Update utfcpp to v2.3.6

Fixes #2657

Incorporates the following utfcpp patches:

1. Sass addition of `retreat`.
   nemtrif/utfcpp#20

2. Fix for `replace_invalid` throwing on incomplete sequence at the end
   of the input.
   nemtrif/utfcpp#21

Author: glebm

src/utf8/checked.h

@@ -1,4 +1,4 @@
-// Copyright 2006 Nemanja Trifunovic
+// Copyright 2006-2016 Nemanja Trifunovic
 
 /*
 Permission is hereby granted, free of charge, to any person or organization
@@ -41,7 +41,7 @@ namespace utf8
     class invalid_code_point : public exception {
         uint32_t cp;
     public:
-        invalid_code_point(uint32_t cp) : cp(cp) {}
+        invalid_code_point(uint32_t codepoint) : cp(codepoint) {}
         virtual const char* what() const throw() { return "Invalid code point"; }
         uint32_t code_point() const {return cp;}
     };
@@ -107,7 +107,9 @@ namespace utf8
                         *out++ = *it;
                     break;
                 case internal::NOT_ENOUGH_ROOM:
-                    throw not_enough_room();
+                    out = utf8::append (replacement, out);
+                    start = end;
+                    break;
                 case internal::INVALID_LEAD:
                     out = utf8::append (replacement, out);
                     ++start;
@@ -194,10 +196,10 @@ namespace utf8
     }
 
     template <typename octet_iterator, typename distance_type>
-    void retreat (octet_iterator& it, distance_type n, octet_iterator start)
+    void retreat (octet_iterator& it, distance_type n, octet_iterator end)
     {
         for (distance_type i = 0; i < n; ++i)
-            utf8::prior(it, start);
+            utf8::prior(it, end);
     }
 
     template <typename octet_iterator>
@@ -240,7 +242,7 @@ namespace utf8
     template <typename u16bit_iterator, typename octet_iterator>
     u16bit_iterator utf8to16 (octet_iterator start, octet_iterator end, u16bit_iterator result)
     {
-        while (start != end) {
+        while (start < end) {
             uint32_t cp = utf8::next(start, end);
             if (cp > 0xffff) { //make a surrogate pair
                 *result++ = static_cast<uint16_t>((cp >> 10)   + internal::LEAD_OFFSET);
@@ -264,7 +266,7 @@ namespace utf8
     template <typename octet_iterator, typename u32bit_iterator>
     u32bit_iterator utf8to32 (octet_iterator start, octet_iterator end, u32bit_iterator result)
     {
-        while (start != end)
+        while (start < end)
             (*result++) = utf8::next(start, end);
 
         return result;
@@ -279,9 +281,9 @@ namespace utf8
       public:
       iterator () {}
       explicit iterator (const octet_iterator& octet_it,
-                         const octet_iterator& range_start,
-                         const octet_iterator& range_end) :
-               it(octet_it), range_start(range_start), range_end(range_end)
+                         const octet_iterator& rangestart,
+                         const octet_iterator& rangeend) :
+               it(octet_it), range_start(rangestart), range_end(rangeend)
       {
           if (it < range_start || it > range_end)
               throw std::out_of_range("Invalid utf-8 iterator position");

src/utf8/core.h

@@ -222,6 +222,9 @@ namespace internal
     template <typename octet_iterator>
     utf_error validate_next(octet_iterator& it, octet_iterator end, uint32_t& code_point)
     {
+        if (it == end)
+            return NOT_ENOUGH_ROOM;
+
         // Save the original value of it so we can go back in case of failure
         // Of course, it does not make much sense with i.e. stream iterators
         octet_iterator original_it = it;