Changeset 83837 in vbox for trunk/src/VBox/Runtime/common

Timestamp:

Apr 20, 2020 12:34:06 AM (5 years ago)

Author:

vboxsync

svn:sync-xref-src-repo-rev:

137366

Message:

IPRT: Added RTUtf16Printf, RTUtf16PrintfV, RTUtf16PrintfEx and RTUtf16PrintfExV. bugref:8489

File:

• trunk/src/VBox/Runtime/common/string/utf-16-printf.cpp (copied) (copied from trunk/src/VBox/Runtime/common/string/strprintf2.cpp ) (5 diffs)

trunk/src/VBox/Runtime/common/string/utf-16-printf.cpp

@@ -1 +1 @@
 /* $Id$ */
 /** @file
- * IPRT - String Formatters, alternative.
+ * IPRT - String Formatters, Outputting UTF-16.
  */
 
@@ -29 +29 @@
 *   Header Files                                                                                                                 *
 *********************************************************************************************************************************/
+#include <iprt/utf16.h>
+#include "internal/iprt.h"
+
+#include <iprt/assert.h>
 #include <iprt/string.h>
-#include "internal/iprt.h"
-
-#include <iprt/assert.h>
+#include <iprt/uni.h>
 
 
@@ -38 +40 @@
 *   Structures and Typedefs                                                                                                      *
 *********************************************************************************************************************************/
-/** rtStrPrintf2Output() argument structure. */
-typedef struct STRPRINTF2OUTPUTARGS
+/** rtUtf16PrintfOutput() argument structure. */
+typedef struct UTF16PRINTFOUTPUTARGS
 {
     /** Pointer to current buffer position. */
-    char   *pszCur;
-    /** Number of bytes left in the buffer (including the trailing zero). */
-    size_t  cbLeft;
+    PRTUTF16    pwszCur;
+    /** Number of RTUTF16 units left in the buffer (including the trailing zero). */
+    size_t      cwcLeft;
     /** Set if we overflowed. */
-    bool    fOverflowed;
-} STRPRINTF2OUTPUTARGS;
-/** Pointer to a rtStrPrintf2Output() argument structure. */
-typedef STRPRINTF2OUTPUTARGS *PSTRPRINTF2OUTPUTARGS;
+    bool        fOverflowed;
+} UTF16PRINTFOUTPUTARGS;
+/** Pointer to a rtUtf16PrintfOutput() argument structure. */
+typedef UTF16PRINTFOUTPUTARGS *PUTF16PRINTFOUTPUTARGS;
 
