Changeset 93103 in vbox

Timestamp:

Dec 30, 2021 11:29:37 PM (3 years ago)

Author:

vboxsync

Message:

IPRT: Implemented RTUtf16NCmp and RTUtf16NCmpUtf8, adding RTUtf16GetCpNEx.

Location:

trunk

Files:

• include/iprt/utf16.h (modified) (2 diffs)
• src/VBox/Runtime/Makefile.kmk (modified) (1 diff)
• src/VBox/Runtime/common/string/RTUtf16NCmp.cpp (added)
• src/VBox/Runtime/common/string/RTUtf16NCmpUtf8.cpp (added)
• src/VBox/Runtime/common/string/utf-16.cpp (modified) (1 diff)

trunk/include/iprt/utf16.h

@@ -1170 +1170 @@
 
 /**
+ * Get the unicode code point at the given string position with length
+ * restriction.
+ *
+ * @returns iprt status code.
+ * @param   ppwsz       Pointer to the string pointer. This will be updated to
+ *                      point to the char following the current code point.
+ * @param   pcwc        Pointer to the max string length. This will be
+ *                      decremented corrsponding to the advancement of @a ppwsz.
+ * @param   pCp         Where to store the code point.
+ *                      RTUNICP_INVALID is stored here on failure.
+ *
+ * @remark  This is an internal worker for RTUtf16GetCpNEx().
+ */
+RTDECL(int) RTUtf16GetCpNExInternal(PCRTUTF16 *ppwsz, size_t *pcwc, PRTUNICP pCp);
+
+/**
  * Get the unicode code point at the given string position, big endian.
  *
@@ -1240 +1256 @@
     }
     return RTUtf16GetCpExInternal(ppwsz, pCp);
+}
+
+/**
+ * Get the unicode code point at the given string position.
+ *
+ * @returns iprt status code.
+ * @param   ppwsz       Pointer to the string pointer. This will be updated to
+ *                      point to the char following the current code point.
+ * @param   pcwc        Pointer to the max string length. This will be
+ *                      decremented corrsponding to the advancement of @a ppwsz.
+ * @param   pCp         Where to store the code point. RTUNICP_INVALID is stored
+ *                      here on failure.
+ *
+ * @remark  We optimize this operation by using an inline function for
+ *          everything which isn't a surrogate pair or and endian indicator.
+ */
+DECLINLINE(int) RTUtf16GetCpNEx(PCRTUTF16 *ppwsz, size_t *pcwc, PRTUNICP pCp)
+{
+    const size_t cwc = *pcwc;
+    if (cwc > 0)
+    {
+        const PCRTUTF16 pwsz = *ppwsz;
+        const RTUTF16   wc   = *pwsz;
+        if (wc < 0xd800 || (wc > 0xdfff && wc < 0xfffe))
+        {
+            *pCp   = wc;
+            *pcwc  = cwc  - 1;
+            *ppwsz = pwsz + 1;
+            return VINF_SUCCESS;
+        }
+    }
+    return RTUtf16GetCpNExInternal(ppwsz, pcwc, pCp);
 }
 

trunk/src/VBox/Runtime/Makefile.kmk

@@ -617 +617 @@
         common/string/RTUtf16ICmpAscii.cpp \
         common/string/RTUtf16End.cpp \
+        common/string/RTUtf16NCmp.cpp \
         common/string/RTUtf16NCmpAscii.cpp \
+        common/string/RTUtf16NCmpUtf8.cpp \
         common/string/RTUtf16NICmpAscii.cpp \
         common/string/RTUtf16NLen.cpp \

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

@@ -1187 +1187 @@
 
 
+RTDECL(int) RTUtf16GetCpNExInternal(PCRTUTF16 *ppwsz, size_t *pcwc, PRTUNICP pCp)
+{
+    int          rc;
+    const size_t cwc = *pcwc;
+    if (cwc > 0)
+    {
+        PCRTUTF16     pwsz = *ppwsz;
+        const RTUTF16 wc   = **ppwsz;
+
+        /* simple */
+        if (wc < 0xd800 || (wc > 0xdfff && wc < 0xfffe))
+        {
+            *pCp   = wc;
+            *pcwc  = cwc  - 1;
+            *ppwsz = pwsz + 1;
+            return VINF_SUCCESS;
+        }
+
+        if (wc < 0xfffe)
+        {
+            /* surrogate pair */
+            if (wc < 0xdc00)
+            {
+                if (cwc >= 2)
+                {
+                    const RTUTF16 wc2 = pwsz[1];
+                    if (wc2 >= 0xdc00 && wc2 <= 0xdfff)
+                    {
+                        *pCp   = 0x10000 + (((wc & 0x3ff) << 10) | (wc2 & 0x3ff));
+                        *pcwc  = cwc  - 2;
+                        *ppwsz = pwsz + 2;
+                        return VINF_SUCCESS;
+                    }
+
+                    RTStrAssertMsgFailed(("wc=%#08x wc2=%#08x - invalid 2nd char in surrogate pair\n", wc, wc2));
+                }
+                else
+                    RTStrAssertMsgFailed(("wc=%#08x - incomplete surrogate pair\n", wc));
+            }
+            else
+                RTStrAssertMsgFailed(("wc=%#08x - invalid surrogate pair order\n", wc));
+            rc = VERR_INVALID_UTF16_ENCODING;
+        }
+        else
+        {
+            RTStrAssertMsgFailed(("wc=%#08x - endian indicator\n", wc));
+            rc = VERR_CODE_POINT_ENDIAN_INDICATOR;
+        }
+        *pcwc  = cwc  - 1;
+        *ppwsz = pwsz + 1;
+    }
+    else
+        rc = VERR_END_OF_STRING;
+    *pCp = RTUNICP_INVALID;
+    return rc;
+}
+RT_EXPORT_SYMBOL(RTUtf16GetCpNExInternal);
+
+
 RTDECL(int) RTUtf16BigGetCpExInternal(PCRTUTF16 *ppwsz, PRTUNICP pCp)
 {