__author__ = 'noway'

import hashlib
import os
import shelve
import unicodedata

#Size of chunks to be hashed when hashing file contents.
#That portion is necessary when hashing huge files,
#which are problematic to be stored in memory
CHUNK_SIZE = 4096


def md5(file_name):
#Compute md5 of file CONTENTS
#   file_name - path to file (ex. /Users/123/Desktop/file.txt)
#Result:
#   128-bit hex string = md5 hash

    hash_md5 = hashlib.md5()
    with open(file_name) as f:
        for chunk in iter(lambda: f.read(CHUNK_SIZE), ''):
            hash_md5.update(chunk)
    return hash_md5.hexdigest()


def update_hash_names(path):
#Produce SHELVE-database with dictionary [HASH:PATH]
#   path - path to the dictionary where all files and subdir should be hashed
#Result:
#   Database named 'names_db'
    names_db = shelve.open('names_db')
    for root, dirs, files in os.walk(path):
        for name in files:
            file = root + "\\" + name
            hash_name = hashlib.md5(file).hexdigest()
            if not names_db.has_key(hash_name):
                names_db[hash_name] = file
    names_db.close()


def compute_dir_hash(path, progress_bar=False):
#Produce SHELVE-database with dictionary [HASH1:HASH2]
#Where HASH1 is hash of path to file, HASH2 is hash of file contents
#   path - path to the directory to be hashed
#   (optional)progress_bar - set True if you want to see
#       approximate progress (MAY USE ADDITIONAL TIME)
#Result:
#   Database named [path to directory]_db
    counter = 0
    counter_saved = 0
    if progress_bar:
        print "Computing all files in directory " + path + " and subdir..."
        for root, dirs, files in os.walk(path):
            for name in files:
                counter += 1
        print "There are " + str(counter) + " files there"
        counter_saved = counter
        counter = 0

    print "Hashing directory " + path + " and all subdir contents"
    dir_db = shelve.open(str(path + '_db').replace(":", '').replace("\\", ''))

    for root, dirs, files in os.walk(path):
        for name in files:
            if progress_bar:
                counter += 1
                print "\r{0} of {1}".format(counter, counter_saved),

            file_path = root + "/" + name        #MAC OS
            #file_name = root + "\\" + name      #WIN

            hash_name = hashlib.md5(file_path).hexdigest()

            try:
                hash_value = md5(file_path)
            except IOError:
                print "Permission Denied to compute md5 hash of " + file_path
            else:
                if not hash_name in dir_db:
                    dir_db[hash_name] = hash_value
    dir_db.close()

#update_hash_names('C:\Windows')
#compute_dir_hash('C:\Windows')
compute_dir_hash('/Users/pontifik/Desktop/Work', progress_bar=True)

print "DONE"