Do not violate UTF-8 in XML output

According to UTF-8 encoding rules, there are no valid one byte long codepoints which start with a bigh bit set (i.e. 0x80 or higher value). Hence such XML output needs to be escaped, same as a control character.
2024-11-17 19:22:25 +01:00 · 2018-03-12 20:34:13 +00:00 · 2018-03-12 20:34:13 +00:00 · 27fa7218ca
commit 27fa7218ca
parent e11508b48a
7 changed files with 57 additions and 8 deletions
--- a/include/internal/catch_xmlwriter.cpp
+++ b/include/internal/catch_xmlwriter.cpp
@ -45,10 +45,14 @@ namespace Catch {
                default:
                    // Escape control chars - based on contribution by @espenalb in PR #465 and
                    // by @mrpi PR #588
-                    if ( ( c >= 0 && c < '\x09' ) || ( c > '\x0D' && c < '\x20') || c=='\x7F' ) {
+                    // Also take care to avoid violation of UTF-8 encoding rules
+                    if ( ( c >= 0 && c < '\x09' )
+                         || ( c > '\x0D' && c < '\x20')
+                         || ( c & 0x80 )
+                         || c=='\x7F') {
                        // see http://stackoverflow.com/questions/404107/why-are-control-characters-illegal-in-xml-1-0
                        os << "\\x" << std::uppercase << std::hex << std::setfill('0') << std::setw(2)
-                           << static_cast<int>( c );
+                           << static_cast<int>( static_cast<unsigned char>(c) );
                    }
                    else
                        os << c;
--- a/projects/SelfTest/Baselines/compact.sw.approved.txt
+++ b/projects/SelfTest/Baselines/compact.sw.approved.txt
@ -905,6 +905,8 @@ Xml.tests.cpp:<line number>: passed: encode( stringWithQuotes, Catch::XmlEncode:
 "don't &quot;quote&quot; me on that"
 Xml.tests.cpp:<line number>: passed: encode( "[/x01]" ) == "[//x01]" for: "[/x01]" == "[/x01]"
 Xml.tests.cpp:<line number>: passed: encode( "[/x7F]" ) == "[//x7F]" for: "[/x7F]" == "[/x7F]"
+Xml.tests.cpp:<line number>: passed: encode( "[/x80]" ) == "[//x80]" for: "[/x80]" == "[/x80]"
+Xml.tests.cpp:<line number>: passed: encode( "[/xFF]" ) == "[//xFF]" for: "[/xFF]" == "[/xFF]"
 ToStringVector.tests.cpp:<line number>: passed: Catch::Detail::stringify( empty ) == "{  }" for: "{  }" == "{  }"
 ToStringVector.tests.cpp:<line number>: passed: Catch::Detail::stringify( oneValue ) == "{ 42 }" for: "{ 42 }" == "{ 42 }"
 ToStringVector.tests.cpp:<line number>: passed: Catch::Detail::stringify( twoValues ) == "{ 42, 250 }" for: "{ 42, 250 }" == "{ 42, 250 }"
--- a/projects/SelfTest/Baselines/console.std.approved.txt
+++ b/projects/SelfTest/Baselines/console.std.approved.txt
@ -1085,5 +1085,5 @@ due to unexpected exception with message:

 ===============================================================================
 test cases:  202 | 149 passed |  49 failed |  4 failed as expected
-assertions: 1015 | 887 passed | 107 failed | 21 failed as expected
+assertions: 1017 | 889 passed | 107 failed | 21 failed as expected

--- a/projects/SelfTest/Baselines/console.sw.approved.txt
+++ b/projects/SelfTest/Baselines/console.sw.approved.txt
@ -7112,6 +7112,25 @@ PASSED:
 with expansion:
  "[\x7F]" == "[\x7F]"

+-------------------------------------------------------------------------------
+XmlEncode
+  string with high bit set (x80 and xFF)
+-------------------------------------------------------------------------------
+Xml.tests.cpp:<line number>
+...............................................................................
+
+Xml.tests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "[\x80]" ) == "[\\x80]" )
+with expansion:
+  "[\x80]" == "[\x80]"
+
+Xml.tests.cpp:<line number>:
+PASSED:
+  REQUIRE( encode( "[\xFF]" ) == "[\\xFF]" )
+with expansion:
+  "[\xFF]" == "[\xFF]"
+
 -------------------------------------------------------------------------------
 array<int, N> -> toString
 -------------------------------------------------------------------------------
@ -8599,5 +8618,5 @@ PASSED:

