vcard_reformat.py

#!/usr/bin/python
# coding=utf8

# reconvert vCard format by filling FN: tag in vCard blocks
# with mail address names if empty
# also, remove vCard blocks not containing any e-mail addresses

import os
import sys
import re

# vars
PARAM = ""
COUNT_ADD = 0
COUNT_SUB = 0

if len(sys.argv) > 1:
    PARAM = sys.argv[1]
else:
    print "error: imput file missing"
    exit(1)

print "* start"

# read .vcf text file
if not os.path.exists(PARAM):
    print "error: invalid file"
    exit(1)
FILE = open(PARAM)
# get lines of text
TEXT = FILE.readlines()

L3 = ""
OUT = ""

# cycle thorugh every line
for i in TEXT:
    # get line without ending new line char
    L = i[:-1]

    # add next block
    if L: L3 += L + "\n"

    # does the line start with "END:" tag?
    L2 = L[:4]
    if L2 == "END:":
        if L3:
            # is there any e-mail address in the block?
            if re.findall("^EMAIL.*:.*@", L3, re.M):

                # is there an empty "FN:" tag?
                if re.findall("^FN:\n", L3, re.M):
                    # if so, then add the mail name to it
                    # get the name part of the e-mail address before "@" char
                    PRE  = re.findall("^EMAIL.*$", L3, re.M)[0]
                    NAME = re.findall("[^:]+$", PRE, re.M)[0]
                    # is it not empty?
                    L4 = ""
                    if NAME:
                        L4 = re.sub("FN:\n", "FN:" + NAME + "\n", L3, re.M)
                    else:
                        L4 = re.sub("FN:\n", "FN:" + "empty" + "\n", L3, re.M)

                else: L4 = L3

                # store output
                OUT += L4 + "\n"

                # count it
                COUNT_ADD += 1

            else:
                COUNT_SUB += 1

        # clear temporary block holder
        L3 = ""


# convert unicode chars back to utf8 in output
# 123=C3=B6=C3=BC=C3=B3=C5=91=C3=BA=C3=A9=C3=A1=C5=B1=C3=AD123=C3=96=C3=9C=C3==0A=93=C5=90=C3=9A=C3=89=C3=81=C5=B0=C3=8D123=0A
# 123öüóőúéáűí123ÖÜÓŐÚÉÁŰÍ123
if OUT:
    TEMP = re.sub("=C3=B6", "ö", OUT); OUT = TEMP
    TEMP = re.sub("=C3=BC", "ü", OUT); OUT = TEMP
    TEMP = re.sub("=C3=B3", "ó", OUT); OUT = TEMP
    TEMP = re.sub("=C5=91", "ő", OUT); OUT = TEMP
    TEMP = re.sub("=C3=BA", "ú", OUT); OUT = TEMP
    TEMP = re.sub("=C3=A9", "é", OUT); OUT = TEMP
    TEMP = re.sub("=C3=A1", "á", OUT); OUT = TEMP
    TEMP = re.sub("=C3=A1", "á", OUT); OUT = TEMP
    TEMP = re.sub("=C5=B1", "ű", OUT); OUT = TEMP
    TEMP = re.sub("=C3=AD", "í", OUT); OUT = TEMP

    TEMP = re.sub("=C3=96", "Ö", OUT); OUT = TEMP
    TEMP = re.sub("=C3=9C", "Ü", OUT); OUT = TEMP
    TEMP = re.sub("=C3==0A", "Ó", OUT);    OUT = TEMP
    TEMP = re.sub("=93=C5", "Ő", OUT); OUT = TEMP
    TEMP = re.sub("=90=C3", "Ú", OUT); OUT = TEMP
    TEMP = re.sub("=9A=C3", "É", OUT); OUT = TEMP
    TEMP = re.sub("=89=C3", "Á", OUT); OUT = TEMP
    TEMP = re.sub("=81=C5", "Ű", OUT); OUT = TEMP
    TEMP = re.sub("=B0=C3=8D", "Í", OUT);  OUT = TEMP

    TEMP = re.sub("==0A", "", OUT);     OUT = TEMP
    TEMP = re.sub("=0A", "", OUT);      OUT = TEMP


# file operations
FILE.close()
FILE = open(PARAM, "w")
FILE.writelines(OUT)
FILE.close()

# print info
print "* " + str(COUNT_ADD) + " vCards added, " + str(COUNT_SUB) + " removed"
print "* exit"

exit(0)