core, ui: support trailing punctuation for mentions, URIs (also support domains), and email addresses (#5888)

* core: improve markdown parser for mentions, URIs, and email addresses

* ui

apps/ios/Shared/Views/Chat/ChatItem/MsgContentView.swift

@@ -288,7 +288,11 @@ func messageText(
             case .uri:
                 attrs = linkAttrs()
                 if !preview {
-                    attrs[linkAttrKey] = NSURL(string: ft.text)
+                    let s = t.lowercased()
+                    let link = s.hasPrefix("http://") || s.hasPrefix("https://")
+                                ? t
+                                : "https://" + t
+                    attrs[linkAttrKey] = NSURL(string: link)
                     attrs[webLinkAttrKey] = true
                     handleTaps = true
                 }
@@ -314,9 +318,9 @@ func messageText(
                         if m.memberId == userMemberId {
                             attrs[.foregroundColor] = UIColor.tintColor
                         }
-                        t = "@'\(name)'"
+                        t = mentionText(name)
                     } else {
-                        t = "@'\(memberName)'"
+                        t = mentionText(memberName)
                     }
                 }
             case .email:
@@ -351,6 +355,11 @@ func messageText(
     }
 }
 
+@inline(__always)
+private func mentionText(_ name: String) -> String {
+    name.contains(" @") ? "@'\(name)'" : "@\(name)"
+}
+
 func simplexLinkText(_ linkType: SimplexLinkType, _ smpHosts: [String]) -> String {
     linkType.description + " " + "(via \(smpHosts.first ?? "?"))"
 }

apps/ios/Shared/Views/Chat/Group/GroupMentions.swift

@@ -196,7 +196,9 @@ struct GroupMentionsView: View {
             newName = composeState.mentionMemberName(member.wrapped.memberProfile.displayName)
         }
         mentions[newName] = CIMention(groupMember: member.wrapped)
-        var msgMention = "@" + (newName.contains(" ") ? "'\(newName)'" : newName)
+        var msgMention = newName.contains(" ") || newName.last?.isPunctuation == true
+                        ? "@'\(newName)'"
+                        : "@\(newName)"
         var newPos = r.location + msgMention.count
         let newMsgLength = composeState.message.count + msgMention.count - r.length
         print(newPos)

apps/multiplatform/common/src/commonMain/kotlin/chat/simplex/common/model/ChatModel.kt

@@ -3902,7 +3902,7 @@ enum class MsgContentTag {
 class FormattedText(val text: String, val format: Format? = null) {
   // TODO make it dependent on simplexLinkMode preference
   fun link(mode: SimplexLinkMode): String? = when (format) {
-    is Format.Uri -> text
+    is Format.Uri -> if (text.startsWith("http://", ignoreCase = true) || text.startsWith("https://", ignoreCase = true)) text else "https://$text"
     is Format.SimplexLink -> if (mode == SimplexLinkMode.BROWSER) text else format.simplexUri
     is Format.Email -> "mailto:$text"
     is Format.Phone -> "tel:$text"

apps/multiplatform/common/src/commonMain/kotlin/chat/simplex/common/views/chat/group/GroupMentions.kt

@@ -23,6 +23,12 @@ import chat.simplex.common.views.chatlist.setGroupMembers
 import chat.simplex.common.views.helpers.*
 import chat.simplex.res.MR
 import kotlinx.coroutines.launch
+import kotlin.text.CharCategory.*
+
+val punctuation = setOf(
+  DASH_PUNCTUATION, START_PUNCTUATION, END_PUNCTUATION,
+  CONNECTOR_PUNCTUATION, OTHER_PUNCTUATION
+)
 
 private val PICKER_ROW_SIZE = MEMBER_ROW_AVATAR_SIZE + (MEMBER_ROW_VERTICAL_PADDING * 2f)
 private val MAX_PICKER_HEIGHT = (PICKER_ROW_SIZE * 4) + (MEMBER_ROW_AVATAR_SIZE + MEMBER_ROW_VERTICAL_PADDING - 4.dp)
@@ -126,7 +132,9 @@ fun GroupMentions(
     }
     val newName = existingMention?.key ?: composeState.value.mentionMemberName(member.memberProfile.displayName)
     mentions[newName] = CIMention(member)
-    var msgMention = "@" + if (newName.contains(" ")) "'$newName'" else newName
+    var msgMention = if (newName.contains(" ") || (newName.lastOrNull()?.category in punctuation))
+                      "@'$newName'"
+                      else "@$newName"
     var newPos = range.start + msgMention.length
     val newMsgLength = composeState.value.message.text.length + msgMention.length - range.length
     if (newPos == newMsgLength) {

apps/simplex-directory-service/src/Directory/Events.hs

@@ -16,7 +16,6 @@ module Directory.Events
     SDirectoryRole (..),
     crDirectoryEvent,
     directoryCmdTag,
-    viewName,
   )
 where
 
