Changeset 95336 in vbox

Timestamp:

Jun 21, 2022 8:48:58 PM (2 years ago)

Author:

vboxsync

Message:

IPRT/RTCString: Added a truncate() method that takes UTF-8 encoding into account when truncating the string. bugref:10185

Location:

trunk

Files:

• include/iprt/cpp/ministring.h (modified) (1 diff)
• src/VBox/Runtime/common/string/ministring.cpp (modified) (1 diff)
• src/VBox/Runtime/testcase/tstIprtMiniString.cpp (modified) (1 diff)

trunk/include/iprt/cpp/ministring.h

@@ -853 +853 @@
 
     /**
+     * Truncates the string to a max length of @a cchMax.
+     *
+     * If the string is shorter than @a cchMax characters, no change is made.
+     *
+     * If the @a cchMax is not at the start of a UTF-8 sequence, it will be adjusted
+     * down to the start of the UTF-8 sequence.  Thus, after a truncation, the
+     * length() may be smaller than @a cchMax.
+     *
+     */
+    RTCString &truncate(size_t cchMax) RT_NOEXCEPT;
+
+    /**
      * Index operator.
      *

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

@@ -774 +774 @@
 
 
+RTCString &RTCString::truncate(size_t cchMax) RT_NOEXCEPT
+{
+    if (cchMax < m_cch)
+    {
+        /*
+         * Make sure the truncated string ends with a correctly encoded
+         * codepoint and is not missing a few bytes.
+         */
+        if (cchMax > 0)
+        {
+            char chTail = m_psz[cchMax];
+            if (   (chTail & 0x80) == 0         /* single byte codepoint */
+                || (chTail & 0xc0) == 0xc0)     /* first byte of codepoint sequence. */
+            { /* likely */ }
+            else
+            {
+                /* We need to find the start of the codepoint sequence: */
+                do
+                    cchMax -= 1;
+                while (   cchMax > 0
+                       && (m_psz[cchMax] & 0xc0) != 0xc0);
+            }
+        }
+
+        /*
+         * Do the truncating.
+         */
+        m_psz[cchMax] = '\0';
+        m_cch = cchMax;
+    }
+    return *this;
+}
+
+
 size_t RTCString::find(const char *pszNeedle, size_t offStart /*= 0*/) const RT_NOEXCEPT
 {

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

@@ -361 +361 @@
         RTTESTI_CHECK(StrAssign.length() == 0);
     }
+
+    /* truncation */
+    RTCString StrTruncate1("abcdef");
+    RTTESTI_CHECK(StrTruncate1.length() == 6);
+    for (int i = 5; i >= 0; i--)
+    {
+        StrTruncate1.truncate(i);
+        RTTESTI_CHECK(StrTruncate1.length() == (size_t)i);
+    }
+
+    RTCString StrTruncate2("01ßä6");
+    CHECK_EQUAL(StrTruncate2, "01ßä6");
+    StrTruncate2.truncate(6);
+    CHECK_EQUAL(StrTruncate2, "01ßä");
+    StrTruncate2.truncate(5);
+    CHECK_EQUAL(StrTruncate2, "01ß");
+    StrTruncate2.truncate(10);
+    CHECK_EQUAL(StrTruncate2, "01ß");
+    StrTruncate2.truncate(4);
+    CHECK_EQUAL(StrTruncate2, "01ß");
+    StrTruncate2.truncate(3);
+    CHECK_EQUAL(StrTruncate2, "01");
+    StrTruncate2.truncate(1);
+    CHECK_EQUAL(StrTruncate2, "0");
+    StrTruncate2.truncate(0);
+    CHECK_EQUAL(StrTruncate2, "");
 
 #undef CHECK