Add and use utf8_encode.

td/telegram/SecureValue.cpp

@@ -713,7 +713,7 @@ static Result<int32> to_int32(Slice str) {
   int32 integer_value = 0;
   for (auto c : str) {
     if (!is_digit(c)) {
-      return Status::Error(400, PSLICE() << "Can't parse \"" << str << "\" as number");
+      return Status::Error(400, PSLICE() << "Can't parse \"" << utf8_encode(str.str()) << "\" as number");
     }
     integer_value = integer_value * 10 + c - '0';
   }
@@ -725,12 +725,12 @@ static Result<td_api::object_ptr<td_api::date>> get_date_object(Slice date) {
     return nullptr;
   }
   if (date.size() > 10u || date.size() < 8u) {
-    return Status::Error(400, PSLICE() << "Date \"" << date << "\" has wrong length");
+    return Status::Error(400, PSLICE() << "Date \"" << utf8_encode(date.str()) << "\" has wrong length");
   }
   auto parts = full_split(date, '.');
   if (parts.size() != 3 || parts[0].size() > 2 || parts[1].size() > 2 || parts[2].size() != 4 || parts[0].empty() ||
       parts[1].empty()) {
-    return Status::Error(400, PSLICE() << "Date \"" << date << "\" has wrong parts");
+    return Status::Error(400, PSLICE() << "Date \"" << utf8_encode(date.str()) << "\" has wrong parts");
   }
   TRY_RESULT(day, to_int32(parts[0]));
   TRY_RESULT(month, to_int32(parts[1]));

td/telegram/files/FileManager.cpp

@@ -42,6 +42,7 @@
 #include "td/utils/Time.h"
 #include "td/utils/tl_helpers.h"
 #include "td/utils/tl_parsers.h"
+#include "td/utils/utf8.h"
 
 #include <algorithm>
 #include <cmath>
@@ -989,27 +990,28 @@ Status FileManager::check_local_location(FullLocalFileLocation &location, int64
   } else if (!are_modification_times_equal(location.mtime_nsec_, stat.mtime_nsec_)) {
     VLOG(file_loader) << "File \"" << location.path_ << "\" was modified: old mtime = " << location.mtime_nsec_
                       << ", new mtime = " << stat.mtime_nsec_;
-    return Status::Error(400, PSLICE() << "File \"" << location.path_ << "\" was modified");
+    return Status::Error(400, PSLICE() << "File \"" << utf8_encode(location.path_) << "\" was modified");
   }
   if (skip_file_size_checks) {
     return Status::OK();
   }
   if ((location.file_type_ == FileType::Thumbnail || location.file_type_ == FileType::EncryptedThumbnail) &&
       size > MAX_THUMBNAIL_SIZE && !begins_with(PathView(location.path_).file_name(), "map")) {
-    return Status::Error(400, PSLICE() << "File \"" << location.path_ << "\" is too big for a thumbnail "
-                                       << tag("size", format::as_size(size)));
+    return Status::Error(400, PSLICE() << "File \"" << utf8_encode(location.path_) << "\" of size " << size
+                                       << " is too big for a thumbnail");
   }
   if (size > MAX_FILE_SIZE) {
-    return Status::Error(400, PSLICE() << "File \"" << location.path_ << "\" of size " << size << " bytes is too big");
+    return Status::Error(
+        400, PSLICE() << "File \"" << utf8_encode(location.path_) << "\" of size " << size << " bytes is too big");
   }
   if (location.file_type_ == FileType::Photo && size > MAX_PHOTO_SIZE) {
-    return Status::Error(
-        400, PSLICE() << "File \"" << location.path_ << "\" of size " << size << " bytes is too big for a photo");
+    return Status::Error(400, PSLICE() << "File \"" << utf8_encode(location.path_) << "\" of size " << size
+                                       << " bytes is too big for a photo");
   }
   if (location.file_type_ == FileType::VideoNote &&
       size > G()->shared_config().get_option_integer("video_note_size_max", DEFAULT_VIDEO_NOTE_SIZE_MAX)) {
-    return Status::Error(
-        400, PSLICE() << "File \"" << location.path_ << "\" of size " << size << " bytes is too big for a video note");
+    return Status::Error(400, PSLICE() << "File \"" << utf8_encode(location.path_) << "\" of size " << size
+                                       << " bytes is too big for a video note");
   }
   return Status::OK();
 }

tdutils/td/utils/utf8.cpp

@@ -7,6 +7,8 @@
 #include "td/utils/utf8.h"
 
 #include "td/utils/logging.h"
+#include "td/utils/misc.h"
+#include "td/utils/SliceBuilder.h"
 #include "td/utils/unicode.h"
 
 namespace td {
@@ -121,4 +123,11 @@ string utf8_to_lower(Slice str) {
   return result;
 }
 
+string utf8_encode(CSlice data) {
+  if (check_utf8(data)) {
+    return data.str();
+  }
+  return PSTRING() << "url_decode(" << url_encode(data) << ')';
+}
+
 }  // namespace td

tdutils/td/utils/utf8.h

@@ -118,4 +118,7 @@ T utf8_utf16_substr(T str, size_t offset, size_t length) {
 /// Returns UTF-8 string converted to lower case.
 string utf8_to_lower(Slice str);
 
+/// Returns valid UTF-8 representation of the string.
+string utf8_encode(CSlice data);
+
 }  // namespace td