feat: switch text to UTF-8; add Unicode key events, dead keys & Czech layout; fix numpad navigation; add NumLock support

kernel/src/Api/Keyboard.hpp

@@ -12,6 +12,107 @@
 #include "Common.hpp"
 
 namespace montauk::abi {
+    // ---- Key text ----------------------------------------------------------
+    // Keyboard layouts are applied in userspace (desktop, terminal), which send
+    // KeyEvents carrying the Unicode codepoint. Binaries built before that send
+    // 9-byte KeyEventV1s whose `ascii` is a Windows-1252 byte; both kinds share
+    // a redirected child's key mailbox, told apart by message length.
+
+    static uint32_t KeyCp1252ToUnicode(uint8_t b) {
+        static constexpr uint16_t high[32] = {
+            0x20AC, 0x81,   0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021,
+            0x02C6, 0x2030, 0x0160, 0x2039, 0x0152, 0x8D,   0x017D, 0x8F,
+            0x90,   0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014,
+            0x02DC, 0x2122, 0x0161, 0x203A, 0x0153, 0x9D,   0x017E, 0x0178,
+        };
+        return (b >= 0x80 && b <= 0x9F) ? high[b - 0x80] : b;
+    }
+
+    // The character a key event types, or 0 for control/navigation keys.
+    static uint32_t KeyTextCodepoint(const KeyEvent& ev) {
+        if (ev.codepoint) return ev.codepoint;
+        uint8_t b = (uint8_t)ev.ascii;
+        if (b < 0x20 || b == 0x7F) return 0;
+        return KeyCp1252ToUnicode(b);
+    }
+
+    static int KeyEncodeUtf8(uint32_t cp, uint8_t* out) {
+        if (cp > 0x10FFFF || (cp >= 0xD800 && cp <= 0xDFFF)) cp = 0xFFFD;
+        if (cp < 0x80) { out[0] = (uint8_t)cp; return 1; }
+        if (cp < 0x800) {
+            out[0] = (uint8_t)(0xC0 | (cp >> 6));
+            out[1] = (uint8_t)(0x80 | (cp & 0x3F));
+            return 2;
+        }
+        if (cp < 0x10000) {
+            out[0] = (uint8_t)(0xE0 | (cp >> 12));
+            out[1] = (uint8_t)(0x80 | ((cp >> 6) & 0x3F));
+            out[2] = (uint8_t)(0x80 | (cp & 0x3F));
+            return 3;
+        }
+        out[0] = (uint8_t)(0xF0 | (cp >> 18));
+        out[1] = (uint8_t)(0x80 | ((cp >> 12) & 0x3F));
+        out[2] = (uint8_t)(0x80 | ((cp >> 6) & 0x3F));
+        out[3] = (uint8_t)(0x80 | (cp & 0x3F));
+        return 4;
+    }
+
+    static void KeyFromDriver(const Drivers::PS2::Keyboard::KeyEvent& k, KeyEvent* out) {
+        memset(out, 0, sizeof(KeyEvent));
+        out->scancode = k.Scancode;
+        out->ascii    = k.Ascii;
+        out->pressed  = k.Pressed;
+        out->shift    = k.Shift;
+        out->ctrl     = k.Ctrl;
+        out->alt      = k.Alt;
+        out->capslock = k.CapsLock;
+        out->altgr    = k.AltGr;
+        out->extended = k.Extended;
+        // The kernel's own scancode table is US ASCII.
+        uint8_t a = (uint8_t)k.Ascii;
+        out->codepoint = (a >= 0x20 && a < 0x7F) ? a : 0;
+    }
+
+    // Receive one key from a redirected child's mailbox as a full KeyEvent,
+    // upgrading legacy 9-byte messages. Returns MailboxRecv's result.
+    static int KeyMailboxRecv(Ipc::Mailbox* mailbox, KeyEvent* out) {
+        memset(out, 0, sizeof(KeyEvent));
+        uint16_t len = sizeof(KeyEvent);
+        int rc = Ipc::MailboxRecv(mailbox, nullptr, out, &len, true);
+        if (rc > 0 && len < sizeof(KeyEvent)) {
+            memset((uint8_t*)out + len, 0, sizeof(KeyEvent) - len);
+            out->codepoint = KeyTextCodepoint(*out);
+        }
+        return rc;
+    }
+
+    // Turn a pressed key into the next getchar byte. Characters above ASCII are
+    // returned as UTF-8, one byte per call; the rest is parked in the process.
+    static bool KeyTakeChar(Sched::Process* proc, const KeyEvent& ev, char* out) {
+        if (!ev.pressed) return false;
+        uint32_t cp = KeyTextCodepoint(ev);
+        if (cp >= 0x80 && proc != nullptr) {
+            int n = KeyEncodeUtf8(cp, proc->getcharUtf8);
+            proc->getcharUtf8Len = (uint8_t)n;
+            proc->getcharUtf8Pos = 1;
+            *out = (char)proc->getcharUtf8[0];
+            return true;
+        }
+        if (ev.ascii != 0 && ev.ascii != KEY_ASCII_SUBSTITUTE) {
+            *out = ev.ascii;
+            return true;
+        }
+        return false;
+    }
+
+    static bool KeyPendingChar(Sched::Process* proc, char* out) {
+        if (proc == nullptr || proc->getcharUtf8Pos >= proc->getcharUtf8Len) return false;
+        *out = (char)proc->getcharUtf8[proc->getcharUtf8Pos++];
+        return true;
+    }
+
+    // ---- Syscalls ------------------------------------------------------------
+
     static bool Sys_IsKeyAvailable() {
         auto* proc = Sched::GetCurrentProcessPtr();
         if (proc && proc->redirected) {
@@ -27,8 +128,8 @@ namespace montauk::abi {
         return Drivers::PS2::Keyboard::IsKeyAvailable();
     }
 
