diff --git a/internal/links/links.go b/internal/links/links.go index cc64b426..753519d3 100644 --- a/internal/links/links.go +++ b/internal/links/links.go @@ -135,6 +135,24 @@ func (b AppLinkBuilder) BuildUsername(username string, query url.Values) string return b.Build("resolve", query) } +// AcceptsEntityURL reports whether a raw custom-scheme URL belongs to this +// server's configured app-link namespace. The legacy route-as-host scheme is +// accepted for rollout compatibility; a distinct host-based scheme is limited +// to the exact configured host so arbitrary same-scheme links aren't promoted +// to clickable message entities by the server. +func (b AppLinkBuilder) AcceptsEntityURL(raw string) bool { + parsed, err := url.Parse(raw) + if err != nil || parsed.Opaque != "" || parsed.User != nil || parsed.Hostname() == "" || parsed.Port() != "" { + return false + } + if strings.EqualFold(parsed.Scheme, b.legacyScheme) { + return true + } + return b.baseHost != "" && + strings.EqualFold(parsed.Scheme, b.baseScheme) && + strings.EqualFold(parsed.Host, b.baseHost) +} + // MatchesRoute accepts the exact configured host-path form and the retained // legacy route-as-host form. Query validation remains the caller's concern. func (b AppLinkBuilder) MatchesRoute(parsed *url.URL, route string) bool { diff --git a/internal/links/links_test.go b/internal/links/links_test.go index 3ba0df30..2b41995c 100644 --- a/internal/links/links_test.go +++ b/internal/links/links_test.go @@ -161,6 +161,34 @@ func TestAppLinkBuilderPreservesLegacyAndSupportsHostBase(t *testing.T) { } } +func TestAppLinkBuilderAcceptsEntityURL(t *testing.T) { + hosted, err := NewAppLinkBuilder("telesrv", "owpg://links.example.test") + if err != nil { + t.Fatal(err) + } + for _, tc := range []struct { + name string + raw string + want bool + }{ + {name: "legacy route", raw: "telesrv://resolve?domain=Alice", want: true}, + {name: "legacy case insensitive", raw: "TELESRV://resolve?domain=Alice", want: true}, + {name: "configured host base", raw: "owpg://links.example.test/Alice", want: true}, + {name: "configured host case insensitive", raw: "OWPG://LINKS.EXAMPLE.TEST/Alice", want: true}, + {name: "same scheme wrong host", raw: "owpg://other.example.test/Alice", want: false}, + {name: "credentials", raw: "telesrv://user@resolve/path", want: false}, + {name: "port", raw: "telesrv://resolve:443/path", want: false}, + {name: "unconfigured scheme", raw: "other://resolve", want: false}, + {name: "missing route host", raw: "telesrv://", want: false}, + } { + t.Run(tc.name, func(t *testing.T) { + if got := hosted.AcceptsEntityURL(tc.raw); got != tc.want { + t.Fatalf("AcceptsEntityURL(%q) = %v, want %v", tc.raw, got, tc.want) + } + }) + } +} + func TestValidateAppName(t *testing.T) { if got, err := ValidateAppName(" Example Chat "); err != nil || got != "Example Chat" { t.Fatalf("ValidateAppName valid = %q, %v", got, err) diff --git a/internal/rpc/message_entities_autodetect.go b/internal/rpc/message_entities_autodetect.go index 472a448a..326627ed 100644 --- a/internal/rpc/message_entities_autodetect.go +++ b/internal/rpc/message_entities_autodetect.go @@ -6,6 +6,8 @@ import ( "unicode/utf8" "github.com/iamxvbaba/td/tg" + + "telesrv/internal/links" ) // 服务端自动实体检测:@mention / #hashtag / $cashtag / bot command。 @@ -24,16 +26,15 @@ import ( // augmentAutoEntities 在客户端已发实体基础上补充服务端检测的自动实体。补充项与任何 // 已有实体(客户端富文本意图实体或先补入的自动实体)区间相交时丢弃,避免把 mention/ // hashtag 打进 code/pre/textUrl/已有 mentionName 内部或彼此重叠(对齐官方不重复打实体)。 -// 客户端已带 url/textUrl 实体(自行检测过,如 DrKLO)时跳过 url 检测,沿用既有口径。 -// 客户端实体保持在前(超过上限裁剪时优先保留),结果裁剪到实体上限。 -func augmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg.MessageEntityClass { +// URL 按跨度逐条补缺,不能因客户端带了某一条 HTTP URL 就跳过同消息中客户端不认识的 +// app-link。客户端实体保持在前(超过上限裁剪时优先保留),结果裁剪到实体上限。 +func augmentAutoEntities(message string, entities []tg.MessageEntityClass, appLinks links.AppLinkBuilder) []tg.MessageEntityClass { // 快路径:绝大多数消息不含任何可自动识别的触发字符。单次 ContainsAny 扫描即短路返回, // 跳过下面各检测器对全文的扫描与区间分配(纯文本发送零额外开销)。所有 http(s) 链接 // 都含 '/',故 "@#$/" 一并覆盖 url 检测;email/phone 未实现故不在触发集内。 