From c6556eb6b86d4267bb35b7a27c34a0046668e3b9 Mon Sep 17 00:00:00 2001
From: Bohdan Horbeshko <bodqhrohro@gmail.com>
Date: Fri, 11 Mar 2022 20:27:15 -0500
Subject: [PATCH] Clasp directives to the following span as required by
 XEP-0393

---
 telegram/formatter/formatter.go      | 51 +++++++++++++++++-
 telegram/formatter/formatter_test.go | 80 ++++++++++++++++++++++++++++
 2 files changed, 130 insertions(+), 1 deletion(-)

diff --git a/telegram/formatter/formatter.go b/telegram/formatter/formatter.go
index eca514a..d6541cb 100644
--- a/telegram/formatter/formatter.go
+++ b/telegram/formatter/formatter.go
@@ -2,6 +2,7 @@ package formatter
 
 import (
 	"sort"
+	"unicode"
 
 	log "github.com/sirupsen/logrus"
 	"github.com/zelenin/go-tdlib/client"
@@ -118,6 +119,54 @@ func MergeAdjacentEntities(entities []*client.TextEntity) []*client.TextEntity {
 	return mergedEntities
 }
 
+// ClaspDirectives to the following span as required by XEP-0393
+func ClaspDirectives(text string, entities []*client.TextEntity) []*client.TextEntity {
+	alignedEntities := make([]*client.TextEntity, len(entities))
+	copy(alignedEntities, entities)
+
+	// transform the source text into a form with uniform runes and code points,
+	// by duplicating the Basic Multilingual Plane
+	doubledRunes := make([]rune, 0, len(text)*2)
+
+	for _, cp := range text {
+		if cp > 0x0000ffff {
+			doubledRunes = append(doubledRunes, cp, cp)
+		} else {
+			doubledRunes = append(doubledRunes, cp)
+		}
+	}
+	for i, entity := range alignedEntities {
+		var dirty bool
+		endOffset := entity.Offset + entity.Length
+
+		if unicode.IsSpace(doubledRunes[entity.Offset]) {
+			for j, r := range doubledRunes[entity.Offset+1:endOffset] {
+				if !unicode.IsSpace(r) {
+					dirty = true
+					entity.Offset += int32(j+1)
+					entity.Length -= int32(j+1)
+					break
+				}
+			}
+		}
+		if unicode.IsSpace(doubledRunes[endOffset-1]) {
+			for j := endOffset-2; j >= entity.Offset; j-- {
+				if !unicode.IsSpace(doubledRunes[j]) {
+					dirty = true
+					entity.Length = j+1-entity.Offset
+					break
+				}
+			}
+		}
+
+		if dirty {
+			alignedEntities[i] = entity
+		}
+	}
+
+	return alignedEntities
+}
+
 func markupBraces(entity *client.TextEntity, lbrace, rbrace []rune) (*Insertion, *Insertion) {
 	return &Insertion{
 			Offset: entity.Offset,
@@ -191,7 +240,7 @@ func Format(
 		return sourceText
 	}
 
-	mergedEntities := SortEntities(MergeAdjacentEntities(SortEntities(entities)))
+	mergedEntities := SortEntities(ClaspDirectives(sourceText, MergeAdjacentEntities(SortEntities(entities))))
 
 	startStack := make(InsertionStack, 0, len(sourceText))
 	endStack := make(InsertionStack, 0, len(sourceText))
diff --git a/telegram/formatter/formatter_test.go b/telegram/formatter/formatter_test.go
index c988b10..e4bdd23 100644
--- a/telegram/formatter/formatter_test.go
+++ b/telegram/formatter/formatter_test.go
@@ -392,3 +392,83 @@ func TestFormattingXEP0393Strikethrough(t *testing.T) {
 		t.Errorf("Wrong strikethrough formatting: %v", markup)
 	}
 }
+
+func TestClaspLeft(t *testing.T) {
+	text := "a b c"
+	entities := []*client.TextEntity{
+		&client.TextEntity{
+			Offset: 1,
+			Length: 2,
+		},
+	}
+	entities = ClaspDirectives(text, entities)
+	if !(len(entities) == 1 &&
+		entities[0].Offset == 2 && entities[0].Length == 1) {
+		t.Errorf("Wrong claspleft: %#v", entities)
+	}
+}
+
+func TestClaspBoth(t *testing.T) {
+	text := "a b c"
+	entities := []*client.TextEntity{
+		&client.TextEntity{
+			Offset: 1,
+			Length: 3,
+		},
+	}
+	entities = ClaspDirectives(text, entities)
+	if !(len(entities) == 1 &&
+		entities[0].Offset == 2 && entities[0].Length == 1) {
+		t.Errorf("Wrong claspboth: %#v", entities)
+	}
+}
+
+func TestClaspNotNeeded(t *testing.T) {
+	text := " abc "
+	entities := []*client.TextEntity{
+		&client.TextEntity{
+			Offset: 1,
+			Length: 3,
+		},
+	}
+	entities = ClaspDirectives(text, entities)
+	if !(len(entities) == 1 &&
+		entities[0].Offset == 1 && entities[0].Length == 3) {
+		t.Errorf("Wrong claspnotneeded: %#v", entities)
+	}
+}
+
+func TestClaspNested(t *testing.T) {
+	text := "a b c"
+	entities := []*client.TextEntity{
+		&client.TextEntity{
+			Offset: 1,
+			Length: 3,
+		},
+		&client.TextEntity{
+			Offset: 2,
+			Length: 2,
+		},
+	}
+	entities = ClaspDirectives(text, entities)
+	if !(len(entities) == 2 &&
+		entities[0].Offset == 2 && entities[0].Length == 1 &&
+		entities[1].Offset == 2 && entities[1].Length == 1) {
+		t.Errorf("Wrong claspnested: %#v", entities)
+	}
+}
+
+func TestClaspEmoji(t *testing.T) {
+	text := "a 🐖 c"
+	entities := []*client.TextEntity{
+		&client.TextEntity{
+			Offset: 1,
+			Length: 4,
+		},
+	}
+	entities = ClaspDirectives(text, entities)
+	if !(len(entities) == 1 &&
+		entities[0].Offset == 2 && entities[0].Length == 2) {
+		t.Errorf("Wrong claspemoji: %#v", entities)
+	}
+}