Remove old per-object crypto stuff we dont need.

CHANGES.md

@@ -2,6 +2,13 @@ Changes in PDFio
 ================
 
 
+v1.5.3 - YYYY-MM-DD
+-------------------
+
+- Fixed decryption of PDF files "protected" by 40-bit RC4 (Issue #42)
+- Fixed decryption of UTF-16 strings (Issue #42)
+
+
 v1.5.2 - 2025-04-12
 -------------------
 

pdfio-crypto.c

@@ -409,13 +409,6 @@ _pdfioCryptoMakeReader(
   uint8_t	data[21];		// Key data
   _pdfio_md5_t	md5;			// MD5 state
   uint8_t	digest[16];		// MD5 digest value
-#if PDFIO_OBJ_CRYPT
-  pdfio_array_t	*id_array;		// Object ID array
-  unsigned char	*id_value;		// Object ID value
-  size_t	id_len;			// Length of object ID
-  uint8_t	temp_key[16];		// File key for object
-#endif // PDFIO_OBJ_CRYPT
-  uint8_t	*file_key;		// Computed file key to use
 
 
   PDFIO_DEBUG("_pdfioCryptoMakeReader(pdf=%p, obj=%p(%d), ctx=%p, iv=%p, ivlen=%p(%d))\n", pdf, obj, (int)obj->number, ctx, iv, ivlen, (int)*ivlen);
@@ -427,60 +420,6 @@ _pdfioCryptoMakeReader(
     return (NULL);
   }
 
-#if PDFIO_OBJ_CRYPT
-  if ((id_array = pdfioDictGetArray(pdfioObjGetDict(obj), "ID")) != NULL)
-  {
-    // Object has its own ID that will get used for encryption...
-    _pdfio_md5_t md5;			// MD5 context
-    uint8_t	file_digest[16];	// MD5 digest of file ID and pad
-    uint8_t	user_pad[32],		// Padded user password
-		own_user_key[32],	// Calculated user key
-		pdf_user_key[32];	// Decrypted user key
-
-    PDFIO_DEBUG("_pdfioCryptoMakeReader: Per-object file ID.\n");
-
-    if ((id_value = pdfioArrayGetBinary(id_array, 0, &id_len)) == NULL)
-    {
-      *ivlen = 0;
-      return (NULL);
-    }
-
-    _pdfioCryptoMD5Init(&md5);
-    _pdfioCryptoMD5Append(&md5, pdf_passpad, 32);
-    _pdfioCryptoMD5Append(&md5, id_value, id_len);
-    _pdfioCryptoMD5Finish(&md5, file_digest);
-
-    make_owner_key(pdf->encryption, pdf->password, pdf->owner_key, user_pad);
-    make_file_key(pdf->encryption, pdf->permissions, id_value, id_len, user_pad, pdf->owner_key, temp_key);
-    make_user_key(id_value, id_len, own_user_key);
-
-    if (memcmp(own_user_key, pdf->user_key, sizeof(own_user_key)))
-    {
-      PDFIO_DEBUG("_pdfioCryptoMakeReader: Not user password, trying owner password.\n");
-
-      make_file_key(pdf->encryption, pdf->permissions, id_value, id_len, pdf->password, pdf->owner_key, temp_key);
-      make_user_key(id_value, id_len, own_user_key);
-
-      memcpy(pdf_user_key, pdf->user_key, sizeof(pdf_user_key));
-      decrypt_user_key(pdf->encryption, temp_key, pdf_user_key);
-
-      if (memcmp(pdf->password, pdf_user_key, 32) && memcmp(own_user_key, pdf_user_key, 16))
-      {
-        _pdfioFileError(pdf, "Unable to unlock file.");
-	*ivlen = 0;
-	return (NULL);
-      }
-    }
-
-    file_key = temp_key;
-  }
-  else
-#endif // PDFIO_OBJ_CRYPT
-  {
-    // Use the default file key...
-    file_key = pdf->file_key;
-  }
-
   switch (pdf->encryption)
   {
     default :
@@ -490,20 +429,20 @@ _pdfioCryptoMakeReader(
 
     case PDFIO_ENCRYPTION_RC4_40 :
 	// Copy the key data for the MD5 hash.
-	memcpy(data, file_key, 16);
-	data[16] = (uint8_t)obj->number;
-	data[17] = (uint8_t)(obj->number >> 8);
-	data[18] = (uint8_t)(obj->number >> 16);
-	data[19] = (uint8_t)obj->generation;
-	data[20] = (uint8_t)(obj->generation >> 8);
+	memcpy(data, pdf->file_key, 5);
+	data[5] = (uint8_t)obj->number;
+	data[6] = (uint8_t)(obj->number >> 8);
+	data[7] = (uint8_t)(obj->number >> 16);
+	data[8] = (uint8_t)obj->generation;
+	data[9] = (uint8_t)(obj->generation >> 8);
 
