From b21bc5825d1e500acb39ad347ba583515c19da2a Mon Sep 17 00:00:00 2001
From: gutmet <me.git@gutmet.org>
Date: Mon, 12 Oct 2020 10:57:08 +0200
Subject: [PATCH] fix bugs in status split (wrong character limit, omission of
 word parts)

---
 drivel.go | 40 +++++++++++++++++++++++++---------------
 1 file changed, 25 insertions(+), 15 deletions(-)

diff --git a/drivel.go b/drivel.go
index 7ad2962..2c68dcd 100644
--- a/drivel.go
+++ b/drivel.go
@@ -340,23 +340,33 @@ func kind(path string) mediaKind {
 }
 
 func splitStatus(status string) []string {
-	split := []string{}
-	words := strings.Split(status, " ")
-	s := ""
-	for _, word := range words {
-		asRunes := []rune(word)
-		if s == "" && len(asRunes) <= CHARACTER_LIMIT {
-			s = word
-		} else if len(s)+1+len(asRunes) <= CHARACTER_LIMIT {
-			s = s + " " + word
-		} else {
-			split = append(split, s)
-			bound := goutil.IntMin(len(asRunes), CHARACTER_LIMIT)
-			s = string(asRunes[:bound])
+	characterLimit := CHARACTER_LIMIT
+	// Twitter has an insane definition of what counts as a character
+	// ( see https://developer.twitter.com/en/docs/counting-characters )
+	// - as a crude approximation, anything outside LATIN-1 halfs the limit
+	for _, ch := range status {
+		if ch > 0x10FF {
+			characterLimit = CHARACTER_LIMIT / 2
+			break
 		}
 	}
-	if s != "" {
-		split = append(split, s)
+	asRunes := []rune(status)
+	split := []string{}
+	for len(asRunes) != 0 {
+		var limit int
+		if len(asRunes) <= characterLimit {
+			limit = len(asRunes)
+		} else {
+			tmp := asRunes[0:characterLimit]
+			lastSpace := strings.LastIndex(string(tmp), " ")
+			if lastSpace == -1 {
+				limit = characterLimit
+			} else {
+				limit = lastSpace + 1
+			}
+		}
+		split = append(split, string(asRunes[0:limit]))
+		asRunes = asRunes[limit:]
 	}
 	return split
 }