dab/vcf_cleaner
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

comment all code

Browse source
This commit is contained in:
diegantobass 2025-05-08 09:12:14 +02:00
parent 988f662265
commit c3496a499c
1 changed files with 17 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
vcf_cleaner.py
View file

@ -6,18 +6,24 @@ import pickle
import phonenumbers import phonenumbers
from pythonvCard4.vcard import Contact from pythonvCard4.vcard import Contact
# input should be a valid .vcf file, output in current dir
input_file = open(sys.argv[1], 'r', encoding="latin-1").readlines() input_file = open(sys.argv[1], 'r', encoding="latin-1").readlines()
output = open('output.vcf', 'w') output = open('output.vcf', 'w')
# this only outputs pictures of contacts with multiples in a dir for manual check
# current code stage ignores the problem and chooses the first picture anyway
check_for_multiple_pictures = False check_for_multiple_pictures = False
# a contact is defined by its END:VCARD markup
current_card = "" current_card = ""
for line in input_file: for line in input_file:
current_card += line current_card += line
if "END:VCARD" in line: if "END:VCARD" in line:
# the pythonvCard4 parsing assumes a LOT of things
contact = Contact.from_vcard(current_card) contact = Contact.from_vcard(current_card)
# TODO : isolate contacts with multiple pictures
if check_for_multiple_pictures: if check_for_multiple_pictures:
if "PHOTO" in contact.custom and len(contact.custom["PHOTO"]) > 1: if "PHOTO" in contact.custom and len(contact.custom["PHOTO"]) > 1:
os.makedirs("multiple_pictures/" + contact.fn, exist_ok=True) os.makedirs("multiple_pictures/" + contact.fn, exist_ok=True)
@ -26,6 +32,7 @@ for line in input_file:
with open("photo/" + contact.fn + "/" + str(image) + ".jpg", "wb") as f: with open("photo/" + contact.fn + "/" + str(image) + ".jpg", "wb") as f:
f.write(base64.decodebytes(str.encode(contact.custom["PHOTO"][image]))) f.write(base64.decodebytes(str.encode(contact.custom["PHOTO"][image])))
# choose the first pictures and ignores multiples
if "PHOTO" in contact.custom: if "PHOTO" in contact.custom:
os.makedirs("pictures/" + contact.fn, exist_ok=True) os.makedirs("pictures/" + contact.fn, exist_ok=True)
with open("pictures/" + contact.fn + "/profile.jpg", "wb") as f: with open("pictures/" + contact.fn + "/profile.jpg", "wb") as f:
@ -33,6 +40,8 @@ for line in input_file:
contact.photo_path = "pictures/" + contact.fn + "/profile.jpg" contact.photo_path = "pictures/" + contact.fn + "/profile.jpg"
contact.custom = {} contact.custom = {}
# reformat phone numbers to international
# TODO : distinguish phone numbers in final vcard (hint : "type" ?!)
clean_tel = [] clean_tel = []
for number in contact.tel: for number in contact.tel:
try: try:
@ -44,25 +53,30 @@ for line in input_file:
clean_tel.append(number) clean_tel.append(number)
except phonenumbers.phonenumberutil.NumberParseException: except phonenumbers.phonenumberutil.NumberParseException:
continue continue
contact.tel = [{"value": x, "type": []} for x in clean_tel] contact.tel = [{"value": x, "type": []} for x in clean_tel]
# handle the horrible quoted-printable string format
# TODO : doesn't work for list of strings in vobject
# full-name OK
contact.fn = quopri.decodestring(contact.fn).decode() contact.fn = quopri.decodestring(contact.fn).decode()
# name list NOT
clean_n = [] clean_n = []
for name in contact.n: for name in contact.n:
clean = quopri.decodestring(name).decode() clean = quopri.decodestring(name).decode()
clean_n.append(clean) clean_n.append(clean)
contact.name = clean_n contact.name = clean_n
print(contact.name)
# nickname NOT USED
clean_nickname = [] clean_nickname = []
for nick in contact.nickname: for nick in contact.nickname:
nick = quopri.decodestring(nick).decode() nick = quopri.decodestring(nick).decode()
clean_n.append(nick) clean_n.append(nick)
contact.nickname = clean_nickname contact.nickname = clean_nickname
# rewrite contact as vcard
# TODO : override the function that encode lists of strings
vcf_text = contact.to_vcard() vcf_text = contact.to_vcard()
output.write(vcf_text) output.write(vcf_text)
current_card = "" current_card = ""
continue