From 6dbef6ef768d54a984388d304c6030ea891973ce Mon Sep 17 00:00:00 2001
From: levlam <levlam@telegram.org>
Date: Tue, 9 Oct 2018 16:27:40 +0300
Subject: [PATCH] Fix returning invalid Unicode from parse_html.

GitOrigin-RevId: 9277022a53d23ac5c3eb1427eb48bd886b789d09
---
 td/telegram/MessageEntity.cpp | 6 ++++++
 td/telegram/cli.cpp           | 4 ++++
 2 files changed, 10 insertions(+)

diff --git a/td/telegram/MessageEntity.cpp b/td/telegram/MessageEntity.cpp
index 3324400c8..3bf24fb9d 100644
--- a/td/telegram/MessageEntity.cpp
+++ b/td/telegram/MessageEntity.cpp
@@ -1616,6 +1616,12 @@ Result<vector<MessageEntity>> parse_html(string &text) {
       utf16_offset += utf16_entity_length;
     }
   }
+  if (!check_utf8(result)) {
+    return Status::Error(400,
+                         "Text contains invalid Unicode characters after decoding HTML entities, check for unmatched "
+                         "surrogate code units");
+  }
+
   text = result;
   return entities;
 }
diff --git a/td/telegram/cli.cpp b/td/telegram/cli.cpp
index f17b10456..8bd4bbd06 100644
--- a/td/telegram/cli.cpp
+++ b/td/telegram/cli.cpp
@@ -2365,8 +2365,12 @@ class CliClient final : public Actor {
       execute(make_tl_object<td_api::getTextEntities>(args));
     } else if (op == "pte") {
       send_request(make_tl_object<td_api::parseTextEntities>(args, make_tl_object<td_api::textParseModeMarkdown>()));
+    } else if (op == "pteh") {
+      send_request(make_tl_object<td_api::parseTextEntities>(args, make_tl_object<td_api::textParseModeHTML>()));
     } else if (op == "ptes") {
       execute(make_tl_object<td_api::parseTextEntities>(args, make_tl_object<td_api::textParseModeMarkdown>()));
+    } else if (op == "ptehs") {
+      execute(make_tl_object<td_api::parseTextEntities>(args, make_tl_object<td_api::textParseModeHTML>()));
     } else if (op == "gfmt") {
       send_request(make_tl_object<td_api::getFileMimeType>(trim(args)));
     } else if (op == "gfe") {