         // Hash it...
         _pdfioCryptoMD5Init(&md5);
-	_pdfioCryptoMD5Append(&md5, data, sizeof(data));
+	_pdfioCryptoMD5Append(&md5, data, 10);
 	_pdfioCryptoMD5Finish(&md5, digest);
 
-        // Initialize the RC4 context using 40 bits of the digest...
-	_pdfioCryptoRC4Init(&ctx->rc4, digest, 5);
+        // Initialize the RC4 context using 80 bits of the digest...
+	_pdfioCryptoRC4Init(&ctx->rc4, digest, 10);
 	*ivlen = 0;
 	return ((_pdfio_crypto_cb_t)_pdfioCryptoRC4Crypt);
 
@@ -517,7 +456,7 @@ _pdfioCryptoMakeReader(
 
     case PDFIO_ENCRYPTION_RC4_128 :
 	// Copy the key data for the MD5 hash.
-	memcpy(data, file_key, 16);
+	memcpy(data, pdf->file_key, 16);
 	data[16] = (uint8_t)obj->number;
 	data[17] = (uint8_t)(obj->number >> 8);
 	data[18] = (uint8_t)(obj->number >> 16);

pdfio-dict.c

@@ -465,127 +465,12 @@ pdfioDictGetString(pdfio_dict_t *dict,	// I - Dictionary
   else if (value && value->type == PDFIO_VALTYPE_BINARY && value->value.binary.datalen < 4096)
   {
     // Convert binary string to regular string...
-    char	temp[4096],		// Temporary string
-		*tempptr;		// Pointer into temporary string
-    unsigned char *dataptr;		// Pointer into the data string
+    char	temp[4096];		// Temporary UTF-8 string
 
-    if (!(value->value.binary.datalen & 1) && !memcmp(value->value.binary.data, "\376\377", 2))
+    if (!(value->value.binary.datalen & 1) && (!memcmp(value->value.binary.data, "\376\377", 2) || !memcmp(value->value.binary.data, "\377\376", 2)))
     {
-      // Copy UTF-16 BE
-      int	ch;			// Unicode character
-      size_t	remaining;		// Remaining bytes
-
-      for (dataptr = value->value.binary.data + 2, remaining = value->value.binary.datalen - 2, tempptr = temp; remaining > 1 && tempptr < (temp + sizeof(temp) - 5); dataptr += 2, remaining -= 2)
-      {
-        ch = (dataptr[0] << 8) | dataptr[1];
-
-        if (ch >= 0xd800 && ch <= 0xdbff && remaining > 3)
-	{
-	  // Multi-word UTF-16 char...
-          int lch;			// Lower bits
-
-	  lch = (dataptr[2] << 8) | dataptr[3];
-
-          if (lch < 0xdc00 || lch >= 0xdfff)
-	    break;
-
-          ch        = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000;
-          dataptr   += 2;
-          remaining -= 2;
-	}
-	else if (ch >= 0xfffe)
-	{
-	  continue;
-	}
-
-        if (ch < 128)
-        {
-          // ASCII
-          *tempptr++ = (char)ch;
-        }
-        else if (ch < 4096)
-        {
-          // 2-byte UTF-8
-          *tempptr++ = (char)(0xc0 | (ch >> 6));
-          *tempptr++ = (char)(0x80 | (ch & 0x3f));
-        }
-        else if (ch < 65536)
-        {
-          // 3-byte UTF-8
-          *tempptr++ = (char)(0xe0 | (ch >> 12));
-          *tempptr++ = (char)(0x80 | ((ch >> 6) & 0x3f));
-          *tempptr++ = (char)(0x80 | (ch & 0x3f));
-        }
-        else
-        {
-          // 4-byte UTF-8
-          *tempptr++ = (char)(0xe0 | (ch >> 18));
-          *tempptr++ = (char)(0x80 | ((ch >> 12) & 0x3f));
-          *tempptr++ = (char)(0x80 | ((ch >> 6) & 0x3f));
-          *tempptr++ = (char)(0x80 | (ch & 0x3f));
-        }
-      }
-
-      *tempptr = '\0';
-    }
-    else if (!(value->value.binary.datalen & 1) && !memcmp(value->value.binary.data, "\377\376", 2))
-    {
-      // Copy UTF-16 LE
-      int	ch;			// Unicode character
-      size_t	remaining;		// Remaining bytes
-
-      for (dataptr = value->value.binary.data + 2, remaining = value->value.binary.datalen - 2, tempptr = temp; remaining > 1 && tempptr < (temp + sizeof(temp) - 5); dataptr += 2, remaining -= 2)
-      {
-        ch = (dataptr[1] << 8) | dataptr[0];
-
-        if (ch >= 0xd800 && ch <= 0xdbff && remaining > 3)
-	{
-	  // Multi-word UTF-16 char...
-          int lch;			// Lower bits
-
-	  lch = (dataptr[3] << 8) | dataptr[2];
-
-          if (lch < 0xdc00 || lch >= 0xdfff)
-	    break;
-
-          ch        = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000;
-          dataptr   += 2;
-          remaining -= 2;
-	}
-	else if (ch >= 0xfffe)
-	{
-	  continue;
-	}
-
-        if (ch < 128)
-        {
-          // ASCII
-          *tempptr++ = (char)ch;
-        }
-        else if (ch < 4096)
-        {
-          // 2-byte UTF-8
-          *tempptr++ = (char)(0xc0 | (ch >> 6));
-          *tempptr++ = (char)(0x80 | (ch & 0x3f));
-        }
-        else if (ch < 65536)
-        {
-          // 3-byte UTF-8
-          *tempptr++ = (char)(0xe0 | (ch >> 12));
-          *tempptr++ = (char)(0x80 | ((ch >> 6) & 0x3f));
-          *tempptr++ = (char)(0x80 | (ch & 0x3f));
-        }
-        else
-        {
-          // 4-byte UTF-8
-          *tempptr++ = (char)(0xe0 | (ch >> 18));
-          *tempptr++ = (char)(0x80 | ((ch >> 12) & 0x3f));
-          *tempptr++ = (char)(0x80 | ((ch >> 6) & 0x3f));
-          *tempptr++ = (char)(0x80 | (ch & 0x3f));
-        }
-      }
-
-      *tempptr = '\0';
+      // Copy UTF-16...
+      _pdfio_utf16cpy(temp, value->value.binary.data, value->value.binary.datalen, sizeof(temp));
     }
     else
     {

