CTest: Do not munge UTF-8 output in XML files

CTest filters the output from tools and tests to ensure that the XML
build/test result documents it generates have valid characters.
Previously we just converted all non-ASCII bytes into XML-escaped
Unicode characters of the corresponding index.  This does not preserve
tool output encoded in UTF-8.

We now assume UTF-8 output from tools and implement decoding as
specified in RFC 3629.  Valid characters are preserved, possibly with
XML escaping.  Invalid byte sequences and characters are converted to
human-readable hex values with distinguishing tags.  See issue #10003.

Source/CMakeLists.txt

@@ -246,6 +246,9 @@ SET(SRCS
   cmake.h
   cmakewizard.cxx
   cmakewizard.h
+
+  cm_utf8.h
+  cm_utf8.c
   )
 
 # Kdevelop only works on UNIX and not windows

Source/cmXMLSafe.cxx

@@ -11,6 +11,8 @@
 ============================================================================*/
 #include "cmXMLSafe.h"
 
+#include "cm_utf8.h"
+
 #include <cmsys/ios/iostream>
 #include <cmsys/ios/sstream>
 
@@ -53,44 +55,47 @@ cmsys_ios::ostream& operator<<(cmsys_ios::ostream& os, cmXMLSafe const& self)
 {
   char const* first = self.Data;
   char const* last = self.Data + self.Size;
-  for(char const* ci = first; ci != last; ++ci)
+  while(first != last)
     {
-    unsigned char c = static_cast<unsigned char>(*ci);
-    switch(c)
+    unsigned int ch;
+    if(const char* next = cm_utf8_decode_character(first, last, &ch))
       {
-      case '&': os << "&amp;"; break;
-      case '<': os << "&lt;"; break;
-      case '>': os << "&gt;"; break;
-      case '"': os << (self.DoQuotes? "&quot;" : "\""); break;
-      case '\'': os << (self.DoQuotes? "&apos;" : "'"); break;
-      case '\t': os << "\t"; break;
-      case '\n': os << "\n"; break;
-      case '\r': break; // Ignore CR
-      default:
-        if(c >= 0x20 && c <= 0x7f)
-          {
-          os.put(static_cast<char>(c));
-          }
-        else
+      // http://www.w3.org/TR/REC-xml/#NT-Char
+      if((ch >= 0x20 && ch <= 0xD7FF) ||
+         (ch >= 0xE000 && ch <= 0xFFFD) ||
+         (ch >= 0x10000 && ch <= 0x10FFFF) ||
+          ch == 0x9 || ch == 0xA || ch == 0xD)
+        {
+        switch(ch)
           {
-          // TODO: More complete treatment of program output character
-          // encoding.  Instead of escaping these bytes, we should
-          // handle the current locale and its encoding.
-          char buf[16];
-          // http://www.w3.org/TR/REC-xml/#NT-Char
-          if(c >= 0x80)
-            {
-            sprintf(buf, "&#x%hx;", static_cast<unsigned short>(c));
-            }
-          else
-            {
-            // We cannot use "&#x%hx;" here because this value is not
-            // valid in XML.  Instead use a human-readable hex value.
-            sprintf(buf, "&lt;0x%hx&gt;", static_cast<unsigned short>(c));
-            }
-          os << buf;
+          // Escape XML control characters.
+          case '&': os << "&amp;"; break;
+          case '<': os << "&lt;"; break;
+          case '>': os << "&gt;"; break;
+          case '"': os << (self.DoQuotes? "&quot;" : "\""); break;
+          case '\'': os << (self.DoQuotes? "&apos;" : "'"); break;
+          case '\r': break; // Ignore CR
+          // Print the UTF-8 character.
+          default: os.write(first, next-first); break;
           }
-        break;
+        }
+      else
+        {
+        // Use a human-readable hex value for this invalid character.
+        char buf[16];
+        sprintf(buf, "%X", ch);
+        os << "[NON-XML-CHAR-0x" << buf << "]";
+        }
+
+      first = next;
+      }
+    else
+      {
+      ch = static_cast<unsigned char>(*first++);
+      // Use a human-readable hex value for this invalid byte.
+      char buf[16];
+      sprintf(buf, "%X", ch);
+      os << "[NON-UTF-8-BYTE-0x" << buf << "]";
       }
     }
   return os;

