import sys
from os import walk
from os.path import basename, isdir, splitext
from operator import itemgetter

if len(sys.argv) < 2:
    print('Usage: %s <directory>'%basename(sys.argv[0]))
    sys.exit(1)

dirname = sys.argv[1]
if not isdir(dirname):
    print('\'%s\'\n is not a valid directory'%dirname)

debugging = False
if len(sys.argv) >= 3:
    if sys.argv[2][1:] == 'd':
        debugging = True

valid_exts = ('cpp','c','h','hpp','sql','py','java','asm',
              'bat','cmd','bas','vbs','vbe','inc','rb','php',
              'asp','htm','html','ini','cfg','hs','bf','s',
              'js','idl','pro')
data = {}
for ext in valid_exts:
    data[ext] = [0,0]
other_data = {}
total_nof_files = 0
total_nof_lines = 0
try:
    for root, dirs, files in walk(dirname):
        for filename in files:
            ext = splitext(filename)[1][1:]
            if not ext in valid_exts:
                other_data[ext] = other_data.get(ext,0)+1
                continue
            with open("%s\\%s"%(root,filename)) as file:
                try:
                    nof_lines = sum([1 for line in file])
                except UnicodeDecodeError:
                    if debugging:
                        print('%\'s\' is not a valid text only file.'%filename)
                    continue
                data[ext][0] = data[ext][0]+1
                data[ext][1] = data[ext][1]+nof_lines
                total_nof_files = total_nof_files+1
                total_nof_lines = total_nof_lines+nof_lines
                if debugging:
                    print('\'%s\' have %d lines.'%(filename,nof_lines))
except KeyboardInterrupt:
    if debugging:
        print('#interrupted.')
finally:
    for ext, data in sorted(data.items(),key=itemgetter(1),reverse=True):
        [nof_files, nof_lines] = data
        if nof_lines:
            print('[%s]\n\t%d files - %.2f%%\n\t%.2f%% of code'
                  '\n\t%.2f lines of code per file'
                  %(ext.upper(),nof_files,nof_files/total_nof_files*100,
                    nof_lines/total_nof_lines*100,nof_lines/nof_files))

    print('\nfiles: %d'%total_nof_files)
    print('lines of code: %d'%total_nof_lines)
    print('lines of code per file: %.2f'%(total_nof_lines/total_nof_files))

    print('\n[OTHER SIGNIFICANT EXTENSIONS]\n')
    for ext in sorted(other_data.items(),key=itemgetter(1),reverse=True)[:5]:
        print(ext)