 
@@ -55 +57 @@
  * Output callback.
  *
- * @returns cbChars
+ * @returns Number of RTUTF16 units we (would have) outputted.
  *
  * @param   pvArg       Pointer to a STRBUFARG structure.
@@ -61 +63 @@
  * @param   cbChars     Number of bytes in the character array pointed to by pachChars.
  */
-static DECLCALLBACK(size_t) rtStrPrintf2Output(void *pvArg, const char *pachChars, size_t cbChars)
-{
-    PSTRPRINTF2OUTPUTARGS pArgs = (PSTRPRINTF2OUTPUTARGS)pvArg;
-    char *pszCur = pArgs->pszCur; /* We actually have to spell this out for VS2010, or it will load it for each case. */
-
-    if (cbChars < pArgs->cbLeft)
+static DECLCALLBACK(size_t) rtUtf16PrintfOutput(void *pvArg, const char *pachChars, size_t cbChars)
+{
+    PUTF16PRINTFOUTPUTARGS pArgs  = (PUTF16PRINTFOUTPUTARGS)pvArg;
+    size_t                 cwcRet = 0;
+
+    size_t cwcLeft = pArgs->cwcLeft;
+    if (cwcLeft > 1)
     {
-        pArgs->cbLeft -= cbChars;
-
-        /* Note! For VS2010/64 we need at least 7 case statements before it generates a jump table. */
-        switch (cbChars)
+        Assert(!pArgs->fOverflowed);
+
+        PRTUTF16 pwszCur = pArgs->pwszCur;
+        for (;;)
         {
-            default:
-                memcpy(pszCur, pachChars, cbChars);
-                break;
-            case 8: pszCur[7] = pachChars[7]; RT_FALL_THRU();
-            case 7: pszCur[6] = pachChars[6]; RT_FALL_THRU();
-            case 6: pszCur[5] = pachChars[5]; RT_FALL_THRU();
-            case 5: pszCur[4] = pachChars[4]; RT_FALL_THRU();
-            case 4: pszCur[3] = pachChars[3]; RT_FALL_THRU();
-            case 3: pszCur[2] = pachChars[2]; RT_FALL_THRU();
-            case 2: pszCur[1] = pachChars[1]; RT_FALL_THRU();
-            case 1: pszCur[0] = pachChars[0]; RT_FALL_THRU();
-            case 0:
-                break;
+            if (cbChars > 0)
+            {
+                RTUNICP uc;
+                int rc = RTStrGetCpNEx(&pachChars, &cbChars, &uc);
+                AssertRCStmt(rc, uc = 0x7f);
+
+                /* Simple: */
+                if (RTUniCpIsBMP(uc))
+                {
+                    cwcRet += 1;
+                    if (RT_LIKELY(cwcLeft > 1))
+                        *pwszCur++ = uc;
+                    else
+                        break;
+                    cwcLeft--;
+                }
+                /* Surrogate pair: */
+                else if (uc >= 0x10000 && uc <= 0x0010ffff)
+                {
+                    cwcRet += 2;
+                    if (RT_LIKELY(cwcLeft > 2))
+                        *pwszCur++ = 0xd800 | (uc >> 10);
+                    else
+                    {
+                        if (cwcLeft > 1)
+                        {
+                            cwcLeft = 1;
+                            pwszCur[1] = '\0';
+                        }
+                        break;
+                    }
+                    *pwszCur++ = 0xdc00 | (uc & 0x3ff);
+                    cwcLeft -= 2;
+                }
+                else
+                {
+                    AssertMsgFailed(("uc=%#x\n", uc));
+                    cwcRet += 1;
+                    if (RT_LIKELY(cwcLeft > 1))
+                        *pwszCur++ = 0x7f;
+                    else
+                        break;
+                    cwcLeft--;
+                }
+            }
+            else
+            {
+                *pwszCur = '\0';
+                pArgs->pwszCur = pwszCur;
+                pArgs->cwcLeft = cwcLeft;
+                return cwcRet;
+            }
         }
-        pArgs->pszCur = pszCur += cbChars;
-        *pszCur = '\0';
+
+        /*
+         * We only get here if we run out of buffer space.
+         */
+        Assert(cwcLeft == 1);
+        *pwszCur = '\0';
+        pArgs->pwszCur = pwszCur;
+        pArgs->cwcLeft = cwcLeft;
     }
-    else
+    /*
+     * We get a special zero byte call at the end for the formatting operation.
+     *
+     * Make sure we don't turn that into an overflow and that we'll terminate
+     * empty result strings.
+     */
+    else if (cbChars == 0 && cwcLeft > 0)
     {
-        size_t cbLeft = pArgs->cbLeft;
-        if (cbLeft-- > 1)
+        *pArgs->pwszCur = '\0';
+        return 0;
+    }
+
+    /*
+     * Overflow handling.  Calc needed space.
+     */
+    pArgs->fOverflowed = true;
+
+    while (cbChars > 0)
+    {
+        RTUNICP uc;
+        int rc = RTStrGetCpNEx(&pachChars, &cbChars, &uc);
+        AssertRCStmt(rc, uc = 0x7f);
+
+        if (RTUniCpIsBMP(uc))
+            cwcRet += 1;
+        else if (uc >= 0x10000 && uc <= 0x0010ffff)
+            cwcRet += 2;
+        else
         {
-            memcpy(pszCur, pachChars, cbLeft);
-            pArgs->pszCur = pszCur += cbLeft;
-            *pszCur = '\0';
-            pArgs->cbLeft = 1;
+            AssertMsgFailed(("uc=%#x\n", uc));
+            cwcRet += 1;
         }
-        pArgs->fOverflowed = true;
     }
 
-    return cbChars;
-}
-
-
-RTDECL(ssize_t) RTStrPrintf2(char *pszBuffer, size_t cchBuffer, const char *pszFormat, ...)
-{
-    /* Explicitly inline RTStrPrintf2V + RTStrPrintf2ExV here because this is a frequently use API. */
-    STRPRINTF2OUTPUTARGS Args;
-    size_t cchRet;
+    return cwcRet;
+}
+
+
+RTDECL(ssize_t) RTUtf16Printf(PRTUTF16 pwszBuffer, size_t cwcBuffer, const char *pszFormat, ...)
+{
+    /* Explicitly inline RTStrPrintfV + RTStrPrintfExV here because this is a frequently use API. */
+    UTF16PRINTFOUTPUTARGS Args;
+    size_t cwcRet;
     va_list args;
-    AssertMsg(cchBuffer > 0, ("Excellent idea! Format a string with no space for the output!\n"));
-
-    Args.pszCur      = pszBuffer;
-    Args.cbLeft      = cchBuffer;
+    AssertMsg(cwcBuffer > 0, ("Excellent idea! Format a string with no space for the output!\n"));
+
+    Args.pwszCur     = pwszBuffer;
+    Args.cwcLeft     = cwcBuffer;
     Args.fOverflowed = false;
 
     va_start(args, pszFormat);
-    cchRet = RTStrFormatV(rtStrPrintf2Output, &Args, NULL, NULL, pszFormat, args);
+    cwcRet = RTStrFormatV(rtUtf16PrintfOutput, &Args, NULL, NULL, pszFormat, args);
     va_end(args);
 
