function autoclose_tag_cleanup(&$sample_html, $offset, $tags_raw, $nested, $close)
{
$tags = mk_open_tag_regexp($tags_raw);
skip_to($sample_html, $offset, $tags_raw);
while (preg_match("#^(.*?)({$tags})#is", substr($sample_html, $offset), $matches)) {
// convert tag name found to lower case
$tag = strtolower($matches[3]);
// calculate position of the tag found
$tag_start = $offset + strlen($matches[1]);
$tag_end = $tag_start + strlen($matches[2]);
if ($tag == $close) {
return $tag_end;
}
// REQ: PHP 4.0.5
if (isset($nested[$tag])) {
$offset = $nested[$tag]($sample_html, $tag_end);
} else {
$to_be_inserted = "<" . $close . ">";
$sample_html = substr_replace($sample_html, $to_be_inserted, $tag_start, 0);
return $tag_start + strlen($to_be_inserted);
}
skip_to($sample_html, $offset, $tags_raw);
}
return $offset;
}
function process_p($sample_html)
{
$open_regexp = implode("|", array("p", "dl", "div", "noscript", "blockquote", "form", "hr", "table", "fieldset", "address", "ul", "ol", "li", "h1", "h2", "h3", "h4", "h5", "h6", "pre", "frameset", "noframes"));
$close_regexp = implode("|", array("dl", "div", "noscript", "blockquote", "form", "hr", "table", "fieldset", "address", "ul", "ol", "li", "h1", "h2", "h3", "h4", "h5", "h6", "pre", "frameset", "noframes", "body"));
$open = mk_open_tag_regexp("(" . $open_regexp . ")");
$close = mk_close_tag_regexp("(" . $close_regexp . ")");
$offset = 0;
while (preg_match("#^(.*?)(<\\s*p(\\s+[^>]*?)?>)(.*?)({$open}|{$close})#is", substr($sample_html, $offset), $matches)) {
if (!preg_match("#<\\s*/\\s*p\\s*>#is", $matches[3])) {
$cutpos = $offset + strlen($matches[1]) + strlen($matches[2]) + strlen($matches[4]);
$sample_html = substr_replace($sample_html, "</p>", $cutpos, 0);
$offset = $cutpos + 4;
} else {
$offset += strlen($matches[1]) + 1;
}
}
return $sample_html;
}
function process_body($html)
{
$open = mk_open_tag_regexp("body");
$close = mk_close_tag_regexp("body");
$ohtml = mk_open_tag_regexp("html");
$chtml = mk_close_tag_regexp("html");
$chead = mk_close_tag_regexp("head");
if (!preg_match("#{$open}#is", $html)) {
if (preg_match("#{$chead}#is", $html)) {
$html = preg_replace("#({$chead})#is", "\\1<body>", $html);
} else {
$html = preg_replace("#({$ohtml})#is", "\\1<body>", $html);
}
}
if (!preg_match("#{$close}#is", $html)) {
$html = preg_replace("#({$chtml})#is", "</body>\\1", $html);
}
// Now check is there any data between </head> and <body>.
$html = preg_replace("#({$chead})(.+)({$open})#is", "\\1\\3\\2", $html);
// Check if there's any data between </body> and </html>
$html = preg_replace("#({$close})(.+)({$chtml})#is", "\\2\\1\\3", $html);
return $html;
}