if message == "" || !strings.ContainsAny(message, "@#$/") { return entities } - hasClientURL := false type interval struct{ start, end int } occupied := make([]interval, 0, len(entities)+8) for _, e := range entities { @@ -41,10 +42,6 @@ func augmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg. off := e.GetOffset() occupied = append(occupied, interval{off, off + ln}) } - switch e.(type) { - case *tg.MessageEntityURL, *tg.MessageEntityTextURL: - hasClientURL = true - } } overlaps := func(s, e int) bool { for _, iv := range occupied { @@ -75,9 +72,9 @@ func augmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg. } // URL 跨度始终计算并加入排除区(occupied),使 @mention/#hashtag 等不会落进 URL 路径内部 - // (如 https://t.me/@scam 的 @scam,既不符官方语义也是钓鱼风险);但仅在客户端未带任何 - // url/textUrl 实体时才作为实体下发,沿用 all-or-nothing(DrKLO 一带即全带;TDesktop 不带、依赖服务端)。 - for _, u := range detectURLEntities(message) { + // (如 https://t.me/@scam 的 @scam,既不符官方语义也是钓鱼风险)。逐跨度补缺可覆盖 + // “客户端带 HTTP entity、但不认识 telesrv://”的混合消息,同时仍避免重复实体。 + for _, u := range detectURLEntities(message, appLinks) { ln := u.GetLength() if ln <= 0 { continue @@ -87,7 +84,7 @@ func augmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg. continue } occupied = append(occupied, interval{off, off + ln}) - if !hasClientURL && len(entities)+len(extra) < maxMessageEntityCount { + if len(entities)+len(extra) < maxMessageEntityCount { extra = append(extra, u) } } @@ -114,6 +111,10 @@ func augmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg. return append(out, extra...) } +func (r *Router) augmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg.MessageEntityClass { + return augmentAutoEntities(message, entities, r.appLinks) +} + // isWordRune 判定「单词字符」(用于实体前导边界:前一个字符是单词字符时不是新实体起点, // 借此排除 email 的 local@domain、路径里的 and/or 等)。 func isWordRune(r rune) bool { diff --git a/internal/rpc/messages_edit.go b/internal/rpc/messages_edit.go index 2e09ca6a..3af6790c 100644 --- a/internal/rpc/messages_edit.go +++ b/internal/rpc/messages_edit.go @@ -50,7 +50,7 @@ func (r *Router) onMessagesEditMessage(ctx context.Context, req *tg.MessagesEdit if hasMessage && richMessage == nil { // 编辑后的文本同样补服务端自动实体(url/@mention/#hashtag/bot command),与发送一致; // 覆盖频道/私聊编辑与各自的定时编辑分支(editScheduledMessage 仅由本处调用)。 - entities = augmentAutoEntities(message, entities) + entities = r.augmentAutoEntities(message, entities) } userID, _, err := r.currentUserID(ctx) if err != nil { diff --git a/internal/rpc/messages_edit_rpc_test.go b/internal/rpc/messages_edit_rpc_test.go index a930e97b..312043ee 100644 --- a/internal/rpc/messages_edit_rpc_test.go +++ b/internal/rpc/messages_edit_rpc_test.go @@ -62,7 +62,7 @@ func TestMessagesEditMessageReturnsUpdateAndRecordsOwnerContext(t *testing.T) { Peer: &tg.InputPeerUser{UserID: peerID, AccessHash: 22}, ID: 3, } - req.SetMessage("edited") + req.SetMessage("edited telesrv://resolve?domain=Alice") req.SetEntities([]tg.MessageEntityClass{&tg.MessageEntityBold{Offset: 0, Length: 6}}) var in bin.Buffer if err := req.Encode(&in); err != nil { @@ -85,14 +85,16 @@ func TestMessagesEditMessageReturnsUpdateAndRecordsOwnerContext(t *testing.T) { t.Fatalf("edit update = %#v, want pts=7 count=1", got.Updates[0]) } msg, ok := edit.Message.(*tg.Message) - if !ok || msg.ID != 3 || msg.Message != "edited" { - t.Fatalf("edited message = %#v, want id=3 text edited", edit.Message) + if !ok || msg.ID != 3 || msg.Message != "edited telesrv://resolve?domain=Alice" { + t.Fatalf("edited message = %#v, want id=3 text with app-link", edit.Message) } if messages.editReq.OwnerUserID != userID || messages.editReq.Peer.ID != peerID || messages.editReq.ID != 3 || messages.editReq.OriginAuthKeyID != authKeyID || messages.editReq.OriginSessionID != 77 { t.Fatalf("edit request = %+v, want owner peer message id and origin", messages.editReq) } - if len(messages.editReq.Entities) != 1 || messages.editReq.Entities[0].Type != domain.MessageEntityBold { - t.Fatalf("edit entities = %+v, want bold", messages.editReq.Entities) + if len(messages.editReq.Entities) != 2 || messages.editReq.Entities[0].Type != domain.MessageEntityBold || + messages.editReq.Entities[1].Type != domain.MessageEntityURL || messages.editReq.Entities[1].Offset != 7 || + messages.editReq.Entities[1].Length != utf16CodeUnitLen("telesrv://resolve?domain=Alice") { + t.Fatalf("edit entities = %+v, want bold plus configured app-link", messages.editReq.Entities) } } diff --git a/internal/rpc/messages_monoforum_rpc_test.go b/internal/rpc/messages_monoforum_rpc_test.go index b893699d..f83de819 100644 --- a/internal/rpc/messages_monoforum_rpc_test.go +++ b/internal/rpc/messages_monoforum_rpc_test.go @@ -303,7 +303,7 @@ func TestMonoforumSendMessageWritePath(t *testing.T) { } // TDesktop 的订阅者请求不携带 InputReplyToMonoForum;服务端必须从调用者推导 saved_peer=self。 - subReq := &tg.MessagesSendMessageRequest{Peer: monoInput, Message: "hi from sub", RandomID: 555} + subReq := &tg.MessagesSendMessageRequest{Peer: monoInput, Message: "hi from sub telesrv://resolve?domain=Owner", RandomID: 555} subReq.ClearDraft = true subReq.SetAllowPaidStars(20) suggestedInput := tg.SuggestedPost{} @@ -325,6 +325,15 @@ func TestMonoforumSendMessageWritePath(t *testing.T) { if message, ok := newMessage.Message.(*tg.Message); ok { subMessageID = message.ID subPaidStars, _ = message.GetPaidMessageStars() + var hasAppLink bool + for _, entity := range message.Entities { + if url, ok := entity.(*tg.MessageEntityURL); ok && url.Offset == utf16CodeUnitLen("hi from sub ") && url.Length == utf16CodeUnitLen("telesrv://resolve?domain=Owner") { + hasAppLink = true + } + } + if !hasAppLink { + t.Fatalf("monoforum message missing configured app-link entity: %+v", message.Entities) + } } } if balance, ok := update.(*tg.UpdateStarsBalance); ok { diff --git a/internal/rpc/messages_quick_replies.go b/internal/rpc/messages_quick_replies.go index 3c2574fe..94e11a21 100644 --- a/internal/rpc/messages_quick_replies.go +++ b/internal/rpc/messages_quick_replies.go @@ -272,7 +272,7 @@ func (r *Router) onMessagesSaveQuickReplyText(ctx context.Context, req *tg.Messa Date: int(r.clock.Now().Unix()), Message: req.Message, // 快速回复模板与普通发送一致补服务端自动实体(url/@mention/#hashtag/bot command)。 - Entities: domainMessageEntitiesForViewer(userID, augmentAutoEntities(req.Message, req.Entities)), + Entities: domainMessageEntitiesForViewer(userID, r.augmentAutoEntities(req.Message, req.Entities)), }) if err != nil { return nil, businessAutomationErr(err) diff --git a/internal/rpc/messages_send.go b/internal/rpc/messages_send.go index e3beb3d9..2eb8f30d 100644 --- a/internal/rpc/messages_send.go +++ b/internal/rpc/messages_send.go @@ -65,6 +65,12 @@ func (r *Router) onMessagesSendMessage(ctx context.Context, req *tg.MessagesSend sendErr = internalErr() return nil, sendErr } + // 自动实体必须在普通/频道/monoforum 分流前完成,保证所有文本写路径持久化同一份 + // 服务端补全结果。指纹仍基于原始请求,派生实体不改变 random_id 幂等语义。 + entities := req.Entities + if req.RichMessage == nil { + entities = r.augmentAutoEntities(req.Message, entities) + } suggestedInput, hasSuggestedPost := req.GetSuggestedPost() // monoforum 普通用户发送不带 reply_to,saved_peer 必须由服务端推导为自己;管理员回复才必须 // 显式携带 monoforum_peer_id。仅凭 reply_to 判路由会把用户请求误送进普通 megagroup 路径。 @@ -127,7 +133,7 @@ func (r *Router) onMessagesSendMessage(ctx context.Context, req *tg.MessagesSend IdempotencyFingerprint: idempotencyFingerprint, IdempotencyPreflighted: replay.checked, Message: req.Message, - Entities: domainMessageEntities(req.Entities), + Entities: domainMessageEntities(entities), ReplyTo: replyTo, Silent: req.Silent, NoForwards: req.Noforwards, @@ -198,16 +204,11 @@ func (r *Router) onMessagesSendMessage(ctx context.Context, req *tg.MessagesSend sendErr = messageEmptyErr() return nil, sendErr } - // 自动实体高亮:客户端未带 url/@mention/#hashtag/bot command 等「可自动识别」实体时,服务端 - // 检测原文补充(官方服务端行为),否则 @username/链接等不渲染为可点蓝色。富文本走独立结构,不处理。 - if richMessage == nil { - req.Entities = augmentAutoEntities(req.Message, req.Entities) - } // 链接预览:纯文本消息(私聊或频道)含可预览 URL 且未抑制时,挂 pending 占位,异步解析回填。 // 富文本消息有独立媒体语义,不叠加。 var previewMedia *domain.MessageMedia if richMessage == nil { - previewMedia = r.webPageMediaFromText(ctx, req.Message, req.Entities, req.NoWebpage, req.InvertMedia) + previewMedia = r.webPageMediaFromText(ctx, req.Message, entities, req.NoWebpage, req.InvertMedia) } if req.ScheduleDate != 0 && !scheduleDateIsImmediate(req.ScheduleDate, int(r.clock.Now().Unix())) { updates, err := r.scheduleOutgoing(ctx, userID, peer, outgoingSend{ @@ -215,7 +216,7 @@ func (r *Router) onMessagesSendMessage(ctx context.Context, req *tg.MessagesSend idempotencyFingerprint: idempotencyFingerprint, idempotencyPreflighted: replay.checked, message: req.Message, - entities: req.Entities, + entities: entities, media: previewMedia, silent: req.Silent, noforwards: req.Noforwards, @@ -236,7 +237,7 @@ func (r *Router) onMessagesSendMessage(ctx context.Context, req *tg.MessagesSend idempotencyFingerprint: idempotencyFingerprint, idempotencyPreflighted: replay.checked, message: req.Message, - entities: req.Entities, + entities: entities, media: previewMedia, silent: req.Silent, noforwards: req.Noforwards, diff --git a/internal/rpc/messages_send_webpage_rpc_test.go b/internal/rpc/messages_send_webpage_rpc_test.go index 47b56109..1751850b 100644 --- a/internal/rpc/messages_send_webpage_rpc_test.go +++ b/internal/rpc/messages_send_webpage_rpc_test.go @@ -227,6 +227,110 @@ func TestSendMessageHighlightsBareURL(t *testing.T) { } } +// TestSendMessageHighlightsConfiguredAppLink locks the server-only compatibility +// contract: clients may omit the custom-scheme entity, while the persisted/echoed +// message still carries MessageEntityURL and never starts webpage resolution. +func TestSendMessageHighlightsConfiguredAppLink(t *testing.T) { + ctx := context.Background() + r, owner, friend := newMediaTestRouter(t) + r.deps.Files.(*fakeFiles).webPagePreviewOn = true + message := "👍 telesrv://resolve?domain=Alice" + + updates, err := r.onMessagesSendMessage(WithUserID(ctx, owner.ID), &tg.MessagesSendMessageRequest{ + Peer: &tg.InputPeerUser{UserID: friend.ID, AccessHash: friend.AccessHash}, + Message: message, + RandomID: 5302, + }) + if err != nil { + t.Fatalf("sendMessage: %v", err) + } + msg := newMessageFromUpdates(t, updates) + if msg.Media != nil { + t.Fatalf("custom app-link media = %T, want nil", msg.Media) + } + for _, entity := range msg.Entities { + if url, ok := entity.(*tg.MessageEntityURL); ok && url.Offset == 3 && url.Length == utf16CodeUnitLen("telesrv://resolve?domain=Alice") { + return + } + } + t.Fatalf("sent message missing configured app-link entity: %+v", msg.Entities) +} + +// TestSendMessageFillsCustomLinkBesideClientHTTPEntity covers the partial-client +// case exposed by Android: an existing HTTP entity must not suppress app-link detection. +func TestSendMessageFillsCustomLinkBesideClientHTTPEntity(t *testing.T) { + ctx := context.Background() + r, owner, friend := newMediaTestRouter(t) + message := "https://x telesrv://resolve?domain=Alice" + + updates, err := r.onMessagesSendMessage(WithUserID(ctx, owner.ID), &tg.MessagesSendMessageRequest{ + Peer: &tg.InputPeerUser{UserID: friend.ID, AccessHash: friend.AccessHash}, + Message: message, + Entities: []tg.MessageEntityClass{&tg.MessageEntityURL{Offset: 0, Length: utf16CodeUnitLen("https://x")}}, + RandomID: 5303, + }) + if err != nil { + t.Fatalf("sendMessage: %v", err) + } + msg := newMessageFromUpdates(t, updates) + if len(msg.Entities) != 2 { + t.Fatalf("message entities = %+v, want client HTTP plus server app-link", msg.Entities) + } + custom, ok := msg.Entities[1].(*tg.MessageEntityURL) + if !ok || custom.Offset != utf16CodeUnitLen("https://x ") || custom.Length != utf16CodeUnitLen("telesrv://resolve?domain=Alice") { + t.Fatalf("custom entity = %#v", msg.Entities[1]) + } +} + +func TestAutoEntityDerivationDoesNotMutateSendRequests(t *testing.T) { + t.Run("send-message-replay", func(t *testing.T) { + ctx := context.Background() + r, owner, friend := newMediaTestRouter(t) + req := &tg.MessagesSendMessageRequest{ + Peer: &tg.InputPeerUser{UserID: friend.ID, AccessHash: friend.AccessHash}, + Message: "telesrv://resolve?domain=Alice", + RandomID: 5304, + } + first, err := r.onMessagesSendMessage(WithUserID(ctx, owner.ID), req) + if err != nil { + t.Fatalf("first sendMessage: %v", err) + } + if len(req.Entities) != 0 { + t.Fatalf("sendMessage request was mutated: %+v", req.Entities) + } + if _, err := r.onMessagesSendMessage(WithUserID(ctx, owner.ID), req); err != nil { + t.Fatalf("sendMessage replay: %v", err) + } + if got := newMessageFromUpdates(t, first); len(got.Entities) != 1 { + t.Fatalf("first send entities = %+v, want derived app-link", got.Entities) + } + }) + + t.Run("send-media-replay", func(t *testing.T) { + ctx := context.Background() + r, owner, friend := newMediaTestRouter(t) + req := &tg.MessagesSendMediaRequest{ + Peer: &tg.InputPeerUser{UserID: friend.ID, AccessHash: friend.AccessHash}, + Media: &tg.InputMediaContact{PhoneNumber: "+15550005305", FirstName: "Alice"}, + Message: "telesrv://resolve?domain=Alice", + RandomID: 5305, + } + first, err := r.onMessagesSendMedia(WithUserID(ctx, owner.ID), req) + if err != nil { + t.Fatalf("first sendMedia: %v", err) + } + if len(req.Entities) != 0 { + t.Fatalf("sendMedia request was mutated: %+v", req.Entities) + } + if _, err := r.onMessagesSendMedia(WithUserID(ctx, owner.ID), req); err != nil { + t.Fatalf("sendMedia replay: %v", err) + } + if got := newMessageFromUpdates(t, first); len(got.Entities) != 1 { + t.Fatalf("first media caption entities = %+v, want derived app-link", got.Entities) + } + }) +} + // TestSendMessageNoURLNoPlaceholder 验证无 URL 实体时不挂占位。 func TestSendMessageNoURLNoPlaceholder(t *testing.T) { ctx := context.Background() diff --git a/internal/rpc/send_media.go b/internal/rpc/send_media.go index a1caf325..9d7cffff 100644 --- a/internal/rpc/send_media.go +++ b/internal/rpc/send_media.go @@ -269,7 +269,9 @@ func (r *Router) onMessagesSendMedia(ctx context.Context, req *tg.MessagesSendMe // 原始 TL 指纹;编码失败则留空,由 store 使用 domain fallback。 idempotencyFingerprint, _ := sendMediaIdempotencyFingerprint(req) // 媒体 caption 里的链接/@mention/#hashtag 等同样补自动高亮实体(客户端未带时)。 - req.Entities = augmentAutoEntities(req.Message, req.Entities) + // 派生结果保持在局部变量中,不能回写原始 TL request,否则同一请求对象重试时指纹会 + // 从“客户端输入”变成“服务端派生输入”,错误触发 RANDOM_ID_DUPLICATE。 + entities := r.augmentAutoEntities(req.Message, req.Entities) userID, _, err := r.currentUserID(ctx) if err != nil { return nil, internalErr() @@ -351,7 +353,7 @@ func (r *Router) onMessagesSendMedia(ctx context.Context, req *tg.MessagesSendMe IdempotencyFingerprint: idempotencyFingerprint, IdempotencyPreflighted: replay.checked, Message: req.Message, - Entities: domainMessageEntities(req.Entities), + Entities: domainMessageEntities(entities), Media: media, ReplyTo: replyTo, Silent: req.Silent, @@ -414,7 +416,7 @@ func (r *Router) onMessagesSendMedia(ctx context.Context, req *tg.MessagesSendMe idempotencyFingerprint: idempotencyFingerprint, idempotencyPreflighted: replay.checked, message: req.Message, - entities: req.Entities, + entities: entities, media: media, silent: req.Silent, noforwards: req.Noforwards, @@ -429,7 +431,7 @@ func (r *Router) onMessagesSendMedia(ctx context.Context, req *tg.MessagesSendMe idempotencyFingerprint: idempotencyFingerprint, idempotencyPreflighted: replay.checked, message: req.Message, - entities: req.Entities, + entities: entities, media: media, silent: req.Silent, noforwards: req.Noforwards, @@ -583,7 +585,7 @@ func (r *Router) onMessagesSendMultiMedia(ctx context.Context, req *tg.MessagesS idempotencyFingerprint: idempotencyFingerprint, idempotencyPreflighted: replays[i].checked, message: item.Message, - entities: augmentAutoEntities(item.Message, item.Entities), + entities: r.augmentAutoEntities(item.Message, item.Entities), media: media, silent: req.Silent, noforwards: req.Noforwards, diff --git a/internal/rpc/webpage_url_extract.go b/internal/rpc/webpage_url_extract.go index 155eaca0..6b89e77f 100644 --- a/internal/rpc/webpage_url_extract.go +++ b/internal/rpc/webpage_url_extract.go @@ -9,19 +9,22 @@ import ( "github.com/iamxvbaba/td/tg" "telesrv/internal/domain" + "telesrv/internal/links" ) -// urlInTextRe 匹配原始文本里的 http(s) 链接(取到首个空白或尖括号/引号为止)。 -var urlInTextRe = regexp.MustCompile(`https?://[^\s<>"')】]+`) +// urlInTextRe 匹配原始文本里的带 scheme 链接(取到首个空白或尖括号/引号为止)。 +// 是否接纳由 detectURLEntities 再按 http(s) 或当前 app-link 配置收口,不能把任意 +// foo:// 都提升为服务端认证的可点击实体。 +var urlInTextRe = regexp.MustCompile(`(?i)[a-z][a-z0-9+.-]*://[^\s<>"')】]+`) // urlTrailingPunct 是不属于 URL 的句末标点('/' 是合法路径末尾,保留)。 const urlTrailingPunct = ".,;:!?)]}'\"。,、!?" -// detectURLEntities 服务端扫描消息文本生成 url 高亮实体(MessageEntityURL)。TDesktop 等客户端 -// 发消息不带 url 实体、依赖服务端检测原文(官方服务端行为),否则链接不高亮。偏移/长度按 -// UTF-16 码元(Telegram 实体口径)。 -func detectURLEntities(message string) []tg.MessageEntityClass { - if !strings.Contains(message, "http") { +// detectURLEntities 服务端扫描消息文本生成 url 高亮实体(MessageEntityURL)。除 http(s) +// 外,仅接受当前 Router 配置允许的 app-link scheme/host。偏移/长度按 UTF-16 码元 +// (Telegram 实体口径)。自定义 scheme 只参与 entity,不改变网页预览的 http(s) 边界。 +func detectURLEntities(message string, appLinks links.AppLinkBuilder) []tg.MessageEntityClass { + if !strings.Contains(message, "://") { return nil } locs := urlInTextRe.FindAllStringIndex(message, -1) @@ -34,6 +37,14 @@ func detectURLEntities(message string) []tg.MessageEntityClass { if raw == "" { continue } + schemeEnd := strings.Index(raw, "://") + if schemeEnd <= 0 { + continue + } + scheme := raw[:schemeEnd] + if !strings.EqualFold(scheme, "http") && !strings.EqualFold(scheme, "https") && !appLinks.AcceptsEntityURL(raw) { + continue + } out = append(out, &tg.MessageEntityURL{ Offset: utf16CodeUnitLen(message[:loc[0]]), Length: utf16CodeUnitLen(raw), @@ -69,8 +80,9 @@ func firstPreviewableURL(message string, entities []tg.MessageEntityClass) (stri return normalized, true } } - // 回退扫原始文本:绝大多数消息无链接,无 "http" 子串则直接跳过正则与分配。 - if !strings.Contains(message, "http") { + // 回退扫原始文本:绝大多数消息无链接,无 "://" 子串则直接跳过正则与分配。 + // firstURLInText 会继续限定为 http(s),自定义 app-link 永不进入网页预览。 + if !strings.Contains(message, "://") { return "", false } if raw, ok := firstURLInText(message); ok { @@ -83,16 +95,15 @@ func firstPreviewableURL(message string, entities []tg.MessageEntityClass) (stri // firstURLInText 扫描原始文本里的首个 http(s) 链接,剥掉句末标点。 func firstURLInText(message string) (string, bool) { - match := urlInTextRe.FindString(message) - if match == "" { - return "", false + for _, match := range urlInTextRe.FindAllString(message, -1) { + // 句末标点不属于 URL("见 https://example.com。" / "...go).")。'/' 是合法路径末尾,保留。 + match = strings.TrimRight(match, urlTrailingPunct) + schemeEnd := strings.Index(match, "://") + if schemeEnd > 0 && (strings.EqualFold(match[:schemeEnd], "http") || strings.EqualFold(match[:schemeEnd], "https")) { + return match, true + } } - // 句末标点不属于 URL("见 https://example.com。" / "...go).")。'/' 是合法路径末尾,保留。 - match = strings.TrimRight(match, urlTrailingPunct) - if match == "" { - return "", false - } - return match, true + return "", false } // sliceUTF16 按 UTF-16 码元偏移/长度从已编码序列取子串;越界返回空串。 diff --git a/internal/rpc/webpage_url_extract_test.go b/internal/rpc/webpage_url_extract_test.go index 990d99f2..abd4d893 100644 --- a/internal/rpc/webpage_url_extract_test.go +++ b/internal/rpc/webpage_url_extract_test.go @@ -4,8 +4,31 @@ import ( "testing" "github.com/iamxvbaba/td/tg" + + "telesrv/internal/links" ) +var testDefaultAppLinks = func() links.AppLinkBuilder { + appLinks, err := links.NewAppLinkBuilder("telesrv", "") + if err != nil { + panic(err) + } + return appLinks +}() + +func testAugmentAutoEntities(message string, entities []tg.MessageEntityClass) []tg.MessageEntityClass { + return augmentAutoEntities(message, entities, testDefaultAppLinks) +} + +func testAugmentAutoEntitiesWithAppLinks(t *testing.T, message string, entities []tg.MessageEntityClass, scheme, base string) []tg.MessageEntityClass { + t.Helper() + appLinks, err := links.NewAppLinkBuilder(scheme, base) + if err != nil { + t.Fatal(err) + } + return augmentAutoEntities(message, entities, appLinks) +} + func TestFirstPreviewableURL(t *testing.T) { t.Run("text-url-entity", func(t *testing.T) { got, ok := firstPreviewableURL("click here", []tg.MessageEntityClass{ @@ -74,6 +97,26 @@ func TestFirstPreviewableURL(t *testing.T) { t.Fatalf("ftp URL should be rejected") } }) + + t.Run("custom-scheme-entity-does-not-preview", func(t *testing.T) { + message := "telesrv://resolve?domain=Alice" + entities := testAugmentAutoEntities(message, nil) + if len(entities) != 1 { + t.Fatalf("entities = %d, want 1", len(entities)) + } + if got, ok := firstPreviewableURL(message, entities); ok { + t.Fatalf("custom app-link must not become a webpage preview, got %q", got) + } + }) + + t.Run("custom-scheme-before-http-still-previews-http", func(t *testing.T) { + message := "telesrv://resolve?domain=Alice then https://example.com/x" + entities := testAugmentAutoEntities(message, nil) + got, ok := firstPreviewableURL(message, entities) + if !ok || got != "https://example.com/x" { + t.Fatalf("got (%q,%v), want https://example.com/x", got, ok) + } + }) } func urlEntity(t *testing.T, e tg.MessageEntityClass) *tg.MessageEntityURL { @@ -89,7 +132,7 @@ func urlEntity(t *testing.T, e tg.MessageEntityClass) *tg.MessageEntityURL { // (高亮),UTF-16 偏移正确,多链接全检测,客户端已带 url 实体则不重复。 