Fixup utf-8 and meta handling in input

5 files changed, 65 insertions, 45 deletions

diff --git a/src/binding.c b/src/binding.c
index 191bc0d..5147c97 100644
--- a/src/binding.c
+++ b/src/binding.c
@@ -42,7 +42,7 @@ struct lookup_result lookup_key(struct keymap *keymaps, uint32_t nkeymaps,
 
     for (uint32_t bi = 0; bi < keymap->nbindings; ++bi) {
       struct binding *binding = &keymap->bindings[bi];
-      if (key->c == binding->key.c && key->mod == binding->key.mod) {
+      if (key_equal(key, &binding->key)) {
         if (binding->type == BindingType_Command) {
           return (struct lookup_result){
               .found = true,
diff --git a/src/binding.h b/src/binding.h
index 8a703c3..f00efed 100644
--- a/src/binding.h
+++ b/src/binding.h
@@ -11,14 +11,14 @@ enum binding_type { BindingType_Command, BindingType_Keymap };
 
 #define BINDING(mod_, c_, command_)                                            \
   (struct binding) {                                                           \
-    .key = {.mod = mod_, .c = c_}, .type = BindingType_Command,                \
-    .command = hash_command_name(command_)                                     \
+    .key = {.mod = mod_, .bytes[0] = c_, .nbytes = 1},                         \
+    .type = BindingType_Command, .command = hash_command_name(command_)        \
   }
 
 #define PREFIX(mod_, c_, keymap_)                                              \
   (struct binding) {                                                           \
-    .key = {.mod = mod_, .c = c_}, .type = BindingType_Keymap,                 \
-    .keymap = keymap_                                                          \
+    .key = {.mod = mod_, .bytes[0] = c_, .nbytes = 1},                         \
+    .type = BindingType_Keymap, .keymap = keymap_                              \
   }
 
 struct binding {
diff --git a/src/keyboard.c b/src/keyboard.c
index aaeccd2..b4c0c6d 100644
--- a/src/keyboard.c
+++ b/src/keyboard.c
@@ -1,6 +1,7 @@
 #include "keyboard.h"
 #include "reactor.h"
 #include "stdio.h"
+#include "utf8.h"
 
 #include <ctype.h>
 #include <errno.h>
@@ -13,31 +14,48 @@ struct keyboard keyboard_create(struct reactor *reactor) {
       .reactor_event_id =
           reactor_register_interest(reactor, STDIN_FILENO, ReadInterest),
       .has_data = false,
+      .last_key = {0},
   };
 }
 
 void parse_keys(uint8_t *bytes, uint32_t nbytes, struct key *out_keys,
-                uint32_t *out_nkeys) {
+                uint32_t *out_nkeys, struct key *previous_key) {
   uint32_t nkps = 0;
+  struct key *prevkp = previous_key;
   for (uint32_t bytei = 0; bytei < nbytes; ++bytei) {
     uint8_t b = bytes[bytei];
 
     struct key *kp = &out_keys[nkps];
+    kp->nbytes = 1;
 
     if (b == 0x1b) { // meta
       kp->mod |= Meta;
     } else if (b >= 0x00 && b <= 0x1f) { // ctrl char
       kp->mod |= Ctrl;
-      kp->c = b | 0x40;
-
+      kp->bytes[0] = b | 0x40;
+      ++nkps;
+      prevkp = kp;
     } else if (b == 0x7f) { // ^?
       kp->mod |= Ctrl;
-      kp->c = '?';
-    } else { // normal char (or part of char)
-      kp->c = b;
+      kp->bytes[0] = '?';
+      ++nkps;
+      prevkp = kp;
+    } else if (utf8_byte_is_unicode_start((uint8_t)b)) {
+      kp->bytes[0] = b;
+      ++nkps;
+      prevkp = kp;
+    } else if (utf8_byte_is_unicode_continuation((uint8_t)b)) {
+      prevkp->bytes[prevkp->nbytes] = b;
+      ++prevkp->nbytes;
+    } else { /* ascii char */
+      if (prevkp->mod & Meta) {
+        prevkp->bytes[0] = b;
+      } else {
+        kp->bytes[0] = b;
+      }
+      ++nkps;
+      prevkp = kp;
     }
-
-    ++nkps;
   }
 
   *out_nkeys = nkps;
@@ -61,7 +79,7 @@ struct keyboard_update keyboard_update(struct keyboard *kbd,
   int nbytes = read(STDIN_FILENO, bytes, 32);
 
   if (nbytes > 0) {
-    parse_keys(bytes, nbytes, upd.keys, &upd.nkeys);
+    parse_keys(bytes, nbytes, upd.keys, &upd.nkeys, &kbd->last_key);
   } else if (nbytes == EAGAIN) {
     kbd->has_data = false;
   }
@@ -69,8 +87,13 @@ struct keyboard_update keyboard_update(struct keyboard *kbd,
   return upd;
 }
 
