[tools 1/2] testbot/SetWinLocale: Add support for UTF-8 system locales.

[Francois Gouget]

Unicode-only locales, such as Hindi (India), can only be used as a
system locale if activating Windows 10's beta support for the UTF-8
code page.
In such cases the Powershell's Set-WinSystemLocale API automatically
and unconditionally sets the ACP&co code pages to UTF-8. In contrast
intl.cpl has no support for UTF-8 (or for explicitly modifying the
code pages) and refuses to use a Unicode-only system locale.
So to keep a consistent SetWinLocale behavior, reset the system locale to
en-US if it requires UTF-8 support and the --utf8 option was not
specified.

Also note that UTF-8 is only supported on Windows 10+ and that forcing
UTF-8 through the registry makes Windows unbootable.

Signed-off-by: Francois Gouget 
---
Interestingly even English locales can be Unicode-only: see en-AE.
---
 testbot/bin/SetWinLocale     | 11 ++++++++++-
 testbot/bin/SetWinLocale.ps1 | 30 ++++++++++++++++++++++++------
 2 files changed, 34 insertions(+), 7 deletions(-)

diff --git a/testbot/bin/SetWinLocale b/testbot/bin/SetWinLocale
index 9f89280ab9..78c838465f 100755
--- a/testbot/bin/SetWinLocale
+++ b/testbot/bin/SetWinLocale
@@ -524,6 +524,9 @@ if (defined $Usage)
   print "  --system SYS   Specifies the system locale (see --defaults).\n";
   print "                 . This requires elevated privileges.\n";
   print "                 . Only takes effect after a reboot.\n";
+  print "                 . Windows only allows using a Unicode-only locale (such as\n";
+  print "                   hi-IN) as the system locale when setting the code pages to\n";
+  print "                   UTF-8.\n";
   print "                 . Windows 10 GUI: Time & language -> Language ->\n";
   print "                   Administrative Language Settings -> Change system locale.\n";
   print "                 . APIs: GetSystemDefaultLCID(), GetThreadLocale().\n";
@@ -841,7 +844,7 @@ sub SetWinLocales($$$$$$$$)

my $Cmd = ["powershell.exe", "-ExecutionPolicy", "ByPass", "-File",
              "$name0.ps1", "locales", $Locale || ".", $CountryId || ".",
-             $System || ".", $UTF8 ? "true" : "false", $MUI || ".",
+             $System || ".", $UTF8 ? "force" : "no", $MUI || ".",
              $KeyboardIds ? $KeyboardIds->[0] : ".",
              $SysCopy ? "true" : "false", $DefCopy ? "true" : "false",
              $OptUseIntl ? "true" : "false"];
@@ -853,6 +856,12 @@ sub SetWinLocales($$$$$$$$)
   {
     $Line =~ s/\r$//;
     Info("$Line\n");
+    if ($Line eq "Setting the system locale to en-US instead")
+    {
+      # For the final checks
+      $OptSystem = "en-US";
+      $LCIDSystem = "0409";
+    }
   }
 }

diff --git a/testbot/bin/SetWinLocale.ps1 b/testbot/bin/SetWinLocale.ps1
index 00e9a1af14..a79d3c92fd 100644
--- a/testbot/bin/SetWinLocale.ps1
+++ b/testbot/bin/SetWinLocale.ps1
@@ -190,7 +190,7 @@ function SetCodePages($Value)

$SWC_Success = $True

-function SetWinCulture([string]$Locale, [string]$CountryId, [string]$System, [bool]$UTF8, [string]$MUI, [string]$KeyboardId, [bool]$SysCopy, [bool]$DefCopy)
+function SetWinCulture([string]$Locale, [string]$CountryId, [string]$System, [string]$UTF8, [string]$MUI, [string]$KeyboardId, [bool]$SysCopy, [bool]$DefCopy)
 {
   $Cmd = Get-Command Set-Culture -ErrorAction SilentlyContinue
   if ($Cmd -eq $null) {
@@ -201,8 +201,22 @@ function SetWinCulture([string]$Locale, [string]$CountryId, [string]$System, [bo

if ($Locale) { Set-Culture $Locale }
   if ($CountryId) { Set-WinHomeLocation $CountryId }
-  if ($System) { Set-WinSystemLocale $System }
-  if ($UTF8) { SetCodePages(65001) }
+  if ($System)
+  {
+    Set-WinSystemLocale $System
+    $ACP = Get-ItemProperty -Path $HKLM_CODE_PAGE -Name "ACP" -ErrorAction SilentlyContinue
+    if ($ACP.ACP -eq "65001")
+    {
+      Write-Output "$System is a UTF-8 system locale"
+      if ($UTF8 -eq "no")
+      {
+        Write-Output "Setting the system locale to en-US instead"
+        Set-WinSystemLocale "en-US"
+      }
+      $UTF8 = ""
+    }
+  }
+  if ($UTF8 -eq "force") { SetCodePages(65001) }
   if ($MUI) { Set-WinUILanguageOverride $MUI }
   if ($KeyboardId) { Set-WinDefaultInputMethodOverride $KeyboardId }
   if ($SysCopy -or $DefCopy)
@@ -253,7 +267,7 @@ function SetLocales($Argv)
   $Locale = GetStringArg($Argv[1])
   $CountryId = GetStringArg($Argv[2])
   $System = GetStringArg($Argv[3])
-  $UTF8 = $Argv[4] -eq "true"
+  $UTF8 = GetStringArg($Argv[4])
   $MUI = GetStringArg($Argv[5])
   $KeyboardId = GetStringArg($Argv[6])
   $SysCopy = $Argv[7] -ne "false"
@@ -272,7 +286,7 @@ function SetLocales($Argv)
   RunIntlCpl "$Name0.xml"
   Remove-Item -Path "$Name0.xml"

-  if ($UTF8) { SetCodePages(65001) }
+  if ($UTF8 -eq "force") { SetCodePages(65001) }
   exit 0
 }

@@ -298,7 +312,11 @@ function ShowUsage()
   Write-Output "  LOCALE     Is the BCP-47 locale to use for formats, date and time."
   Write-Output "  COUNTRYID  Is the numerical country code."
   Write-Output "  SYSTEM     Is the BCP-47 locale to use as the system locale."
-  Write-Output "  UTF8       If set to 'true' the code pages will be changed to UTF-8."
+  Write-Output "  UTF8       If set to 'force' the code pages will be changed to UTF-8"
+  Write-Output "             regardless of the system locale."
+  Write-Output "             If set to 'allow' the code pages will be set to match the system"
+  Write-Output "             locale even if that implies using UTF-8. This is the default."
+  Write-Output "             If set to 'no' Unicode-only locales will be replaced with en-US."
   Write-Output "  MUI        Is the BCP-47 locale to use for the display language."
   Write-Output "  KEYBOARDID Is the keyboard identifier to set as the default."
   Write-Output "  SYSCOPY    If 'true' the locale settings will be copied to the system accounts"