Changeset 62916 in vbox for trunk

Timestamp:

Aug 3, 2016 2:05:01 PM (8 years ago)

Author:

vboxsync

Message:

RTStrPurgeEncoding: Optimized it a little, adding debug assertion for bad pairs.

Location:

trunk

Files:

• include/VBox/vmm/pdmifs.h (modified) (1 diff)
• include/iprt/string.h (modified) (1 diff)
• src/VBox/Runtime/common/string/utf-8.cpp (modified) (1 diff)
• src/VBox/Runtime/testcase/tstUtf8.cpp (modified) (1 diff)

trunk/include/VBox/vmm/pdmifs.h

@@ -1197 +1197 @@
      * @thread  Any thread.
      */
-    DECLR3CALLBACKMEMBER(int, pfnSetParameters,(PPDMICHARCONNECTOR pInterface, unsigned Bps, char chParity, unsigned cDataBits, unsigned cStopBits));
+    DECLR3CALLBACKMEMBER(int, pfnSetParameters,(PPDMICHARCONNECTOR pInterface, unsigned Bps, char chParity,
+                                                unsigned cDataBits, unsigned cStopBits));
 
     /**

trunk/include/iprt/string.h

@@ -726 +726 @@
 
 /**
- * Sanitise a (valid) UTF-8 string by replacing all characters outside a white
+ * Sanitizes a (valid) UTF-8 string by replacing all characters outside a white
  * list in-place by an ASCII replacement character.  Multi-byte characters will
  * be replaced byte by byte.
  *
- * @returns The number of code points replaced, or a negative value if the
- *          string is not correctly encoded.  In this last case the string
- *          may be partially processed.
+ * @returns The number of code points replaced.  In the case of an incorrectly
+ *          encoded string -1 will be returned, and the string is not completely
+ *          processed.  In the case of puszValidPairs having an odd number of
+ *          code points, -1 will be also return but without any modification to
+ *          the string.
  * @param   psz            The string to sanitise.
- * @param   puszValidSet   A zero-terminated array of pairs of Unicode points.
+ * @param   puszValidPairs A zero-terminated array of pairs of Unicode points.
  *                         Each pair is the start and end point of a range,
  *                         and the union of these ranges forms the white list.
  * @param   chReplacement  The ASCII replacement character.
  */
-RTDECL(ssize_t) RTStrPurgeComplementSet(char *psz, PCRTUNICP puszValidSet, char chReplacement);
+RTDECL(ssize_t) RTStrPurgeComplementSet(char *psz, PCRTUNICP puszValidPairs, char chReplacement);
 
 /**

trunk/src/VBox/Runtime/common/string/utf-8.cpp

@@ -364 +364 @@
 
 
-RTDECL(ssize_t) RTStrPurgeComplementSet(char *psz, PCRTUNICP puszValidSet, char chReplacement)
-{
-    size_t cReplacements = 0;
+/**
+ * Helper for RTStrPurgeComplementSet.
+ *
+ * @returns true if @a Cp is valid, false if not.
+ * @param   Cp              The code point to validate.
+ * @param   puszValidPairs  Pair of valid code point sets.
+ * @param   cValidPairs     Number of pairs.
+ */
+DECLINLINE(bool) rtStrPurgeIsInSet(RTUNICP Cp, PCRTUNICP puszValidPairs, uint32_t cValidPairs)
+{
+    while (cValidPairs-- > 0)
+    {
+        if (   Cp >= puszValidPairs[0]
+            && Cp <= puszValidPairs[1])
+            return true;
+        puszValidPairs += 2;
+    }
+    return false;
+}
+
+
+RTDECL(ssize_t) RTStrPurgeComplementSet(char *psz, PCRTUNICP puszValidPairs, char chReplacement)
+{
     AssertReturn(chReplacement && (unsigned)chReplacement < 128, -1);
+
+    /*
+     * Calc valid pairs and check that we've got an even number.
+     */
+    uint32_t cValidPairs = 0;
+    while (puszValidPairs[cValidPairs * 2])
+    {
+        AssertReturn(puszValidPairs[cValidPairs * 2 + 1], -1);
+        AssertMsg(puszValidPairs[cValidPairs * 2] <= puszValidPairs[cValidPairs * 2 + 1],
+                  ("%#x vs %#x\n", puszValidPairs[cValidPairs * 2], puszValidPairs[cValidPairs * 2 + 1]));
+        cValidPairs++;
+    }
+
+    /*
+     * Do the replacing.
+     */
+    ssize_t cReplacements = 0;
     for (;;)
     {
-        RTUNICP Cp;
-        PCRTUNICP pCp;
-        char *pszOld = psz;
-        if (RT_FAILURE(RTStrGetCpEx((const char **)&psz, &Cp)))
+        char    *pszCur = psz;
+        RTUNICP  Cp;
+        int rc = RTStrGetCpEx((const char **)&psz, &Cp);
+        if (RT_SUCCESS(rc))
+        {
+            if (Cp)
+            {
+                if (!rtStrPurgeIsInSet(Cp, puszValidPairs, cValidPairs))
+                {
+                    for (; pszCur != psz; ++pszCur)
+                        *pszCur = chReplacement;
+                    ++cReplacements;
+                }
+            }
+            else
+                break;
+        }
+        else
             return -1;
-        if (!Cp)
-            break;
-        for (pCp = puszValidSet; *pCp; pCp += 2)
-        {
-            AssertReturn(*(pCp + 1), -1);
-            if (*pCp <= Cp && *(pCp + 1) >= Cp) /* No, I won't do * and ++. */
-                break;
-        }
-        if (!*pCp)
-        {
-            for (; pszOld != psz; ++pszOld)
-                *pszOld = chReplacement;
-            ++cReplacements;
-        }
     }
     return cReplacements;

trunk/src/VBox/Runtime/testcase/tstUtf8.cpp

@@ -874 +874 @@
         ssize_t cReplacements;
         AssertRC(RTStrCopy(szCopy, RT_ELEMENTS(szCopy), aTests[i].pcszIn));
-        cReplacements = RTStrPurgeComplementSet(szCopy, aTests[i].pcCpSet,
-                                                aTests[i].chReplacement);
+        RTTestDisableAssertions(hTest);
+        cReplacements = RTStrPurgeComplementSet(szCopy, aTests[i].pcCpSet, aTests[i].chReplacement);
+        RTTestRestoreAssertions(hTest);
         if (cReplacements != aTests[i].cExpected)
             RTTestFailed(hTest, "#%u: expected %lld, actual %lld\n", i,