Edit File by line

// The start tag cannot contain another start tag, if so add as text

[2000] Fix | Delete

// i.e. "<<html>"

[2001] Fix | Delete

if ($pos = strpos($tag, '<') !== false) {

[2002] Fix | Delete

$tag = '<' . substr($tag, 0, -1);

[2003] Fix | Delete

$node->_[HDOM_INFO_TEXT] = $tag;

[2004] Fix | Delete

$this->link_nodes($node, false);

[2005] Fix | Delete

$this->char = $this->doc[--$this->pos]; // prev

[2006] Fix | Delete

return true;

[2007] Fix | Delete

}

[2008] Fix | Delete

[2009] Fix | Delete

// Handle invalid tag names (i.e. "<html#doc>")

[2010] Fix | Delete

if (!preg_match('/^\w[\w:-]*$/', $tag)) {

[2011] Fix | Delete

$node->_[HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until('<>');

[2012] Fix | Delete

[2013] Fix | Delete

// Next char is the beginning of a new tag, don't touch it.

[2014] Fix | Delete

if ($this->char === '<') {

[2015] Fix | Delete

$this->link_nodes($node, false);

[2016] Fix | Delete

return true;

[2017] Fix | Delete

}

[2018] Fix | Delete

[2019] Fix | Delete

// Next char closes current tag, add and be done with it.

[2020] Fix | Delete

if ($this->char === '>') { $node->_[HDOM_INFO_TEXT] .= '>'; }

[2021] Fix | Delete

$this->link_nodes($node, false);

[2022] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2023] Fix | Delete

return true;

[2024] Fix | Delete

}

[2025] Fix | Delete

[2026] Fix | Delete

// begin tag, add new node

[2027] Fix | Delete

$node->nodetype = HDOM_TYPE_ELEMENT;

[2028] Fix | Delete

$tag_lower = strtolower($tag);

[2029] Fix | Delete

$node->tag = ($this->lowercase) ? $tag_lower : $tag;

[2030] Fix | Delete

[2031] Fix | Delete

// handle optional closing tags

[2032] Fix | Delete

if (isset($this->optional_closing_tags[$tag_lower])) {

[2033] Fix | Delete

// Traverse ancestors to close all optional closing tags

[2034] Fix | Delete

while (isset($this->optional_closing_tags[$tag_lower][strtolower($this->parent->tag)])) {

[2035] Fix | Delete

$this->parent->_[HDOM_INFO_END] = 0;

[2036] Fix | Delete

$this->parent = $this->parent->parent;

[2037] Fix | Delete

}

[2038] Fix | Delete

$node->parent = $this->parent;

[2039] Fix | Delete

}

[2040] Fix | Delete

[2041] Fix | Delete

$guard = 0; // prevent infinity loop

[2042] Fix | Delete

[2043] Fix | Delete

// [0] Space between tag and first attribute

[2044] Fix | Delete

$space = array($this->copy_skip($this->token_blank), '', '');

[2045] Fix | Delete

[2046] Fix | Delete

// attributes

[2047] Fix | Delete

do {

[2048] Fix | Delete

// Everything until the first equal sign should be the attribute name

[2049] Fix | Delete

$name = $this->copy_until($this->token_equal);

[2050] Fix | Delete

[2051] Fix | Delete

if ($name === '' && $this->char !== null && $space[0] === '') {

[2052] Fix | Delete

break;

[2053] Fix | Delete

}

[2054] Fix | Delete

[2055] Fix | Delete

if ($guard === $this->pos) { // Escape infinite loop

[2056] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2057] Fix | Delete

continue;

[2058] Fix | Delete

}

[2059] Fix | Delete

[2060] Fix | Delete

$guard = $this->pos;

[2061] Fix | Delete

[2062] Fix | Delete

// handle endless '<'

