baseline-diff.py
# Jun Go [email protected]
# Last updated: 2016-08-03

# Given a baseline file that contains a list of rpm's
# from a RHEL installation DVD, compare this rpm list with
# the list of rpm's currently installed on a RHEL 5.X or
# RHEL 6.X machine and write any differences to a file.
# Note that the Python 2 version on RHEL 5.11 is 2.4.3
# while the version on RHEL 6.8 is 2.6.6

# This python script should work in both python 2 and 3
# as it does not use any 2-or-3-specific syntax or libraries
# and only uses language features available up to Python 2.4


# USAGE: ./rhel-baseline-diff.py [baseline file]
# This script will send output to the file 'rhel-diff.txt'

import optparse
import subprocess
import re
# Although 'optparse' has been deprecated in favor of 'argparse'
# starting from Python 2.7.X, the RHEL machines this script
# will be executed on use Python 2.6.X and 2.4.X


def chk_rhel_rel():
    """
    -> string

    Search /etc/redat-release for 'Tikanga' (RHEL 5),
    'Santiago' (RHEL 6), or 'Maipo' (RHEL 7) and return
    string with release codename
    """
    rhel5 = r"Tikanga"
    rhel6 = r"Santiago"
    rhel7 = r"Maipo"
    rhel5_obj = re.compile(rhel5)
    rhel6_obj = re.compile(rhel6)
    rhel7_obj = re.compile(rhel7)

    rhrel = open("/etc/redhat-release", 'r')
    try:
        for line in rhrel:
            match5 = rhel5_obj.search(line)
            match6 = rhel6_obj.search(line)
            match7 = rhel7_obj.search(line)
            if match5:
                return "Tikanga"
            elif match6:
                return "Santiago"
            elif match7:
                return "Maipo"
    finally:
        rhrel.close()


def check_rpm_macro():
    """
    -> string

    Look for %query_all_fmt line in /usr/lib/rpm/macros and verify
    that 'architecture' is printed. The format for this setting
    differs between RHEL 5.X or 6.X, however, so the search patterns
    are different. If the pattern is not found, the function returns
    the strings '5' or '6'. If the pattern is found, the function
    returns blank string ''.

    TODO: need to handle edge case in which search pattern is found
    but the entire line is commented out with '#'
    """
    # /usr/lib/rpm/macros pattern for RHEL5.X
    pat1 = r'%%{name}-%%{version}-%%{release}.%%{arch}'
    # /usr/lib/rpm/macros pattern for RHEL6.X
    pat2 = r'%%{nvra}'
    pat1_obj = re.compile(pat1)
    pat2_obj = re.compile(pat2)
    write_flag = ''
    match_count = 0

    # RHEL version string
    rhel_ver = chk_rhel_rel()

    f = open("/usr/lib/rpm/macros", 'r')
    try:
        if rhel_ver == "Tikanga":
            for line in f:
                match5 = pat1_obj.search(line)
                if match5:
                    match_count += 1
            if match_count == 0:
                write_flag = '5'
            return write_flag
        elif rhel_ver == "Santiago":
            for line in f:
                match6 = pat2_obj.search(line)
                if match6:
                    match_count += 1
            if match_count == 0:
                write_flag = '6'
            return write_flag
    finally:
        f.close()


def write_rpm_macro(write_flag):
    """
    String -> Write to file

    Depending on the string 'write_flag', the proper '%query_all_fmt'
    string for RHEL5.X / RHEL6.X will be appended to
    /usr/lib/rpm/macros or not written at all
    """

    f = open("/usr/lib/rpm/macros", 'a')
    try:
        if write_flag == '5':
            f.write(
                "\n%_query_all_fmt %%{name}-%%{version}-%%{release}.%%{arch}\n"
            )
        elif write_flag == '6':
            f.write("\n%%{nvra}\n")
    finally:
        f.close


def load_baseline(filename):
    """
    file -> LoS

    Given a LF-delimited file containing rpm package names, this
    function will read each line into a list and strip out newlines
    """
    f = open(filename, 'r')
    try:
        baselineL = [line.strip() for line in f]
        return baselineL
    finally:
        f.close()


def write_file(los, filename):
    """
    ListOfString, text -> file

    Given a list of strings and a filename, write each list element
    to 'filename' and add a newline char with each element.
    """
    g = open(filename, 'w')
    try:
        for elem in los:
            g.write(elem + "\n")
    finally:
        g.close()


def current_rpmlist():
    """
    No input -> LoS

    This function will execute 'rpm -qa' on the target system
    (which must be a RHEL variant) and insert the output into
    a ListOfString while stripping out `\n' chars.
    """
    rpmqa = subprocess.Popen("rpm -qa", shell=True, stdout=subprocess.PIPE)
    out = rpmqa.stdout.readlines()
    currentL = [line.strip() for line in out]

    return currentL


def main():
    p = optparse.OptionParser(description="",
                              prog="rhel-baseline-diff.py",
                              version="0.1",
                              usage="%prog [baseline-file]")
    options, arguments = p.parse_args()

    if len(arguments) == 1:
        # Check if /usr/lib/rpm/macros is correctly set up for rpm -q
        write_rpm_macro(check_rpm_macro())
        # Filename of rpm baseline pkg list
        filename = arguments[0]
        # List of baseline rpm pkg's
        baseL = load_baseline(filename)
        # List of rpm pkg's currently installed on machine
        currL = current_rpmlist()
        # List of rpm's on machine that are identical to baseline
        intersectL = list(set(currL).intersection(baseL))
        # List of rpm's on machine that are different from baseline
        diffL = list(set(currL) - set(intersectL))

        print("No. of local pkgs identical to baseline:\n %s"
              % len(intersectL))
        print("No. of local pkgs different from baseline:\n %s"
              % len(diffL))

        # Sanity check - the number of installed pkg's must equal
        # no. of identical pkg's + no. of different pkg's
        assert len(intersectL) + len(diffL) == len(currL)
#        print("Local pkgs identical to baseline:\n %s" % intersectL)
#        print("Local packages different from baseline: \n %s" % diffL)
        write_file(intersectL, "pkg-intersect.txt")
        write_file(diffL, "pkg-diff.txt")
    # print help message if no arguments are passed
    else:
        p.print_help()


if __name__ == '__main__':
    main()