-- Copyright © 2010-2011 Jon Kristensen. See the LICENSE file in the Pontarius -- XMPP distribution for more details. {-# OPTIONS_HADDOCK hide #-} -- TODO: When no longer using stringprep, do appropriate testing. (Including -- testing addresses like a@b/c@d/e, a/b@c, a@/b, a/@b...) -- TODO: Unicode 3.2 should be used. -- This module deals with XMPP addresses (also known as JIDs and JabberIDs). For -- more information on XMPP addresses, see RFC 6122: XMPP: Address Format. -- -- This module does not internationalize hostnames. module Network.XMPP.Address (fromString, fromStrings, isBare, isFull) where import Network.XMPP.Types import Data.Maybe (fromJust, isJust) import Text.Parsec ((<|>), anyToken, char, eof, many, noneOf, parse) import Text.Parsec.ByteString (GenParser) import Text.StringPrep (StringPrepProfile (..), a1, b1, b2, c11, c12, c21, c22, c3, c4, c5, c6, c7, c8, c9, runStringPrep) import Text.NamePrep (namePrepProfile) import Network.URI (isIPv4address, isIPv6address) import qualified Data.ByteString.Char8 as DBC (pack) import qualified Data.Text as DT (pack, unpack) -- | -- Converts a string to an XMPP address. fromString :: String -> Maybe Address fromString s = fromStrings localpart domainpart resourcepart where Right (localpart, domainpart, resourcepart) = parse addressParts "" (DBC.pack s) -- | -- Converts localpart, domainpart, and resourcepart strings to an XMPP address. -- Runs the appropriate stringprep profiles and validates the parts. fromStrings :: Maybe String -> String -> Maybe String -> Maybe Address fromStrings l s r | domainpart == Nothing = Nothing | otherwise = if validateNonDomainpart localpart && isJust domainpart' && validateNonDomainpart resourcepart then Just (Address localpart (fromJust domainpart') resourcepart) else Nothing where -- Applies the nodeprep profile on the localpart string, if any. localpart :: Maybe String localpart = case l of Just l' -> case runStringPrep nodeprepProfile (DT.pack l') of Just l'' -> Just $ DT.unpack l'' Nothing -> Nothing Nothing -> Nothing -- Applies the nameprep profile on the domainpart string. -- TODO: Allow unassigned? domainpart :: Maybe String domainpart = case runStringPrep (namePrepProfile False) (DT.pack s) of Just s' -> Just $ DT.unpack s' Nothing -> Nothing -- Applies the resourceprep profile on the resourcepart string, if any. resourcepart :: Maybe String resourcepart = case r of Just r' -> case runStringPrep resourceprepProfile (DT.pack r') of Just r'' -> Just $ DT.unpack r'' Nothing -> Nothing Nothing -> Nothing -- Returns the domainpart if it was a valid IP or if the toASCII -- function was successful, or Nothing otherwise. domainpart' :: Maybe String domainpart' | isIPv4address s || isIPv6address s = Just s | validHostname s = Just s | otherwise = Nothing -- Validates that non-domainpart strings have an appropriate length. validateNonDomainpart :: Maybe String -> Bool validateNonDomainpart Nothing = True validateNonDomainpart (Just l) = validPartLength l where validPartLength :: String -> Bool validPartLength p = length p > 0 && length p < 1024 -- Validates a host name validHostname :: String -> Bool validHostname _ = True -- TODO -- | Returns True if the address is `bare', and False otherwise. isBare :: Address -> Bool isBare j | resourcepart j == Nothing = True | otherwise = False -- | Returns True if the address is `full', and False otherwise. isFull :: Address -> Bool isFull jid = not $ isBare jid -- Parses an address string and returns its three parts. It performs no -- validation or transformations. We are using Parsec to parse the address. -- There is no input for which 'addressParts' fails. addressParts :: GenParser Char st (Maybe String, String, Maybe String) addressParts = do -- Read until we reach an '@', a '/', or EOF. a <- many $ noneOf ['@', '/'] -- Case 1: We found an '@', and thus the localpart. At least the domainpart -- is remaining. Read the '@' and until a '/' or EOF. do char '@' b <- many $ noneOf ['/'] -- Case 1A: We found a '/' and thus have all the address parts. Read the -- '/' and until EOF. do char '/' -- Resourcepart remaining c <- many $ anyToken -- Parse resourcepart eof return (Just a, b, Just c) -- Case 1B: We have reached EOF; the address is in the form -- localpart@domainpart. <|> do eof return (Just a, b, Nothing) -- Case 2: We found a '/'; the address is in the form -- domainpart/resourcepart. <|> do char '/' b <- many $ anyToken eof return (Nothing, a, Just b) -- Case 3: We have reached EOF; we have an address consisting of only a -- domainpart. <|> do eof return (Nothing, a, Nothing) nodeprepProfile :: StringPrepProfile nodeprepProfile = Profile { maps = [b1, b2] , shouldNormalize = True , prohibited = [a1] ++ [c11, c12, c21, c22, c3, c4, c5, c6, c7, c8, c9] , shouldCheckBidi = True } -- These needs to be checked for after normalization. We could also look up the -- Unicode mappings and include a list of characters in the prohibited field -- above. Let's defer that until we know that we are going to use stringprep. nodeprepExtraProhibitedCharacters = ['\x22', '\x26', '\x27', '\x2F', '\x3A', '\x3C', '\x3E', '\x40'] resourceprepProfile :: StringPrepProfile resourceprepProfile = Profile { maps = [b1] , shouldNormalize = True , prohibited = [a1] ++ [c12, c21, c22, c3, c4, c5, c6, c7, c8, c9] , shouldCheckBidi = True }