[2063] Fix | Delete

// Out of bounds before the tag ended

[2064] Fix | Delete

if ($this->pos >= $this->size - 1 && $this->char !== '>') {

[2065] Fix | Delete

$node->nodetype = HDOM_TYPE_TEXT;

[2066] Fix | Delete

$node->_[HDOM_INFO_END] = 0;

[2067] Fix | Delete

$node->_[HDOM_INFO_TEXT] = '<' . $tag . $space[0] . $name;

[2068] Fix | Delete

$node->tag = 'text';

[2069] Fix | Delete

$this->link_nodes($node, false);

[2070] Fix | Delete

return true;

[2071] Fix | Delete

}

[2072] Fix | Delete

[2073] Fix | Delete

// handle mismatch '<'

[2074] Fix | Delete

// Attributes cannot start after opening tag

[2075] Fix | Delete

if ($this->doc[$this->pos - 1] == '<') {

[2076] Fix | Delete

$node->nodetype = HDOM_TYPE_TEXT;

[2077] Fix | Delete

$node->tag = 'text';

[2078] Fix | Delete

$node->attr = array();

[2079] Fix | Delete

$node->_[HDOM_INFO_END] = 0;

[2080] Fix | Delete

$node->_[HDOM_INFO_TEXT] = substr(

[2081] Fix | Delete

$this->doc,

[2082] Fix | Delete

$begin_tag_pos,

[2083] Fix | Delete

$this->pos - $begin_tag_pos - 1

[2084] Fix | Delete

);

[2085] Fix | Delete

$this->pos -= 2;

[2086] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2087] Fix | Delete

$this->link_nodes($node, false);

[2088] Fix | Delete

return true;

[2089] Fix | Delete

}

[2090] Fix | Delete

[2091] Fix | Delete

if ($name !== '/' && $name !== '') { // this is a attribute name

[2092] Fix | Delete

// [1] Whitespace after attribute name

[2093] Fix | Delete

$space[1] = $this->copy_skip($this->token_blank);

[2094] Fix | Delete

[2095] Fix | Delete

$name = $this->restore_noise($name); // might be a noisy name

[2096] Fix | Delete

[2097] Fix | Delete

if ($this->lowercase) { $name = strtolower($name); }

[2098] Fix | Delete

[2099] Fix | Delete

if ($this->char === '=') { // attribute with value

[2100] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2101] Fix | Delete

$this->parse_attr($node, $name, $space); // get attribute value

[2102] Fix | Delete

} else {

[2103] Fix | Delete

//no value attr: nowrap, checked selected...

[2104] Fix | Delete

$node->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_NO;

[2105] Fix | Delete

$node->attr[$name] = true;

[2106] Fix | Delete

if ($this->char != '>') { $this->char = $this->doc[--$this->pos]; } // prev

[2107] Fix | Delete

}

[2108] Fix | Delete

[2109] Fix | Delete

$node->_[HDOM_INFO_SPACE][] = $space;

[2110] Fix | Delete

[2111] Fix | Delete

// prepare for next attribute

[2112] Fix | Delete

$space = array(

[2113] Fix | Delete

$this->copy_skip($this->token_blank),

[2114] Fix | Delete

'',

[2115] Fix | Delete

[2116] Fix | Delete

);

[2117] Fix | Delete

} else { // no more attributes

[2118] Fix | Delete

break;

[2119] Fix | Delete

}

[2120] Fix | Delete

} while ($this->char !== '>' && $this->char !== '/'); // go until the tag ended

[2121] Fix | Delete

[2122] Fix | Delete

$this->link_nodes($node, true);

[2123] Fix | Delete

$node->_[HDOM_INFO_ENDSPACE] = $space[0];

[2124] Fix | Delete

[2125] Fix | Delete

// handle empty tags (i.e. "<div/>")

[2126] Fix | Delete

if ($this->copy_until_char('>') === '/') {

[2127] Fix | Delete

$node->_[HDOM_INFO_ENDSPACE] .= '/';

