Noissue - [emscripten] Replace sprintf() usage by safe alternatives

compiled/filter/ActiveFilter.cpp

Index: compiled/filter/ActiveFilter.cpp
===================================================================
--- a/compiled/filter/ActiveFilter.cpp
+++ b/compiled/filter/ActiveFilter.cpp
@@ -21,18 +21,19 @@
 #include "../StringScanner.h"
 
 namespace
 {
   const DependentString DEFAULT_DOMAIN(u""_str);
 
   OwnedString to_string(unsigned int i)
   {
-    char buffer[11];
-    int len = sprintf(buffer, "%u", i);
+    const std::size_t BUFFER_SIZE = 11;
+    char buffer[BUFFER_SIZE];
+    int len = snprintf(buffer, BUFFER_SIZE, "%u", i);
 
     OwnedString result(len);

[hub, 2017/04/06 10:31:23]

From `man snprintf`
So we shouldn't use len to initialise result that way.
Nor as a counter for iterating through the buffer.

[Wladimir Palant, 2017/04/06 15:14:40]

We have the same issue in the subscription classes. I think we should just avoid
unnecessary buffers by writing our own conversion function that can deal with
wide chars.

     for (String::size_type i = 0; i < len; i++)
       result[i] = buffer[i];
     return result;
   }
 }