pdfio-private.h

@@ -339,6 +339,7 @@ struct _pdfio_stream_s			// Stream
 
 extern size_t		_pdfio_strlcpy(char *dst, const char *src, size_t dstsize) _PDFIO_INTERNAL;
 extern double		_pdfio_strtod(pdfio_file_t *pdf, const char *s) _PDFIO_INTERNAL;
+extern void		_pdfio_utf16cpy(char *dst, const unsigned char *src, size_t srclen, size_t dstsize) _PDFIO_INTERNAL;
 extern ssize_t		_pdfio_vsnprintf(pdfio_file_t *pdf, char *buffer, size_t bufsize, const char *format, va_list ap) _PDFIO_INTERNAL;
 
 extern bool		_pdfioArrayDecrypt(pdfio_file_t *pdf, pdfio_obj_t *obj, pdfio_array_t *a, size_t depth) _PDFIO_INTERNAL;

pdfio-string.c

@@ -158,6 +158,89 @@ _pdfio_strtod(pdfio_file_t *pdf,	// I - PDF file
 }
 
 
+//
+// '_pdfio_utf16cpy()' - Convert UTF-16 to UTF-8.
+//
+
+void
+_pdfio_utf16cpy(
+    char                *dst,		// I - Destination buffer for UTF-8
+    const unsigned char *src,		// I - Source UTF-16
+    size_t              srclen,		// I - Length of UTF-16
+    size_t              dstsize)	// I - Destination buffer size
+{
+  char	*dstptr = dst,			// Pointer into buffer
+	*dstend = dst + dstsize - 5;	// End of buffer
+  int	ch;				// Unicode character
+  bool	is_be = !memcmp(src, "\376\377", 2);
+					// Big-endian strings?
+
+
+  // Loop through the UTF-16 string, converting to Unicode then UTF-8...
+  for (src += 2, srclen -= 2; srclen > 1 && dstptr < dstend; src += 2, srclen -= 2)
+  {
+    // Initial character...
+    if (is_be)
+      ch = (src[0] << 8) | src[1];
+    else
+      ch = (src[1] << 8) | src[0];
+
+    if (ch >= 0xd800 && ch <= 0xdbff && srclen > 3)
+    {
+      // Multi-word UTF-16 char...
+      int lch;			// Lower bits
+
+      if (is_be)
+	lch = (src[2] << 8) | src[3];
+      else
+	lch = (src[3] << 8) | src[2];
+
+      if (lch < 0xdc00 || lch >= 0xdfff)
+	break;
+
+      ch     = (((ch & 0x3ff) << 10) | (lch & 0x3ff)) + 0x10000;
+      src    += 2;
+      srclen -= 2;
+    }
+    else if (ch >= 0xfffe)
+    {
+      continue;
+    }
+
+    // Convert Unicode to UTF-8...
+    if (ch < 128)
+    {
+      // ASCII
+      *dstptr++ = (char)ch;
+    }
+    else if (ch < 4096)
+    {
+      // 2-byte UTF-8
+      *dstptr++ = (char)(0xc0 | (ch >> 6));
+      *dstptr++ = (char)(0x80 | (ch & 0x3f));
+    }
+    else if (ch < 65536)
+    {
+      // 3-byte UTF-8
+      *dstptr++ = (char)(0xe0 | (ch >> 12));
+      *dstptr++ = (char)(0x80 | ((ch >> 6) & 0x3f));
+      *dstptr++ = (char)(0x80 | (ch & 0x3f));
+    }
+    else
+    {
+      // 4-byte UTF-8
+      *dstptr++ = (char)(0xe0 | (ch >> 18));
+      *dstptr++ = (char)(0x80 | ((ch >> 12) & 0x3f));
+      *dstptr++ = (char)(0x80 | ((ch >> 6) & 0x3f));
+      *dstptr++ = (char)(0x80 | (ch & 0x3f));
+    }
+  }
+
+  // Nul-terminate the UTF-8 string...
+  *dstptr = '\0';
+}
+
+
 //
 // '_pdfio_vsnprintf()' - Format a string.
 //