func TestAugmentAutoEntitiesURL(t *testing.T) { t.Run("detect-when-no-client-entities", func(t *testing.T) { - got := augmentAutoEntities("see https://example.com/x now", nil) + got := testAugmentAutoEntities("see https://example.com/x now", nil) if len(got) != 1 { t.Fatalf("entities = %d, want 1", len(got)) } @@ -99,7 +142,7 @@ func TestAugmentAutoEntitiesURL(t *testing.T) { } }) t.Run("utf16-offset-with-emoji", func(t *testing.T) { - got := augmentAutoEntities("\U0001f44d https://x", nil) // 👍=2 units, space=1 → url at 3 + got := testAugmentAutoEntities("\U0001f44d https://x", nil) // 👍=2 units, space=1 → url at 3 if len(got) != 1 { t.Fatalf("entities = %d, want 1", len(got)) } @@ -108,28 +151,64 @@ func TestAugmentAutoEntitiesURL(t *testing.T) { } }) t.Run("multiple-urls", func(t *testing.T) { - if got := augmentAutoEntities("https://a.com and https://b.com", nil); len(got) != 2 { + if got := testAugmentAutoEntities("https://a.com and https://b.com", nil); len(got) != 2 { t.Fatalf("entities = %d, want 2", len(got)) } }) t.Run("respect-client-url-entity", func(t *testing.T) { ents := []tg.MessageEntityClass{&tg.MessageEntityURL{Offset: 0, Length: 9}} - if got := augmentAutoEntities("https://x more https://y", ents); len(got) != 1 { - t.Fatalf("client url entity present → no server detection, got %d", len(got)) + if got := testAugmentAutoEntities("https://x more https://y", ents); len(got) != 2 { + t.Fatalf("server should fill the missing URL span, got %d", len(got)) } }) t.Run("trailing-punct-not-in-entity", func(t *testing.T) { - got := augmentAutoEntities("见 https://example.com。", nil) // 见 ...。 + got := testAugmentAutoEntities("见 https://example.com。", nil) // 见 ...。 e := urlEntity(t, got[0]) if e.Length != utf16CodeUnitLen("https://example.com") { t.Errorf("length = %d, want %d (trailing 。 excluded)", e.Length, utf16CodeUnitLen("https://example.com")) } }) t.Run("no-url-no-entities", func(t *testing.T) { - if got := augmentAutoEntities("plain text", nil); len(got) != 0 { + if got := testAugmentAutoEntities("plain text", nil); len(got) != 0 { t.Fatalf("entities = %d, want 0", len(got)) } }) + t.Run("configured-custom-scheme", func(t *testing.T) { + message := "👍 TELESRV://resolve?domain=Alice。" + got := testAugmentAutoEntities(message, nil) + if len(got) != 1 { + t.Fatalf("entities = %d, want 1", len(got)) + } + e := urlEntity(t, got[0]) + wantURL := "TELESRV://resolve?domain=Alice" + if e.Offset != 3 || e.Length != utf16CodeUnitLen(wantURL) { + t.Fatalf("offset/length = %d/%d, want 3/%d", e.Offset, e.Length, utf16CodeUnitLen(wantURL)) + } + }) + t.Run("mixed-client-http-and-missing-custom-scheme", func(t *testing.T) { + message := "https://x telesrv://resolve?domain=Alice" + client := []tg.MessageEntityClass{&tg.MessageEntityURL{Offset: 0, Length: utf16CodeUnitLen("https://x")}} + got := testAugmentAutoEntities(message, client) + if len(got) != 2 { + t.Fatalf("entities = %d, want client HTTP plus server app-link", len(got)) + } + e := urlEntity(t, got[1]) + if e.Offset != utf16CodeUnitLen("https://x ") || e.Length != utf16CodeUnitLen("telesrv://resolve?domain=Alice") { + t.Fatalf("custom entity offset/length = %d/%d", e.Offset, e.Length) + } + }) + t.Run("unconfigured-scheme-rejected", func(t *testing.T) { + if got := testAugmentAutoEntities("foo://resolve telesrv://resolve", nil); len(got) != 1 { + t.Fatalf("entities = %d, want only configured telesrv link", len(got)) + } + }) + t.Run("host-base-is-exact", func(t *testing.T) { + message := "owpg://links.example.test/Alice owpg://other.example.test/Bob telesrv://resolve?domain=Carol" + got := testAugmentAutoEntitiesWithAppLinks(t, message, nil, "telesrv", "owpg://links.example.test") + if len(got) != 2 { + t.Fatalf("entities = %d, want configured host-base and legacy link", len(got)) + } + }) } // BenchmarkAugmentAutoEntities 量化发送热路径:纯文本(无触发字符)应零分配走快路径短路; @@ -139,14 +218,14 @@ func BenchmarkAugmentAutoEntities(b *testing.B) { msg := "hey everyone, just wanted to share some thoughts about the meeting today and tomorrow" b.ReportAllocs() for i := 0; i < b.N; i++ { - _ = augmentAutoEntities(msg, nil) + _ = testAugmentAutoEntities(msg, nil) } }) b.Run("with-mention-hashtag-url", func(b *testing.B) { msg := "hi @alice please check #golang docs at https://example.com/x thanks" b.ReportAllocs() for i := 0; i < b.N; i++ { - _ = augmentAutoEntities(msg, nil) + _ = testAugmentAutoEntities(msg, nil) } }) } @@ -167,16 +246,16 @@ func mentionAt(t *testing.T, got []tg.MessageEntityClass, off, ln int) { func TestAugmentAutoEntitiesMention(t *testing.T) { t.Run("bare-mentions", func(t *testing.T) { // 对齐官方抓包:纯 "@G0ldenMods\n@NGame_Official" → 两个裸 messageEntityMention,含前导 @。 - got := augmentAutoEntities("@G0ldenMods\n@NGame_Official", nil) + got := testAugmentAutoEntities("@G0ldenMods\n@NGame_Official", nil) mentionAt(t, got, 0, 11) // @G0ldenMods = 10+1 mentionAt(t, got, 12, 15) // @NGame_Official = 14+1 }) t.Run("mention-mid-text", func(t *testing.T) { - got := augmentAutoEntities("hi @alice see you", nil) + got := testAugmentAutoEntities("hi @alice see you", nil) mentionAt(t, got, 3, 6) // @alice }) t.Run("email-not-a-mention", func(t *testing.T) { - for _, e := range augmentAutoEntities("mail me at bob@example.com please", nil) { + for _, e := range testAugmentAutoEntities("mail me at bob@example.com please", nil) { if _, ok := e.(*tg.MessageEntityMention); ok { t.Fatalf("email local@domain must not yield a mention: %#v", e) } @@ -184,13 +263,13 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { }) t.Run("utf16-offset-with-emoji", func(t *testing.T) { // 👍(2 units) + space(1) → @bob 起于 offset 3。 - got := augmentAutoEntities("\U0001f44d @bob", nil) + got := testAugmentAutoEntities("\U0001f44d @bob", nil) mentionAt(t, got, 3, 4) }) t.Run("no-overlap-with-client-entity", func(t *testing.T) { // 客户端把 "@bob" 区间标成 textUrl([offset 0,len 4)),服务端不得再补 mention。 ents := []tg.MessageEntityClass{&tg.MessageEntityTextURL{Offset: 0, Length: 4, URL: "https://x"}} - for _, e := range augmentAutoEntities("@bob", ents) { + for _, e := range testAugmentAutoEntities("@bob", ents) { if _, ok := e.(*tg.MessageEntityMention); ok { t.Fatalf("mention must not overlap client entity: %#v", e) } @@ -201,7 +280,7 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { // 裸 URL 路径里的 @scam / #frag 不得被误标成 mention/hashtag(钓鱼风险)。 msg := "Docs see https://t.me/@scam and #promo" ents := []tg.MessageEntityClass{&tg.MessageEntityTextURL{Offset: 0, Length: 4, URL: "https://x"}} - got := augmentAutoEntities(msg, ents) + got := testAugmentAutoEntities(msg, ents) for _, e := range got { if m, ok := e.(*tg.MessageEntityMention); ok { t.Fatalf("mention must not be synthesised inside a raw URL: offset=%d len=%d", m.Offset, m.Length) @@ -220,7 +299,7 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { }) t.Run("hashtag", func(t *testing.T) { var found bool - for _, e := range augmentAutoEntities("love #golang here", nil) { + for _, e := range testAugmentAutoEntities("love #golang here", nil) { if h, ok := e.(*tg.MessageEntityHashtag); ok && h.Offset == 5 && h.Length == 7 { found = true } @@ -230,7 +309,7 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { } }) t.Run("hashtag-all-digits-skipped", func(t *testing.T) { - for _, e := range augmentAutoEntities("number #123 here", nil) { + for _, e := range testAugmentAutoEntities("number #123 here", nil) { if _, ok := e.(*tg.MessageEntityHashtag); ok { t.Fatalf("leading-digit hashtag must be skipped: %#v", e) } @@ -238,7 +317,7 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { }) t.Run("bot-command", func(t *testing.T) { var found bool - for _, e := range augmentAutoEntities("/start@MyBot now", nil) { + for _, e := range testAugmentAutoEntities("/start@MyBot now", nil) { if c, ok := e.(*tg.MessageEntityBotCommand); ok && c.Offset == 0 && c.Length == 12 { found = true } @@ -248,7 +327,7 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { } }) t.Run("slash-in-path-not-command", func(t *testing.T) { - for _, e := range augmentAutoEntities("see and/or maybe", nil) { + for _, e := range testAugmentAutoEntities("see and/or maybe", nil) { if _, ok := e.(*tg.MessageEntityBotCommand); ok { t.Fatalf("and/or must not be a bot command: %#v", e) } @@ -256,7 +335,7 @@ func TestAugmentAutoEntitiesMention(t *testing.T) { }) t.Run("cashtag", func(t *testing.T) { var found bool - for _, e := range augmentAutoEntities("buy $USD now", nil) { + for _, e := range testAugmentAutoEntities("buy $USD now", nil) { if c, ok := e.(*tg.MessageEntityCashtag); ok && c.Offset == 4 && c.Length == 4 { found = true }