01/01: [CONUTILS] Use the currently active console codepage to output text to files. If you want to output UTF-8 run "chcp 65001" before. CORE-12122

[Hermès Bélusca-Maïto]

https://git.reactos.org/?p=reactos.git;a=commitdiff;h=2b3fd31bc55ddb2185a64…

commit 2b3fd31bc55ddb2185a64c33bf2dd8b753de44e2
Author:     Hermès Bélusca-Maïto <hermes.belusca-maito(a)reactos.org>
AuthorDate: Sat May 5 01:36:19 2018 +0200
Commit:     Hermès Bélusca-Maïto <hermes.belusca-maito(a)reactos.org>
CommitDate: Sat May 5 01:37:30 2018 +0200

    [CONUTILS] Use the currently active console codepage to output text to files. If you
want to output UTF-8 run "chcp 65001" before.
    CORE-12122
---
 sdk/lib/conutils/outstream.c      | 34 +++++++++++++++++++---------------
 sdk/lib/conutils/stream.c         | 20 +++++++++-----------
 sdk/lib/conutils/stream.h         | 14 ++++----------
 sdk/lib/conutils/stream_private.h |  2 +-
 4 files changed, 33 insertions(+), 37 deletions(-)

diff --git a/sdk/lib/conutils/outstream.c b/sdk/lib/conutils/outstream.c
index a36a5ddaa2..645452f3ef 100644
--- a/sdk/lib/conutils/outstream.c
+++ b/sdk/lib/conutils/outstream.c
@@ -103,7 +103,7 @@ ConWrite(
     // if (IsConsoleHandle(Stream->hHandle))
     if (Stream->IsConsole)
     {
-        // TODO: Check if (ConStream->Mode == WideText or UTF16Text) ??
+        // TODO: Check if (Stream->Mode == WideText or UTF16Text) ??
 
         /*
          * This code is inspired from _cputws, in particular from the fact that,
@@ -136,15 +136,15 @@ ConWrite(
      *
      * Implementation NOTE:
      *   MultiByteToWideChar (resp. WideCharToMultiByte) are equivalent to
-     *   OemToCharBuffW (resp. CharToOemBuffW), but the latters uselessly
-     *   depend on user32.dll, while MultiByteToWideChar and WideCharToMultiByte
-     *   only need kernel32.dll.
+     *   OemToCharBuffW (resp. CharToOemBuffW), but these latter functions
+     *   uselessly depend on user32.dll, while MultiByteToWideChar and
+     *   WideCharToMultiByte only need kernel32.dll.
      */
     if ((Stream->Mode == WideText) || (Stream->Mode == UTF16Text))
     {
 #ifndef _UNICODE // UNICODE means that TCHAR == WCHAR == UTF-16
-        /* Convert from the current process/thread's codepage to UTF-16 */
-        WCHAR *buffer = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, len *
sizeof(WCHAR));
+        /* Convert from the current process/thread's code page to UTF-16 */
+        PWCHAR buffer = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, len *
sizeof(WCHAR));
         if (!buffer)
         {
             SetLastError(ERROR_NOT_ENOUGH_MEMORY);
@@ -200,18 +200,22 @@ ConWrite(
     }
     else if ((Stream->Mode == UTF8Text) || (Stream->Mode == AnsiText))
     {
-        CHAR *buffer;
+        UINT CodePage;
+        PCHAR buffer;
 
         /*
-         * Resolve the codepage cache if it was not assigned yet
-         * (only if the stream is in ANSI mode; in UTF8 mode the
-         * codepage was already set to CP_UTF8).
+         * Resolve the current code page if it has not been assigned yet
+         * (we do this only if the stream is in ANSI mode; in UTF8 mode
+         * the code page is always set to CP_UTF8). Otherwise use the
+         * current stream's code page.
          */
         if (/*(Stream->Mode == AnsiText) &&*/ (Stream->CodePage ==
INVALID_CP))
-            Stream->CodePage = GetConsoleOutputCP(); // CP_ACP, CP_OEMCP
+            CodePage = GetConsoleOutputCP(); // CP_ACP, CP_OEMCP
+        else
+            CodePage = Stream->CodePage;
 
 #ifdef _UNICODE // UNICODE means that TCHAR == WCHAR == UTF-16
-        /* Convert from UTF-16 to either UTF-8 or ANSI, using stream codepage */
+        /* Convert from UTF-16 to either UTF-8 or ANSI, using the stream code page */
         // NOTE: MB_LEN_MAX defined either in limits.h or in stdlib.h .
         buffer = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, len * MB_LEN_MAX);
         if (!buffer)
@@ -219,14 +223,14 @@ ConWrite(
             SetLastError(ERROR_NOT_ENOUGH_MEMORY);
             return 0;
         }
-        len = WideCharToMultiByte(Stream->CodePage, 0,
+        len = WideCharToMultiByte(CodePage, 0,
                                   szStr, len, buffer, len * MB_LEN_MAX,
                                   NULL, NULL);
         szStr = (PVOID)buffer;
 #else
         /*
-         * Convert from the current process/thread's codepage to either
-         * UTF-8 or ANSI, using stream codepage.
+         * Convert from the current process/thread's code page to either
+         * UTF-8 or ANSI, using the stream code page.
          * We need to perform a double conversion, by going through UTF-16.
          */
         // TODO!
diff --git a/sdk/lib/conutils/stream.c b/sdk/lib/conutils/stream.c
index 0b137ea2f7..7246453df4 100644
--- a/sdk/lib/conutils/stream.c
+++ b/sdk/lib/conutils/stream.c
@@ -73,9 +73,6 @@ static int ConToCRTMode[] =
     _O_U16TEXT, // UTF16Text (UTF16 without BOM; translated)
     _O_U8TEXT,  // UTF8Text  (UTF8  without BOM; translated)
 };
-#endif
-
-#ifdef USE_CRT
 