@@ -291,9 +290,6 @@ directoryCmdP =
         -- wordP = spacesP *> A.takeTill isSpace
         spacesP = A.takeWhile1 isSpace
 
-viewName :: Text -> Text
-viewName n = if T.any isSpace n then "'" <> n <> "'" else n
-
 directoryCmdTag :: DirectoryCmd r -> Text
 directoryCmdTag = \case
   DCHelp _ -> "help"

apps/simplex-directory-service/src/Directory/Service.hs

@@ -48,7 +48,7 @@ import Simplex.Chat.Bot
 import Simplex.Chat.Bot.KnownContacts
 import Simplex.Chat.Controller
 import Simplex.Chat.Core
-import Simplex.Chat.Markdown (FormattedText (..), Format (..), parseMaybeMarkdownList)
+import Simplex.Chat.Markdown (FormattedText (..), Format (..), parseMaybeMarkdownList, viewName)
 import Simplex.Chat.Messages
 import Simplex.Chat.Options
 import Simplex.Chat.Protocol (MsgContent (..))

src/Simplex/Chat/Markdown.hs

@@ -4,6 +4,7 @@
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE TemplateHaskell #-}
+{-# LANGUAGE TupleSections #-}
 {-# OPTIONS_GHC -Wno-unrecognised-pragmas #-}
 
 {-# HLINT ignore "Use newtype instead of data" #-}
@@ -16,7 +17,7 @@ import qualified Data.Aeson as J
 import qualified Data.Aeson.TH as JQ
 import Data.Attoparsec.Text (Parser)
 import qualified Data.Attoparsec.Text as A
-import Data.Char (isDigit, isPunctuation, isSpace)
+import Data.Char (isAlpha, isAscii, isDigit, isPunctuation, isSpace)
 import Data.Either (fromRight)
 import Data.Functor (($>))
 import Data.List (foldl', intercalate)
@@ -204,17 +205,18 @@ markdownP = mconcat <$> A.many' fragmentP
         else pure $ markdown (colored clr) s
     mentionP = do
       c <- A.char '@' *> A.peekChar'
-      name <- displayNameTextP
+      (name, punct) <- displayNameTextP_
       let sName = if c == '\'' then '\'' `T.cons` name `T.snoc` '\'' else name
-      pure $ markdown (Mention name) ('@' `T.cons` sName)
+          mention = markdown (Mention name) ('@' `T.cons` sName)
+      pure $ if T.null punct then mention else mention :|: unmarked punct
     colorP =
       A.anyChar >>= \case
-        'r' -> "ed" $> Red <|> pure Red
-        'g' -> "reen" $> Green <|> pure Green
-        'b' -> "lue" $> Blue <|> pure Blue
-        'y' -> "ellow" $> Yellow <|> pure Yellow
-        'c' -> "yan" $> Cyan <|> pure Cyan
-        'm' -> "agenta" $> Magenta <|> pure Magenta
+        'r' -> optional "ed" $> Red
+        'g' -> optional "reen" $> Green
+        'b' -> optional "lue" $> Blue
+        'y' -> optional "ellow" $> Yellow
+        'c' -> optional "yan" $> Cyan
+        'm' -> optional "agenta" $> Magenta
         '1' -> pure Red
         '2' -> pure Green
         '3' -> pure Blue
@@ -236,12 +238,14 @@ markdownP = mconcat <$> A.many' fragmentP
     wordMD :: Text -> Markdown
     wordMD s
       | T.null s = unmarked s
-      | isUri s =
-          let t = T.takeWhileEnd isPunctuation' s
-              uri = uriMarkdown $ T.dropWhileEnd isPunctuation' s
-           in if T.null t then uri else uri :|: unmarked t
-      | isEmail s = markdown Email s
+      | isUri s' = res $ uriMarkdown s'
+      | isDomain s' = res $ markdown Uri s'
+      | isEmail s' = res $ markdown Email s'
       | otherwise = unmarked s
+      where
+        punct = T.takeWhileEnd isPunctuation' s
+        s' = T.dropWhileEnd isPunctuation' s
+        res md' = if T.null punct then md' else md' :|: unmarked punct
     isPunctuation' = \case
       '/' -> False
       ')' -> False
@@ -250,6 +254,16 @@ markdownP = mconcat <$> A.many' fragmentP
       Right cLink -> markdown (simplexUriFormat cLink) s
       _ -> markdown Uri s
     isUri s = T.length s >= 10 && any (`T.isPrefixOf` s) ["http://", "https://", "simplex:/"]
+    -- matches what is likely to be a domain, not all valid domain names
+    isDomain s = case T.splitOn "." s of
+      [name, tld] -> isDomain_ name tld
+      [sub, name, tld] -> T.length sub >= 3 && T.length sub <= 8 && isDomain_ name tld
+      _ -> False
+      where
+        isDomain_ name tld =
+          (let n = T.length name in n >= 1 && n <= 24)
+            && (let n = T.length tld in n >= 2 && n <= 8)
+            && (let p c = isAscii c && isAlpha c in T.all p name && T.all p tld)
     isEmail s = T.any (== '@') s && Email.isValid (encodeUtf8 s)
     noFormat = pure . unmarked
     simplexUriFormat :: AConnectionLink -> Format
@@ -307,16 +321,22 @@ markdownText (FormattedText f_ t) = case f_ of
         White -> Nothing
 
