Support PreCode entities in getMarkdownText.

td/telegram/MessageEntity.cpp

@@ -2924,6 +2924,14 @@ FormattedText get_markdown_v3(FormattedText text) {
   int32 utf16_offset = 0;
   int32 utf16_added = 0;
 
+  auto is_valid_language_code = [](Slice code) {
+    for (auto c : code) {
+      if (c < 33 || c > 126) {
+        return false;
+      }
+    }
+    return true;
+  };
   for (size_t pos = 0; pos <= text.text.size(); pos++) {
     auto c = static_cast<unsigned char>(text.text[pos]);
     if (is_utf8_character_first_code_unit(c)) {
@@ -2936,6 +2944,7 @@ FormattedText get_markdown_v3(FormattedText text) {
 
         CHECK(utf16_offset == entity_end);
 
+        bool need_entity = false;
         switch (entity->type) {
           case MessageEntity::Type::Italic:
             result.text += "__";
@@ -2967,11 +2976,22 @@ FormattedText get_markdown_v3(FormattedText text) {
             result.text += "```";
             utf16_added += 3;
             break;
+          case MessageEntity::Type::PreCode:
+            if (is_valid_language_code(entity->argument)) {
+              result.text += "```";
+              utf16_added += 3;
+            } else {
+              need_entity = true;
+            }
+            break;
           default:
+            need_entity = true;
+            break;
+        }
+        if (need_entity) {
           result.entities.push_back(*entity);
           result.entities.back().offset += nested_entities_stack.back().utf16_added_before;
           result.entities.back().length += utf16_added - nested_entities_stack.back().utf16_added_before;
-            break;
         }
         nested_entities_stack.pop_back();
       }
@@ -3006,6 +3026,24 @@ FormattedText get_markdown_v3(FormattedText text) {
           case MessageEntity::Type::Pre:
             result.text += "```";
             utf16_added += 3;
+            if (c != '\n') {
+              result.text += "\n";
+              utf16_added++;
+            }
+            break;
+          case MessageEntity::Type::PreCode:
+            if (is_valid_language_code(text.entities[current_entity].argument)) {
+              result.text += "```";
+              utf16_added += 3;
+              if (!text.entities[current_entity].argument.empty()) {
+                result.text += text.entities[current_entity].argument;
+                utf16_added += text.entities[current_entity].argument.size();
+              }
+              if (c != '\n') {
+                result.text += "\n";
+                utf16_added++;
+              }
+            }
             break;
           default:
             // keep as is

test/message_entities.cpp

@@ -1901,7 +1901,7 @@ static void check_get_markdown_v3(const td::string &result_text, const td::vecto
 }
 
 TEST(MessageEntities, get_markdown_v3) {
-  check_get_markdown_v3("``` ```", {}, " ", {{td::MessageEntity::Type::Pre, 0, 1}});
+  check_get_markdown_v3("```\n ```", {}, " ", {{td::MessageEntity::Type::Pre, 0, 1}});
   check_get_markdown_v3("` `", {}, " ", {{td::MessageEntity::Type::Code, 0, 1}});
   check_get_markdown_v3("`\n`", {}, "\n", {{td::MessageEntity::Type::Code, 0, 1}});
   check_get_markdown_v3("ab", {{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Pre, 1, 1}}, "ab",
@@ -1922,10 +1922,12 @@ TEST(MessageEntities, get_markdown_v3) {
                          {td::MessageEntity::Type::Strikethrough, 4, 1},
                          {td::MessageEntity::Type::Spoiler, 6, 1},
                          {td::MessageEntity::Type::PreCode, 8, 1, " C++"}});
-  check_get_markdown_v3("`ab` ```cd``` ef", {{td::MessageEntity::Type::PreCode, 14, 2, "C++"}}, "ab cd ef",
+  check_get_markdown_v3("```cpp\ngh```\n`ab`\n```\ncd```\nef", {{td::MessageEntity::Type::PreCode, 28, 2, " C++"}},
-                        {{td::MessageEntity::Type::Code, 0, 2},
+                        "gh\nab\ncd\nef",
-                         {td::MessageEntity::Type::Pre, 3, 2},
+                        {{td::MessageEntity::Type::PreCode, 0, 2, "cpp"},
-                         {td::MessageEntity::Type::PreCode, 6, 2, "C++"}});
+                         {td::MessageEntity::Type::Code, 3, 2},
+                         {td::MessageEntity::Type::Pre, 6, 2},
+                         {td::MessageEntity::Type::PreCode, 9, 2, " C++"}});
   check_get_markdown_v3("__asd__[__ab__cd](http://t.me/)", {}, "asdabcd",
                         {{td::MessageEntity::Type::Italic, 0, 3},
                          {td::MessageEntity::Type::TextUrl, 3, 4, "http://t.me/"},
@@ -1944,4 +1946,19 @@ TEST(MessageEntities, get_markdown_v3) {
                         {{td::MessageEntity::Type::TextUrl, 0, 16, "http://example.com/"},
                          {td::MessageEntity::Type::Bold, 0, 16},
                          {td::MessageEntity::Type::Italic, 0, 16}});
+  check_get_markdown_v3("```\nsome code\n```", {}, "some code\n", {{td::MessageEntity::Type::Pre, 0, 10}});
+  check_get_markdown_v3("asd\n```\nsome code\n```cabab", {}, "asd\nsome code\ncabab",
+                        {{td::MessageEntity::Type::Pre, 4, 10}});
+  check_get_markdown_v3("asd\naba```\nsome code\n```cabab", {}, "asd\nabasome code\ncabab",
+                        {{td::MessageEntity::Type::Pre, 7, 10}});
+  check_get_markdown_v3("asd\naba```\nsome code\n```\ncabab", {}, "asd\nabasome code\n\ncabab",
+                        {{td::MessageEntity::Type::Pre, 7, 10}});
+  check_get_markdown_v3("asd\naba```\na b\nsome code\n```\ncabab", {}, "asd\nabaa b\nsome code\n\ncabab",
+                        {{td::MessageEntity::Type::Pre, 7, 14}});
+  check_get_markdown_v3("asd\n```\na b\nsome code\n```\ncabab", {}, "asd\na b\nsome code\n\ncabab",
+                        {{td::MessageEntity::Type::Pre, 4, 14}});
+  check_get_markdown_v3("asd\naba```a!@#$%^&*(b\nsome code\n```\ncabab", {}, "asd\nabasome code\n\ncabab",
+                        {{td::MessageEntity::Type::PreCode, 7, 10, "a!@#$%^&*(b"}});
+  check_get_markdown_v3("```\naba\n```", {}, "aba\n", {{td::MessageEntity::Type::Pre, 0, 4}});
+  check_get_markdown_v3("```\n```", {}, "\n", {{td::MessageEntity::Type::Pre, 0, 1}});
 }