codecvt: Re-implement do_out and do_unshift

The previous implementation assumed that only one byte would be given
in the `from` buffer by the caller at a time.  This may be true for
MSVC but is not for the GNU library on Windows.  Re-implement these
methods to handle more than one byte per call.

Also simplify the state management by keeping all state between calls
directly in the `mbstate_t` argument instead of using it to index our
own heap-allocated state.

Fixes: #16893

1 files changed, 18 insertions, 10 deletions

diff --git a/Source/cm_codecvt.hxx b/Source/cm_codecvt.hxx
index b9b52eccf8..30c6d5407d 100644
--- a/Source/cm_codecvt.hxx
+++ b/Source/cm_codecvt.hxx
@@ -6,7 +6,6 @@
 #include "cmConfigure.h"
 
 #include <locale>
-#include <vector>
 #include <wchar.h>
 
 class codecvt : public std::codecvt<char, char, mbstate_t>
@@ -35,21 +34,30 @@ protected:
   int do_encoding() const throw() CM_OVERRIDE;
 
 private:
-  typedef struct
+  // The mbstate_t argument to do_out and do_unshift is responsible
+  // for storing state between calls.  We cannot control the type
+  // since we want to imbue on standard streams.  However, we do
+  // know that it is a trivial type.  Define our own type to overlay
+  // on it safely with no alignment requirements.
+  struct State
   {
-    bool used;
-    unsigned char totalBytes;
-    unsigned char bytesLeft;
-    char bytes[4];
-  } State;
+    // Buffer bytes we have consumed from a partial codepoint.
+    char partial[3];
 
-  unsigned int findStateId() const;
+    // Number of bytes we have buffered from a partial codepoint.
+    unsigned char buffered : 4;
+
+    // Size of the current codepoint in bytes.
+    unsigned char size : 4;
+  };
 
   bool m_noconv;
-  mutable std::vector<State> m_states;
-  mutable unsigned int m_lastState;
 #if defined(_WIN32)
   unsigned int m_codepage;
+  result Decode(mbstate_t& state, int need, const char*& from_next,
+                char*& to_next, char* to_end) const;
+  result DecodePartial(mbstate_t& state, char*& to_next, char* to_end) const;
+  void BufferPartial(mbstate_t& state, int need, const char*& from_next) const;
 #endif
 
 #endif