Fixed Xml encoding

- don't encode apostrophes
- only encode quotes in attributes
- encode control characters (as in PR #465)

include/internal/catch_xmlwriter.hpp

@@ -14,9 +14,65 @@
 #include <sstream>
 #include <string>
 #include <vector>
+#include <iomanip>
 
 namespace Catch {
 
+    class XmlEncode {
+    public:
+        enum ForWhat { ForTextNodes, ForAttributes };
+        
+        XmlEncode( std::string const& str, ForWhat forWhat = ForTextNodes )
+        :   m_str( str ),
+            m_forWhat( forWhat )
+        {}
+        
+        void encodeTo( std::ostream& os ) const {
+
+            // Apostrophe escaping not necessary if we always use " to write attributes
+            // (see: http://www.w3.org/TR/xml/#syntax)
+            
+            for( std::size_t i = 0; i < m_str.size(); ++ i ) {
+                char c = m_str[i];
+                switch( c ) {
+                    case '<':   os << "&lt;"; break;
+                    case '&':   os << "&amp;"; break;
+                    
+                    case '>':
+                        // See: http://www.w3.org/TR/xml/#syntax
+                        if( i > 2 && m_str[i-1] == ']' && m_str[i-2] == ']' )
+                            os << "&gt;";
+                        else
+                            os << c;
+                        break;
+                        
+                    case '\"':
+                        if( m_forWhat == ForAttributes )
+                            os << "&quot;";
+                        else
+                            os << c;
+                        break;
+                        
+                    default:
+                        // Escape control chars - based on contribution by @espenalb in PR #465
+                        if ( ( c < '\x09' ) || ( c > '\x0D' && c < '\x20') || c=='\x7F' )
+                            os << "&#x" << std::uppercase << std::hex << static_cast<int>( c );
+                        else
+                            os << c;
+                }
+            }
+        }
+
+        friend std::ostream& operator << ( std::ostream& os, XmlEncode const& xmlEncode ) {
+            xmlEncode.encodeTo( os );
+            return os;
+        }
+        
+    private:
+        std::string m_str;
+        ForWhat m_forWhat;
+    };
+    
     class XmlWriter {
     public:
 
@@ -99,11 +155,8 @@ namespace Catch {
         }
 
         XmlWriter& writeAttribute( std::string const& name, std::string const& attribute ) {
-            if( !name.empty() && !attribute.empty() ) {
+            if( !name.empty() && !attribute.empty() )
-                stream() << " " << name << "=\"";
+                stream() << " " << name << "=\"" << XmlEncode( attribute, XmlEncode::ForAttributes ) << "\"";
-                writeEncodedText( attribute );
-                stream() << "\"";
-            }
             return *this;
         }
 
@@ -114,9 +167,9 @@ namespace Catch {
 
         template<typename T>
         XmlWriter& writeAttribute( std::string const& name, T const& attribute ) {
-            if( !name.empty() )
+            std::ostringstream oss;
-                stream() << " " << name << "=\"" << attribute << "\"";
+            oss << attribute;
-            return *this;
+            return writeAttribute( name, oss.str() );
         }
 
         XmlWriter& writeText( std::string const& text, bool indent = true ) {
@@ -125,7 +178,7 @@ namespace Catch {
                 ensureTagClosed();
                 if( tagWasOpen && indent )
                     stream() << m_indent;
-                writeEncodedText( text );
+                stream() << XmlEncode( text );
                 m_needsNewline = true;
             }
             return *this;
@@ -170,30 +223,6 @@ namespace Catch {
             }
         }
 