-    static void Sys_GetKey(KeyEvent* outEvent) {
-        if (outEvent == nullptr) return;
+    // Blocking read of the next key event, in the full format.
+    static void ReadKeyEvent(KeyEvent* outEvent) {
         auto* proc = Sched::GetCurrentProcessPtr();
         if (proc && proc->redirected) {
             Ipc::HandleSnapshot snapshot;
@@ -36,8 +137,7 @@ namespace montauk::abi {
             if (mailbox != nullptr) {
                 for (;;) {
                     uint64_t observedWake = Sched::ObserveObjectWake(mailbox);
-                    uint16_t len = sizeof(KeyEvent);
-                    int rc = Ipc::MailboxRecv(mailbox, nullptr, outEvent, &len, true);
+                    int rc = KeyMailboxRecv(mailbox, outEvent);
                     if (rc > 0) return;
                     if (rc < 0) {
                         memset(outEvent, 0, sizeof(KeyEvent));
@@ -47,20 +147,33 @@ namespace montauk::abi {
                 }
             }
         }
-        auto k = Drivers::PS2::Keyboard::GetKey();
-        outEvent->scancode = k.Scancode;
-        outEvent->ascii    = k.Ascii;
-        outEvent->pressed  = k.Pressed;
-        outEvent->shift    = k.Shift;
-        outEvent->ctrl     = k.Ctrl;
-        outEvent->alt      = k.Alt;
-        outEvent->capslock = k.CapsLock;
-        outEvent->altgr    = k.AltGr;
-        outEvent->extended = k.Extended;
+        KeyFromDriver(Drivers::PS2::Keyboard::GetKey(), outEvent);
+    }
+
+    // SYS_GETKEY: legacy 9-byte event. A new-format event's first 9 bytes are
+    // already a valid KeyEventV1 (ascii holds the 1252 byte or SUB).
+    static void Sys_GetKey(KeyEventV1* outEvent) {
+        if (outEvent == nullptr) return;
+        KeyEvent ev;
+        ReadKeyEvent(&ev);
+        memcpy(outEvent, &ev, sizeof(KeyEventV1));
+    }
+
+    // SYS_GETKEY_EX: size-versioned. The caller's size was validated by the
+    // dispatcher (>= sizeof(KeyEventV1), writable).
+    static int Sys_GetKeyEx(void* outEvent, uint64_t size) {
+        KeyEvent ev;
+        ReadKeyEvent(&ev);
+        uint64_t n = size < sizeof(KeyEvent) ? size : sizeof(KeyEvent);
+        memcpy(outEvent, &ev, n);
+        if (size > n) memset((uint8_t*)outEvent + n, 0, size - n);
+        return Success;
     }
 
     static char Sys_GetChar() {
         auto* proc = Sched::GetCurrentProcessPtr();
+        char pending;
+        if (KeyPendingChar(proc, &pending)) return pending;
         if (proc && proc->redirected) {
             Ipc::HandleSnapshot inputSnapshot;
             Ipc::HandleSnapshot mailboxSnapshot;
@@ -78,11 +191,11 @@ namespace montauk::abi {
                     }
 
                     if (mailbox != nullptr) {
-                        KeyEvent ev{};
-                        uint16_t len = sizeof(ev);
-                        int rc = Ipc::MailboxRecv(mailbox, nullptr, &ev, &len, true);
+                        KeyEvent ev;
+                        int rc = KeyMailboxRecv(mailbox, &ev);
                         if (rc > 0) {
-                            if (ev.pressed && ev.ascii != 0) return ev.ascii;
+                            char c;
+                            if (KeyTakeChar(proc, ev, &c)) return c;
                             continue;
                         }
                         if (rc < 0 && input == nullptr) return 0;
@@ -115,6 +228,8 @@ namespace montauk::abi {
      */
     static char Sys_GetCharNb() {
         auto* proc = Sched::GetCurrentProcessPtr();
+        char pending;
+        if (KeyPendingChar(proc, &pending)) return pending;
         if (proc && proc->redirected) {
             Ipc::HandleSnapshot inputSnapshot;
             Ipc::HandleSnapshot mailboxSnapshot;
@@ -127,10 +242,10 @@ namespace montauk::abi {
                 }
 
                 while (mailbox != nullptr) {
-                    KeyEvent ev{};
-                    uint16_t len = sizeof(ev);
-                    if (Ipc::MailboxRecv(mailbox, nullptr, &ev, &len, true) <= 0) break;
-                    if (ev.pressed && ev.ascii != 0) return ev.ascii;
+                    KeyEvent ev;
+                    if (KeyMailboxRecv(mailbox, &ev) <= 0) break;
+                    char c;
+                    if (KeyTakeChar(proc, ev, &c)) return c;
                 }
                 return 0;
             }