Github user mike-jumper commented on a diff in the pull request:
https://github.com/apache/guacamole-server/pull/154#discussion_r178485119
--- Diff: src/protocols/rdp/guac_rdpdr/rdpdr_printer.c ---
@@ -141,18 +143,25 @@ static void
guac_rdpdr_device_printer_announce_handler(guac_rdpdr_device* device
Stream_Write(output_stream, "PRN1\0\0\0\0", 8); /* DOS name */
/* Printer data */
- Stream_Write_UINT32(output_stream, 24 + GUAC_PRINTER_DRIVER_LENGTH +
GUAC_PRINTER_NAME_LENGTH);
+ int settings_length = guac_utf8_strlen(device->device_name);
+ int printer_name_length = (settings_length + 1) * 2;
+ char* printer_name = malloc(printer_name_length);
+ guac_rdp_utf8_to_utf16((const unsigned char*)device->device_name,
+ settings_length, printer_name, printer_name_length);
+ printer_name[printer_name_length - 2] = '\0';
+ printer_name[printer_name_length - 1] = '\0';
+ Stream_Write_UINT32(output_stream, 24 + GUAC_PRINTER_DRIVER_LENGTH +
printer_name_length);
Stream_Write_UINT32(output_stream,
RDPDR_PRINTER_ANNOUNCE_FLAG_DEFAULTPRINTER
| RDPDR_PRINTER_ANNOUNCE_FLAG_NETWORKPRINTER);
Stream_Write_UINT32(output_stream, 0); /* reserved - must be 0 */
Stream_Write_UINT32(output_stream, 0); /* PnPName length (PnPName is
ultimately ignored) */
Stream_Write_UINT32(output_stream, GUAC_PRINTER_DRIVER_LENGTH); /*
DriverName length */
- Stream_Write_UINT32(output_stream, GUAC_PRINTER_NAME_LENGTH); /*
PrinterName length */
+ Stream_Write_UINT32(output_stream, printer_name_length); /*
PrinterName length */
Stream_Write_UINT32(output_stream, 0); /*
CachedFields length */
Stream_Write(output_stream, GUAC_PRINTER_DRIVER,
GUAC_PRINTER_DRIVER_LENGTH);
- Stream_Write(output_stream, GUAC_PRINTER_NAME,
GUAC_PRINTER_NAME_LENGTH);
+ Stream_Write(output_stream, printer_name, printer_name_length);
--- End diff --
I think some of the complexity we're dealing with is due to the concept of
the `announce_handler` and overlapping responsibilities between the devices and
the top-level handling of RDPDR and the announce packet.
What if, instead of an `announce_handler`, we:
* Stored the device type (`RDPDR_DTYP_PRINT`, etc.) within the
`guac_rdpdr_device`
* Stored the arbitrary device data and the length of that data within the
`guac_rdpdr_device`
* Relied upon that within
`guac_rdpdr_send_client_device_list_announce_request()`, such that it's truly
only the responsibility of that function to assemble the announce packet
I think that should simplify things greatly.
If it helps, I'd be glad to take a stab at the above refactor against the
current codebase, though I'm not sure how painful the rebases of these
name-related PRs would be after that.
---
