fix #606: HTMLForm constructor read application/x-www-form-urlencoded UTF-8 request body first parameter with BOM in name

Foundation/include/Poco/UTF8String.h

@@ -35,6 +35,9 @@ struct Foundation_API UTF8
 	///
 	/// toUpper(), toUpperInPlace(), toLower() and toLowerInPlace() provide
 	/// Unicode-based character case transformation for UTF-8 encoded strings.
+	///
+	/// removeBOM() removes the UTF-8 Byte Order Mark sequence (0xEF, 0xBB, 0xBF)
+	/// from the beginning of the given string, if it's there.
 {
 	static int icompare(const std::string& str, std::string::size_type pos, std::string::size_type n, std::string::const_iterator it2, std::string::const_iterator end2);
 	static int icompare(const std::string& str1, const std::string& str2);
@@ -51,6 +54,10 @@ struct Foundation_API UTF8
 	static std::string& toUpperInPlace(std::string& str);
 	static std::string toLower(const std::string& str);
 	static std::string& toLowerInPlace(std::string& str);
+	
+	static void removeBOM(std::string& str);
+		/// Remove the UTF-8 Byte Order Mark sequence (0xEF, 0xBB, 0xBF)
+		/// from the beginning of the string, if it's there.
 };
 
 

Foundation/src/UTF8String.cpp

@@ -160,4 +160,16 @@ std::string& UTF8::toLowerInPlace(std::string& str)
 }
 
 
+void UTF8::removeBOM(std::string& str)
+{
+	if (str.size() >= 3 
+		&& static_cast<unsigned char>(str[0]) == 0xEF 
+		&& static_cast<unsigned char>(str[1]) == 0xBB 
+		&& static_cast<unsigned char>(str[2]) == 0xBF)
+	{
+		str.erase(0, 3);
+	}
+}
+
+
 } // namespace Poco

Net/src/HTMLForm.cpp

@@ -28,6 +28,7 @@
 #include "Poco/URI.h"
 #include "Poco/String.h"
 #include "Poco/CountingStream.h"
+#include "Poco/UTF8String.h"
 #include <sstream>
 
 
@@ -42,20 +43,31 @@ namespace Poco {
 namespace Net {
 
 
-const std::string HTMLForm::ENCODING_URL				= "application/x-www-form-urlencoded";
-const std::string HTMLForm::ENCODING_MULTIPART			= "multipart/form-data";
-const int         HTMLForm::UNKNOWN_CONTENT_LENGTH		= -1;
+const std::string HTMLForm::ENCODING_URL           = "application/x-www-form-urlencoded";
+const std::string HTMLForm::ENCODING_MULTIPART     = "multipart/form-data";
+const int         HTMLForm::UNKNOWN_CONTENT_LENGTH = -1;
 
 
-class HTMLFormCountingOutputStream : public CountingOutputStream
+class HTMLFormCountingOutputStream: public CountingOutputStream
 {
 public:
-	HTMLFormCountingOutputStream() : _isvalid(true)  {}
+	HTMLFormCountingOutputStream():
+		_valid(true)
+	{
+	}
+
+	bool isValid() const 
+	{
+		return _valid;
+	}
+	
+	void setValid(bool v)
+	{
+		_valid = v;
+	}
 
-	bool getIsValid() const { return _isvalid; }
-	void setIsValid(bool v) { _isvalid = v; }
 private:
-	bool _isvalid;
+	bool _valid;
 };
 
 
@@ -238,7 +250,7 @@ std::streamsize HTMLForm::calculateContentLength()
 
 	HTMLFormCountingOutputStream c;
 	write(c);
-	if (c.getIsValid())
+	if (c.isValid())
 		return c.chars();
 	else
 		return UNKNOWN_CONTENT_LENGTH;
@@ -274,6 +286,7 @@ void HTMLForm::readUrl(std::istream& istr)
 
 	int fields = 0;
 	int ch = istr.get();
+	bool isFirst = true;
 	while (ch != eof)
 	{
 		if (_fieldLimit > 0 && fields == _fieldLimit)
@@ -296,6 +309,11 @@ void HTMLForm::readUrl(std::istream& istr)
 				ch = istr.get();
 			}
 		}
+		// remove UTF-8 byte order mark from first name, if present
+		if (isFirst)
+		{
+			UTF8::removeBOM(name);
+		}
 		std::string decodedName;
 		std::string decodedValue;
 		URI::decode(name, decodedName);
@@ -303,6 +321,7 @@ void HTMLForm::readUrl(std::istream& istr)
 		add(decodedName, decodedValue);
 		++fields;
 		if (ch == '&') ch = istr.get();
+		isFirst = false;
 	}
 }
 
@@ -402,7 +421,7 @@ void HTMLForm::writeMultipart(std::ostream& ostr)
 			if (partlen != PartSource::UNKNOWN_CONTENT_LENGTH)
 				costr->addChars(static_cast<int>(partlen));
 			else
-				costr->setIsValid(false);
+				costr->setValid(false);
 		}
 		else
 			StreamCopier::copyStream(ita->pSource->stream(), ostr);

Net/testsuite/src/HTMLFormTest.cpp

@@ -191,6 +191,20 @@ void HTMLFormTest::testReadUrlPUT()
 }
 
 
+void HTMLFormTest::testReadUrlBOM()
+{
+	HTTPRequest req("PUT", "/form.cgi?field0=value0");
+	std::istringstream istr("\357\273\277field1=value1&field2=value%202&field3=value%3D3&field4=value%264");
+	HTMLForm form(req, istr);
+	assert (form.size() == 5);
+	assert (form["field0"] == "value0");
+	assert (form["field1"] == "value1");
+	assert (form["field2"] == "value 2");
+	assert (form["field3"] == "value=3");
+	assert (form["field4"] == "value&4");
+}
+
+
 void HTMLFormTest::testReadMultipart()
 {
 	std::istringstream istr(
@@ -359,6 +373,7 @@ CppUnit::Test* HTMLFormTest::suite()
 	CppUnit_addTest(pSuite, HTMLFormTest, testReadUrlGET);
 	CppUnit_addTest(pSuite, HTMLFormTest, testReadUrlPOST);
 	CppUnit_addTest(pSuite, HTMLFormTest, testReadUrlPUT);
+	CppUnit_addTest(pSuite, HTMLFormTest, testReadUrlBOM);
 	CppUnit_addTest(pSuite, HTMLFormTest, testReadMultipart);
 	CppUnit_addTest(pSuite, HTMLFormTest, testSubmit1);
 	CppUnit_addTest(pSuite, HTMLFormTest, testSubmit2);

Net/testsuite/src/HTMLFormTest.h

@@ -31,6 +31,7 @@ public:
 	void testReadUrlGET();
 	void testReadUrlPOST();
 	void testReadUrlPUT();
+	void testReadUrlBOM();
 	void testReadMultipart();
 	void testSubmit1();
 	void testSubmit2();