-        void writeEncodedText( std::string const& text ) {
-            static const char* charsToEncode = "<&\"";
-            std::string mtext = text;
-            std::string::size_type pos = mtext.find_first_of( charsToEncode );
-            while( pos != std::string::npos ) {
-                stream() << mtext.substr( 0, pos );
-
-                switch( mtext[pos] ) {
-                    case '<':
-                        stream() << "&lt;";
-                        break;
-                    case '&':
-                        stream() << "&amp;";
-                        break;
-                    case '\"':
-                        stream() << "&quot;";
-                        break;
-                }
-                mtext = mtext.substr( pos+1 );
-                pos = mtext.find_first_of( charsToEncode );
-            }
-            stream() << mtext;
-        }
-
         bool m_tagIsOpen;
         bool m_needsNewline;
         std::vector<std::string> m_tags;

projects/SelfTest/Baselines/console.std.approved.txt

@@ -797,6 +797,6 @@ with expansion:
   "first" == "second"
 
 ===============================================================================
-test cases: 157 | 117 passed | 39 failed |  1 failed as expected
+test cases: 158 | 118 passed | 39 failed |  1 failed as expected
-assertions: 773 | 680 passed | 80 failed | 13 failed as expected
+assertions: 783 | 690 passed | 80 failed | 13 failed as expected
 

projects/SelfTest/Baselines/console.sw.approved.txt

@@ -3762,6 +3762,128 @@ PASSED:
 with expansion:
   ""wide load"" == ""wide load""
 
+-------------------------------------------------------------------------------
+XmlEncode
+  normal string
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "normal string" ) == "normal string" )
+with expansion:
+  "normal string" == "normal string"
+
+-------------------------------------------------------------------------------
+XmlEncode
+  empty string
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "" ) == "" )
+with expansion:
+  "" == ""
+
+-------------------------------------------------------------------------------
+XmlEncode
+  string with ampersand
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "smith & jones" ) == "smith &amp; jones" )
+with expansion:
+  "smith &amp; jones" == "smith &amp; jones"
+
+-------------------------------------------------------------------------------
+XmlEncode
+  string with less-than
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "smith < jones" ) == "smith &lt; jones" )
+with expansion:
+  "smith &lt; jones" == "smith &lt; jones"
+
+-------------------------------------------------------------------------------
+XmlEncode
+  string with greater-than
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "smith > jones" ) == "smith > jones" )
+with expansion:
+  "smith > jones" == "smith > jones"
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "smith ]]> jones" ) == "smith ]]&gt; jones" )
+with expansion:
+  "smith ]]&gt; jones"
+  ==
+  "smith ]]&gt; jones"
+
+-------------------------------------------------------------------------------
+XmlEncode
+  string with quotes
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( stringWithQuotes ) == stringWithQuotes )
+with expansion:
+  "don't "quote" me on that"
+  ==
+  "don't "quote" me on that"
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( stringWithQuotes, Catch::XmlEncode::ForAttributes ) == "don't &quot;quote&quot; me on that" )
+with expansion:
+  "don't &quot;quote&quot; me on that"
+  ==
+  "don't &quot;quote&quot; me on that"
+
+-------------------------------------------------------------------------------
+XmlEncode
+  string with control char (1)
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "[\x01]" ) == "[&#x1]" )
+with expansion:
+  "[&#x1]" == "[&#x1]"
+
+-------------------------------------------------------------------------------
+XmlEncode
+  string with control char (x7F)
+-------------------------------------------------------------------------------
+MiscTests.cpp:<line number>
+...............................................................................
+
+MiscTests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "[\x7F]" ) == "[&#x7F]" )
+with expansion:
+  "[&#x7F]" == "[&#x7F]"
+
 -------------------------------------------------------------------------------
 Process can be configured on command line
   default - no arguments
@@ -8004,6 +8126,6 @@ with expansion:
   true
 
 ===============================================================================
-test cases: 157 | 101 passed |  55 failed |  1 failed as expected
+test cases: 158 | 102 passed |  55 failed |  1 failed as expected
-assertions: 793 | 680 passed | 100 failed | 13 failed as expected
+assertions: 803 | 690 passed | 100 failed | 13 failed as expected
 