 displayNameTextP :: Parser Text
-displayNameTextP = quoted '\'' <|> takeNameTill (== ' ')
+displayNameTextP = displayNameTextP_ >>= \(t, sfx) -> if T.null sfx then pure t else fail "Name ends with punctuation"
+{-# INLINE displayNameTextP #-}
+
+displayNameTextP_ :: Parser (Text, Text)
+displayNameTextP_ = (,"") <$> quoted '\'' <|> splitPunctuation <$> takeNameTill isSpace
   where
     takeNameTill p =
       A.peekChar' >>= \c ->
         if refChar c then A.takeTill p else fail "invalid first character in display name"
+    splitPunctuation s = (T.dropWhileEnd isPunctuation s, T.takeWhileEnd isPunctuation s)
     quoted c = A.char c *> takeNameTill (== c) <* A.char c
     refChar c = c > ' ' && c /= '#' && c /= '@' && c /= '\''
 
+-- quotes names that contain spaces or end on punctuation
 viewName :: Text -> Text
-viewName s = if T.any isSpace s then "'" <> s <> "'" else s
+viewName s = if T.any isSpace s || maybe False (isPunctuation . snd) (T.unsnoc s) then "'" <> s <> "'" else s
 
 $(JQ.deriveJSON (enumJSON $ dropPrefix "XL") ''SimplexLinkType)
 

src/Simplex/Chat/Store/SQLite/Migrations/agent_query_plans.txt

@@ -1119,7 +1119,7 @@ Query: UPDATE rcv_messages SET user_ack = ? WHERE conn_id = ? AND internal_id =
 Plan:
 SEARCH rcv_messages USING COVERING INDEX idx_rcv_messages_conn_id_internal_id (conn_id=? AND internal_id=?)
 
-Query: UPDATE rcv_queues SET last_broker_ts = ? WHERE conn_id = ? AND rcv_queue_id = ? AND last_broker_ts < ?
+Query: UPDATE rcv_queues SET last_broker_ts = ? WHERE conn_id = ? AND rcv_queue_id = ? AND (last_broker_ts IS NULL OR last_broker_ts < ?)
 Plan:
 SEARCH rcv_queues USING INDEX idx_rcv_queue_id (conn_id=? AND rcv_queue_id=?)
 

tests/Bots/DirectoryTests.hs

@@ -14,7 +14,6 @@ import Control.Exception (finally)
 import Control.Monad (forM_, when)
 import qualified Data.Text as T
 import Directory.Captcha
-import qualified Directory.Events as DE
 import Directory.Options
 import Directory.Service
 import Directory.Store
@@ -22,6 +21,7 @@ import GHC.IO.Handle (hClose)
 import Simplex.Chat.Bot.KnownContacts
 import Simplex.Chat.Controller (ChatConfig (..), ChatHooks (..), defaultChatHooks)
 import Simplex.Chat.Core
+import qualified Simplex.Chat.Markdown as MD
 import Simplex.Chat.Options (CoreChatOpts (..))
 import Simplex.Chat.Options.DB
 import Simplex.Chat.Types (Profile (..))
@@ -111,7 +111,7 @@ serviceDbPrefix :: FilePath
 serviceDbPrefix = "directory_service"
 
 viewName :: String -> String
-viewName = T.unpack . DE.viewName . T.pack
+viewName = T.unpack . MD.viewName . T.pack
 
 testDirectoryService :: HasCallStack => TestParams -> IO ()
 testDirectoryService ps =

