From ebcc1d0dc3acc146a759d9a6490f03600b83506c Mon Sep 17 00:00:00 2001
From: levlam <levlam@telegram.org>
Date: Fri, 4 Oct 2019 18:00:51 +0300
Subject: [PATCH] Strip empty characters in file names.

GitOrigin-RevId: 896c899c9d0893b467844b5640d8f65b240bc759
---
 td/telegram/DocumentsManager.cpp |  3 +++
 td/telegram/misc.cpp             | 11 +++++++----
 td/telegram/misc.h               |  2 +-
 test/string_cleaning.cpp         |  9 +++++++--
 4 files changed, 18 insertions(+), 7 deletions(-)

diff --git a/td/telegram/DocumentsManager.cpp b/td/telegram/DocumentsManager.cpp
index ee7c8d8d..51ee96b1 100644
--- a/td/telegram/DocumentsManager.cpp
+++ b/td/telegram/DocumentsManager.cpp
@@ -13,6 +13,7 @@
 #include "td/telegram/files/FileLocation.h"
 #include "td/telegram/files/FileManager.h"
 #include "td/telegram/files/FileType.h"
+#include "td/telegram/misc.h"
 #include "td/telegram/net/DcId.h"
 #include "td/telegram/Photo.h"
 #include "td/telegram/StickersManager.h"
@@ -337,6 +338,8 @@ Document DocumentsManager::on_get_document(RemoteDocument remote_document, Dialo
     return {};
   }
 
+  file_name = strip_empty_characters(file_name, 255, true);
+
   auto suggested_file_name = file_name;
   if (suggested_file_name.empty()) {
     suggested_file_name = to_string(static_cast<uint64>(id));
diff --git a/td/telegram/misc.cpp b/td/telegram/misc.cpp
index 5f5bbbd7..9d619d5d 100644
--- a/td/telegram/misc.cpp
+++ b/td/telegram/misc.cpp
@@ -136,11 +136,11 @@ bool clean_input_string(string &str) {
   return true;
 }
 
-string strip_empty_characters(string str, size_t max_length) {
+string strip_empty_characters(string str, size_t max_length, bool strip_rtlo) {
   static const char *space_characters[] = {u8"\u1680", u8"\u180E", u8"\u2000", u8"\u2001", u8"\u2002",
                                            u8"\u2003", u8"\u2004", u8"\u2005", u8"\u2006", u8"\u2007",
-                                           u8"\u2008", u8"\u2009", u8"\u200A", u8"\u200B", u8"\u202F",
-                                           u8"\u205F", u8"\u3000", u8"\uFEFF", u8"\uFFFC"};
+                                           u8"\u2008", u8"\u2009", u8"\u200A", u8"\u200B", u8"\u202E",
+                                           u8"\u202F", u8"\u205F", u8"\u3000", u8"\uFEFF", u8"\uFFFC"};
   static bool can_be_first[std::numeric_limits<unsigned char>::max() + 1];
   static bool can_be_first_inited = [&] {
     for (auto space_ch : space_characters) {
@@ -162,7 +162,10 @@ string strip_empty_characters(string str, size_t max_length) {
       bool found = false;
       for (auto space_ch : space_characters) {
         if (space_ch[0] == str[i] && space_ch[1] == str[i + 1] && space_ch[2] == str[i + 2]) {
-          found = true;
+          if (static_cast<unsigned char>(str[i + 2]) != 0xAE || static_cast<unsigned char>(str[i + 1]) != 0x80 ||
+              static_cast<unsigned char>(str[i]) != 0xE2 || strip_rtlo) {
+            found = true;
+          }
           break;
         }
       }
diff --git a/td/telegram/misc.h b/td/telegram/misc.h
index f25d5ea4..62804318 100644
--- a/td/telegram/misc.h
+++ b/td/telegram/misc.h
@@ -22,7 +22,7 @@ string clean_username(string str) TD_WARN_UNUSED_RESULT;
 bool clean_input_string(string &str) TD_WARN_UNUSED_RESULT;
 
 // strips empty characters and ensures that string length is no more than max_length
-string strip_empty_characters(string str, size_t max_length) TD_WARN_UNUSED_RESULT;
+string strip_empty_characters(string str, size_t max_length, bool strip_rtlo = false) TD_WARN_UNUSED_RESULT;
 
 // checks if string is empty after strip_empty_characters
 bool is_empty_string(const string &str) TD_WARN_UNUSED_RESULT;
diff --git a/test/string_cleaning.cpp b/test/string_cleaning.cpp
index 160b7b03..b35cee1c 100644
--- a/test/string_cleaning.cpp
+++ b/test/string_cleaning.cpp
@@ -69,8 +69,8 @@ TEST(StringCleaning, clean_input_string) {
   check_clean_input_string("\xcc\xb3\xcc\xbf\xcc\x8a", "", true);
 }
 
-static void check_strip_empty_characters(string str, size_t max_length, string expected) {
-  ASSERT_EQ(expected, strip_empty_characters(str, max_length));
+static void check_strip_empty_characters(string str, size_t max_length, string expected, bool strip_rtlo = false) {
+  ASSERT_EQ(expected, strip_empty_characters(str, max_length, strip_rtlo));
 }
 
 TEST(StringCleaning, strip_empty_characters) {
@@ -82,9 +82,14 @@ TEST(StringCleaning, strip_empty_characters) {
       u8"\u1680\u180E\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u200B\u202F\u205F\u3000\uFEFF"
       u8"\uFFFC\uFFFC";
   string spaces_replace = "                    ";
+  string rtlo = u8"\u202E";
   string empty = "\xE2\x80\x8C\xE2\x80\x8D\xE2\x80\xAE\xC2\xA0\xC2\xA0";
 
   check_strip_empty_characters(spaces, 1000000, "");
+  check_strip_empty_characters(spaces + rtlo, 1000000, "");
+  check_strip_empty_characters(spaces + rtlo, 1000000, "", true);
+  check_strip_empty_characters(spaces + rtlo + "a", 1000000, rtlo + "a");
+  check_strip_empty_characters(spaces + rtlo + "a", 1000000, "a", true);
   check_strip_empty_characters(empty, 1000000, "");
   check_strip_empty_characters(empty + "a", 1000000, empty + "a");
   check_strip_empty_characters(spaces + empty + spaces + "abc" + spaces, 1000000, empty + spaces_replace + "abc");