static warc_string_t
_warc_rduri(const char *buf, size_t bsz)
{
static const char _key[] = "\r\nWARC-Target-URI:";
const char *const eob = buf + bsz;
const char *val;
const char *uri;
const char *eol;
warc_string_t res = {0U, NULL};
if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) {
/* no bother */
return res;
}
/* overread whitespace */
for (val += sizeof(_key) - 1U; val < eob && isspace(*val); val++);
/* overread URL designators */
if ((uri = xmemmem(val, eob - val, "://", 3U)) == NULL) {
/* not touching that! */
return res;
} else if ((eol = memchr(uri, '\n', eob - uri)) == NULL) {
/* no end of line? :O */
return res;
}
/* massage uri to point to after :// */
uri += 3U;
/* also massage eol to point to the first whitespace
* after the last non-whitespace character before
* the end of the line */
for (; eol > uri && isspace(eol[-1]); eol--);
/* now then, inspect the URI */
if (memcmp(val, "file", 4U) == 0) {
/* perfect, nothing left to do here */
} else if (memcmp(val, "http", 4U) == 0 ||
memcmp(val, "ftp", 3U) == 0) {
/* overread domain, and the first / */
while (uri < eol && *uri++ != '/');
} else {
/* not sure what to do? best to bugger off */
return res;
}
res.str = uri;
res.len = eol - uri;
return res;
}
static unsigned int
_warc_rdtyp(const char *buf, size_t bsz)
{
static const char _key[] = "\r\nWARC-Type:";
const char *const eob = buf + bsz;
const char *val;
if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) {
/* no bother */
return WT_NONE;
}
/* overread whitespace */
for (val += sizeof(_key) - 1U; val < eob && isspace(*val); val++);
if (val + 8U > eob) {
;
} else if (memcmp(val, "resource", 8U) == 0) {
return WT_RSRC;
} else if (memcmp(val, "warcinfo", 8U) == 0) {
return WT_INFO;
} else if (memcmp(val, "metadata", 8U) == 0) {
return WT_META;
} else if (memcmp(val, "request", 7U) == 0) {
return WT_REQ;
} else if (memcmp(val, "response", 8U) == 0) {
return WT_RSP;
} else if (memcmp(val, "conversi", 8U) == 0) {
return WT_CONV;
} else if (memcmp(val, "continua", 8U) == 0) {
return WT_CONT;
}
return WT_NONE;
}
static int
deinfix1(const char *s, size_t z)
{
if (xmemmem(pool, npool, s, z)) {
return -1;
}
if (npool + z >= zpool) {
/* resize */
zpool = zpool * 2U ?: 4096U;
pool = realloc(pool, zpool);
}
static const char*
_warc_find_eoh(const char *buf, size_t bsz)
{
static const char _marker[] = "\r\n\r\n";
const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U);
if (hit != NULL) {
hit += sizeof(_marker) - 1U;
}
return hit;
}
static bool
read_line(mux_ctx_t ctx, ff_msg_t msg)
{
static const char cmd_pmsg[] = "PRIVMSG";
static const size_t cmd_pmsg_sz = sizeof(cmd_pmsg) - 1;
char *line;
size_t llen;
const char *cursor;
char *p;
llen = prchunk_getline(ctx->rdr, &line);
/* we parse the line in 3 steps, receive time stamp, symbol, values */
cursor = line;
msg->chan = NULL;
msg->json = NULL;
msg->msglen = 0UL;
/* receive time stamp, always first on line */
if (UNLIKELY(parse_rcv_stmp(msg, &cursor) < 0)) {
goto bugger;
}
/* parse the rest */
llen -= cursor - line;
/* message types */
if ((p = xmemmem(cursor, llen, cmd_pmsg, cmd_pmsg_sz))) {
msg->ty = FF_MSG_JSON;
msg->chan = p + cmd_pmsg_sz + 1;
if ((p = strchr(msg->chan, ' '))) {
*p++ = '\0';
if (*p++ == ':') {
/* ah a normal channel message */
llen -= p - cursor;
msg->json = p;
msg->msglen = llen;
p[llen] = '\0';
return true;
}
}
}
return false;
bugger:
/* declare the line f****d */
fputs("line b0rked\n> ", stderr);
fputs(line, stderr);
fputc('\n', stderr);
return false;
}
static time_t
_warc_rdmtm(const char *buf, size_t bsz)
{
static const char _key[] = "\r\nLast-Modified:";
const char *val;
char *on = NULL;
time_t res;
if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) {
/* no bother */
return (time_t)-1;
}
/* xstrpisotime() kindly overreads whitespace for us, so use that */
val += sizeof(_key) - 1U;
res = xstrpisotime(val, &on);
if (on == NULL || !isspace(*on)) {
/* hm, can we trust that number? Best not. */
return (time_t)-1;
}
return res;
}
static ssize_t
_warc_rdlen(const char *buf, size_t bsz)
{
static const char _key[] = "\r\nContent-Length:";
const char *val;
char *on = NULL;
long int len;
if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) {
/* no bother */
return -1;
}
/* strtol kindly overreads whitespace for us, so use that */
val += sizeof(_key) - 1U;
len = strtol(val, &on, 10);
if (on == NULL || !isspace(*on)) {
/* hm, can we trust that number? Best not. */
return -1;
}
return (size_t)len;
}
