Allow tg URLs in text links and url buttons.

GitOrigin-RevId: 9708c3016375564e737bf5a8f79494807d61c10f

td/generate/scheme/td_api.tl

@@ -582,7 +582,7 @@ keyboardButton text:string type:KeyboardButtonType = KeyboardButton;
 
 //@class InlineKeyboardButtonType @description Describes the type of an inline keyboard button
 
-//@description A button that opens a specified URL @url URL to open
+//@description A button that opens a specified URL @url HTTP or tg:// URL to open
 inlineKeyboardButtonTypeUrl url:string = InlineKeyboardButtonType;
 
 //@description A button that sends a special callback query to a bot @data Data to be sent to the bot via a callback query
@@ -1179,7 +1179,7 @@ textEntityTypePre = TextEntityType;
 //@description Text that must be formatted as if inside pre, and code HTML tags @language Programming language of the code; as defined by the sender
 textEntityTypePreCode language:string = TextEntityType;
 
-//@description A text description shown instead of a raw URL @url URL to be opened when the link is clicked
+//@description A text description shown instead of a raw URL @url HTTP or tg:// URL to be opened when the link is clicked
 textEntityTypeTextUrl url:string = TextEntityType;
 
 //@description A text shows instead of a raw mention of the user (e.g., when the user has no username) @user_id Identifier of the mentioned user

td/telegram/MessageEntity.cpp

@@ -9,7 +9,6 @@
 #include "td/telegram/ContactsManager.h"
 #include "td/telegram/misc.h"
 
-#include "td/utils/HttpUrl.h"
 #include "td/utils/logging.h"
 #include "td/utils/misc.h"
 #include "td/utils/unicode.h"
@@ -1329,10 +1328,10 @@ Result<vector<MessageEntity>> parse_markdown(string &text) {
           if (user_id.is_valid()) {
             entities.emplace_back(utf16_offset, utf16_entity_length, user_id);
           } else {
-            auto r_http_url = parse_url(url);
+            auto r_url = check_url(url);
-            if (r_http_url.is_ok() && url.find('.') != string::npos) {
+            if (r_url.is_ok()) {
               entities.emplace_back(MessageEntity::Type::TextUrl, utf16_offset, utf16_entity_length,
-                                    r_http_url.ok().get_url());
+                                    r_url.move_as_ok());
             }
           }
           break;
@@ -1585,10 +1584,9 @@ Result<vector<MessageEntity>> parse_html(string &text) {
         if (user_id.is_valid()) {
           entities.emplace_back(utf16_offset, utf16_entity_length, user_id);
         } else {
-          auto r_http_url = parse_url(url);
+          auto r_url = check_url(url);
-          if (r_http_url.is_ok() && url.find('.') != string::npos) {
+          if (r_url.is_ok()) {
-            entities.emplace_back(MessageEntity::Type::TextUrl, utf16_offset, utf16_entity_length,
+            entities.emplace_back(MessageEntity::Type::TextUrl, utf16_offset, utf16_entity_length, r_url.move_as_ok());
-                                  r_http_url.ok().get_url());
           }
         }
       } else if (tag_name == "pre") {
@@ -1746,12 +1744,11 @@ Result<vector<MessageEntity>> get_message_entities(const ContactsManager *contac
         if (!clean_input_string(entity_text_url->url_)) {
           return Status::Error(400, "MessageEntityTextUrl.url must be encoded in UTF-8");
         }
-        auto r_http_url = parse_url(entity_text_url->url_);
+        auto r_url = check_url(entity_text_url->url_);
-        if (r_http_url.is_error()) {
+        if (r_url.is_error()) {
-          return Status::Error(400, PSTRING() << "Wrong message entity: " << r_http_url.error().message());
+          return Status::Error(400, PSTRING() << "Wrong message entity: " << r_url.error().message());
         }
-        entities.emplace_back(MessageEntity::Type::TextUrl, entity->offset_, entity->length_,
+        entities.emplace_back(MessageEntity::Type::TextUrl, entity->offset_, entity->length_, r_url.move_as_ok());
-                              r_http_url.ok().get_url());
         break;
       }
       case td_api::textEntityTypeMentionName::ID: {
@@ -1843,14 +1840,14 @@ vector<MessageEntity> get_message_entities(const ContactsManager *contacts_manag
       case telegram_api::messageEntityTextUrl::ID: {
         // TODO const telegram_api::messageEntityTextUrl *
         auto entity_text_url = static_cast<telegram_api::messageEntityTextUrl *>(entity.get());
-        auto r_http_url = parse_url(entity_text_url->url_);
+        auto r_url = check_url(entity_text_url->url_);
-        if (r_http_url.is_error()) {
+        if (r_url.is_error()) {
-          LOG(ERROR) << "Wrong URL entity: \"" << entity_text_url->url_ << "\": " << r_http_url.error().message()
+          LOG(ERROR) << "Wrong URL entity: \"" << entity_text_url->url_ << "\": " << r_url.error().message() << " from "
-                     << " from " << source;
+                     << source;
           continue;
         }
         entities.emplace_back(MessageEntity::Type::TextUrl, entity_text_url->offset_, entity_text_url->length_,
-                              r_http_url.ok().get_url());
+                              r_url.move_as_ok());
         break;
       }
       case telegram_api::messageEntityMentionName::ID: {
@@ -1947,13 +1944,13 @@ vector<MessageEntity> get_message_entities(vector<tl_object_ptr<secret_api::Mess
           LOG(WARNING) << "Wrong URL entity: \"" << entity_text_url->url_ << '"';
           continue;
         }
-        auto r_http_url = parse_url(entity_text_url->url_);
+        auto r_url = check_url(entity_text_url->url_);
-        if (r_http_url.is_error()) {
+        if (r_url.is_error()) {
-          LOG(WARNING) << "Wrong URL entity: \"" << entity_text_url->url_ << "\": " << r_http_url.error().message();
+          LOG(WARNING) << "Wrong URL entity: \"" << entity_text_url->url_ << "\": " << r_url.error().message();
           continue;
         }
         entities.emplace_back(MessageEntity::Type::TextUrl, entity_text_url->offset_, entity_text_url->length_,
-                              r_http_url.ok().get_url());
+                              r_url.move_as_ok());
         break;
       }
       case secret_api::messageEntityMentionName::ID:

