Fix resolving of Internationalized Domain Names (on Windows only).

GitOrigin-RevId: 10b81d8b00a4cb6bb6c06e4b66831461ef0cc286

tdutils/CMakeLists.txt

@@ -215,7 +215,7 @@ if (WIN32)
   # find_library(WS2_32_LIBRARY ws2_32)
   # find_library(MSWSOCK_LIBRARY Mswsock)
   # target_link_libraries(tdutils PRIVATE ${WS2_32_LIBRARY} ${MSWSOCK_LIBRARY})
-  target_link_libraries(tdutils PRIVATE ws2_32 Mswsock)
+  target_link_libraries(tdutils PRIVATE ws2_32 Mswsock Normaliz)
 endif()
 if (NOT CMAKE_CROSSCOMPILING)
   add_dependencies(tdutils tdmime_auto)

tdutils/td/utils/port/IPAddress.cpp

@@ -12,8 +12,11 @@
 #include "td/utils/port/SocketFd.h"
 #include "td/utils/port/thread_local.h"
 #include "td/utils/ScopeGuard.h"
+#include "td/utils/utf8.h"
 
-#if !TD_WINDOWS
+#if TD_WINDOWS
+#include "td/utils/port/wstring_convert.h"
+#else
 #include <netdb.h>
 #include <netinet/in.h>
 #include <netinet/tcp.h>
@@ -25,6 +28,44 @@
 
 namespace td {
 
+static bool is_ascii_host_char(char c) {
+  return is_alnum(c) || c == '-';
+}
+
+static bool is_ascii_host(Slice host) {
+  for (auto c : host) {
+    // ':' and '@' are not allowed in a host name anyway, so we can skip them
+    if (!is_ascii_host_char(c) && c != '.' && c != ':' && c != '@') {
+      return false;
+    }
+  }
+  return true;
+}
+
+Result<string> idn_to_ascii(CSlice host) {
+  if (is_ascii_host(host)) {
+    return to_lower(host);
+  }
+  if (!check_utf8(host)) {
+    return Status::Error("Host name must be encoded in UTF-8");
+  }
+
+#if TD_WINDOWS
+  TRY_RESULT(whost, to_wstring(host));
+  wchar_t punycode[256];
+  int result_length = IdnToAscii(IDN_ALLOW_UNASSIGNED, whost.c_str(), whost.size(), punycode, 255);
+  if (result_length == 0) {
+    return Status::Error("Host can't be punycoded");
+  }
+
+  TRY_RESULT(idn_host, from_wstring(punycode, result_length));
+  return idn_host;
+#else
+  // TODO
+  return Status::Error("Internationalized Domain Names are not supported");
+#endif
+}
+
 IPAddress::IPAddress() : is_valid_(false) {
 }
 
@@ -84,12 +125,14 @@ IPAddress IPAddress::get_any_addr() const {
   }
   return res;
 }
+
 void IPAddress::init_ipv4_any() {
   is_valid_ = true;
   ipv4_addr_.sin_family = AF_INET;
   ipv4_addr_.sin_addr.s_addr = INADDR_ANY;
   ipv4_addr_.sin_port = 0;
 }
+
 void IPAddress::init_ipv6_any() {
   is_valid_ = true;
   ipv6_addr_.sin6_family = AF_INET6;
@@ -151,18 +194,21 @@ Status IPAddress::init_host_port(CSlice host, CSlice port) {
     return Status::Error("Host is invalid");
   }
 #endif
+  TRY_RESULT(ascii_host, idn_to_ascii(host));
+  host = ascii_host;
+
   addrinfo hints;
   addrinfo *info = nullptr;
   std::memset(&hints, 0, sizeof(hints));
   hints.ai_family = AF_INET;  // TODO AF_UNSPEC;
   hints.ai_socktype = SOCK_STREAM;
   LOG(INFO) << "Try to init IP address of " << host << " with port " << port;
-  auto s = getaddrinfo(host.c_str(), port.c_str(), &hints, &info);
-  if (s != 0) {
+  auto err = getaddrinfo(host.c_str(), port.c_str(), &hints, &info);
+  if (err != 0) {
 #if TD_WINDOWS
     return OS_SOCKET_ERROR("Failed to resolve host");
 #else
-    return Status::Error(PSLICE() << "Failed to resolve host: " << gai_strerror(s));
+    return Status::Error(PSLICE() << "Failed to resolve host: " << gai_strerror(err));
 #endif
   }
   SCOPE_EXIT {

