Convert vCard to CSV (2021-06-31)

convert-vcard-to-csv.py

# -*- coding: utf-8 -*-
import vobject
import re
import sys

# TODO gérer les entrées qui contiennent plus d'un numéro de téléphone, ou au moins écrire un warning

vcf = "".join(open(sys.argv[1]).readlines()).strip()
vcfs = [e for e in vcf.split("END:VCARD") if e != ""]
vcfs = [e+"END:VCARD" for e in vcfs]

contacts = []
for i,s in enumerate(vcfs):
  try:
    contacts.append(vobject.readOne(s))
  except Exception as e:
    print("error when parsing #%d" % (i+1))
    print(s)
    raise e

def transform_name(n):
  return re.sub(re.compile("\\s+"), " ", n).strip()

def transform_tel(t):
  if len(t) < 5:
   return t
  r = t.replace("-", "")
  if r.startswith("00"): # numéro avec indicatif
    return "+"+r[2:]
  elif r.startswith("+"): # numéro avec indicatif
    return r
  return "+33"+r[1:] # numéro français

def dot_every_two_digits(ds):
  return ".".join([a+b for a, b in zip(ds[::2], ds[1::2])])

def pretty_phone(t):
  for ind in ["+33", "+213"]:
    if t.startswith(ind):
      return "{} (0){}.{}".format(ind, t[len(ind)], dot_every_two_digits(t[len(ind)+1:]))
  return t
    
  #if t.startswith("+33"):
  #  return "+33 (0)" + t[3] + "." + dot_every_two_digits(t[4:])
  #elif t.startswith("+213"):
  #  return "+213 (0)" + t[4] + "." + dot_every_two_digits(t[5:])
  #return t

print("phone;phone_pretty;name")
for c in contacts:
  name = transform_name(str(c.n.value))
  tel = transform_tel(str(c.tel.value)) if "tel" in c.contents.keys() else ""
  if len(tel) > 0 and len(tel) < 5: # oublie les services type "messagerie", "info conso" etc
    continue
  rtel = pretty_phone(tel)
  print("{};{};{}".format(tel, rtel, name))