handle UTF-16BE fully

git-svn-id: svn+q:///qpdf/trunk@639 71b93d88-0707-0410-a8cf-f5a4172ac649

ChangeLog

@@ -1,3 +1,9 @@
+2008-11-23  Jay Berkenbilt  <ejb@ql.org>
+
+	* libqpdf/QPDF_String.cc (QPDF_String::getUTF8Val): handle
+	UTF-16BE properly rather than just treating the string as a string
+	of 16-bit characters.
+
 2008-06-30  Jay Berkenbilt  <ejb@ql.org>
 
 	* 2.0.2: release

libqpdf/QPDF_String.cc

@@ -2,6 +2,8 @@
 #include <qpdf/QPDF_String.hh>
 
 #include <qpdf/QUtil.hh>
+#include <qpdf/QTC.hh>
+
 // DO NOT USE ctype -- it is locale dependent for some things, and
 // it's not worth the risk of including it in case it may accidentally
 // be used.
@@ -159,12 +161,42 @@ QPDF_String::getUTF8Val() const
 	(this->val[0] == '\xfe') && (this->val[1] == '\xff'))
     {
 	// This is a Unicode string using big-endian UTF-16.  This
-	// code is not actually correct as it doesn't properly handle
-	// characters past 0xffff.
+	// code uses unsigned long and unsigned short to hold
+	// codepoint values.  It requires unsigned long to be at least
+	// 32 bits and unsigned short to be at least 16 bits, but it
+	// will work fine if they are larger.
+	unsigned long codepoint = 0L;
 	for (unsigned int i = 2; i < len; i += 2)
 	{
-	    result += QUtil::toUTF8(((unsigned char) this->val[i] << 8) +
-				    ((unsigned char) this->val[i+1]));
+	    // Convert from UTF16-BE.  If we get a malformed
+	    // codepoint, this code will generate incorrect output
+	    // without giving a warning.  Specifically, a high
+	    // codepoint not followed by a low codepoint will be
+	    // discarded, and a low codepoint not preceded by a high
+	    // codepoint will just get its low 10 bits output.
+	    unsigned short bits =
+		(((unsigned char) this->val[i]) << 8) +
+		((unsigned char) this->val[i+1]);
+	    if ((bits & 0xFC00) == 0xD800)
+	    {
+		codepoint = 0x10000 + ((bits & 0x3FF) << 10);
+		continue;
+	    }
+	    else if ((bits & 0xFC00) == 0xDC00)
+	    {
+		if (codepoint != 0)
+		{
+		    QTC::TC("qpdf", "QPDF_String non-trivial UTF-16");
+		}
+		codepoint += bits & 0x3FF;
+	    }
+	    else
+	    {
+		codepoint = bits;
+	    }
+
+	    result += QUtil::toUTF8(codepoint);
+	    codepoint = 0;
 	}
     }
     else

qpdf/qpdf.testcov

@@ -115,3 +115,4 @@ QPDF_Stream pipeStreamData with null pipeline 0
 QPDFWriter not recompressing /FlateDecode 0
 QPDF piping xref stream from encrypted file 0
 unable to filter 0
+QPDF_String non-trivial UTF-16 0

qpdf/qtest/qpdf/misc-3.out

@@ -8,6 +8,7 @@ QStrings:
 No Special Characters
 These: ¿÷¢þ and no more
 πωτατω
+treble clef: 𝄠; sixteenth note: 𝅘𝅥𝅮
 QNumbers:
 1.000
 3.142

qpdf/qtest/qpdf/misc-3.pdf

@@ -13,6 +13,7 @@
     (No Special Characters)
     (These: ¿÷¢þ and no more)
     <feff03c003c903c403b103c403c9>
+    <feff0074007200650062006c006500200063006c00650066003a0020d834dd20003b0020007300690078007400650065006e007400680020006e006f00740065003a0020d834dd60>
   ]
   /Type /Catalog
 >>
@@ -109,19 +110,19 @@ xref
 0 10
 0000000000 65535 f 
 0000000025 00000 n 
-0000000226 00000 n 
-0000000308 00000 n 
-0000000543 00000 n 
-0000000642 00000 n 
-0000000684 00000 n 
-0000000782 00000 n 
-0000000801 00000 n 
-0000000919 00000 n 
+0000000377 00000 n 
+0000000459 00000 n 
+0000000694 00000 n 
+0000000793 00000 n 
+0000000835 00000 n 
+0000000933 00000 n 
+0000000952 00000 n 
+0000001070 00000 n 
 trailer <<
   /Root 1 0 R
   /Size 10
   /ID [<e017d8dc1fe53a81e40aa79bcb43fdec><76269ee0b6579446b731e060af8ef436>]
 >>
 startxref
-954
+1105
 %%EOF