From 73fc3d6a13b89c1dd4f6d6178d64859267096f1b Mon Sep 17 00:00:00 2001
From: Alexander Mai <amai@users.sf.net>
Date: Sun, 13 Apr 2014 09:50:57 +0200
Subject: [PATCH] #5300 - Invalid encoding in XML output (use escape sequence
 string for non-printable characters)

---
 lib/errorlogger.cpp      | 24 +++++++++++++++++++++++-
 test/testerrorlogger.cpp |  8 ++++++++
 2 files changed, 31 insertions(+), 1 deletion(-)

diff --git a/lib/errorlogger.cpp b/lib/errorlogger.cpp
index 48732b850..b5ccd0a74 100644
--- a/lib/errorlogger.cpp
+++ b/lib/errorlogger.cpp
@@ -25,6 +25,7 @@
 #include <tinyxml2.h>
 
 #include <cassert>
+#include <iomanip>
 #include <sstream>
 #include <vector>
 
@@ -211,6 +212,27 @@ std::string ErrorLogger::ErrorMessage::getXMLFooter(int xml_version)
     return (xml_version<=1) ? "</results>" : "    </errors>\n</results>";
 }
 
+// There is no utf-8 support around but the strings should at least be safe for to tinyxml2.
+// See #5300 "Invalid encoding in XML output"
+static std::string fixInvalidChars(const std::string& raw)
+{
+    std::string result;
+    result.reserve(raw.length());
+    std::string::const_iterator from=raw.begin();
+    while (from!=raw.end()) {
+        if (std::isprint(*from)) {
+            result.push_back(*from);
+        } else {
+            std::ostringstream es;
+            // straight cast to (unsigned) doesn't work out.
+            es << '\\' << std::setbase(8) << std::setw(3) << std::setfill('0') << (unsigned)(unsigned char)*from;
+            result += es.str();
+        }
+        ++from;
+    }
+    return result;
+}
+
 std::string ErrorLogger::ErrorMessage::toXML(bool verbose, int version) const
 {
     // The default xml format
@@ -239,7 +261,7 @@ std::string ErrorLogger::ErrorMessage::toXML(bool verbose, int version) const
         printer.PushAttribute("id", _id.c_str());
         printer.PushAttribute("severity", Severity::toString(_severity).c_str());
         printer.PushAttribute("msg", _shortMessage.c_str());
-        printer.PushAttribute("verbose", _verboseMessage.c_str());
+        printer.PushAttribute("verbose", fixInvalidChars(_verboseMessage).c_str());
         if (_inconclusive)
             printer.PushAttribute("inconclusive", "true");
 
diff --git a/test/testerrorlogger.cpp b/test/testerrorlogger.cpp
index 7d9ccbaa2..0b275466a 100644
--- a/test/testerrorlogger.cpp
+++ b/test/testerrorlogger.cpp
@@ -46,6 +46,7 @@ private:
         TEST_CASE(ToVerboseXmlLocations);
         TEST_CASE(ToXmlV2);
         TEST_CASE(ToXmlV2Locations);
+        TEST_CASE(ToXmlV2Encoding);
 
         // Inconclusive results in xml reports..
         TEST_CASE(InconclusiveXml);
@@ -215,6 +216,13 @@ private:
         ASSERT_EQUALS(message, msg.toXML(false, 2));
     }
 
+    void ToXmlV2Encoding() const {
+        std::list<ErrorLogger::ErrorMessage::FileLocation> locs;
+        ErrorMessage msg(locs, Severity::error, "Programming error.\nComparing \"\203\" with \"\003\"", "errorId", false);
+        const std::string message("        <error id=\"errorId\" severity=\"error\" msg=\"Programming error.\" verbose=\"Comparing &quot;\\203&quot; with &quot;\\003&quot;\"/>");
+        ASSERT_EQUALS(message, msg.toXML(false, 2));
+    }
+
     void InconclusiveXml() const {
         // Location
         std::list<ErrorLogger::ErrorMessage::FileLocation> locs(1, fooCpp5);