Source/cm_utf8.c

@@ -0,0 +1,84 @@
+/*============================================================================
+  CMake - Cross Platform Makefile Generator
+  Copyright 2000-2009 Kitware, Inc., Insight Software Consortium
+
+  Distributed under the OSI-approved BSD License (the "License");
+  see accompanying file Copyright.txt for details.
+
+  This software is distributed WITHOUT ANY WARRANTY; without even the
+  implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+  See the License for more information.
+============================================================================*/
+#include "cm_utf8.h"
+
+/*
+  RFC 3629
+  07-bit: 0xxxxxxx
+  11-bit: 110xxxxx 10xxxxxx
+  16-bit: 1110xxxx 10xxxxxx 10xxxxxx
+  21-bit: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+
+  Pre-RFC Compatibility
+  26-bit: 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
+  31-bit: 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
+*/
+
+/* Number of leading ones before a zero in the byte.  */
+static unsigned char const cm_utf8_ones[256] = {
+  0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+  0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+  0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+  0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
+  1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
+  1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
+  2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
+  3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6,7,8
+};
+
+/* Mask away control bits from bytes with n leading ones.  */
+static unsigned char const cm_utf8_mask[7] = {
+  0xEF, 0x3F, 0x1F, 0x0F, 0x07, 0x03, 0x01
+};
+
+/* Minimum allowed value when first byte has n leading ones.  */
+static unsigned int const cm_utf8_min[7] = {
+  0, 0, 1u<<7, 1u<<11, 1u<<16, 1u<<21, 1u<<26 /*, 1u<<31 */
+};
+
+/*--------------------------------------------------------------------------*/
+const char* cm_utf8_decode_character(const char* first, const char* last,
+                                     unsigned int* pc)
+{
+  /* Count leading ones in the first byte.  */
+  unsigned char c = *first++;
+  unsigned char const ones = cm_utf8_ones[c];
+  switch(ones)
+    {
+    case 0: *pc = c; return first;    /* One-byte character.  */
+    case 1: case 7: case 8: return 0; /* Invalid leading byte.  */
+    default: break;
+    }
+
+  /* Extract bits from this multi-byte character.  */
+  {
+  unsigned int uc = c & cm_utf8_mask[ones];
+  unsigned char left;
+  for(left = ones-1; left && first != last; --left)
+    {
+    c = *first++;
+    if(cm_utf8_ones[c] != 1)
+      {
+      return 0;
+      }
+    uc = (uc << 6) | (c & cm_utf8_mask[1]);
+    }
+
+  if(left > 0 || uc < cm_utf8_min[ones])
+    {
+    return 0;
+    }
+
+  *pc = uc;
+  return first;
+  }
+}

Source/cm_utf8.h

@@ -0,0 +1,29 @@
+/*============================================================================
+  CMake - Cross Platform Makefile Generator
+  Copyright 2000-2009 Kitware, Inc., Insight Software Consortium
+
+  Distributed under the OSI-approved BSD License (the "License");
+  see accompanying file Copyright.txt for details.
+
+  This software is distributed WITHOUT ANY WARRANTY; without even the
+  implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+  See the License for more information.
+============================================================================*/
+#ifndef cm_utf8_h
+#define cm_utf8_h
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/** Decode one UTF-8 character from the input byte range.  On success,
+    stores the unicode character number in *pc and returns the first
+    position not extracted.  On failure, returns 0.  */
+const char* cm_utf8_decode_character(const char* first, const char* last,
+                                     unsigned int* pc);
+
+#ifdef __cplusplus
+} /* extern "C" */
+#endif
+
+#endif