static int methodline_parse(struct buffer *bu, struct methodline_state *ms,
struct buffer *method, struct buffer *uri, struct buffer *version)
{
const char *s = bu->data;
unsigned rem = bu->cur;
char ch;
while (rem > 0) {
ch = *s;
s++;
rem--;
switch (ms->state) {
case 0: /* method */
if (ch == ' ')
ms->state = 1;
else
buffer_addch(method, ch);
break;
case 1: /* uri */
if (ch == ' ')
ms->state = 2;
else
buffer_addch(uri, ch);
break;
case 2: /* version */
if (ch == '\r')
ms->state = 3;
else if (isspace(ch))
goto parse_error;
else
buffer_addch(version, ch);
break;
case 3: /* CR */
if (ch == '\n')
goto complete;
else
goto parse_error;
break;
default:
assert(0);
}
}
buffer_consume(bu, s - bu->data);
return 0;
complete:
ms->done = true;
buffer_consume(bu, s - bu->data);
return 0;
parse_error:
Error("%s():parse failure (%.*s)\n",
__func__, bu->cur, bu->data);
buffer_consume(bu, s - bu->data);
Error("%s():parse failure @ %d (state=%d)\n",
__func__, bu->cur, ms->state);
return -1;
}
static int _csv_escaped_dquote(struct csv_parser *cp)
{
if (buffer_addch(&cp->buf, '"'))
return -1;
cp->s = CSV_ESCAPED;
return 0;
}
static int headers_parse(struct buffer *bu, struct headers_state *hs,
struct buffer *name, struct buffer *value, struct env *env)
{
const char *s = bu->data;
unsigned rem = bu->cur;
char ch;
// TODO: 413 Entity Too Large
while (rem > 0) {
ch = *s;
s++;
rem--;
switch (hs->state) {
case 0: /* first char */
if (ch == '\r') {
if (name->cur)
env_set(env, name->data, value->data);
hs->state = 6; /* CR/LF - end of headers */
} else if (ch == ' ' || ch == '\t') { /* indented continuation line */
// TODO: this is completely untested .. device a test for it!
// TODO: insert single SP for these leading LWS
hs->state = 4; /* append to existing value data */
} else {
if (name->cur)
env_set(env, name->data, value->data);
buffer_reset(name);
buffer_reset(value);
buffer_addch(name, ch);
hs->state = 11;
}
break;
case 11: /* name */
if (ch == ' ' || ch == '\t') { /* LWS after name */
hs->state = 1;
} else if (ch == ':') {
hs->state = 2;
} else if (isspace(ch)) {
goto parse_error;
} else {
buffer_addch(name, ch);
}
break;
case 1: /* LWS after name */
if (ch == ' ' || ch == '\t')
; /* ignore */
else if (ch == ':')
hs->state = 2;
else
goto parse_error;
break;
case 2: /* ':' */
if (ch == ' ' || ch == '\t') {
hs->state = 3;
} else {
buffer_addch(value, ch);
hs->state = 4;
}
break;
case 3: /* LWS before value */
if (ch == ' ' || ch == '\t') {
/* ignore */
} else if (isspace(ch)) {
goto parse_error;
} else {
buffer_addch(value, ch);
hs->state = 4;
}
break;
case 4: /* value */
if (ch == '\r')
hs->state = 5;
else
buffer_addch(value, ch);
break;
case 5: /* CR after value */
if (ch == '\n')
hs->state = 0;
else
goto parse_error;
break;
case 6: /* CR on first column */
if (ch == '\n')
goto complete;
else
goto parse_error;
break;
default:
assert(0);
}
}
buffer_consume(bu, s - bu->data);
return 0;
complete:
hs->done = true;
buffer_consume(bu, s - bu->data);
return 0;
parse_error:
Error("%s():parse failure (%.*s)\n",
__func__, bu->cur, bu->data);
buffer_consume(bu, s - bu->data);
Error("%s():parse failure @ %d (ch=%c state=%d)\n",
__func__, bu->cur, ch, hs->state);
return -1;
}
static int csv(struct csv_parser *cp, const char *buf, size_t len)
{
while (len > 0) {
char ch = *buf;
buf++;
len--;
switch (cp->s) {
case CSV_START:
if (ch == '\r')
cp->s = CSV_NEWLINE;
else if (ch == '\n')
_csv_next_row(cp);
else if (ch == '"')
cp->s = CSV_ESCAPED;
else if (ch == ',')
_csv_next_field(cp);
else
goto unescaped;
break;
case CSV_ESCAPED:
if (ch == '"')
cp->s = CSV_ESCAPED_DQUOTE;
else if (buffer_addch(&cp->buf, ch))
return -1;
break;
case CSV_ESCAPED_DQUOTE:
if (ch == '\r')
cp->s = CSV_NEWLINE;
else if (ch == '\n')
_csv_next_row(cp);
else if (ch == '"') {
if (_csv_escaped_dquote(cp))
return -1;
} else if (ch == ',')
_csv_next_field(cp);
else
goto unescaped; /* deviates from RFC 4180 */
break;
unescaped:
cp->s = CSV_UNESCAPED;
case CSV_UNESCAPED:
if (ch == '\r')
cp->s = CSV_NEWLINE;
else if (ch == '\n')
_csv_next_row(cp);
else if (ch == '"')
cp->s = CSV_ESCAPED; /* deviates from RFC 4180 */
else if (ch == ',')
_csv_next_field(cp);
else if (buffer_addch(&cp->buf, ch))
return -1;
break;
case CSV_NEWLINE:
if (ch == '\n')
_csv_next_row(cp);
else
return -1;
case CSV_ERROR:
return -1;
}
}
return cp->s == CSV_ERROR ? -1 : 0;
}