tests/MarkdownTests.hs

@@ -192,10 +192,23 @@ textWithUri = describe "text with Uri" do
     "https://github.com/simplex-chat/ - SimpleX on GitHub" <==> uri "https://github.com/simplex-chat/" <> " - SimpleX on GitHub"
     -- "SimpleX on GitHub (https://github.com/simplex-chat/)" <==> "SimpleX on GitHub (" <> uri "https://github.com/simplex-chat/" <> ")"
     "https://en.m.wikipedia.org/wiki/Servo_(software)" <==> uri "https://en.m.wikipedia.org/wiki/Servo_(software)"
+    "example.com" <==> uri "example.com"
+    "example.com." <==> uri "example.com" <> "."
+    "example.com..." <==> uri "example.com" <> "..."
+    "www.example.com" <==> uri "www.example.com"
+    "example.academy" <==> uri "example.academy"
+    "this is example.com" <==> "this is " <> uri "example.com"
+    "x.com" <==> uri "x.com"
   it "ignored as markdown" do
     "_https://simplex.chat" <==> "_https://simplex.chat"
     "this is _https://simplex.chat" <==> "this is _https://simplex.chat"
     "this is https://" <==> "this is https://"
+    "example.c" <==> "example.c"
+    "www.www.example.com" <==> "www.www.example.com"
+    "www.example1.com" <==> "www.example1.com"
+    "www." <==> "www."
+    ".com" <==> ".com"
+    "example.academytoolong" <==> "example.academytoolong"
   it "SimpleX links" do
     let inv = "/invitation#/?v=1&smp=smp%3A%2F%2F1234-w%3D%3D%40smp.simplex.im%3A5223%2F3456-w%3D%3D%23%2F%3Fv%3D1-2%26dh%3DMCowBQYDK2VuAyEAjiswwI3O_NlS8Fk3HJUW870EY2bAwmttMBsvRB9eV3o%253D&e2e=v%3D2%26x3dh%3DMEIwBQYDK2VvAzkAmKuSYeQ_m0SixPDS8Wq8VBaTS1cW-Lp0n0h4Diu-kUpR-qXx4SDJ32YGEFoGFGSbGPry5Ychr6U%3D%2CMEIwBQYDK2VvAzkAmKuSYeQ_m0SixPDS8Wq8VBaTS1cW-Lp0n0h4Diu-kUpR-qXx4SDJ32YGEFoGFGSbGPry5Ychr6U%3D"
     ("https://simplex.chat" <> inv) <==> simplexLink XLInvitation ("simplex:" <> inv) ["smp.simplex.im"] ("https://simplex.chat" <> inv)
@@ -220,12 +233,14 @@ textWithEmail = describe "text with Email" do
     "test chat.chat+123@simplex.chat" <==> "test " <> email "chat.chat+123@simplex.chat"
     "chat@simplex.chat test" <==> email "chat@simplex.chat" <> " test"
     "test1 chat@simplex.chat test2" <==> "test1 " <> email "chat@simplex.chat" <> " test2"
-  it "ignored as markdown" do
+    "test chat@simplex.chat." <==> "test " <> email "chat@simplex.chat" <> "."
+    "test chat@simplex.chat..." <==> "test " <> email "chat@simplex.chat" <> "..."
+  it "ignored as email markdown" do
     "chat @simplex.chat" <==> "chat " <> mention "simplex.chat" "@simplex.chat"
     "this is chat @simplex.chat" <==> "this is chat " <> mention "simplex.chat" "@simplex.chat"
-    "this is chat@ simplex.chat" <==> "this is chat@ simplex.chat"
-    "this is chat @ simplex.chat" <==> "this is chat @ simplex.chat"
-    "*this* is chat @ simplex.chat" <==> bold "this" <> " is chat @ simplex.chat"
+    "this is chat@ simplex.chat" <==> "this is chat@ " <> uri "simplex.chat"
+    "this is chat @ simplex.chat" <==> "this is chat @ " <> uri "simplex.chat"
+    "*this* is chat @ simplex.chat" <==> bold "this" <> " is chat @ " <> uri "simplex.chat"
 
 phone :: Text -> Markdown
 phone = Markdown $ Just Phone
@@ -258,8 +273,13 @@ textWithMentions = describe "text with mentions" do
     "@alice" <==> mention "alice" "@alice"
     "hello @alice" <==> "hello " <> mention "alice" "@alice"
     "hello @alice !" <==> "hello " <> mention "alice" "@alice" <> " !"
+    "hello @alice!" <==> "hello " <> mention "alice" "@alice" <> "!"
+    "hello @alice..." <==> "hello " <> mention "alice" "@alice" <> "..."
+    "hello @alice@example.com" <==> "hello " <> mention "alice@example.com" "@alice@example.com"
+    "hello @'alice @ example.com'" <==> "hello " <> mention "alice @ example.com" "@'alice @ example.com'"
     "@'alice jones'" <==> mention "alice jones" "@'alice jones'"
     "hello @'alice jones'!" <==> "hello " <> mention "alice jones" "@'alice jones'" <> "!"
+    "hello @'a.j.'!" <==> "hello " <> mention "a.j." "@'a.j.'" <> "!"
   it "ignored as markdown" $ do
     "hello @'alice jones!" <==> "hello @'alice jones!"
     "hello @bob @'alice jones!" <==> "hello " <> mention "bob" "@bob" <> " @'alice jones!"