add regex parsing meta formatting

mautrix · Jul 30, 2024 · d4cea85 · d4cea85
1 parent 56b475a
commit d4cea85
Showing 1 changed file with 58 additions and 41 deletions.
diff --git a/pkg/connector/msgconv/mentions.go b/pkg/connector/msgconv/mentions.go
@@ -18,6 +18,7 @@ package msgconv
 
 import (
 	"context"
+	"regexp"
 	"slices"
 
 	//"log"
@@ -42,6 +43,14 @@ func (u UTF16String) String() string {
 	return string(utf16.Decode(u))
 }
 
+var (
+	META_BOLD_REGEX            = regexp.MustCompile(`\*([^*]+)\*`)
+	META_ITALIC_REGEX          = regexp.MustCompile(`_([^_]+)_`)
+	META_STRIKE_REGEX          = regexp.MustCompile(`~([^~]+)~`)
+	META_MONOSPACE_REGEX       = regexp.MustCompile("`([^`]+)`")
+	META_MONOSPACE_BLOCK_REGEX = regexp.MustCompile("```([^`]+)```")
+)
+
 func (mc *MessageConverter) metaToMatrixText(ctx context.Context, text string, rawMentions *socket.MentionData, portal *bridgev2.Portal) (content *event.MessageEventContent) {
 	content = &event.MessageEventContent{
 		MsgType:  event.MsgText,
@@ -52,53 +61,61 @@ func (mc *MessageConverter) metaToMatrixText(ctx context.Context, text string, r
 	if err != nil {
 		zerolog.Ctx(ctx).Err(err).Msg("Failed to parse mentions")
 	}
-	if mentions == nil {
-		return
-	}
-	utf16Text := NewUTF16String(text)
-	prevEnd := 0
-	var output strings.Builder
-	for _, mention := range mentions {
-		if mention.Offset < prevEnd {
-			zerolog.Ctx(ctx).Warn().Msg("Ignoring overlapping mentions in message")
-			continue
-		} else if mention.Offset >= len(utf16Text) {
-			zerolog.Ctx(ctx).Warn().Msg("Ignoring mention outside of message")
-			continue
-		}
-		end := mention.Offset + mention.Length
-		if end > len(utf16Text) {
-			end = len(utf16Text)
-		}
-		var mentionLink string
-		switch mention.Type {
-		case socket.MentionTypePerson:
-			info, err := mc.getBasicUserInfo(ctx, portal, ids.MakeUserID(mention.ID))
-			if err != nil {
-				zerolog.Ctx(ctx).Err(err).Msg("Failed to get user info for mention")
+
+	outputString := text
+
+	if mentions != nil {
+		utf16Text := NewUTF16String(text)
+		prevEnd := 0
+		var output strings.Builder
+		for _, mention := range mentions {
+			if mention.Offset < prevEnd {
+				zerolog.Ctx(ctx).Warn().Msg("Ignoring overlapping mentions in message")
+				continue
+			} else if mention.Offset >= len(utf16Text) {
+				zerolog.Ctx(ctx).Warn().Msg("Ignoring mention outside of message")
 				continue
 			}
-			if !slices.Contains(content.Mentions.UserIDs, info.MXID) {
-				content.Mentions.UserIDs = append(content.Mentions.UserIDs, info.MXID)
+			end := mention.Offset + mention.Length
+			if end > len(utf16Text) {
+				end = len(utf16Text)
 			}
-			mentionLink = info.MXID.URI().MatrixToURL()
-		case socket.MentionTypeThread:
-			// TODO: how does one send thread mentions?
-		}
-		if mentionLink == "" {
-			continue
+			var mentionLink string
+			switch mention.Type {
+			case socket.MentionTypePerson:
+				info, err := mc.getBasicUserInfo(ctx, portal, ids.MakeUserID(mention.ID))
+				if err != nil {
+					zerolog.Ctx(ctx).Err(err).Msg("Failed to get user info for mention")
+					continue
+				}
+				if !slices.Contains(content.Mentions.UserIDs, info.MXID) {
+					content.Mentions.UserIDs = append(content.Mentions.UserIDs, info.MXID)
+				}
+				mentionLink = info.MXID.URI().MatrixToURL()
+			case socket.MentionTypeThread:
+				// TODO: how does one send thread mentions?
+			}
+			if mentionLink == "" {
+				continue
+			}
+
+			output.WriteString(utf16Text[prevEnd:mention.Offset].String() + `<a href="` + mentionLink + `">` + utf16Text[mention.Offset:end].String() + `</a>`)
+			prevEnd = end
 		}
-		output.WriteString(utf16Text[prevEnd:mention.Offset].String())
-		output.WriteString(`<a href="`)
-		output.WriteString(mentionLink)
-		output.WriteString(`">`)
-		output.WriteString(utf16Text[mention.Offset:end].String())
-		output.WriteString(`</a>`)
-		prevEnd = end
+		output.WriteString(utf16Text[prevEnd:].String())
+
+		outputString = output.String()
 	}
-	output.WriteString(utf16Text[prevEnd:].String())
+
+	// Second parsing pass, replacing other formatting:
+	outputString = META_BOLD_REGEX.ReplaceAllString(outputString, "<strong>$1</strong>")
+	outputString = META_ITALIC_REGEX.ReplaceAllString(outputString, "<em>$1</em>")
+	outputString = META_STRIKE_REGEX.ReplaceAllString(outputString, "<del>$1</del>")
+	outputString = META_MONOSPACE_REGEX.ReplaceAllString(outputString, "<code>$1</code>")
+	outputString = META_MONOSPACE_BLOCK_REGEX.ReplaceAllString(outputString, "<pre>$1</pre>")
+
 	content.Format = event.FormatHTML
-	content.FormattedBody = output.String()
+	content.FormattedBody = outputString
 
 	log := zerolog.Ctx(ctx)
 	log.Debug().Str("text", text).Str("formatted_body", content.FormattedBody).Msg("Converted message to Matrix text")