Changeset 40071 in vbox for trunk/include

Timestamp:

Feb 10, 2012 9:35:27 PM (13 years ago)

Author:

vboxsync

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

76205

Message:

Runtime/strings: add Utf-8 and Utf-16 sanitising to a white list of characters.

Location:

trunk/include/iprt

Files:

• mangling.h (modified) (2 diffs)
• string.h (modified) (2 diffs)

trunk/include/iprt/mangling.h

@@ -1306 +1306 @@
 # define RTStrPrintHexBytes                             RT_MANGLER(RTStrPrintHexBytes)
 # define RTStrPurgeEncoding                             RT_MANGLER(RTStrPurgeEncoding)
+# define RTStrPurgeComplementSet                        RT_MANGLER(RTStrPurgeComplimentSet)
 # define RTStrPutCpInternal                             RT_MANGLER(RTStrPutCpInternal)
 # define RTStrReallocTag                                RT_MANGLER(RTStrReallocTag)
@@ -1630 +1631 @@
 # define RTUtf16ToLower                                 RT_MANGLER(RTUtf16ToLower)
 # define RTUtf16ToUpper                                 RT_MANGLER(RTUtf16ToUpper)
+# define RTUtf16PurgeComplementSet                      RT_MANGLER(RTUtf16PurgeComplementSet)
 # define RTUtf16ToUtf8ExTag                             RT_MANGLER(RTUtf16ToUtf8ExTag)
 # define RTUtf16ToUtf8Tag                               RT_MANGLER(RTUtf16ToUtf8Tag)

trunk/include/iprt/string.h

@@ -704 +704 @@
  */
 RTDECL(size_t) RTStrPurgeEncoding(char *psz);
+
+/**
+ * Sanitise 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 case the string will not
+ *          be changed.
+ * @param   psz            The string to sanitise.
+ * @param   puszValidSet   A zero-terminated array of the Unicode code points
+ *                         in the white list.
+ * @param   chReplacement  The ASCII replacement character.
+ */
+RTDECL(ssize_t) RTStrPurgeComplementSet(char *psz, PCRTUNICP puszValidSet, char chReplacement);
 
 /**
@@ -3305 +3320 @@
 
 /**
+ * Sanitise a (valid) UTF-16 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 case the string will not
+ *          be changed.
+ * @param   pwsz           The string to sanitise.
+ * @param   puszValidSet   A zero-terminated array of the Unicode code points
+ *                         in the white list.
+ * @param   chReplacement  The ASCII replacement character.
+ */
+RTDECL(ssize_t) RTUtf16PurgeComplementSet(PRTUTF16 pwsz, PCRTUNICP puszValidSet, char chReplacement);
+
+/**
  * Translate a UTF-16 string into a UTF-8 allocating the result buffer (default
  * tag).