Understanding vCard spec by reading the RFC spec(rfc6350)

Bharat Kalluri / 2021-01-01

This is also called as the vCard format. The spec is located on the IETF website (rfc6350). The complete spec is 75 pages! 75 pages of rules defining how to store contact information!


Specs are interesting, a group of people took upon a task of forming a set of rules for defining how contacts should be stored and transferred throughout multiple organizations. This sounds like a tremendously hard task (and it is). There must have been a lot of research done before deciding on a specification meant for the world.

Specs also show the breadth and depth of the problem. They showcase pieces which you and I might not think of. I think this also helps in appreciating the complexity of the problem.

I thought a good starting point for me would be a spec of a standard which is not too complicated. So I picked up the vCard spec. Well, turns out what I expected to happen was wrong. This spec is 75 pages! But then, I can't back down now. So here is what I have been able to get out reading the spec. Hope you also find this interesting.

Notes while reading the spec

  • mime type: text/vcard. Can contain one contact or a group of contacts
  • lines are delimited by line breaks, which is a CRLF sequence (U+000D followed by U+000A)
  • Line folding should be done at 75 octets (75 letters? since octets is 8 bits = byte?). Folding can be done by inserting a CRLF immediately followed by a single white space character (space (U+0020) or horizontal tab (U+0009))
  • The spec uses ABNF Format Definition (which has its own spec, a shorter description can be found on wikipedia). This could take time, but is a pretty powerful way of representing rules. Here is the complete vCard spec in ABNF
vcard-entity = 1*vcard
; A vCard object MUST include the VERSION and FN properties.
; VERSION MUST come immediately after BEGIN:VCARD.
contentline = [group "."] name *(";" param) ":" value CRLF
; When parsing a content line, folded lines must first
; be unfolded according to the unfolding procedure
; described in Section 3.2.
; When generating a content line, lines longer than 75
; characters SHOULD be folded according to the folding
; procedure described in Section 3.2.
group = 1*(ALPHA / DIGIT / "-")
name = "SOURCE" / "KIND" / "FN" / "N" / "NICKNAME"
/ "EMAIL" / "IMPP" / "LANG" / "TZ" / "GEO" / "TITLE" / "ROLE"
/ "URL" / "KEY" / "FBURL" / "CALADRURI" / "CALURI" / "XML"
/ iana-token / x-name
; Parsing of the param and value is based on the "name" as
; defined in ABNF sections below.
; Group and name are case-insensitive.
iana-token = 1*(ALPHA / DIGIT / "-")
; identifier registered with IANA
x-name = "x-" 1*(ALPHA / DIGIT / "-")
; Names that begin with "x-" or "X-" are
; reserved for experimental use, not intended for released
; products, or for use in bilateral agreements.
param = language-param / value-param / pref-param / pid-param
/ type-param / geo-parameter / tz-parameter / sort-as-param
/ calscale-param / any-param
; Allowed parameters depend on property name.
any-param = (iana-token / x-name) "=" param-value *("," param-value)
NON-ASCII = UTF8-2 / UTF8-3 / UTF8-4
; UTF8-{2,3,4} are defined in [RFC3629]
QSAFE-CHAR = WSP / "!" / %x23-7E / NON-ASCII
; Any character except CTLs, DQUOTE
SAFE-CHAR = WSP / "!" / %x23-39 / %x3C-7E / NON-ASCII
; Any character except CTLs, DQUOTE, ";", ":"
; Any textual character
  • Folding and unfolding is a critical piece. A line that begins with a white space character is a continuation of the previous line and should be treated accordingly.
  • Here is the ABNF spec for value
value = text
/ text-list
/ date-list
/ time-list
/ date-time-list
/ date-and-or-time-list
/ timestamp-list
/ boolean
/ integer-list
/ float-list
/ URI ; from Section 3 of [RFC3986]
/ utc-offset
/ Language-Tag
/ iana-valuespec
; Actual value type depends on property name and VALUE parameter.
text = *TEXT-CHAR
TEXT-CHAR = "\\" / "\," / "\n" / WSP / NON-ASCII
/ %x21-2B / %x2D-5B / %x5D-7E
; Backslashes, commas, and newlines must be encoded.
component = "\\" / "\," / "\;" / "\n" / WSP / NON-ASCII
/ %x21-2B / %x2D-3A / %x3C-5B / %x5D-7E
list-component = component *("," component)
text-list = text *("," text)
date-list = date *("," date)
time-list = time *("," time)
date-time-list = date-time *("," date-time)
date-and-or-time-list = date-and-or-time *("," date-and-or-time)
timestamp-list = timestamp *("," timestamp)
integer-list = integer *("," integer)
float-list = float *("," float)
boolean = "TRUE" / "FALSE"
integer = [sign] 1*DIGIT
float = [sign] 1*DIGIT ["." 1*DIGIT]
sign = "+" / "-"
year = 4DIGIT ; 0000-9999
month = 2DIGIT ; 01-12
day = 2DIGIT ; 01-28/29/30/31 depending on month and leap year
hour = 2DIGIT ; 00-23
minute = 2DIGIT ; 00-59
second = 2DIGIT ; 00-58/59/60 depending on leap second
zone = utc-designator / utc-offset
utc-designator = %x5A ; uppercase "Z"
date = year [month day]
/ year "-" month
/ "--" month [day]
/ "--" "-" day
date-noreduc = year month day
/ "--" month day
/ "--" "-" day
date-complete = year month day
time = hour [minute [second]] [zone]
/ "-" minute [second] [zone]
/ "-" "-" second [zone]
time-notrunc = hour [minute [second]] [zone]
time-complete = hour minute second [zone]
time-designator = %x54 ; uppercase "T"
date-time = date-noreduc time-designator time-notrunc
timestamp = date-complete time-designator time-complete
date-and-or-time = date-time / date / time-designator time
utc-offset = sign hour [minute]
Language-Tag = <Language-Tag, defined in [RFC5646], Section 2.1>
iana-valuespec = <value-spec, see Section 12>
; a publicly defined valuetype format, registered
; with IANA, as defined in Section 12 of this
; document.