[2128] Fix | Delete

$node->_[HDOM_INFO_END] = 0;

[2129] Fix | Delete

} else {

[2130] Fix | Delete

// reset parent

[2131] Fix | Delete

if (!isset($this->self_closing_tags[strtolower($node->tag)])) {

[2132] Fix | Delete

$this->parent = $node;

[2133] Fix | Delete

}

[2134] Fix | Delete

}

[2135] Fix | Delete

[2136] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2137] Fix | Delete

[2138] Fix | Delete

// If it's a BR tag, we need to set it's text to the default text.

[2139] Fix | Delete

// This way when we see it in plaintext, we can generate formatting that the user wants.

[2140] Fix | Delete

// since a br tag never has sub nodes, this works well.

[2141] Fix | Delete

if ($node->tag === 'br') {

[2142] Fix | Delete

$node->_[HDOM_INFO_INNER] = $this->default_br_text;

[2143] Fix | Delete

}

[2144] Fix | Delete

[2145] Fix | Delete

return true;

[2146] Fix | Delete

}

[2147] Fix | Delete

[2148] Fix | Delete

protected function parse_attr($node, $name, &$space)

[2149] Fix | Delete

{

[2150] Fix | Delete

$is_duplicate = isset($node->attr[$name]);

[2151] Fix | Delete

[2152] Fix | Delete

if (!$is_duplicate) // Copy whitespace between "=" and value

[2153] Fix | Delete

$space[2] = $this->copy_skip($this->token_blank);

[2154] Fix | Delete

[2155] Fix | Delete

switch ($this->char) {

[2156] Fix | Delete

case '"':

[2157] Fix | Delete

$quote_type = HDOM_QUOTE_DOUBLE;

[2158] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2159] Fix | Delete

$value = $this->copy_until_char('"');

[2160] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2161] Fix | Delete

break;

[2162] Fix | Delete

case '\'':

[2163] Fix | Delete

$quote_type = HDOM_QUOTE_SINGLE;

[2164] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2165] Fix | Delete

$value = $this->copy_until_char('\'');

[2166] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2167] Fix | Delete

break;

[2168] Fix | Delete

default:

[2169] Fix | Delete

$quote_type = HDOM_QUOTE_NO;

[2170] Fix | Delete

$value = $this->copy_until($this->token_attr);

[2171] Fix | Delete

}

[2172] Fix | Delete

[2173] Fix | Delete

$value = $this->restore_noise($value);

[2174] Fix | Delete

[2175] Fix | Delete

// PaperG: Attributes should not have \r or \n in them, that counts as

[2176] Fix | Delete

// html whitespace.

[2177] Fix | Delete

$value = str_replace("\r", '', $value);

[2178] Fix | Delete

$value = str_replace("\n", '', $value);

[2179] Fix | Delete

[2180] Fix | Delete

// PaperG: If this is a "class" selector, lets get rid of the preceeding

[2181] Fix | Delete

// and trailing space since some people leave it in the multi class case.

[2182] Fix | Delete

if ($name === 'class') {

[2183] Fix | Delete

$value = trim($value);

[2184] Fix | Delete

}

[2185] Fix | Delete

[2186] Fix | Delete

if (!$is_duplicate) {

[2187] Fix | Delete

$node->_[HDOM_INFO_QUOTE][] = $quote_type;

[2188] Fix | Delete

$node->attr[$name] = $value;

[2189] Fix | Delete

}

[2190] Fix | Delete

}

[2191] Fix | Delete

[2192] Fix | Delete

protected function link_nodes(&$node, $is_child)

[2193] Fix | Delete

{

[2194] Fix | Delete

$node->parent = $this->parent;

[2195] Fix | Delete

$this->parent->nodes[] = $node;

[2196] Fix | Delete

if ($is_child) {

[2197] Fix | Delete

$this->parent->children[] = $node;