pdfio-value.c

@@ -172,7 +172,7 @@ _pdfioValueDecrypt(pdfio_file_t   *pdf,	// I - PDF file
 	// Copy the decrypted string back to the value and adjust the length...
 	memcpy(v->value.binary.data, temp, templen);
 
-	if (pdf->encryption >= PDFIO_ENCRYPTION_AES_128)
+	if (pdf->encryption >= PDFIO_ENCRYPTION_AES_128 && temp[templen - 1] <= templen)
 	  v->value.binary.datalen = templen - temp[templen - 1];
 	else
 	  v->value.binary.datalen = templen;
@@ -183,20 +183,48 @@ _pdfioValueDecrypt(pdfio_file_t   *pdf,	// I - PDF file
     case PDFIO_VALTYPE_STRING :
         // Decrypt regular string...
         templen = strlen(v->value.string);
-	if (templen > (sizeof(temp) - 33))
+	if (templen > (PDFIO_MAX_STRING - 1))
 	{
 	  _pdfioFileError(pdf, "Unable to read encrypted string - too long.");
 	  return (false);
 	}
+	else if ((temp = (uint8_t *)_pdfioStringAllocBuffer(pdf)) == NULL)
+	{
+	  _pdfioFileError(pdf, "Unable to read encrypted binary string - out of memory.");
+	  return (false);
+	}
 
 	ivlen = templen;
 	if ((cb = _pdfioCryptoMakeReader(pdf, obj, &ctx, (uint8_t *)v->value.string, &ivlen)) == NULL)
 	  return (false);
 
 	templen = (cb)(&ctx, temp, (uint8_t *)v->value.string + ivlen, templen - ivlen);
+
+	if (pdf->encryption >= PDFIO_ENCRYPTION_AES_128 && temp[templen - 1] <= templen)
+	  templen -= temp[templen - 1];
+
 	temp[templen] = '\0';
 
-        if ((timeval = get_date_time((char *)temp)) != 0)
+        if ((templen & 1) == 0 && (!memcmp(temp, "\376\377", 2) || !memcmp(temp, "\377\376", 2)))
+        {
+          // Convert UTF-16 to UTF-8...
+          char	utf8[4096];		// Temporary string
+
+          _pdfio_utf16cpy(utf8, temp, templen, sizeof(utf8));
+
+	  if ((timeval = get_date_time((char *)utf8)) != 0)
+	  {
+	    // Change the type to date...
+	    v->type       = PDFIO_VALTYPE_DATE;
+	    v->value.date = timeval;
+	  }
+	  else
+	  {
+	    // Copy the decrypted string back to the value...
+	    v->value.string = pdfioStringCreate(pdf, utf8);
+	  }
+        }
+        else if ((timeval = get_date_time((char *)temp)) != 0)
         {
           // Change the type to date...
           v->type       = PDFIO_VALTYPE_DATE;
@@ -207,6 +235,8 @@ _pdfioValueDecrypt(pdfio_file_t   *pdf,	// I - PDF file
           // Copy the decrypted string back to the value...
 	  v->value.string = pdfioStringCreate(pdf, (char *)temp);
 	}
+
+        _pdfioStringFreeBuffer(pdf, (char *)temp);
         break;
   }