 /*
  * See http://archives.miloush.net/michkap/archive/2008/03/18/8306597.html
@@ -83,7 +80,7 @@ static int ConToCRTMode[] =
  * for more details.
  */
 
-// NOTE: May the translated mode be cached somehow?
+// NOTE1: May the translated mode be cached somehow?
 // NOTE2: We may also call IsConsoleHandle to directly set the mode to
 //        _O_U16TEXT if it's ok??
 // NOTE3: _setmode returns the previous mode, or -1 if failure.
@@ -99,16 +96,17 @@ do { \
 #else /* defined(USE_CRT) */
 
 /*
- * We set Stream->CodePage to INVALID_CP (= -1) to signal that the codepage
+ * We set Stream->CodePage to INVALID_CP (== -1) to signal that the code page
  * is either not assigned (if the mode is Binary, WideText, or UTF16Text), or
- * is not cached yet (if the mode is AnsiText). In this latter case the cache
- * is resolved inside ConWrite. Finally, if the mode is UTF8Text, the codepage
- * cache is set to CP_UTF8.
- * The codepage cache can be reset by an explicit call to CON_STREAM_SET_MODE
+ * is not cached (if the mode is AnsiText). In this latter case the code page
+ * is resolved inside ConWrite. Finally, if the mode is UTF8Text, the code page
+ * cache is always set to CP_UTF8.
+ * The code page cache can be reset by an explicit call to CON_STREAM_SET_MODE
  * (i.e. by calling ConStreamSetMode, or by reinitializing the stream with
  * ConStreamInit(Ex)).
  *
- * NOTE: the magic value could not be '0' since it is reserved for CP_ACP.
+ * NOTE: the reserved values are: 0 (CP_ACP), 1 (CP_OEMCP), 2 (CP_MACCP),
+ * 3 (CP_THREAD_ACP), 42 (CP_SYMBOL), 65000 (CP_UTF7) and 65001 (CP_UTF8).
  */
 #define CON_STREAM_SET_MODE(Stream, Mode, CacheCodePage)    \
 do { \
@@ -229,7 +227,7 @@ ConStreamSetCacheCodePage(
         return FALSE;
 
     /*
-     * Keep the original stream mode but set the correct file codepage
+     * Keep the original stream mode but set the correct file code page
      * (will be reset only if Mode == AnsiText).
      */
     Mode = Stream->Mode;
diff --git a/sdk/lib/conutils/stream.h b/sdk/lib/conutils/stream.h
index 8a2746af5c..031ae56d30 100644
--- a/sdk/lib/conutils/stream.h
+++ b/sdk/lib/conutils/stream.h
@@ -113,20 +113,14 @@ do { \
 } while(0)
 #endif /* defined(USE_CRT) */
 
-#ifdef _UNICODE
 /*
- * Use UTF8 by default for file output, because this mode is back-compatible
- * with ANSI, and it displays nice on terminals that support UTF8 by default
- * (not many terminals support UTF16 on the contrary).
+ * Use ANSI by default for file output, with no cached code page.
+ * Note that setting the stream mode to AnsiText and the code page value
+ * to CP_UTF8 sets the stream to UTF8 mode, and has the same effect as if
+ * the stream mode UTF8Text had been specified instead.
  */
-#define ConInitStdStreams() \
-    ConInitStdStreamsAndMode(UTF8Text, INVALID_CP)
-    /* Note that here the cache code page is unused */
-#else
-/* Use ANSI by default for file output */
 #define ConInitStdStreams() \
     ConInitStdStreamsAndMode(AnsiText, INVALID_CP)
-#endif /* defined(_UNICODE) */
 
 /* Stream translation modes */
 BOOL
diff --git a/sdk/lib/conutils/stream_private.h b/sdk/lib/conutils/stream_private.h
index 6b801dfb23..181cf77fc3 100644
--- a/sdk/lib/conutils/stream_private.h
+++ b/sdk/lib/conutils/stream_private.h
@@ -46,7 +46,7 @@ typedef struct _CON_STREAM
      * when 'hHandle' refers to a file or a pipe.
      */
     CON_STREAM_MODE Mode;
-    UINT CodePage;  // Used to convert UTF-16 text to some ANSI codepage.
+    UINT CodePage;  // Used to convert UTF-16 text to some ANSI code page.
 #endif /* defined(USE_CRT) */
 } CON_STREAM, *PCON_STREAM;