tdutils/td/utils/port/IPAddress.h

@@ -19,7 +19,11 @@
 #endif
 
 namespace td {
+
+Result<string> idn_to_ascii(CSlice host);
+
 class SocketFd;
+
 class IPAddress {
  public:
   IPAddress();

tdutils/test/misc.cpp

@@ -10,6 +10,7 @@
 #include "td/utils/misc.h"
 #include "td/utils/port/EventFd.h"
 #include "td/utils/port/FileFd.h"
+#include "td/utils/port/IPAddress.h"
 #include "td/utils/port/path.h"
 #include "td/utils/port/sleep.h"
 #include "td/utils/port/Stat.h"
@@ -260,3 +261,40 @@ TEST(Misc, get_url_query_file_name) {
     test_get_url_query_file_name_one("/", suffix, "\\a\\1\\2\\3\\a\\s\\a\\das");
   }
 }
+
+static void test_idn_to_ascii_one(string host, string result) {
+  if (result != idn_to_ascii(host).ok()) {
+    LOG(ERROR) << "Failed to convert " << host << " to " << result << ", got \"" << idn_to_ascii(host).ok() << "\"";
+  }
+}
+
+TEST(Misc, idn_to_ascii) {
+  test_idn_to_ascii_one("::::::::::::::::::::::::::::::::::::::@/", "::::::::::::::::::::::::::::::::::::::@/");
+  test_idn_to_ascii_one("%30", "%30");
+  test_idn_to_ascii_one("%30", "%30");
+  test_idn_to_ascii_one("127.0.0.1", "127.0.0.1");
+  test_idn_to_ascii_one("fe80::", "fe80::");
+  test_idn_to_ascii_one("fe80:0:0:0:200:f8ff:fe21:67cf", "fe80:0:0:0:200:f8ff:fe21:67cf");
+  test_idn_to_ascii_one("2001:0db8:11a3:09d7:1f34:8a2e:07a0:765d", "2001:0db8:11a3:09d7:1f34:8a2e:07a0:765d");
+  test_idn_to_ascii_one("::ffff:192.0.2.1", "::ffff:192.0.2.1");
+  test_idn_to_ascii_one("ABCDEF", "abcdef");
+  test_idn_to_ascii_one("abcdef", "abcdef");
+  test_idn_to_ascii_one("abæcdöef", "xn--abcdef-qua4k");
+  test_idn_to_ascii_one("schön", "xn--schn-7qa");
+  test_idn_to_ascii_one("ยจฆฟคฏข", "xn--22cdfh1b8fsa");
+  test_idn_to_ascii_one("☺", "xn--74h");
+  test_idn_to_ascii_one("правда", "xn--80aafi6cg");
+  test_idn_to_ascii_one("büücher", "xn--bcher-kvaa");
+  test_idn_to_ascii_one("BüüCHER", "xn--bcher-kvaa");
+  test_idn_to_ascii_one("bücüher", "xn--bcher-kvab");
+  test_idn_to_ascii_one("bücherü", "xn--bcher-kvae");
+  test_idn_to_ascii_one("ýbücher", "xn--bcher-kvaf");
+  test_idn_to_ascii_one("übücher", "xn--bcher-jvab");
+  test_idn_to_ascii_one("bücher.tld", "xn--bcher-kva.tld");
+  test_idn_to_ascii_one("кто.рф", "xn--j1ail.xn--p1ai");
+  test_idn_to_ascii_one("wіkіреdіа.org", "xn--wkd-8cdx9d7hbd.org");
+  test_idn_to_ascii_one("cnwin2k8中国.avol.com", "xn--cnwin2k8-sd0mx14e.avol.com");
+  test_idn_to_ascii_one("win-2k12r2-addc.阿伯测阿伯测ad.hai.com", "win-2k12r2-addc.xn--ad-tl3ca3569aba8944eca.hai.com");
+  test_idn_to_ascii_one("✌️.ws", "xn--7bi.ws");
+  test_idn_to_ascii_one("⛧", "xn--59h");
+}