Better support for creating Unicode strings

3 files changed, 73 insertions, 0 deletions

diff --git a/libqpdf/QPDFObjectHandle.cc b/libqpdf/QPDFObjectHandle.cc
index 5c111cc8..da609cc2 100644
--- a/libqpdf/QPDFObjectHandle.cc
+++ b/libqpdf/QPDFObjectHandle.cc
@@ -1221,6 +1221,20 @@ QPDFObjectHandle::unparseResolved()
     return this->m->obj->unparse();
 }
 
+std::string
+QPDFObjectHandle::unparseBinary()
+{
+    if (this->isString())
+    {
+        return dynamic_cast<QPDF_String*>(
+            this->m->obj.getPointer())->unparse(true);
+    }
+    else
+    {
+        return unparse();
+    }
+}
+
 QPDFObjectHandle
 QPDFObjectHandle::parse(std::string const& object_str,
                         std::string const& object_description)
@@ -1846,6 +1860,12 @@ QPDFObjectHandle::newString(std::string const& str)
 }
 
 QPDFObjectHandle
+QPDFObjectHandle::newUnicodeString(std::string const& utf8_str)
+{
+    return QPDFObjectHandle(QPDF_String::new_utf16(utf8_str));
+}
+
+QPDFObjectHandle
 QPDFObjectHandle::newOperator(std::string const& value)
 {
     return QPDFObjectHandle(new QPDF_Operator(value));
diff --git a/libqpdf/QPDF_String.cc b/libqpdf/QPDF_String.cc
index 60a3e0df..eb31a808 100644
--- a/libqpdf/QPDF_String.cc
+++ b/libqpdf/QPDF_String.cc
@@ -64,6 +64,58 @@ QPDF_String::~QPDF_String()
 {
 }
 
+QPDF_String*
+QPDF_String::new_utf16(std::string const& utf8_val)
+{
+    std::string result = "\xfe\xff";
+    size_t len = utf8_val.length();
+    for (size_t i = 0; i < len; ++i)
+    {
+        unsigned char ch = static_cast<unsigned char>(utf8_val.at(i));
+        if (ch < 128)
+        {
+            result += QUtil::toUTF16(ch);
+        }
+        else
+        {
+            size_t bytes_needed = 0;
+            unsigned bit_check = 0x40;
+            unsigned char to_clear = 0x80;
+            while (ch & bit_check)
+            {
+                ++bytes_needed;
+                to_clear |= bit_check;
+                bit_check >>= 1;
+            }
+
+            if (((bytes_needed > 5) || (bytes_needed < 1)) ||
+                ((i + bytes_needed) >= len))
+            {
+                result += "\xff\xfd";
+            }
+            else
+            {
+                unsigned long codepoint = (ch & ~to_clear);
+                while (bytes_needed > 0)
+                {
+                    --bytes_needed;
+                    ch = utf8_val.at(++i);
+                    if ((ch & 0xc0) != 0x80)
+                    {
+                        --i;
+                        codepoint = 0xfffd;
+                        break;
+                    }
+                    codepoint <<= 6;
+                    codepoint += (ch & 0x3f);
+                }
+                result += QUtil::toUTF16(codepoint);
+            }
+        }
+    }
+    return new QPDF_String(result);
+}
+
 std::string
 QPDF_String::unparse()
 {
diff --git a/libqpdf/qpdf/QPDF_String.hh b/libqpdf/qpdf/QPDF_String.hh
index abf8291a..b4858c49 100644
--- a/libqpdf/qpdf/QPDF_String.hh
+++ b/libqpdf/qpdf/QPDF_String.hh
@@ -9,6 +9,7 @@ class QPDF_String: public QPDFObject
 {
   public:
     QPDF_String(std::string const& val);
+    static QPDF_String* new_utf16(std::string const& utf8_val);
     virtual ~QPDF_String();
     virtual std::string unparse();
     virtual QPDFObject::object_type_e getTypeCode() const;