-bool key_equal(struct key *key, uint8_t mod, uint8_t c) {
-  return key->c == c && key->mod == mod;
+bool key_equal_char(struct key *key, uint8_t mod, uint8_t c) {
+  return key->bytes[0] == c && key->mod == mod;
+}
+
+bool key_equal(struct key *key1, struct key *key2) {
+  return memcmp(key1->bytes, key2->bytes, key1->nbytes) == 0 &&
+         key1->mod == key2->mod && key1->nbytes == key2->nbytes;
 }
 
 void key_name(struct key *key, char *buf, size_t capacity) {
@@ -84,5 +107,10 @@ void key_name(struct key *key, char *buf, size_t capacity) {
     break;
   }
 
-  snprintf(buf, capacity, "%s%c", mod, tolower((char)key->c));
+  uint8_t lower[6];
+  for (uint32_t bytei = 0; bytei < key->nbytes; ++bytei) {
+    lower[bytei] = tolower(key->bytes[bytei]);
+  }
+
+  snprintf(buf, capacity, "%s%.*s", mod, key->nbytes, lower);
 }
diff --git a/src/keyboard.h b/src/keyboard.h
index 4078213..18630c3 100644
--- a/src/keyboard.h
+++ b/src/keyboard.h
@@ -7,16 +7,16 @@ enum modifiers {
   Meta = 1 << 1,
 };
 
-// note that unicode chars are split over multiple keypresses
-// TODO: make unicode chars nicer to deal with
 struct key {
-  uint8_t c;
+  uint8_t bytes[6];
+  uint8_t nbytes;
   uint8_t mod;
 };
 
 struct keyboard {
   uint32_t reactor_event_id;
   bool has_data;
+  struct key last_key;
 };
 
 struct keyboard_update {
@@ -31,5 +31,6 @@ struct keyboard keyboard_create(struct reactor *reactor);
 struct keyboard_update keyboard_update(struct keyboard *kbd,
                                        struct reactor *reactor);
 
-bool key_equal(struct key *key, uint8_t mod, uint8_t c);
+bool key_equal_char(struct key *key, uint8_t mod, uint8_t c);
+bool key_equal(struct key *key1, struct key *key2);
 void key_name(struct key *key, char *buf, size_t capacity);
diff --git a/src/main.c b/src/main.c
index a1d26e5..dc7cf63 100644
--- a/src/main.c
+++ b/src/main.c
@@ -58,13 +58,19 @@ void exit_editor(struct command_ctx ctx, int argc, const char *argv[]) {
   terminate();
 }
 
-static struct command GLOBAL_COMMANDS[] = {{
-                                               .name = "find-file",
-                                               .fn = unimplemented_command,
-                                               .userdata = (void *)"find-file",
-                                           },
-                                           {.name = "abort", .fn = _abort},
-                                           {.name = "exit", .fn = exit_editor}};
+static struct command GLOBAL_COMMANDS[] = {
+    {
+        .name = "find-file",
+        .fn = unimplemented_command,
+        .userdata = (void *)"find-file",
+    },
+    {
+        .name = "run-command-interactive",
+        .fn = unimplemented_command,
+        .userdata = (void *)"run-command-interactive",
+    },
+    {.name = "abort", .fn = _abort},
+    {.name = "exit", .fn = exit_editor}};
 
 uint64_t calc_frame_time_ns(struct timespec *timers, uint32_t num_timer_pairs) {
   uint64_t total = 0;
@@ -151,6 +157,7 @@ int main(int argc, char *argv[]) {
   struct binding global_binds[] = {
       PREFIX(Ctrl, 'X', &ctrlx_map),
       BINDING(Ctrl, 'G', "abort"),
+      BINDING(Meta, 'x', "run-command-interactive"),
   };
   struct binding ctrlx_bindings[] = {
       BINDING(Ctrl, 'C', "exit"),
@@ -245,9 +252,6 @@ int main(int argc, char *argv[]) {
     uint32_t nbuffer_keymaps = buffer_keymaps(&curbuf, &local_keymaps);
     struct keyboard_update kbd_upd = keyboard_update(&kbd, &reactor);
 
-    // buffer up chars to handle utf-8 "correctly"
-    uint8_t chars[kbd_upd.nkeys];
-    uint8_t nchars;
     for (uint32_t ki = 0; ki < kbd_upd.nkeys; ++ki) {
       struct key *k = &kbd_upd.keys[ki];
 
@@ -263,10 +267,6 @@ int main(int argc, char *argv[]) {
       }
 
       if (res.found) {
-        if (nchars > 0) {
-          buffer_add_text(active_window->buffer, chars, nchars);
-          nchars = 0;
-        }
         switch (res.type) {
         case BindingType_Command: {
           execute_command(res.command, active_window->buffer, 0, NULL);
@@ -282,21 +282,12 @@ int main(int argc, char *argv[]) {
         }
         }
       } else if (current_keymap != NULL) {
-        if (nchars > 0) {
-          buffer_add_text(active_window->buffer, chars, nchars);
-          nchars = 0;
-        }
         minibuffer_echo_timeout(4, "key is not bound!");
         current_keymap = NULL;
       } else {
-        chars[nchars] = k->c;
-        ++nchars;
+        buffer_add_text(active_window->buffer, k->bytes, k->nbytes);
       }
     }
-    if (nchars > 0) {
-      buffer_add_text(active_window->buffer, chars, nchars);
-      nchars = 0;
-    }
     clock_gettime(CLOCK_MONOTONIC, &keyboard_end);
 
     // calculate frame time