Make the charsets code useable in kittens

kitty/charsets.c

@@ -5,9 +5,13 @@
  * Distributed under terms of the GPL3 license.
  */
 
-#include "data-types.h"
 // Taken from consolemap.c in the linux vt driver sourcecode
 
+#include <stddef.h>
+#include <stdint.h>
+#define UTF8_ACCEPT 0
+#define UTF8_REJECT 1
+
 static uint32_t charset_translations[5][256] = {
   /* 8-bit Latin-1 mapped to Unicode -- trivial mapping */
   {
@@ -238,6 +242,26 @@ decode_utf8(uint32_t* state, uint32_t* codep, uint8_t byte) {
   return *state;
 }
 
+size_t
+decode_utf8_string(char *src, size_t sz, uint32_t *dest) {
+    // dest must be a zeroed array of size at least sz
+    uint32_t codep = 0, state = 0, prev = UTF8_ACCEPT;
+    size_t i, d;
+    for (i = 0, d = 0; i < sz; i++) {
+        switch(decode_utf8(&state, &codep, src[i])) {
+            case UTF8_ACCEPT:
+                dest[d++] = codep;
+                break;
+            case UTF8_REJECT:
+                state = UTF8_ACCEPT;
+                if (prev != UTF8_ACCEPT && i > 0) i--;
+                break;
+        }
+        prev = state;
+    }
+    return d;
+}
+
 unsigned int
 encode_utf8(uint32_t ch, char* dest) {
     if (ch < 0x80) {

kitty/charsets.h

@@ -0,0 +1,14 @@
+/*
+ * Copyright (C) 2019 Kovid Goyal <kovid at kovidgoyal.net>
+ *
+ * Distributed under terms of the GPL3 license.
+ */
+
+#pragma once
+
+#include <stdint.h>
+#include <stddef.h>
+
+uint32_t decode_utf8(uint32_t*, uint32_t*, uint8_t byte);
+size_t decode_utf8_string(char *src, size_t sz, uint32_t *dest);
+unsigned int encode_utf8(uint32_t ch, char* dest);

kitty/child-monitor.c

@@ -9,6 +9,7 @@
 #include "threading.h"
 #include "screen.h"
 #include "fonts.h"
+#include "charsets.h"
 #include <termios.h>
 #include <unistd.h>
 #include <float.h>

kitty/data-types.h

@@ -268,8 +268,6 @@ ColorProfile* alloc_color_profile();
 PyObject* create_256_color_table();
 PyObject* parse_bytes_dump(PyObject UNUSED *, PyObject *);
 PyObject* parse_bytes(PyObject UNUSED *, PyObject *);
-uint32_t decode_utf8(uint32_t*, uint32_t*, uint8_t byte);
-unsigned int encode_utf8(uint32_t ch, char* dest);
 void cursor_reset(Cursor*);
 Cursor* cursor_copy(Cursor*);
 void cursor_copy_to(Cursor *src, Cursor *dest);

kitty/line.c

@@ -8,6 +8,7 @@
 #include "data-types.h"
 #include "unicode-data.h"
 #include "lineops.h"
+#include "charsets.h"
 
 extern PyTypeObject Cursor_Type;
 

kitty/mouse.c

@@ -8,6 +8,7 @@
 #include "state.h"
 #include "screen.h"
 #include "lineops.h"
+#include "charsets.h"
 #include <limits.h>
 #include <math.h>
 #include "glfw-wrapper.h"

kitty/parser.c

@@ -8,6 +8,7 @@
 #include "control-codes.h"
 #include "screen.h"
 #include "graphics.h"
+#include "charsets.h"
 #include <time.h>
 
 extern PyTypeObject Screen_Type;

kitty/screen.c

@@ -23,6 +23,7 @@
 #include "modes.h"
 #include "wcwidth-std.h"
 #include "control-codes.h"
+#include "charsets.h"
 
 static const ScreenModes empty_modes = {0, .mDECAWM=true, .mDECTCEM=true, .mDECARM=true};
 static Selection EMPTY_SELECTION = {0};