static int cs_complete_http(const char *buf, int len, int head_only)
{
/* deal with HTTP request/response */
int i, content_len = 0, chunked = 0;
/* need at least one line followed by \n or \r .. */
for (i = 0; ; i++)
if (i == len)
return 0; /* incomplete */
else if (buf[i] == '\n' || buf[i] == '\r')
break;
/* check to see if it's a response with content */
if (!head_only && !memcmp(buf, "HTTP/", 5))
{
int j;
for (j = 5; j < i; j++)
if (buf[j] == ' ')
{
++j;
if (buf[j] == '1') /* 1XX */
;
else if (!memcmp(buf + j, "204", 3))
;
else if (!memcmp(buf + j, "304", 3))
;
else
content_len = -1;
break;
}
}
#if 0
printf("len = %d\n", len);
fwrite (buf, 1, len, stdout);
printf("----------\n");
#endif
for (i = 2; i <= len-2; )
{
if (i > 8192)
{
return i; /* do not allow more than 8K HTTP header */
}
if (skip_crlf(buf, len, &i))
{
if (skip_crlf(buf, len, &i))
{
/* inside content */
if (chunked)
return cs_read_chunk(buf, i, len);
else
{ /* not chunked ; inside body */
if (content_len == -1)
return 0; /* no content length */
else if (len >= i + content_len)
{
return i + content_len;
}
}
break;
}
else if (i < len - 20 &&
!yaz_strncasecmp((const char *) buf+i,
"Transfer-Encoding:", 18))
{
i+=18;
while (buf[i] == ' ')
i++;
if (i < len - 8)
if (!yaz_strncasecmp((const char *) buf+i, "chunked", 7))
chunked = 1;
}
else if (i < len - 17 &&
!yaz_strncasecmp((const char *)buf+i,
"Content-Length:", 15))
{
i+= 15;
while (buf[i] == ' ')
i++;
content_len = 0;
while (i <= len-4 && yaz_isdigit(buf[i]))
content_len = content_len*10 + (buf[i++] - '0');
if (content_len < 0) /* prevent negative offsets */
content_len = 0;
}
else
i++;
}
else
i++;
}
return 0;
}
void mp::HTMLParser::Rep::parse_str(HTMLParserEvent &event, const char *cp)
{
const char *text_start = cp;
while (*cp)
{
if (*cp++ != '<')
continue;
if (nest && *cp == '!')
{
int i;
tagText(event, text_start, cp - 1);
if (cp[1] == '-' && cp[2] == '-')
{
for (i = 3; cp[i]; i++)
if (cp[i] == '-' && cp[i+1] == '-' && cp[i+2] == '>')
{
i+= 2;
event.openTagStart(cp, i);
break;
}
}
else
{
for (i = 1; cp[i] && cp[i] != '>'; i++)
;
event.openTagStart(cp, i);
}
if (m_verbose)
printf("------ dtd %.*s\n", i, cp);
i += tagEnd(event, cp, i, cp + i);
cp += i;
text_start = cp;
}
else if (nest && *cp == '?')
{
int i;
tagText(event, text_start, cp - 1);
for (i = 1; cp[i] && cp[i] != '>'; i++)
;
event.openTagStart(cp, i);
if (m_verbose)
printf("------ pi %.*s\n", i, cp);
i += tagEnd(event, cp, i, cp + i);
cp += i;
text_start = cp;
}
else if (*cp == '/' && isAlpha(cp[1]))
{
int i;
i = skipName(++cp);
if (!nest)
{
if (i == 6 && !yaz_strncasecmp(cp, "script", i))
{
int ws = skipSpace(cp + 6);
if (cp[ws + 6] == '>')
nest = true; /* really terminated */
}
if (!nest)
continue;
}
tagText(event, text_start, cp - 2);
event.closeTag(cp, i);
if (m_verbose)
printf("------ tag close %.*s\n", i, cp);
i += tagEnd(event, cp, i, cp + i);
cp += i;
text_start = cp;
}
else if (nest && isAlpha(*cp))
{
int i, j;
tagText(event, text_start, cp - 1);
i = skipName(cp);
event.openTagStart(cp, i);
if (m_verbose)
printf("------ tag open %.*s\n", i, cp);
j = tagAttrs(event, cp, i, cp + i);
j += tagEnd(event, cp, i, cp + i + j);
if (i == 6 && !yaz_strncasecmp(cp, "script", i))
nest = false;
cp += i + j;
text_start = cp;
}
}
tagText(event, text_start, cp);
}