p
/
termo
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Moved single byte and UTF-8 sequence recombiner out of driver-csi.c back into termkey.c

This commit is contained in:
Paul LeoNerd Evans 2008-10-08 00:36:27 +01:00
parent d46f72fc1a
commit ac6cae29bb
3 changed files with 114 additions and 104 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

94
driver-csi.c
View File

@ -120,8 +120,6 @@ static void free_driver(void *private)
free(csi); free(csi);
} }
#define UTF8_INVALID 0xFFFD
#define CHARAT(i) (tk->buffer[tk->buffstart + (i)]) #define CHARAT(i) (tk->buffer[tk->buffstart + (i)])
static termkey_result getkey_csi(termkey_t *tk, size_t introlen, termkey_key *key) static termkey_result getkey_csi(termkey_t *tk, size_t introlen, termkey_key *key)
@ -332,96 +330,8 @@ static termkey_result getkey(termkey_t *tk, termkey_key *key)
else if(b0 == 0x9b) { else if(b0 == 0x9b) {
return getkey_csi(tk, 1, key); return getkey_csi(tk, 1, key);
} }
else if(b0 < 0xa0) { else
// Single byte C0, G0 or C1 - C1 is never UTF-8 initial byte return (*tk->method.getkey_simple)(tk, key);
(*tk->method.emit_codepoint)(tk, b0, key);
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
else if(tk->flags & TERMKEY_FLAG_UTF8) {
// Some UTF-8
int nbytes;
int codepoint;
key->type = TERMKEY_TYPE_UNICODE;
key->modifiers = 0;
if(b0 < 0xc0) {
// Starts with a continuation byte - that's not right
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
else if(b0 < 0xe0) {
nbytes = 2;
codepoint = b0 & 0x1f;
}
else if(b0 < 0xf0) {
nbytes = 3;
codepoint = b0 & 0x0f;
}
else if(b0 < 0xf8) {
nbytes = 4;
codepoint = b0 & 0x07;
}
else if(b0 < 0xfc) {
nbytes = 5;
codepoint = b0 & 0x03;
}
else if(b0 < 0xfe) {
nbytes = 6;
codepoint = b0 & 0x01;
}
else {
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
if(tk->buffcount < nbytes)
return tk->waittime ? TERMKEY_RES_AGAIN : TERMKEY_RES_NONE;
for(int b = 1; b < nbytes; b++) {
unsigned char cb = CHARAT(b);
if(cb < 0x80 || cb >= 0xc0) {
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
(*tk->method.eat_bytes)(tk, b - 1);
return TERMKEY_RES_KEY;
}
codepoint <<= 6;
codepoint |= cb & 0x3f;
}
// Check for overlong sequences
if(nbytes > utf8_seqlen(codepoint))
codepoint = UTF8_INVALID;
// Check for UTF-16 surrogates or invalid codepoints
if((codepoint >= 0xD800 && codepoint <= 0xDFFF) ||
codepoint == 0xFFFE ||
codepoint == 0xFFFF)
codepoint = UTF8_INVALID;
(*tk->method.emit_codepoint)(tk, codepoint, key);
(*tk->method.eat_bytes)(tk, nbytes);
return TERMKEY_RES_KEY;
}
else {
// Non UTF-8 case - just report the raw byte
key->type = TERMKEY_TYPE_UNICODE;
key->code.codepoint = b0;
key->modifiers = 0;
key->utf8[0] = key->code.codepoint;
key->utf8[1] = 0;
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
return TERMKEY_SYM_NONE;
} }
static termkey_keysym register_csi_ss3(termkey_csi *csi, termkey_type type, termkey_keysym sym, unsigned char cmd, const char *name) static termkey_keysym register_csi_ss3(termkey_csi *csi, termkey_type type, termkey_keysym sym, unsigned char cmd, const char *name)

12
termkey-internal.h
View File

@ -49,20 +49,10 @@ struct termkey {
struct { struct {
void (*eat_bytes)(termkey_t *tk, size_t count); void (*eat_bytes)(termkey_t *tk, size_t count);
void (*emit_codepoint)(termkey_t *tk, int codepoint, termkey_key *key); void (*emit_codepoint)(termkey_t *tk, int codepoint, termkey_key *key);
termkey_result (*getkey_simple)(termkey_t *tk, termkey_key *key);
} method; } method;
}; };
extern struct termkey_driver termkey_driver_csi; extern struct termkey_driver termkey_driver_csi;
// Keep this here for now since it's tiny
static inline int utf8_seqlen(int codepoint)
{
if(codepoint < 0x0000080) return 1;
if(codepoint < 0x0000800) return 2;
if(codepoint < 0x0010000) return 3;
if(codepoint < 0x0200000) return 4;
if(codepoint < 0x4000000) return 5;
return 6;
}
#endif #endif