 ===============================================================================
 test cases:  202 | 136 passed |  62 failed |  4 failed as expected
-assertions: 1029 | 887 passed | 121 failed | 21 failed as expected
+assertions: 1031 | 889 passed | 121 failed | 21 failed as expected

--- a/projects/SelfTest/Baselines/junit.sw.approved.txt
+++ b/projects/SelfTest/Baselines/junit.sw.approved.txt
@ -1,7 +1,7 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <testsuitesloose text artifact
 >
-  <testsuite name="<exe-name>" errors="17" failures="105" tests="1030" hostname="tbd" time="{duration}" timestamp="{iso8601-timestamp}">
+  <testsuite name="<exe-name>" errors="17" failures="105" tests="1032" hostname="tbd" time="{duration}" timestamp="{iso8601-timestamp}">
    <testcase classname="<exe-name>.global" name="# A test name that starts with a #" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="#1005: Comparing pointer to int and long (NULL can be either on various systems)" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="#1027" time="{duration}"/>
@ -706,6 +706,7 @@ Exception.tests.cpp:<line number>
    <testcase classname="<exe-name>.global" name="XmlEncode/string with quotes" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="XmlEncode/string with control char (1)" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="XmlEncode/string with control char (x7F)" time="{duration}"/>
+    <testcase classname="<exe-name>.global" name="XmlEncode/string with high bit set (x80 and xFF)" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="array&lt;int, N> -> toString" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="atomic if" time="{duration}"/>
    <testcase classname="<exe-name>.global" name="boolean member" time="{duration}"/>
--- a/projects/SelfTest/Baselines/xml.sw.approved.txt
+++ b/projects/SelfTest/Baselines/xml.sw.approved.txt
@ -7992,6 +7992,25 @@ Message from section two
        </Expression>
        <OverallResults successes="1" failures="0" expectedFailures="0"/>
      </Section>
+      <Section name="string with high bit set (x80 and xFF)" filename="projects/<exe-name>/IntrospectiveTests/Xml.tests.cpp" >
+        <Expression success="true" type="REQUIRE" filename="projects/<exe-name>/IntrospectiveTests/Xml.tests.cpp" >
+          <Original>
+            encode( "[\x80]" ) == "[\\x80]"
+          </Original>
+          <Expanded>
+            "[\x80]" == "[\x80]"
+          </Expanded>
+        </Expression>
+        <Expression success="true" type="REQUIRE" filename="projects/<exe-name>/IntrospectiveTests/Xml.tests.cpp" >
+          <Original>
+            encode( "[\xFF]" ) == "[\\xFF]"
+          </Original>
+          <Expanded>
+            "[\xFF]" == "[\xFF]"
+          </Expanded>
+        </Expression>
+        <OverallResults successes="2" failures="0" expectedFailures="0"/>
+      </Section>
      <OverallResult success="true"/>
    </TestCase>
    <TestCase name="array&lt;int, N> -> toString" tags="[array][containers][toString]" filename="projects/<exe-name>/UsageTests/ToStringVector.tests.cpp" >
@ -9469,7 +9488,7 @@ loose text artifact
      </Section>
      <OverallResult success="true"/>
    </TestCase>
-    <OverallResults successes="887" failures="122" expectedFailures="21"/>
+    <OverallResults successes="889" failures="122" expectedFailures="21"/>
  </Group>
-  <OverallResults successes="887" failures="121" expectedFailures="21"/>
+  <OverallResults successes="889" failures="121" expectedFailures="21"/>
 </Catch>
--- a/projects/SelfTest/IntrospectiveTests/Xml.tests.cpp
+++ b/projects/SelfTest/IntrospectiveTests/Xml.tests.cpp
@ -38,4 +38,8 @@ TEST_CASE( "XmlEncode" ) {
    SECTION( "string with control char (x7F)" ) {
        REQUIRE( encode( "[\x7F]" ) == "[\\x7F]" );
    }
-}
+    SECTION( "string with high bit set (x80 and xFF)" ) {
+        REQUIRE( encode( "[\x80]" ) == "[\\x80]" );
+        REQUIRE( encode( "[\xFF]" ) == "[\\xFF]" );
+    }
+}