Fcalva/cyclone
1
0
Fork 0
mirror of https://github.com/justinethier/cyclone.git synced 2025-05-24 04:25:06 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Updated Cyc_io_read_line to prevent truncation

Browse source 
Ensure last codepoint is fully-read before returning
This commit is contained in:
Justin Ethier 2017-11-06 13:19:31 +00:00
parent 6910e3e4cb
commit d431b2af1c
2 changed files with 23 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
include/cyclone/runtime.h
View file

@ -725,7 +725,7 @@ void Cyc_set_globals_changed(gc_thread_data *thd);
int Cyc_utf8_encode(char *dest, int sz, uint32_t *src, int srcsz); int Cyc_utf8_encode(char *dest, int sz, uint32_t *src, int srcsz);
uint32_t Cyc_utf8_decode(uint32_t* state, uint32_t* codep, uint32_t byte); uint32_t Cyc_utf8_decode(uint32_t* state, uint32_t* codep, uint32_t byte);
int Cyc_utf8_count_code_points(uint8_t* s); int Cyc_utf8_count_code_points(uint8_t* s);
int Cyc_utf8_count_code_points_and_bytes(uint8_t* s, int *cpts, int *bytes); int Cyc_utf8_count_code_points_and_bytes(uint8_t* s, char_type *codepoint, int *cpts, int *bytes);
uint32_t Cyc_utf8_validate_stream(uint32_t *state, char *str, size_t len); uint32_t Cyc_utf8_validate_stream(uint32_t *state, char *str, size_t len);
uint32_t Cyc_utf8_validate(char *str, size_t len); uint32_t Cyc_utf8_validate(char *str, size_t len);
/**@}*/ /**@}*/

32
runtime.c
View file

@ -6365,8 +6365,10 @@ object Cyc_io_read_char(void *data, object cont, object port)
object Cyc_io_read_line(void *data, object cont, object port) object Cyc_io_read_line(void *data, object cont, object port)
{ {
FILE *stream = ((port_type *) port)->fp; FILE *stream = ((port_type *) port)->fp;
char buf[1024]; char buf[1027];
int len, num_cp; int len, num_cp, i = 0;
char_type codepoint;
uint32_t state;
Cyc_check_port(data, port); Cyc_check_port(data, port);
if (stream == NULL) { if (stream == NULL) {
@ -6375,10 +6377,21 @@ object Cyc_io_read_line(void *data, object cont, object port)
set_thread_blocked(data, cont); set_thread_blocked(data, cont);
errno = 0; errno = 0;
if (fgets(buf, 1023, stream) != NULL) { if (fgets(buf, 1023, stream) != NULL) {
// TODO: not good enough for UTF-8, what if we stopped reading in the middle of a code point? state = Cyc_utf8_count_code_points_and_bytes((uint8_t *)buf, &codepoint, &num_cp, &len);
// should reserve 3 extra bytes and, if last code point is not complete, read one byte at a // Check if we stopped reading in the middle of a code point and
// time until it has been read // if so, read one byte at a time until that code point is finished.
Cyc_utf8_count_code_points_and_bytes((uint8_t *)buf, &num_cp, &len); while (state != CYC_UTF8_ACCEPT && i < 3) {
int c = fgetc(stream);
buf[len] = c;
len++;
Cyc_utf8_decode(&state, &codepoint, (uint8_t)c);
if (state == CYC_UTF8_ACCEPT) {
num_cp++;
break;
}
i++;
}
{ {
// Remove any trailing CR / newline chars // Remove any trailing CR / newline chars
while (len > 0 && (buf[len - 1] == '\n' || while (len > 0 && (buf[len - 1] == '\n' ||
@ -6596,19 +6609,18 @@ int Cyc_utf8_count_code_points(uint8_t* s) {
return count; return count;
} }
int Cyc_utf8_count_code_points_and_bytes(uint8_t* s, int *cpts, int *bytes) { int Cyc_utf8_count_code_points_and_bytes(uint8_t* s, char_type *codepoint, int *cpts, int *bytes) {
uint32_t codepoint;
uint32_t state = 0; uint32_t state = 0;
*cpts = 0; *cpts = 0;
*bytes = 0; *bytes = 0;
for (; *s; ++s){ for (; *s; ++s){
*bytes += 1; *bytes += 1;
if (!Cyc_utf8_decode(&state, &codepoint, *s)) if (!Cyc_utf8_decode(&state, codepoint, *s))
*cpts += 1; *cpts += 1;
} }
if (state != CYC_UTF8_ACCEPT) if (state != CYC_UTF8_ACCEPT)
return -1; return state;
return 0; return 0;
} }