function html2xhtml($html)
{
process_pagebreak_commands($html);
// Remove SCRIPT tags from the page being processed, as script content may
// mess the firther html-parsing utilities
$html = process_script($html);
// Remove STYLE tags for the same reason and store them in the temporary variable
// later they will be added back to HEAD section
$styles = process_style($html);
// Convert HTML character references to their Unicode analogues
process_character_references($html);
remove_comments($html);
fix_attrs_spaces($html);
$html = quote_attrs($html);
$html = escape_attrs_entities($html);
$html = lowercase_tags($html);
$html = lowercase_closing_tags($html);
$html = fix_closing_tags($html);
$html = close_tag("area", $html);
$html = close_tag("base", $html);
$html = close_tag("basefont", $html);
$html = close_tag("br", $html);
$html = close_tag("col", $html);
$html = close_tag("embed", $html);
$html = close_tag("frame", $html);
$html = close_tag("hr", $html);
$html = close_tag("img", $html);
$html = close_tag("input", $html);
$html = close_tag("isindex", $html);
$html = close_tag("link", $html);
$html = close_tag("meta", $html);
$html = close_tag("param", $html);
$html = make_attr_value("checked", $html);
$html = make_attr_value("compact", $html);
$html = make_attr_value("declare", $html);
$html = make_attr_value("defer", $html);
$html = make_attr_value("disabled", $html);
$html = make_attr_value("ismap", $html);
$html = make_attr_value("multiple", $html);
$html = make_attr_value("nohref", $html);
$html = make_attr_value("noresize", $html);
$html = make_attr_value("noshade", $html);
$html = make_attr_value("nowrap", $html);
$html = make_attr_value("readonly", $html);
$html = make_attr_value("selected", $html);
$html = process_html($html);
$html = process_body($html);
$html = process_head($html);
$html = process_p($html);
$html = escape_amp($html);
$html = escape_lt($html);
$html = escape_gt($html);
$html = escape_textarea_content($html);
process_tables($html, 0);
process_lists($html, 0);
process_deflists($html, 0);
process_selects($html, 0);
$html = fix_tags($html);
$html = fix_attrs($html);
$html = insert_styles($html, $styles);
return $html;
}
function sn_submit($url)
{
if ($url == "") {
return False;
}
$url = get_redirected_url($url);
if ($url === False) {
privmsg("error: unable to download source (get_redirected_url)");
return False;
}
$host = "";
$uri = "";
$port = 80;
if (get_host_and_uri($url, $host, $uri, $port) == False) {
privmsg("error: unable to download source (get_host_and_uri)");
return False;
}
$response = wget($host, $uri, $port);
if (get_host_and_uri($url, $host, $uri, $port) == False) {
privmsg("error: unable to download source (wget)");
return False;
}
$source_html = strip_headers($response);
$source_title = extract_raw_tag($source_html, "title");
$delimiters = array("--", "|", " - ", " : ", " — ", " • ");
for ($i = 0; $i < count($delimiters); $i++) {
$j = strpos($source_title, $delimiters[$i]);
if ($j !== False) {
$source_title = trim(substr($source_title, 0, $j));
}
}
if ($source_title === False or $source_title == "") {
privmsg("error: title not found or empty");
return False;
}
$source_title = html_decode($source_title);
$source_title = html_decode($source_title);
$source_body = extract_meta_content($source_html, "description");
if ($source_body === False or $source_body == "") {
$source_body = extract_meta_content($source_html, "og:description", "property");
if ($source_body === False or $source_body == "") {
privmsg("error: description meta content not found or empty");
return False;
}
}
$html = $source_html;
$article = extract_raw_tag($html, "article");
if ($article !== False) {
$html = $article;
}
strip_all_tag($html, "head");
strip_all_tag($html, "script");
strip_all_tag($html, "style");
#strip_all_tag($html,"a");
strip_all_tag($html, "strong");
$html = strip_tags($html, "<p>");
$html = lowercase_tags($html);
$html = explode("<p", $html);
$source_body = array();
for ($i = 0; $i < count($html); $i++) {
$parts = explode(">", $html[$i]);
if (count($parts) >= 2) {
array_shift($parts);
$html[$i] = implode(">", $parts);
}
$html[$i] = strip_tags($html[$i]);
$html[$i] = clean_text($html[$i]);
$host_parts = explode(".", $host);
for ($j = 0; $j < count($host_parts); $j++) {
if (strlen($host_parts[$j]) > 3) {
if (strpos(strtolower($html[$i]), strtolower($host_parts[$j])) !== False) {
continue 2;
}
}
}
if (filter($html[$i], "0123456789") != "") {
continue;
}
if (strlen($html[$i]) > 1) {
if ($html[$i][strlen($html[$i]) - 1] != ".") {
continue;
}
while (True) {
$j = strlen($html[$i]) - 1;
if ($j < 0) {
break;
}
$c = $html[$i][$j];
if ($c == ".") {
break;
}
$html[$i] = substr($html[$i], 0, $j);
}
}
if (strlen($html[$i]) > 100) {
$source_body[] = $html[$i];
}
}
$source_body = implode("\n\n", $source_body);
$source_body = html_decode($source_body);
$source_body = html_decode($source_body);
$host = "dev.soylentnews.org";
$port = 443;
$uri = "/submit.pl";
$response = wget($host, $uri, $port, ICEWEASEL_UA);
$html = strip_headers($response);
$reskey = extract_text($html, "<input type=\"hidden\" id=\"reskey\" name=\"reskey\" value=\"", "\">");
if ($reskey === False) {
privmsg("error: unable to extract reskey");
return False;
}
sleep(25);
$params = array();
$params["reskey"] = $reskey;
#$params["name"]=trim(substr($nick,0,50));
$params["name"] = get_bot_nick();
$params["email"] = "";
$params["subj"] = trim(substr($source_title, 0, 100));
$params["primaryskid"] = "1";
$params["tid"] = "6";
$params["sub_type"] = "plain";
$params["story"] = $source_body . "\n\n" . $url . "\n\n-- submitted from IRC";
$params["op"] = "SubmitStory";
$response = wpost($host, $uri, $port, ICEWEASEL_UA, $params);
$html = strip_headers($response);
strip_all_tag($html, "head");
strip_all_tag($html, "script");
strip_all_tag($html, "style");
strip_all_tag($html, "a");
$html = strip_tags($html);
$html = clean_text($html);
if (strpos($html, "Perhaps you would like to enter an email address or a URL next time. Thanks for the submission.") !== False) {
privmsg("submission successful - https://{$host}/submit.pl?op=list");
return True;
} else {
privmsg("error: something went wrong with your submission");
return False;
}
}