All of this is to say the value can take a bunch of values. Different dates, timestamps with offsets, normal text etc..

  • param can take on multiple values as values as well
  • vCard properties are the most interesting and important component. There are
    • 5 general properties (BEGIN, END, SOURCE, KIND, XML)
    • 7 user identification properties (FN, N, NICKNAME, PHOTO, BDAY, ANNIVERSARY, GENDER)
    • 1 user address property (ADR)
    • 4 user communication properties (TEL, EMAIL, IMPP, LANG)
    • 2 user geo properties (TZ, GEO)
    • 6 user organization properties (TITLE, ROLE, LOGO, ORG, MEMBER, RELATED)
    • 1 security propery (KEY)
    • 3 calendar properties (FBURL, CALADURI, CALURI)
    • And some extended properties and parameters (begin with X- and can be an defined bilaterally between two agents)

I can talk about each one of them individually, but then I will just be making a copy of the spec. So if you have questions, I suggest you go through that piece of the spec to clarify your questions.

Trying to decode a .vcf file, following the spec

Let us pick up an example vcf document from icloud contacts as an example

FN:Bharat Kalluri
NOTE:Previously worked at JitFin\, Shubhloans and Wells Fargo.
item1.ADR;TYPE=HOME;TYPE=pref:;;2nd block;Banga
TEL;TYPE=CELL;TYPE=pref;TYPE=VOICE:(961) 199-3894
PRODID:-//Apple Inc.//iCloud Web Address Book 2023B58//EN

Every vCard begins with BEGIN:VCARD and ends with END:VCARD. And the second line must contain the version, in this case VERSION:3.0. Which is a bummer, the latest version of vCard is 4.0.