112
termkey.c
View File

@ -16,6 +16,7 @@ static struct termkey_driver *drivers[] = {
// Forwards for the "protected" methods // Forwards for the "protected" methods
static void eat_bytes(termkey_t *tk, size_t count); static void eat_bytes(termkey_t *tk, size_t count);
static void emit_codepoint(termkey_t *tk, int codepoint, termkey_key *key); static void emit_codepoint(termkey_t *tk, int codepoint, termkey_key *key);
static termkey_result getkey_simple(termkey_t *tk, termkey_key *key);
static termkey_keysym register_c0(termkey_t *tk, termkey_keysym sym, unsigned char ctrl, const char *name); static termkey_keysym register_c0(termkey_t *tk, termkey_keysym sym, unsigned char ctrl, const char *name);
static termkey_keysym register_c0_full(termkey_t *tk, termkey_keysym sym, int modifier_set, int modifier_mask, unsigned char ctrl, const char *name); static termkey_keysym register_c0_full(termkey_t *tk, termkey_keysym sym, int modifier_set, int modifier_mask, unsigned char ctrl, const char *name);
@ -74,8 +75,9 @@ termkey_t *termkey_new_full(int fd, int flags, size_t buffsize, int waittime)
for(i = 0; i < 32; i++) for(i = 0; i < 32; i++)
tk->c0[i].sym = TERMKEY_SYM_NONE; tk->c0[i].sym = TERMKEY_SYM_NONE;
tk->method.eat_bytes = &eat_bytes; tk->method.eat_bytes = &eat_bytes;
tk->method.emit_codepoint = &emit_codepoint; tk->method.emit_codepoint = &emit_codepoint;
tk->method.getkey_simple = &getkey_simple;
register_c0(tk, TERMKEY_SYM_BACKSPACE, 0x08, "Backspace"); register_c0(tk, TERMKEY_SYM_BACKSPACE, 0x08, "Backspace");
register_c0(tk, TERMKEY_SYM_TAB, 0x09, "Tab"); register_c0(tk, TERMKEY_SYM_TAB, 0x09, "Tab");
@ -170,6 +172,16 @@ static void eat_bytes(termkey_t *tk, size_t count)
} }
} }
static inline int utf8_seqlen(int codepoint)
{
if(codepoint < 0x0000080) return 1;
if(codepoint < 0x0000800) return 2;
if(codepoint < 0x0010000) return 3;
if(codepoint < 0x0200000) return 4;
if(codepoint < 0x4000000) return 5;
return 6;
}
static void fill_utf8(termkey_key *key) static void fill_utf8(termkey_key *key)
{ {
int codepoint = key->code.codepoint; int codepoint = key->code.codepoint;
@ -251,6 +263,104 @@ static void emit_codepoint(termkey_t *tk, int codepoint, termkey_key *key)
fill_utf8(key); fill_utf8(key);
} }
#define UTF8_INVALID 0xFFFD
#define CHARAT(i) (tk->buffer[tk->buffstart + (i)])
static termkey_result getkey_simple(termkey_t *tk, termkey_key *key)
{
unsigned char b0 = CHARAT(0);
if(b0 < 0xa0) {
// Single byte C0, G0 or C1 - C1 is never UTF-8 initial byte
(*tk->method.emit_codepoint)(tk, b0, key);
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
else if(tk->flags & TERMKEY_FLAG_UTF8) {
// Some UTF-8
int nbytes;
int codepoint;
key->type = TERMKEY_TYPE_UNICODE;
key->modifiers = 0;
if(b0 < 0xc0) {
// Starts with a continuation byte - that's not right
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
else if(b0 < 0xe0) {
nbytes = 2;
codepoint = b0 & 0x1f;
}
else if(b0 < 0xf0) {
nbytes = 3;
codepoint = b0 & 0x0f;
}
else if(b0 < 0xf8) {
nbytes = 4;
codepoint = b0 & 0x07;
}
else if(b0 < 0xfc) {
nbytes = 5;
codepoint = b0 & 0x03;
}
else if(b0 < 0xfe) {
nbytes = 6;
codepoint = b0 & 0x01;
}
else {
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
if(tk->buffcount < nbytes)
return tk->waittime ? TERMKEY_RES_AGAIN : TERMKEY_RES_NONE;
for(int b = 1; b < nbytes; b++) {
unsigned char cb = CHARAT(b);
if(cb < 0x80 || cb >= 0xc0) {
(*tk->method.emit_codepoint)(tk, UTF8_INVALID, key);
(*tk->method.eat_bytes)(tk, b - 1);
return TERMKEY_RES_KEY;
}
codepoint <<= 6;
codepoint |= cb & 0x3f;
}
// Check for overlong sequences
if(nbytes > utf8_seqlen(codepoint))
codepoint = UTF8_INVALID;
// Check for UTF-16 surrogates or invalid codepoints
if((codepoint >= 0xD800 && codepoint <= 0xDFFF) ||
codepoint == 0xFFFE ||
codepoint == 0xFFFF)
codepoint = UTF8_INVALID;
(*tk->method.emit_codepoint)(tk, codepoint, key);
(*tk->method.eat_bytes)(tk, nbytes);
return TERMKEY_RES_KEY;
}
else {
// Non UTF-8 case - just report the raw byte
key->type = TERMKEY_TYPE_UNICODE;
key->code.codepoint = b0;
key->modifiers = 0;
key->utf8[0] = key->code.codepoint;
key->utf8[1] = 0;
(*tk->method.eat_bytes)(tk, 1);
return TERMKEY_RES_KEY;
}
}
termkey_result termkey_getkey(termkey_t *tk, termkey_key *key) termkey_result termkey_getkey(termkey_t *tk, termkey_key *key)
{ {
return (*tk->driver.getkey)(tk, key); return (*tk->driver.getkey)(tk, key);