td/telegram/ReplyMarkup.cpp

@@ -13,7 +13,6 @@
 
 #include "td/utils/buffer.h"
 #include "td/utils/format.h"
-#include "td/utils/HttpUrl.h"
 #include "td/utils/logging.h"
 
 namespace td {
@@ -363,8 +362,8 @@ static Result<InlineKeyboardButton> get_inline_keyboard_button(tl_object_ptr<td_
   switch (button_type_id) {
     case td_api::inlineKeyboardButtonTypeUrl::ID: {
       current_button.type = InlineKeyboardButton::Type::Url;
-      TRY_RESULT(http_url, parse_url(static_cast<td_api::inlineKeyboardButtonTypeUrl *>(button->type_.get())->url_));
+      TRY_RESULT(url, check_url(static_cast<td_api::inlineKeyboardButtonTypeUrl *>(button->type_.get())->url_));
-      current_button.data = http_url.get_url();
+      current_button.data = std::move(url);
       if (!clean_input_string(current_button.data)) {
         return Status::Error(400, "Inline keyboard button url must be encoded in UTF-8");
       }

td/telegram/misc.cpp

@@ -7,6 +7,7 @@
 #include "td/telegram/misc.h"
 
 #include "td/utils/common.h"
+#include "td/utils/HttpUrl.h"
 #include "td/utils/logging.h"
 #include "td/utils/misc.h"
 #include "td/utils/Slice.h"
@@ -274,4 +275,32 @@ string get_emoji_fingerprint(uint64 num) {
   return emojis[(num & 0x7FFFFFFFFFFFFFFF) % emojis.size()].str();
 }
 
+Result<string> check_url(MutableSlice url) {
+  bool is_tg = false;
+  if (begins_with(url, "tg://")) {
+    url.remove_prefix(5);
+    is_tg = true;
+  } else if (begins_with(url, "tg:")) {
+    url.remove_prefix(3);
+    is_tg = true;
+  } else {
+    is_tg = false;
+  }
+  TRY_RESULT(http_url, parse_url(url));
+  if (is_tg) {
+    if (begins_with(url, "http://") || http_url.protocol_ == HttpUrl::Protocol::HTTPS || !http_url.userinfo_.empty() || http_url.specified_port_ != 0 || http_url.is_ipv6_) {
+      return Status::Error("Wrong tg URL");
+    }
+
+    auto result = http_url.get_url();
+    CHECK(begins_with(result, "http://"));
+    return PSTRING() << "tg" << Slice(result).substr(4);
+  }
+
+  if (url.find('.') == string::npos) {
+    return Status::Error("Wrong HTTP URL");
+  }
+  return http_url.get_url();
+}
+
 }  // namespace td

td/telegram/misc.h

@@ -7,6 +7,8 @@
 #pragma once
 
 #include "td/utils/common.h"
+#include "td/utils/Slice.h"
+#include "td/utils/Status.h"
 
 namespace td {
 
@@ -31,4 +33,7 @@ int32 get_vector_hash(const vector<uint32> &numbers) TD_WARN_UNUSED_RESULT;
 // returns emoji corresponding to the specified number
 string get_emoji_fingerprint(uint64 num);
 
+// checks whether url is a valid tg or HTTP(S) URL and returns its in a canonical form
+Result<string> check_url(MutableSlice url);
+
 }  // namespace td

tdutils/td/utils/HttpUrl.cpp

@@ -29,11 +29,11 @@ string HttpUrl::get_url() const {
     result += userinfo_;
     result += '@';
   }
-  if (is_ipv6) {
+  if (is_ipv6_) {
     result += '[';
   }
   result += host_;
-  if (is_ipv6) {
+  if (is_ipv6_) {
     result += ']';
   }
   if (specified_port_ > 0) {

tdutils/td/utils/HttpUrl.h

@@ -18,7 +18,7 @@ class HttpUrl {
   enum class Protocol { HTTP, HTTPS } protocol_;
   string userinfo_;
   string host_;
-  bool is_ipv6;
+  bool is_ipv6_;
   int specified_port_;
   int port_;
   string query_;