[2198] Fix | Delete

}

[2199] Fix | Delete

}

[2200] Fix | Delete

[2201] Fix | Delete

protected function as_text_node($tag)

[2202] Fix | Delete

{

[2203] Fix | Delete

$node = new simple_html_dom_node($this);

[2204] Fix | Delete

++$this->cursor;

[2205] Fix | Delete

$node->_[HDOM_INFO_TEXT] = '</' . $tag . '>';

[2206] Fix | Delete

$this->link_nodes($node, false);

[2207] Fix | Delete

$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2208] Fix | Delete

return true;

[2209] Fix | Delete

}

[2210] Fix | Delete

[2211] Fix | Delete

protected function skip($chars)

[2212] Fix | Delete

{

[2213] Fix | Delete

$this->pos += strspn($this->doc, $chars, $this->pos);

[2214] Fix | Delete

$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2215] Fix | Delete

}

[2216] Fix | Delete

[2217] Fix | Delete

protected function copy_skip($chars)

[2218] Fix | Delete

{

[2219] Fix | Delete

$pos = $this->pos;

[2220] Fix | Delete

$len = strspn($this->doc, $chars, $pos);

[2221] Fix | Delete

$this->pos += $len;

[2222] Fix | Delete

$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2223] Fix | Delete

if ($len === 0) { return ''; }

[2224] Fix | Delete

return substr($this->doc, $pos, $len);

[2225] Fix | Delete

}

[2226] Fix | Delete

[2227] Fix | Delete

protected function copy_until($chars)

[2228] Fix | Delete

{

[2229] Fix | Delete

$pos = $this->pos;

[2230] Fix | Delete

$len = strcspn($this->doc, $chars, $pos);

[2231] Fix | Delete

$this->pos += $len;

[2232] Fix | Delete

$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next

[2233] Fix | Delete

return substr($this->doc, $pos, $len);

[2234] Fix | Delete

}

[2235] Fix | Delete

[2236] Fix | Delete

protected function copy_until_char($char)

[2237] Fix | Delete

{

[2238] Fix | Delete

if ($this->char === null) { return ''; }

[2239] Fix | Delete

[2240] Fix | Delete

if (($pos = strpos($this->doc, $char, $this->pos)) === false) {

[2241] Fix | Delete

$ret = substr($this->doc, $this->pos, $this->size - $this->pos);

[2242] Fix | Delete

$this->char = null;

[2243] Fix | Delete

$this->pos = $this->size;

[2244] Fix | Delete

return $ret;

[2245] Fix | Delete

}

[2246] Fix | Delete

[2247] Fix | Delete

if ($pos === $this->pos) { return ''; }

[2248] Fix | Delete

[2249] Fix | Delete

$pos_old = $this->pos;

[2250] Fix | Delete

$this->char = $this->doc[$pos];

[2251] Fix | Delete

$this->pos = $pos;

[2252] Fix | Delete

return substr($this->doc, $pos_old, $pos - $pos_old);

[2253] Fix | Delete

}

[2254] Fix | Delete

[2255] Fix | Delete

protected function remove_noise($pattern, $remove_tag = false)

[2256] Fix | Delete

{

[2257] Fix | Delete

global $debug_object;

[2258] Fix | Delete

if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

[2259] Fix | Delete

[2260] Fix | Delete

$count = preg_match_all(

[2261] Fix | Delete

$pattern,

[2262] Fix | Delete

$this->doc,

[2263] Fix | Delete

$matches,

[2264] Fix | Delete

PREG_SET_ORDER | PREG_OFFSET_CAPTURE

[2265] Fix | Delete

);

[2266] Fix | Delete

[2267] Fix | Delete

for ($i = $count - 1; $i > -1; --$i) {

[2268] Fix | Delete

$key = '___noise___' . sprintf('% 5d', count($this->noise) + 1000);

