Changeset 35567 in vbox for trunk/src/VBox/Runtime

Timestamp:

Jan 14, 2011 2:16:45 PM (14 years ago)

Author:

vboxsync

Message:

IPRT: fix rare crash in MiniString::substr(); rename substr() to substrCP() and add a substr that operates on bytes, not codepoints; more to come

Location:

trunk/src/VBox/Runtime

Files:

• common/string/ministring.cpp (modified) (2 diffs)
• testcase/tstIprtMiniString.cpp (modified) (1 diff)

trunk/src/VBox/Runtime/common/string/ministring.cpp

@@ -224 +224 @@
 }
 
-MiniString MiniString::substr(size_t pos /*= 0*/, size_t n /*= npos*/)
+MiniString MiniString::substrCP(size_t pos /*= 0*/, size_t n /*= npos*/)
     const
 {
@@ -256 +256 @@
 
                 size_t cbCopy = psz - pFirst;
-                ret.reserve(cbCopy + 1); // may throw bad_alloc
-#ifndef RT_EXCEPTIONS_ENABLED
-                AssertRelease(capacity() >= cbCopy + 1);
-#endif
-                memcpy(ret.m_psz, pFirst, cbCopy);
-                ret.m_cch = cbCopy;
-                ret.m_psz[cbCopy] = '\0';
+                if (cbCopy)
+                {
+                    ret.reserve(cbCopy + 1); // may throw bad_alloc
+#ifndef RT_EXCEPTIONS_ENABLED
+                    AssertRelease(capacity() >= cbCopy + 1);
+#endif
+                    memcpy(ret.m_psz, pFirst, cbCopy);
+                    ret.m_cch = cbCopy;
+                    ret.m_psz[cbCopy] = '\0';
+                }
             }
         }

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

@@ -211 +211 @@
     CHECK_EQUAL(SubStr15, "cdef");
 
+    /* substr() and substrCP() functions */
+    iprt::MiniString strTest("");
+    CHECK_EQUAL(strTest.substr(0), "");
+    CHECK_EQUAL(strTest.substrCP(0), "");
+    CHECK_EQUAL(strTest.substr(1), "");
+    CHECK_EQUAL(strTest.substrCP(1), "");
+
+    /* now let's have some non-ASCII to chew on */
+    strTest = "abcdefßäbcdef";
+            // 13 codepoints, but 15 bytes (excluding null terminator);
+            // "ß" and "ä" consume two bytes each
+    CHECK_EQUAL(strTest.substr(0),   strTest.c_str());
+    CHECK_EQUAL(strTest.substrCP(0), strTest.c_str());
+
+    CHECK_EQUAL(strTest.substr(2),   "cdefßäbcdef");
+    CHECK_EQUAL(strTest.substrCP(2), "cdefßäbcdef");
+
+    CHECK_EQUAL(strTest.substr(2, 2),   "cd");
+    CHECK_EQUAL(strTest.substrCP(2, 2), "cd");
+
+    CHECK_EQUAL(strTest.substr(6),   "ßäbcdef");
+    CHECK_EQUAL(strTest.substrCP(6), "ßäbcdef");
+
+    CHECK_EQUAL(strTest.substr(6, 2),   "ß");           // UTF-8 "ß" consumes two bytes
+    CHECK_EQUAL(strTest.substrCP(6, 1), "ß");
+
+    CHECK_EQUAL(strTest.substr(8),   "äbcdef");         // UTF-8 "ß" consumes two bytes
+    CHECK_EQUAL(strTest.substrCP(7), "äbcdef");
+
+    CHECK_EQUAL(strTest.substr(8, 3),   "äb");          // UTF-8 "ä" consumes two bytes
+    CHECK_EQUAL(strTest.substrCP(7, 2), "äb");
+
+    CHECK_EQUAL(strTest.substr(14, 1),   "f");
+    CHECK_EQUAL(strTest.substrCP(12, 1), "f");
+
+    CHECK_EQUAL(strTest.substr(15, 1),   "");
+    CHECK_EQUAL(strTest.substrCP(13, 1), "");
+
+    CHECK_EQUAL(strTest.substr(16, 1),   "");
+    CHECK_EQUAL(strTest.substrCP(15, 1), "");
 
     /* special constructor and assignment arguments */