wrap up

vcf_cleaner.py

@@ -1,23 +1,32 @@
 import os
+from re import escape
 import sys
 import quopri
 import base64
 import pickle
+import itertools
 import phonenumbers
+import pythonvCard4
 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()
 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
 
+# a contact is defined by its END:VCARD markup
 current_card = ""
 for line in input_file:
     current_card += line
 
     if "END:VCARD" in line:
+        # the pythonvCard4 parsing assumes a LOT of things
         contact = Contact.from_vcard(current_card)
 
+        # TODO : isolate contacts with multiple pictures
         if check_for_multiple_pictures:
             if "PHOTO" in contact.custom and len(contact.custom["PHOTO"]) > 1:
                 os.makedirs("multiple_pictures/" + contact.fn, exist_ok=True)
@@ -26,15 +35,19 @@ for line in input_file:
                     with open("photo/" + contact.fn + "/" + str(image) + ".jpg", "wb") as f:
                         f.write(base64.decodebytes(str.encode(contact.custom["PHOTO"][image])))
 
+        # choose the first pictures and ignores multiples
         if "PHOTO" in contact.custom:
             os.makedirs("pictures/" + contact.fn, exist_ok=True)
             with open("pictures/" + contact.fn + "/profile.jpg", "wb") as f:
                 f.write(base64.decodebytes(str.encode(contact.custom["PHOTO"][0])))
             contact.photo_path = "pictures/" + contact.fn + "/profile.jpg"
-        contact.custom = {}
+        contact.custom.pop("PHOTO", None)
 
+        # reformat phone numbers to international
         clean_tel = []
         for number in contact.tel:
+            if number["type"] != []:
+                print(number["type"])
             try:
                 number = number["value"]
                 number = number.replace("-", "")
@@ -44,25 +57,45 @@ for line in input_file:
                     clean_tel.append(number)
             except phonenumbers.phonenumberutil.NumberParseException:
                 continue
-
         contact.tel = [{"value": x, "type": []} for x in clean_tel]
 
+        # full-name reencoding
         contact.fn = quopri.decodestring(contact.fn).decode()
+        contact.custom.pop("FN", None)
 
+        # name list reencoding
         clean_n = []
         for name in contact.n:
             clean = quopri.decodestring(name).decode()
             clean_n.append(clean)
-        contact.name = clean_n
+        contact.n = clean_n
-        print(contact.name)
+        contact.custom.pop("N", None)
 
+        # nickname reencoding
         clean_nickname = []
         for nick in contact.nickname:
             nick = quopri.decodestring(nick).decode()
-            clean_n.append(nick)
+            clean_nickname.append(nick)
-        contact.nickname = clean_nickname
 
+        # custom field reencoding
+        for field in contact.custom:
+            for value in contact.custom[field]:
+                contact.custom[field] = quopri.decodestring(value).decode()
+
+        # notes reencoding
+        if contact.note:
+            contact.note = quopri.decodestring(contact.note).decode()
+
+        # deduplicate and reencode address
+        if contact.adr != []:
+            adresses = []
+            for adresse in contact.adr:
+                adresses.append([quopri.decodestring(x).decode() for x in adresse["value"]])
+            clean = list(k for k,_ in itertools.groupby(adresses))
+            clean = [{"value":x, "type":["HOME"]} for x in clean]
+            contact.adr = clean
+
+        # rewrite contact as vcard
         vcf_text = contact.to_vcard()
         output.write(vcf_text)
         current_card = ""
-        continue