Changeset 40123 in vbox for trunk/src

Timestamp:

Feb 14, 2012 11:16:58 AM (13 years ago)

Author:

vboxsync

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

76267

Message:

Runtime/strings: add Utf-8 and Utf-16 sanitising to a white list of characters. Use ranges not individual code points for the white list.

Location:

trunk/src/VBox/Runtime

Files:

• common/string/utf-16.cpp (modified) (1 diff)
• common/string/utf-8.cpp (modified) (1 diff)
• testcase/tstUtf8.cpp (modified) (5 diffs)

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

@@ -264 +264 @@
         if (!Cp)
             break;
-        for (pCp = puszValidSet; ; ++pCp)
-            if (!*pCp || *pCp == Cp)
+        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)
         {

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

@@ -365 +365 @@
         if (!Cp)
             break;
-        for (pCp = puszValidSet; ; ++pCp)
-            if (!*pCp || *pCp == Cp)
+        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)
         {

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

@@ -845 +845 @@
 {
     RTTestSub(hTest, "RTStrPurgeComplementSet");
-    RTUNICP aCpSet[] = { '1', '2', '3', '4', '5', 'w', 'r', 'f', 't', 'e',
-                         '\0' };
+    RTUNICP aCpSet[]    = { '1', '5', 'w', 'w', 'r', 'r', 'e', 'f', 't', 't',
+                            '\0' };
+    RTUNICP aCpBadSet[] = { '1', '5', 'w', 'w', 'r', 'r', 'e', 'f', 't', 't',
+                            '7', '\0' };  /* Contains an incomplete pair. */
     struct
     {
@@ -862 +864 @@
         { "hjhj8766", "????????", aCpSet, '?', 8 },
         { "123\xf0\xa4\xad\xa2""4", "123____4", aCpSet, '_', 1 },
-        { "\xff", "\xff", aCpSet, '_', -1 }
+        { "\xff", "\xff", aCpSet, '_', -1 },
+        { "____", "____", aCpBadSet, '_', -1 }
     };
     enum { MAX_IN_STRING = 256 };
@@ -890 +893 @@
 {
     RTTestSub(hTest, "RTUtf16PurgeComplementSet");
-    RTUNICP aCpSet[] = { '1', '2', '3', '4', '5', 'w', 'r', 'f', 't', 'e',
-                         '\0' };
+    RTUNICP aCpSet[]    = { '1', '5', 'w', 'w', 'r', 'r', 'e', 'f', 't', 't',
+                            '\0' };
+    RTUNICP aCpBadSet[] = { '1', '5', 'w', 'w', 'r', 'r', 'e', 'f', 't', 't',
+                            '7', '\0' };  /* Contains an incomplete pair. */
     struct
     {
@@ -909 +914 @@
         { "123\xf0\xa4\xad\xa2""4", "123__4", 0, aCpSet, '_', 1 },
         { "\xff\xff\0", "\xff\xff\0", 2, aCpSet, '_', -1 },
-        { "\xff\xff\0", "\xff\xff\0", 2, aCpSet, '_', -1 }
+        { "\xff\xff\0", "\xff\xff\0", 2, aCpSet, '_', -1 },
+        { "____", "____", 0, aCpBadSet, '_', -1 }
     };
     enum { MAX_IN_STRING = 256 };
@@ -1508 +1514 @@
     TstRTStrXCmp(hTest);
     TstRTStrPurgeEncoding(hTest);
+    /* TstRT*PurgeComplementSet test conditions which assert. */
+    bool fAreQuiet = RTAssertAreQuiet(), fMayPanic = RTAssertMayPanic();
+    RTAssertSetQuiet(true);
+    RTAssertSetMayPanic(false);
     TstRTStrPurgeComplementSet(hTest);
     TstRTUtf16PurgeComplementSet(hTest);
+    RTAssertSetQuiet(fAreQuiet);
+    RTAssertSetMayPanic(fMayPanic);
     testStrEnd(hTest);
     testStrStr(hTest);