projects/SelfTest/Baselines/junit.sw.approved.txt

@@ -1,5 +1,5 @@
 <testsuites>
-  <testsuite name="all tests" errors="12" failures="88" tests="793" hostname="tbd" time="{duration}" timestamp="tbd">
+  <testsuite name="all tests" errors="12" failures="88" tests="803" hostname="tbd" time="{duration}" timestamp="tbd">
     <testcase classname="global" name="toString(enum)" time="{duration}"/>
     <testcase classname="global" name="toString(enum w/operator&lt;&lt;)" time="{duration}"/>
     <testcase classname="global" name="toString(enum class)" time="{duration}"/>
@@ -463,6 +463,14 @@ MiscTests.cpp:<line number>
     <testcase classname="global" name="toString on const wchar_t pointer returns the string contents" time="{duration}"/>
     <testcase classname="global" name="toString on wchar_t const pointer returns the string contents" time="{duration}"/>
     <testcase classname="global" name="toString on wchar_t returns the string contents" time="{duration}"/>
+    <testcase classname="XmlEncode" name="normal string" time="{duration}"/>
+    <testcase classname="XmlEncode" name="empty string" time="{duration}"/>
+    <testcase classname="XmlEncode" name="string with ampersand" time="{duration}"/>
+    <testcase classname="XmlEncode" name="string with less-than" time="{duration}"/>
+    <testcase classname="XmlEncode" name="string with greater-than" time="{duration}"/>
+    <testcase classname="XmlEncode" name="string with quotes" time="{duration}"/>
+    <testcase classname="XmlEncode" name="string with control char (1)" time="{duration}"/>
+    <testcase classname="XmlEncode" name="string with control char (x7F)" time="{duration}"/>
     <testcase classname="Process can be configured on command line" name="default - no arguments" time="{duration}"/>
     <testcase classname="Process can be configured on command line" name="test lists/1 test" time="{duration}"/>
     <testcase classname="Process can be configured on command line" name="test lists/Specify one test case exclusion using exclude:" time="{duration}"/>

projects/SelfTest/MiscTests.cpp

@@ -7,6 +7,7 @@
  */
 
 #include "catch.hpp"
+#include "catch_xmlwriter.hpp"
 
 #include <iostream>
 
@@ -381,6 +382,42 @@ TEST_CASE( "toString on wchar_t returns the string contents", "[toString]" ) {
 	CHECK( result == "\"wide load\"" );
 }
 
+inline std::string encode( std::string const& str, Catch::XmlEncode::ForWhat forWhat = Catch::XmlEncode::ForTextNodes ) {
+    std::ostringstream oss;
+    oss << Catch::XmlEncode( str, forWhat );
+    return oss.str();
+}
+
+TEST_CASE( "XmlEncode" ) {
+    SECTION( "normal string" ) {
+        REQUIRE( encode( "normal string" ) == "normal string" );
+    }
+    SECTION( "empty string" ) {
+        REQUIRE( encode( "" ) == "" );
+    }
+    SECTION( "string with ampersand" ) {
+        REQUIRE( encode( "smith & jones" ) == "smith &amp; jones" );
+    }
+    SECTION( "string with less-than" ) {
+        REQUIRE( encode( "smith < jones" ) == "smith &lt; jones" );
+    }
+    SECTION( "string with greater-than" ) {
+        REQUIRE( encode( "smith > jones" ) == "smith > jones" );
+        REQUIRE( encode( "smith ]]> jones" ) == "smith ]]&gt; jones" );
+    }
+    SECTION( "string with quotes" ) {
+        std::string stringWithQuotes = "don't \"quote\" me on that";
+        REQUIRE( encode( stringWithQuotes ) == stringWithQuotes );
+        REQUIRE( encode( stringWithQuotes, Catch::XmlEncode::ForAttributes ) == "don't &quot;quote&quot; me on that" );
+    }
+    SECTION( "string with control char (1)" ) {
+        REQUIRE( encode( "[\x01]" ) == "[&#x1]" );
+    }
+    SECTION( "string with control char (x7F)" ) {
+        REQUIRE( encode( "[\x7F]" ) == "[&#x7F]" );
+    }
+}
+
 //TEST_CASE( "Divide by Zero signal handler", "[.][sig]" ) {
 //    int i = 0;
 //    int x = 10/i; // This should cause the signal to fire