-    return !Args.fOverflowed ? (ssize_t)cchRet : -(ssize_t)cchRet - 1;
+    return !Args.fOverflowed ? (ssize_t)cwcRet : -(ssize_t)cwcRet - 1;
 }
 RT_EXPORT_SYMBOL(RTStrPrintf2);
 
 
-RTDECL(ssize_t) RTStrPrintf2ExV(PFNSTRFORMAT pfnFormat, void *pvArg, char *pszBuffer, size_t cchBuffer,
-                                const char *pszFormat,  va_list args)
-{
-    STRPRINTF2OUTPUTARGS Args;
-    size_t cchRet;
-    AssertMsg(cchBuffer > 0, ("Excellent idea! Format a string with no space for the output!\n"));
-
-    Args.pszCur      = pszBuffer;
-    Args.cbLeft      = cchBuffer;
+RTDECL(ssize_t) RTUtf16PrintfExV(PFNSTRFORMAT pfnFormat, void *pvArg, PRTUTF16 pwszBuffer, size_t cwcBuffer,
+                                 const char *pszFormat,  va_list args)
+{
+    UTF16PRINTFOUTPUTARGS Args;
+    size_t cwcRet;
+    AssertMsg(cwcBuffer > 0, ("Excellent idea! Format a string with no space for the output!\n"));
+
+    Args.pwszCur     = pwszBuffer;
+    Args.cwcLeft     = cwcBuffer;
     Args.fOverflowed = false;
-    cchRet = RTStrFormatV(rtStrPrintf2Output, &Args, pfnFormat, pvArg, pszFormat, args);
-    return !Args.fOverflowed ? (ssize_t)cchRet : -(ssize_t)cchRet - 1;
-}
-RT_EXPORT_SYMBOL(RTStrPrintf2ExV);
-
-
-RTDECL(ssize_t) RTStrPrintf2V(char *pszBuffer, size_t cchBuffer, const char *pszFormat, va_list args)
-{
-    return RTStrPrintf2ExV(NULL, NULL, pszBuffer, cchBuffer, pszFormat, args);
-}
-RT_EXPORT_SYMBOL(RTStrPrintf2V);
-
-
-RTDECL(ssize_t) RTStrPrintf2Ex(PFNSTRFORMAT pfnFormat, void *pvArg, char *pszBuffer, size_t cchBuffer, const char *pszFormat, ...)
+    cwcRet = RTStrFormatV(rtUtf16PrintfOutput, &Args, pfnFormat, pvArg, pszFormat, args);
+    return !Args.fOverflowed ? (ssize_t)cwcRet : -(ssize_t)cwcRet - 1;
+}
+RT_EXPORT_SYMBOL(RTUtf16PrintfExV);
+
+
+RTDECL(ssize_t) RTUtf16PrintfV(PRTUTF16 pwszBuffer, size_t cwcBuffer, const char *pszFormat, va_list args)
+{
+    return RTUtf16PrintfExV(NULL, NULL, pwszBuffer, cwcBuffer, pszFormat, args);
+}
+RT_EXPORT_SYMBOL(RTUtf16Printf2V);
+
+
+RTDECL(ssize_t) RTUtf16PrintfEx(PFNSTRFORMAT pfnFormat, void *pvArg, PRTUTF16 pwszBuffer, size_t cwcBuffer,
+                                const char *pszFormat, ...)
 {
     va_list args;
     ssize_t cbRet;
     va_start(args, pszFormat);
-    cbRet = RTStrPrintf2ExV(pfnFormat, pvArg, pszBuffer, cchBuffer, pszFormat, args);
+    cbRet = RTUtf16PrintfExV(pfnFormat, pvArg, pwszBuffer, cwcBuffer, pszFormat, args);
     va_end(args);
     return cbRet;
 }
-RT_EXPORT_SYMBOL(RTStrPrintf2Ex);
-
+RT_EXPORT_SYMBOL(RTUtf16PrintfEx);
+