static char stream_get(stream_t *stream, json_error_t *error)
{
char c;
if(!stream->buffer[stream->buffer_pos])
{
stream->buffer[0] = stream->get(stream->data);
stream->buffer_pos = 0;
c = stream->buffer[0];
if((unsigned char)c >= 0x80 && c != (char)EOF)
{
/* multi-byte UTF-8 sequence */
int i, count;
count = utf8_check_first(c);
if(!count)
goto out;
assert(count >= 2);
for(i = 1; i < count; i++)
stream->buffer[i] = stream->get(stream->data);
if(!utf8_check_full(stream->buffer, count, NULL))
goto out;
stream->stream_pos += count;
stream->buffer[count] = '\0';
}
else {
stream->buffer[1] = '\0';
stream->stream_pos++;
}
}
return stream->buffer[stream->buffer_pos++];
out:
error_set(error, NULL, "unable to decode byte 0x%x at position %d",
(unsigned char)c, stream->stream_pos);
stream->buffer[0] = EOF;
stream->buffer[1] = '\0';
stream->buffer_pos = 1;
return EOF;
}
static void stream_unget(stream_t *stream, int c)
{
if(c == STREAM_STATE_EOF || c == STREAM_STATE_ERROR)
return;
stream->position--;
if(c == '\n') {
stream->line--;
stream->column = stream->last_column;
}
else if(utf8_check_first(c))
stream->column--;
assert(stream->buffer_pos > 0);
stream->buffer_pos--;
assert(stream->buffer[stream->buffer_pos] == c);
}
static int stream_get(stream_t *stream, json_error_t *error)
{
int c;
if(stream->state != STREAM_STATE_OK)
return stream->state;
if(!stream->buffer[stream->buffer_pos])
{
c = stream->get(stream->data);
if(c == EOF) {
stream->state = STREAM_STATE_EOF;
return STREAM_STATE_EOF;
}
stream->buffer[0] = c;
stream->buffer_pos = 0;
if(0x80 <= c && c <= 0xFF)
{
/* multi-byte UTF-8 sequence */
int i, count;
count = utf8_check_first(c);
if(!count)
goto out;
assert(count >= 2);
for(i = 1; i < count; i++)
stream->buffer[i] = stream->get(stream->data);
if(!utf8_check_full(stream->buffer, count, NULL))
goto out;
stream->buffer[count] = '\0';
}
else
stream->buffer[1] = '\0';
}
c = stream->buffer[stream->buffer_pos++];
stream->position++;
if(c == '\n') {
stream->line++;
stream->last_column = stream->column;
stream->column = 0;
}
else if(utf8_check_first(c)) {
/* track the Unicode character column, so increment only if
this is the first character of a UTF-8 sequence */
stream->column++;
}
return c;
out:
stream->state = STREAM_STATE_ERROR;
error_set(error, stream_to_lex(stream), "unable to decode byte 0x%x", c);
return STREAM_STATE_ERROR;
}