bha

#!/usr/bin/python3
# Bash History Analyser
# by Farooq Karimi Zadeh
# farooghkarimizadeh@gmail.com

# Under GNU GPLv3+
# This software is free software and COMES WITHOUT ANY WARRENTY FROM MAIN AUTHOR

# https://github.com/farooqkz/bha

import sys
import os

def fexist(path): # file exist
    try:
        fs = open(path)
    except:
        return 0
    else:
        return 1


def cexist(command): # command exist
    if fexist('/bin/' + command) or fexist('/sbin/' + command): return True
    if fexist('/usr/bin/' + command) or fexist('/usr/games/' + command): return True
    if fexist('/usr/sbin/' + command): return True
    if fexist('/usr/local/bin/' + command): return True
    if fexist('/usr/local/games/' + command): return True
    return False


def usage():
    print('Usage: bha [History_file_path]')


def dictsort(adict): # this function sorts a map by its values (uses bubblesort) 
    keys = list(adict.keys())
    values = list(adict.values())
    swaped = 1
    while swaped:
        swaped = 0
        for i in range(len(values)):
            if i == len(values) - 1: continue
            if values[i] < values[i+1]:
                values[i], values[i+1] = values[i+1], values[i]
                keys[i], keys[i+1] = keys[i+1], keys[i] # swap
                swaped = 1
    return [keys, values]


def excmd(s): # EXtracts CoMMands from a given string
    # first we delete anything between two "
    ns = '' # new string
    inq = False
    for i in s:
        if i == '\"': 
            inq = not inq # toggle
            continue
        if not inq: 
            ns += i

    # then we delete anything between two '
    fs = '' # final string
    inq = False
    for i in ns:
        if i == '\'':
            inq = not inq
            continue
        if not inq:
            fs += i
    cmds = []
    for i in fs.split(' '):
        if i.startswith('./'): # we don't count them
            continue
        if i.startswith('#'): # we don't count comments
            continue
        if i == 'sudo' or i == 'su': # they aren't command.are they?
            continue
        if cexist(i):
            cmds.extend([i])

    return cmds


history_file_path = os.path.expanduser('~') + '/.bash_history'
if len(sys.argv) == 2: history_file_path = sys.argv[1]

fs = open(history_file_path) # opens a File Stream for text reading
history = fs.read() # reads all what is in file
fs.close() # closes the file stream (no longer we need it)

history = history.split('\n') # => a list of strings seprated by a new line
cmds = {} # an empty map


for i in history:
    for cmd in excmd(i):
        if cmd not in cmds:
            cmds[cmd] = 1
        else:
            cmds[cmd] += 1

cmds = dictsort(cmds) # {'ls': 125, 'gcc': 10, 'cat': 100} => [['ls', 'cat', 'gcc'], [125, 100, 10]]

total = sum(cmds[1]) # [0, 1, 2] => 3
othercc = 0 # Other Commands Count (anything lower than 0.25%)
otherc = [] # list of other commands

for i in range(len(cmds[0])):
    cmd = cmds[0][i]
    count = cmds[1][i]
    percent = count / total * 100
    if percent < 0.25:
        othercc += count
        otherc.extend([cmd])
    else:
        print(cmd + ': ', round(percent,2), '%', ' (', count, '/', total, ')', sep='')

print('other commands: ')
for i in otherc:
    print(i, end=', ')
print()
print('% of other: ', round(othercc/total*100, 2),  '%', ' (', othercc, '/', total, ')', sep='')