#!/volume1/@appstore/python/bin/python2
#-*- coding: iso-8859-15 -*-
#
# If a file has been archieved under an ISO-8859 environment and unarchived
# under an UTF8 environment, then you will get an encoding format problem.
# The file will not be readable through SAMBA.
#
# To fix this problem, you must convert the encoding format
# to the UTF8 (default Synology encoding)
#
# NB: in all cases, files will be readable through samba, even if the detection
# failed. But converted characters will not be good
#
# Remark: I guess it should work for any other encoding style. Just replace
# ISO-8859-15 (Western Europe) by the one coresponding to your country:
# http://en.wikipedia.org/wiki/Character_encoding#Common_character_encodings
#

# get library modules
import sys, os

########################
# ----- Functions ---- #
########################

# Special character hex range:
# CP850: 0x80-0xA5 (fortunately not used in ISO-8859-15)
# UTF-8: 1st hex code 0xC2-0xC3 followed by a 2nd hex code 0xA1-0xFF
# ISO-8859-15: 0xA6-0xFF
# The function will detect if fileDirName contains a special character
# If there is special character, detects if it is a UTF-8, CP850 or ISO-8859-15 encoding
def renameFunc(fullPath, fileDirName):
    encodingDetected = False
    Count = 1
    # parsing all files/directories in order to detect if CP850 is used
    for Idx in range(len(fileDirName)):
        # /!\ detection is done 2char by 2char for UTF-8 special character
        if (len(fileDirName) != 1) & (Idx < (len(fileDirName) - 1)):
            # Detect UTF-8
            if ((fileDirName[Idx] == '\xC2') | (fileDirName[Idx] == '\xC3')) & ((fileDirName[Idx+1] >= '\xA0') & (fileDirName[Idx+1] <= '\xFF')):
                print os.path.join(fullPath, fileDirName) + " -> UTF-8 detected: Nothing to be done"
                encodingDetected = True
                break;
            # Detect CP850
            elif ((fileDirName[Idx] >= '\x80') & (fileDirName[Idx] <= '\xA5')):
                utf8Name = fileDirName.decode('cp850')
                utf8Name = utf8Name.encode('utf-8')
#                os.rename(os.path.join(fullPath, fileDirName), os.path.join(fullPath, utf8Name))
                print os.path.join(fullPath, fileDirName)
                print os.path.join(fullPath, utf8Name) + " -> CP850 detected: Renamed"
                encodingDetected = True
                break;
            # Detect ISO-8859-15
            elif (fileDirName[Idx] >= '\xA6') & (fileDirName[Idx] <= '\xFF'):
                utf8Name = fileDirName.decode('iso-8859-15')
                utf8Name = utf8Name.encode('utf-8')
#                os.rename(os.path.join(fullPath, fileDirName), os.path.join(fullPath, utf8Name))
                print os.path.join(fullPath, fileDirName)
                print os.path.join(fullPath, utf8Name) + " -> ISO-8859-15 detected: Renamed"
                encodingDetected = True
                break;
        else:
            # Detect CP850
            if ((fileDirName[Idx] >= '\x80') & (fileDirName[Idx] <= '\xA5')):
                utf8Name = fileDirName.decode('cp850')
                utf8Name = utf8Name.encode('utf-8')
#                os.rename(os.path.join(fullPath, fileDirName), os.path.join(fullPath, utf8Name))
                print os.path.join(fullPath, fileDirName)
                print os.path.join(fullPath, utf8Name) + " -> CP850 detected: Renamed"
                encodingDetected = True
                break;
            # Detect ISO-8859-15
            elif (fileDirName[Idx] >= '\xA6') & (fileDirName[Idx] <= '\xFF'):
                utf8Name = fileDirName.decode('iso-8859-15')
                utf8Name = utf8Name.encode('utf-8')
#                os.rename(os.path.join(fullPath, fileDirName), os.path.join(fullPath, utf8Name))
                print os.path.join(fullPath, fileDirName)
                print os.path.join(fullPath, utf8Name) + " -> ISO-8859-15 detected: Renamed"
                encodingDetected = True
                break;
    if (encodingDetected == False):
        print os.path.join(fullPath, fileDirName) + " -> No special characters detected: Nothing to be done"
    return


###########################
# ----- Main Program ---- #
###########################

StartFolder = "/media/usenet/TV"

# process each sub-folders starting from the deepest level
print 80*'-'
print "Renaming folders to UTF-8 format..."
for dirname, dirnames, filenames in os.walk(StartFolder, topdown=False):
    for subdirname in dirnames:
        renameFunc(dirname, subdirname)
print "Folder renaming Done !"
print 80*'-'
print ""

# process each file recursively
print 80*'-'
print "Renaming files to UTF-8 format..."
for dirname, dirnames, filenames in os.walk(StartFolder):
    for filename in filenames:
        renameFunc(dirname, filename)
print "Files renaming Done !"
print 80*'-'
print ""


print ""
print "Character encoding translation done!"