/*
* Get next tag text, eliminating leading and trailing whitespace
* and leaving only a single space for all internal whitespace.
*/
const char *
getTag(memBuf_t *mp)
{
static char *buf = NULL;
static size_t bufsize = 0;
size_t count = 0;
int inStr = 0, comment = 0, c;
if (memEof(mp)) {
log(("getTag(): returning NULL\n"));
return NULL;
}
while ((c = memGetc(mp)) != EOF && c != '<')
;
if (c == EOF) {
log(("getTag(): returning NULL\n"));
return NULL;
}
/* first char - check for comment */
c = memGetc(mp);
if (c == '>') {
log(("getTag(): returning empty tag\n"));
return "";
} else if (c == EOF) {
log(("getTag(): returning NULL\n"));
return NULL;
}
addchar(buf, bufsize, count, (char)c);
if (c == '!') {
int c2 = memGetc(mp);
if (c2 == '>' || c2 == EOF) {
term(buf, bufsize, count);
log(("getTag(): returning %s\n", buf));
return buf;
}
addchar(buf, bufsize, count, (char)c2);
if (c2 == '-') {
int c3 = memGetc(mp);
if (c3 == '>' || c3 == EOF) {
term(buf, bufsize, count);
log(("getTag(): returning %s\n", buf));
return buf;
}
addchar(buf, bufsize, count, (char)c3);
comment = 1;
}
}
if (comment) {
while ((c = memGetc(mp)) != EOF) {
if (c=='>' && buf[count-1]=='-' && buf[count-2]=='-') {
term(buf, bufsize, count);
log(("getTag(): returning %s\n", buf));
return buf;
}
if (isspace(c) && buf[count-1] == ' ')
continue;
addchar(buf, bufsize, count, (char)c);
}
} else {
while ((c = memGetc(mp)) != EOF) {
switch (c) {
case '\\':
addchar(buf, bufsize, count, (char)c);
c = memGetc(mp);
if (c == EOF) {
term(buf, bufsize, count);
log(("getTag(): returning %s\n", buf));
return buf;
}
addchar(buf, bufsize, count, (char)c);
break;
case '>':
if (inStr)
addchar(buf, bufsize, count, (char)c);
else {
term(buf, bufsize, count);
log(("getTag(): returning %s\n", buf));
return buf;
}
break;
case ' ':
case '\n':
case '\r':
case '\t':
case '\v':
if (inStr)
addchar(buf, bufsize, count, (char)c);
else if (count > 0 && buf[count-1] != ' ')
addchar(buf, bufsize, count, ' ');
break;
case '"':
inStr = !inStr;
/* fall through */
default:
addchar(buf, bufsize, count, (char)c);
}
}
}
term(buf, bufsize, count);
log(("getTag(): returning %s\n", count ? buf : "NULL"));
return count ? buf : NULL;
}
/* get META refresh URL (if any) */
char *
memGetMetaRefresh(memBuf_t *mp)
{
char *cp;
static char *buf = NULL;
char *bufptr;
static size_t bufsize = 0;
char *metaRefresh = NULL;
if (!buf) {
bufsize = 1024;
buf = myMalloc(bufsize);
}
/* look for all "meta" tags until Refresh found */
while (!metaRefresh && (cp = memStr(mp, "<meta")) != NULL) {
int c;
bufptr = buf;
/* copy whole tag to buffer for processing */
for (c = memGetc(mp); c != EOF && c != '>'; c = memGetc(mp)) {
*bufptr++ = (char)c;
if (bufptr > buf + (bufsize -1)) {
bufsize += 1024;
buf = myRealloc(buf, bufsize);
}
}
/* terminate string */
*bufptr = '\0';
log(("found META tag: %s", buf));
cp = strstr(buf, "http-equiv=");
if (!cp) {
log(("no http-equiv, looking for next"));
continue;
}
cp += 11;
if (strncasecmp(cp, "\"Refresh\"", 9)) {
log(("no Refresh, looking for next"));
continue;
}
cp = strstr(buf, "content=\"");
if (!cp) {
log(("no content, looking for next"));
continue;
}
cp += 9;
/* skip delay value (everything until ';') */
while (*cp && *cp != ';') cp++;
/* if not end of string skip ';' */
if (*cp) cp++;
/* and skip whitespace */
while (*cp && isspace(*cp)) cp++;
/* now there should be "url=" with optional whitespace around '=' */
if (strncasecmp(cp, "url", 3)) {
log(("no url key, looking for next"));
continue;
}
cp += 3;
while (*cp && isspace(*cp)) cp++;
if (*cp != '=') {
log(("no = after url, looking for next"));
continue;
}
cp++;
while (*cp && isspace(*cp)) cp++;
/* this is the beginning of the redirection URL */
bufptr = cp;
cp = strchr(bufptr, '"');
if (!cp) {
log(("no closing \", looking for next"));
continue;
}
/* cut off terminating '"' and other trailing garbage */
*cp = '\0';
metaRefresh = bufptr;
}
if (metaRefresh)
log(("found redirection"));
else
log(("no redirection found"));
memReset(mp);
return metaRefresh;
}
/*
* Get next non-tag text, eliminating leading and trailing whitespace
* and leaving only a single space for all internal whitespace.
*/
char *
getNonTag(memBuf_t *mp)
{
static char *buf = NULL;
static size_t bufsize = 0;
size_t count = 0, amp = 0;
int c;
if (memEof(mp)) {
log(("getNonTag(): returning NULL\n"));
return NULL;
}
while ((c = memGetc(mp)) != EOF) {
switch (c) {
case '<':
memUngetc(mp);
if (count) {
if (buf[count-1] == ' ')
--count;
term(buf, bufsize, count);
log(("getNonTag(): returning %s\n", buf));
return buf;
} else
(void)getTag(mp);
break;
case ' ':
case '\n':
case '\r':
case '\t':
case '\v':
case 0x82: /* UTF-8 */
case 0xC2: /* UTF-8 */
case 0xC3: /* UTF-8 */
case 0xA0: /* iso-8859-1 nbsp */
if (count && buf[count-1] != ' ')
addchar(buf, bufsize, count, ' ');
break;
case ';':
if (amp > 0) {
char *cp = &buf[amp];
term(buf, bufsize, count);
if (*cp == '#') {
buf[amp-1] = (char)atoi(cp+1);
count = amp;
} else if (!strcmp(cp, "amp")) {
count = amp;
} else if (!strcmp(cp, "gt")) {
buf[amp-1] = '>';
count = amp;
} else if (!strcmp(cp, "lt")) {
buf[amp-1] = '<';
count = amp;
} else if (!strcmp(cp, "nbsp")) {
buf[amp-1] = ' ';
count = amp;
if (count && buf[count-1] == ' ')
--count;
} else if (!strcmp(cp, "quot")) {
buf[amp-1] = '&';
count = amp;
} else
addchar(buf, bufsize, count, (char)c);
amp = 0;
} else
addchar(buf, bufsize, count, (char)c);
break;
case '&':
amp = count + 1;
/* fall through */
default:
addchar(buf, bufsize, count, (char)c);
}
}
if (count && buf[count-1] == ' ')
--count;
term(buf, bufsize, count);
log(("getNonTag(): returning %s\n", count ? buf : "NULL"));
return count ? buf : NULL;
} /* getNonTag() */
