(draft): telegram markdown v2 string builder
Golang lint / lint (push) Successful in 4m51s

This commit is contained in:
2026-04-30 13:56:36 +03:00
parent 269ccec007
commit 071fc2375e
34 changed files with 1227 additions and 385 deletions
+2
View File
@@ -0,0 +1,2 @@
// Package tgmd provides small helpers for Telegram Markdown text.
package tgmd
+295
View File
@@ -0,0 +1,295 @@
package tgmd
import (
"strings"
"time"
"git.scuroneko.dev/scuroneko/extypes"
"git.scuroneko.dev/scuroneko/laniakea/tgapi"
)
//TODO GoDoc, tests. Maybe escape Markdown v2
// MessageBuilder builds Telegram message text with explicit message entities.
// MessageBuilder is not safe for concurrent use.
type MessageBuilder struct {
str string
offset int
entities extypes.Slice[tgapi.MessageEntity]
entries extypes.Slice[*MessageBuilderEntry]
isDirty bool
}
// NewMessageBuilder returns an empty MessageBuilder.
func NewMessageBuilder() *MessageBuilder {
return &MessageBuilder{
entities: make([]tgapi.MessageEntity, 0),
entries: make(extypes.Slice[*MessageBuilderEntry], 0),
isDirty: false,
}
}
// String returns the built message text.
func (b *MessageBuilder) String() string {
if b.isDirty {
b.update()
}
return b.str
}
// Entities returns a copy of the built message entities.
func (b *MessageBuilder) Entities() []tgapi.MessageEntity {
if b.isDirty {
b.update()
}
return append([]tgapi.MessageEntity(nil), b.entities...)
}
func (b *MessageBuilder) Build() (string, []tgapi.MessageEntity) {
if b.isDirty {
b.update()
}
return b.str, append([]tgapi.MessageEntity(nil), b.entities...)
}
func (b *MessageBuilder) Reset() {
b.str = ""
b.offset = 0
b.entities = b.entities[:0]
b.entries = b.entries[:0]
b.isDirty = false
}
func (b *MessageBuilder) update() *MessageBuilder {
b.offset = 0
var textLen int
var entitiesLen int
for _, e := range b.entries {
textLen += len(e.text) // bytes, для Grow нормально
entitiesLen += len(e.entities)
}
b.entities = make(extypes.Slice[tgapi.MessageEntity], 0, entitiesLen)
var sb strings.Builder
sb.Grow(textLen)
for _, e := range b.entries {
sb.WriteString(e.text)
for _, entity := range e.entities {
entity.Offset += b.offset
b.entities = append(b.entities, entity)
}
b.offset += e.length
}
b.str = sb.String()
b.isDirty = false
return b
}
func (b *MessageBuilder) markDirty() {
b.isDirty = true
}
type MessageBuilderEntry struct {
text string
length int
b *MessageBuilder
entities extypes.Slice[tgapi.MessageEntity]
}
// Add appends plain text to the message and returns its entry for formatting.
func (b *MessageBuilder) Add(text string) *MessageBuilderEntry {
e := &MessageBuilderEntry{
b: b,
entities: make(extypes.Slice[tgapi.MessageEntity], 0),
text: text,
length: telegramTextLen(text),
}
b.entries = b.entries.Push(e)
b.markDirty()
return e
}
func (e *MessageBuilderEntry) Mention() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityMention,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Hashtag() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityHashtag,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Cashtag() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityCashtag,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) BotCommand() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityBotCommand,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Email() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityEmail,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Phone() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityPhoneNumber,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Bold() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityBold,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Italic() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityItalic,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Underline() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityUnderline,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Strikethrough() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityStrike,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Spoiler() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntitySpoiler,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) Quote() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityBlockquote,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) ExpandableQuote() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityExpandableBlockquote,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) InlineCode() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityCode,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) CodeBlock() *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityPre,
Offset: 0, Length: e.length,
})
return e
}
func (e *MessageBuilderEntry) CodeBlockWithLanguage(lang string) *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityPre,
Offset: 0, Length: e.length, Language: lang,
})
return e
}
func (e *MessageBuilderEntry) Link(url string) *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityTextLink,
Offset: 0, Length: e.length, URL: url,
})
return e
}
func (e *MessageBuilderEntry) TextMention(user *tgapi.User) *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityTextMention,
Offset: 0, Length: e.length, User: user,
})
return e
}
func (e *MessageBuilderEntry) CustomEmoji(emojiID string) *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityCustomEmoji,
Offset: 0, Length: e.length, CustomEmojiID: emojiID,
})
return e
}
func (e *MessageBuilderEntry) DateTime(time time.Time) *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityDateTime,
Offset: 0, Length: e.length, UnixTime: time.Unix(),
})
return e
}
func (e *MessageBuilderEntry) DateTimeFormat(time time.Time, format string) *MessageBuilderEntry {
e.addEntity(tgapi.MessageEntity{
Type: tgapi.MessageEntityDateTime,
Offset: 0, Length: e.length,
UnixTime: time.Unix(), DateTimeFormat: format,
})
return e
}
func telegramTextLen(text string) int {
n := 0
for _, r := range text {
if r <= 0xFFFF {
n++
} else {
n += 2
}
}
return n
}
func (e *MessageBuilderEntry) addEntity(entity tgapi.MessageEntity) {
if entity.Length <= 0 {
return
}
e.entities = append(e.entities, entity)
if e.b != nil {
e.b.markDirty()
}
}
+416
View File
@@ -0,0 +1,416 @@
package tgmd
import (
"reflect"
"testing"
"time"
"git.scuroneko.dev/scuroneko/laniakea/tgapi"
)
func TestMessageBuilder_BuildPlainText(t *testing.T) {
b := NewMessageBuilder()
b.Add("Hello")
b.Add(", ")
b.Add("world")
text, entities := b.Build()
if text != "Hello, world" {
t.Fatalf("text = %q, want %q", text, "Hello, world")
}
if len(entities) != 0 {
t.Fatalf("entities len = %d, want 0", len(entities))
}
}
func TestMessageBuilder_EntityOffsetsAreUTF16(t *testing.T) {
b := NewMessageBuilder()
b.Add("Hi ")
b.Add("👋") // 2 UTF-16 code units
b.Add(" ")
b.Add("world").Bold()
text, entities := b.Build()
if text != "Hi 👋 world" {
t.Fatalf("text = %q, want %q", text, "Hi 👋 world")
}
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityBold,
Offset: 6, // H i space = 3, 👋 = 2, space = 1
Length: 5,
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_EntityLengthIsUTF16(t *testing.T) {
b := NewMessageBuilder()
b.Add("👋").Bold()
text, entities := b.Build()
if text != "👋" {
t.Fatalf("text = %q, want %q", text, "👋")
}
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityBold,
Offset: 0,
Length: 2,
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_MultipleEntitiesOnSameEntry(t *testing.T) {
b := NewMessageBuilder()
b.Add("hello").Bold().Italic()
_, entities := b.Build()
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityBold,
Offset: 0,
Length: 5,
},
{
Type: tgapi.MessageEntityItalic,
Offset: 0,
Length: 5,
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_DoesNotDuplicateAfterRepeatedReads(t *testing.T) {
b := NewMessageBuilder()
b.Add("hello").Bold()
text1 := b.String()
entities1 := b.Entities()
text2 := b.String()
entities2 := b.Entities()
if text1 != text2 {
t.Fatalf("texts differ: %q != %q", text1, text2)
}
if !reflect.DeepEqual(entities1, entities2) {
t.Fatalf("entities differ: %#v != %#v", entities1, entities2)