The next property is N. Spec here, cardinality is *1 which means there can be [0,1] entities of N. And N stands for name of the contact. Carefully read the special note, it says "The structured property value corresponds, in sequence, to the Family Names (also known as surnames), Given Names, Additional names, Honorific Prefixes, and Honorific Suffixes.The text components are separated by the SEMICOLON character (U+003B). Individual text components can include multiple text values separated by the COMMA character (U+002C)". There we have everything we need, so if this property was denoted as a json, it would be

"name": {
"content": {
"familyNames": ["Kalluri"],
"givenNames": ["Bharat"],
"additionalNames": ["Kumar"],
"honorificPrefixes": [],
"honorificSuffixes": []

Jumping to the next property, FN. Spec here. This talks about the name of the object the vCard represents. cardinality of 1*, which means only one entity is allowed. THis is straight forward, this is just a text field. The json here would be

"objName": {
"content": "Bharat Kalluri"

To the next one, ORG. Spec here. Its purpose is to specify organizational name and units. cardinality is *, which means [0, infinity]. Entities are separated by colon(;). The property ordered such that the first one is the org name and the consecutive levels (zero or more) are org unit names. One interesting observation, looks like the assumption is that one person can belong to one company? How would freelancers working for multiple corporations be represented? Not sure how that would work.. So, the json would be

"organizationInfo": {
"content": {
"name": "Refyne",
"units": ["Technology"]

The next prop, NICKNAME. Spec here. With a cardinality of *. One or more text values separated by comma. So the json representation is

"nickName": {
"content": "Myself"

Next prop, BDAY. Spec here. This is hard to make sense out of, looks like there can be a VALUE param which can be text/date-and-or-time. So the json here would be (I'm sure I'm missing a bunch of details here, but this works for now)

"birthDate": {
"value": "date",
"content": "1997-10-04"

Next prop, TITLE. Spec here. Talks about the specific job or position. cardinality is *. So, the json representation is

"title": {
"content": "SDE"

Next up, Note. Spec here. Supplemental information. This is just text.

"note": {
"content": "Previously worked at JitFin, Shubhloans and Wells Fargo."

Next, ADR. ignore the item.1, it is just a grouping mechanism. This is a structured field, so that sequence of values is post office box, extended address, street address, locality, region, postal code, country name. These must be specified for sure, it can be empty if needed. Each field can hold multiple values, separated by a comma.

"addresses": [
"type": ["home". "pref"],
"content": {
"postOfficeBox": "",
"extendedAddress": "",
"streetAddress": "House Number 174\n6th main\n2nd block",
"locality": "bangalore",
"region": "karnataka",
"postalCode": "560032",
"country": "India"

Next tag, TEL.Spec here. This can have a param of TYPE which can be text/voice/fax/cell/video/pager/telephone. Also can set PREF=1, to set it as preferred. It's cardinality is *.

"telephonesInfo": [
"type": ["cell", "pref", "voice"],
"content": "(961) 199-3894"

Next tag, URL. Spec here. Again, cardinality is *. It should just have one uri(uniform resource locator).

"URLsInfo": [
"type": ["pref"]
"content": "https://bharatkalluri.com"

Next tag, EMAIL. Spec here. cardinality is *. Just one email here per block.

"emailsInfo": [
"type": ["home", "pref", "internet"],
"content": "bharatkalluri@protonmail.com"

And now for the custom fields, obviously the RFC does not include social profiles like twitter, facebook etc.. So Apple made a choice to include them as extended properties(X-). It has a user, type and the URL. Here is how it will decode to json.

"extendedProperties": [
"socialProfile": {
"user": "bharatkalluri",
"type": ["twitter"],
"content": "http://twitter.com/bharatkalluri"

And finally PRODID. This has a cardinality of *1. This contains metadata about the product that created the vCard.

Well, that's it! This article turned out to much longer than I expected it to be. But hopefully this gives you an idea of what it is like to read a spec.

Some more Notes

  • A lot of the props can take in a type-param/language-param/altid-param etc.. so NICKNAME:Robbie and NICKNAME;TYPE=work:Boss are both valid. So the second string translates to {"nickName": {"type": "work", "content": "boss"}}
  • There is a beautiful article having some notes on the vcard format, check it out for sure
Spotify album cover

NowReading logDashboardUses