}
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityBold,
Offset: 0,
Length: 5,
},
}
if !reflect.DeepEqual(entities2, want) {
t.Fatalf("entities = %#v, want %#v", entities2, want)
}
}
func TestMessageBuilder_AddEntityAfterStringMarksDirty(t *testing.T) {
b := NewMessageBuilder()
entry := b.Add("hello")
if got := b.String(); got != "hello" {
t.Fatalf("String() = %q, want %q", got, "hello")
}
entry.Bold()
entities := b.Entities()
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityBold,
Offset: 0,
Length: 5,
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_EntitiesReturnsCopy(t *testing.T) {
b := NewMessageBuilder()
b.Add("hello").Bold()
entities1 := b.Entities()
entities1[0].Offset = 999
entities2 := b.Entities()
if entities2[0].Offset != 0 {
t.Fatalf("Entities() did not return copy: offset = %d, want 0", entities2[0].Offset)
}
}
func TestMessageBuilder_BuildReturnsEntitiesCopy(t *testing.T) {
b := NewMessageBuilder()
b.Add("hello").Bold()
_, entities1 := b.Build()
entities1[0].Offset = 999
_, entities2 := b.Build()
if entities2[0].Offset != 0 {
t.Fatalf("Build() did not return entities copy: offset = %d, want 0", entities2[0].Offset)
}
}
func TestMessageBuilder_Reset(t *testing.T) {
b := NewMessageBuilder()
b.Add("hello").Bold()
if got := b.String(); got != "hello" {
t.Fatalf("String() before Reset = %q, want %q", got, "hello")
}
b.Reset()
text, entities := b.Build()
if text != "" {
t.Fatalf("text after Reset = %q, want empty", text)
}
if len(entities) != 0 {
t.Fatalf("entities len after Reset = %d, want 0", len(entities))
}
b.Add("world").Italic()
text, entities = b.Build()
if text != "world" {
t.Fatalf("text after reuse = %q, want %q", text, "world")
}
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityItalic,
Offset: 0,
Length: 5,
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities after reuse = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_EmptyEntryDoesNotCreateEntity(t *testing.T) {
b := NewMessageBuilder()
b.Add("").Bold()
b.Add("x")
text, entities := b.Build()
if text != "x" {
t.Fatalf("text = %q, want %q", text, "x")
}
if len(entities) != 0 {
t.Fatalf("entities len = %d, want 0: %#v", len(entities), entities)
}
}
func TestMessageBuilder_Link(t *testing.T) {
b := NewMessageBuilder()
b.Add("OpenAI").Link("https://openai.com")
_, entities := b.Build()
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityTextLink,
Offset: 0,
Length: 6,
URL: "https://openai.com",
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_CodeBlockWithLanguage(t *testing.T) {
b := NewMessageBuilder()
b.Add("fmt.Println(\"hi\")").CodeBlockWithLanguage("go")
_, entities := b.Build()
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityPre,
Offset: 0,
Length: 17,
Language: "go",
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestMessageBuilder_DateTimeFormat(t *testing.T) {
b := NewMessageBuilder()
ts := time.Unix(1772323200, 0)
b.Add("date").DateTimeFormat(ts, "MMMM d, yyyy")
_, entities := b.Build()
want := []tgapi.MessageEntity{
{
Type: tgapi.MessageEntityDateTime,
Offset: 0,
Length: 4,
UnixTime: 1772323200,
DateTimeFormat: "MMMM d, yyyy",
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
}
func TestTelegramTextLen(t *testing.T) {
tests := []struct {
name string
text string
want int
}{
{
name: "ascii",
text: "hello",
want: 5,
},
{
name: "cyrillic",
text: "привет",
want: 6,
},
{
name: "emoji",
text: "👋",
want: 2,
},
{
name: "mixed",
text: "a👋b",
want: 4,
},
{
name: "zwj sequence",
text: "👨‍👩‍👧‍👦",
want: 11,
},
{
name: "flag",
text: "🇫🇮",
want: 4,
},
{
name: "variation selector",
text: "❤️",
want: 2,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
got := telegramTextLen(tt.text)
if got != tt.want {
t.Fatalf("telegramTextLen(%q) = %d, want %d", tt.text, got, tt.want)
}
})
}
}
func TestMessageBuilder_SimpleEntityTypes(t *testing.T) {
tests := []struct {
name string
add func(*MessageBuilderEntry)
want tgapi.MessageEntityType
}{
{"mention", func(e *MessageBuilderEntry) { e.Mention() }, tgapi.MessageEntityMention},
{"hashtag", func(e *MessageBuilderEntry) { e.Hashtag() }, tgapi.MessageEntityHashtag},
{"cashtag", func(e *MessageBuilderEntry) { e.Cashtag() }, tgapi.MessageEntityCashtag},
{"bot command", func(e *MessageBuilderEntry) { e.BotCommand() }, tgapi.MessageEntityBotCommand},
{"email", func(e *MessageBuilderEntry) { e.Email() }, tgapi.MessageEntityEmail},
{"phone", func(e *MessageBuilderEntry) { e.Phone() }, tgapi.MessageEntityPhoneNumber},
{"bold", func(e *MessageBuilderEntry) { e.Bold() }, tgapi.MessageEntityBold},
{"italic", func(e *MessageBuilderEntry) { e.Italic() }, tgapi.MessageEntityItalic},
{"underline", func(e *MessageBuilderEntry) { e.Underline() }, tgapi.MessageEntityUnderline},
{"strikethrough", func(e *MessageBuilderEntry) { e.Strikethrough() }, tgapi.MessageEntityStrike},
{"spoiler", func(e *MessageBuilderEntry) { e.Spoiler() }, tgapi.MessageEntitySpoiler},
{"quote", func(e *MessageBuilderEntry) { e.Quote() }, tgapi.MessageEntityBlockquote},
{"expandable quote", func(e *MessageBuilderEntry) { e.ExpandableQuote() }, tgapi.MessageEntityExpandableBlockquote},
{"inline code", func(e *MessageBuilderEntry) { e.InlineCode() }, tgapi.MessageEntityCode},
{"code block", func(e *MessageBuilderEntry) { e.CodeBlock() }, tgapi.MessageEntityPre},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
b := NewMessageBuilder()
e := b.Add("hello")
tt.add(e)
_, entities := b.Build()
want := []tgapi.MessageEntity{
{
Type: tt.want,
Offset: 0,
Length: 5,
},
}
if !reflect.DeepEqual(entities, want) {
t.Fatalf("entities = %#v, want %#v", entities, want)
}
})
}
}
+71
View File
@@ -0,0 +1,71 @@
package tgmd
import (
"strconv"
"strings"
"git.scuroneko.dev/scuroneko/laniakea"
)
// Helpers in this file generate Telegram MarkdownV2.
// All user-provided text is escaped.
// TODO Markdown v2 escaping. GoDoc and tests
// WithBold returns s wrapped as bold Telegram Markdown text.
func WithBold(s string) string {
return "*" + s + "*"
}
// WithItalic returns s wrapped as italic Telegram Markdown text.
func WithItalic(s string) string {
return "_" + s + "_"
}
func WithUnderline(s string) string {
return "__" + s + "__"
}
func WithStrikethrough(s string) string {
return "~" + s + "~"
}
func WithSpoiler(s string) string {
return "||" + s + "||"
}
// WithLink returns a Telegram Markdown link for text and URL.
func WithLink(text, url string) string {
return "[" + text + "](" + url + ")"
}
func WithMention(text string, userID uint64) string {
return "[" + text + "](tg://user?id=" + strconv.FormatUint(userID, 10) + ")"
}
func WithEmoji(text, emojiID string) string {
return "[" + text + "](tg://emoji?id=" + emojiID + ")"
}
func WithTime(text string, unix uint64) string {
return "![" + text + "](tg://time?unix=" + strconv.FormatUint(unix, 10) + ")"
}
func WithTimeFormat(text string, unix uint64, format string) string {
return "![" + text + "](tg://time?unix=" +
strconv.FormatUint(unix, 10) +
"&format=" + format + ")"
}
// WithInlineCode returns s wrapped as inline code Telegram Markdown text.
func WithInlineCode(s string) string {
return "`" + s + "`"
}
func WithBlockCode(s string) string {
return "```\n" + s + "\n```"
}
func WithBlockCodeLanguage(s, lang string) string {
return "```" + lang + "\n" + s + "\n```"
}
func WithQuote(s string) string {
return ">" + strings.ReplaceAll(laniakea.EscapeMarkdownV2(s), "\n", "\n>")
}
func WithQuoteExpandable(s string) string {
return "**>" + s
}
+24
View File
@@ -0,0 +1,24 @@
package tgmd
import "testing"
func TestFormattingHelpers(t *testing.T) {
tests := []struct {
name string
got string
want string
}{
{name: "bold", got: WithBold("text"), want: "*text*"},
{name: "italic", got: WithItalic("text"), want: "_text_"},
{name: "inline code", got: WithInlineCode("text"), want: "`text`"},
{name: "link", got: WithLink("Laniakea", "https://example.test"), want: "[Laniakea](https://example.test)"},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
if tc.got != tc.want {
t.Fatalf("unexpected formatted text: got %q want %q", tc.got, tc.want)
}
})
}
}