[2269] Fix | Delete

[2270] Fix | Delete

if (is_object($debug_object)) {

[2271] Fix | Delete

$debug_object->debug_log(2, 'key is: ' . $key);

[2272] Fix | Delete

}

[2273] Fix | Delete

[2274] Fix | Delete

$idx = ($remove_tag) ? 0 : 1; // 0 = entire match, 1 = submatch

[2275] Fix | Delete

$this->noise[$key] = $matches[$i][$idx][0];

[2276] Fix | Delete

$this->doc = substr_replace($this->doc, $key, $matches[$i][$idx][1], strlen($matches[$i][$idx][0]));

[2277] Fix | Delete

}

[2278] Fix | Delete

[2279] Fix | Delete

// reset the length of content

[2280] Fix | Delete

$this->size = strlen($this->doc);

[2281] Fix | Delete

[2282] Fix | Delete

if ($this->size > 0) {

[2283] Fix | Delete

$this->char = $this->doc[0];

[2284] Fix | Delete

}

[2285] Fix | Delete

}

[2286] Fix | Delete

[2287] Fix | Delete

function restore_noise($text)

[2288] Fix | Delete

{

[2289] Fix | Delete

global $debug_object;

[2290] Fix | Delete

if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

[2291] Fix | Delete

[2292] Fix | Delete

while (($pos = strpos($text, '___noise___')) !== false) {

[2293] Fix | Delete

// Sometimes there is a broken piece of markup, and we don't GET the

[2294] Fix | Delete

// pos+11 etc... token which indicates a problem outside of us...

[2295] Fix | Delete

[2296] Fix | Delete

// todo: "___noise___1000" (or any number with four or more digits)

[2297] Fix | Delete

// in the DOM causes an infinite loop which could be utilized by

[2298] Fix | Delete

// malicious software

[2299] Fix | Delete

if (strlen($text) > $pos + 15) {

[2300] Fix | Delete

$key = '___noise___'

[2301] Fix | Delete

. $text[$pos + 11]

[2302] Fix | Delete

. $text[$pos + 12]

[2303] Fix | Delete

. $text[$pos + 13]

[2304] Fix | Delete

. $text[$pos + 14]

[2305] Fix | Delete

. $text[$pos + 15];

[2306] Fix | Delete

[2307] Fix | Delete

if (is_object($debug_object)) {

[2308] Fix | Delete

$debug_object->debug_log(2, 'located key of: ' . $key);

[2309] Fix | Delete

}

[2310] Fix | Delete

[2311] Fix | Delete

if (isset($this->noise[$key])) {

[2312] Fix | Delete

$text = substr($text, 0, $pos)

[2313] Fix | Delete

. $this->noise[$key]

[2314] Fix | Delete

. substr($text, $pos + 16);

[2315] Fix | Delete

} else {

[2316] Fix | Delete

// do this to prevent an infinite loop.

[2317] Fix | Delete

$text = substr($text, 0, $pos)

[2318] Fix | Delete

. 'UNDEFINED NOISE FOR KEY: '

[2319] Fix | Delete

. $key

[2320] Fix | Delete

. substr($text, $pos + 16);

[2321] Fix | Delete

}

[2322] Fix | Delete

} else {

[2323] Fix | Delete

// There is no valid key being given back to us... We must get

[2324] Fix | Delete

// rid of the ___noise___ or we will have a problem.

[2325] Fix | Delete

$text = substr($text, 0, $pos)

[2326] Fix | Delete

. 'NO NUMERIC NOISE KEY'

[2327] Fix | Delete

. substr($text, $pos + 11);

[2328] Fix | Delete

}

[2329] Fix | Delete

}

[2330] Fix | Delete

return $text;

[2331] Fix | Delete

}

[2332] Fix | Delete

[2333] Fix | Delete

function search_noise($text)

[2334] Fix | Delete

{

[2335] Fix | Delete

global $debug_object;

[2336] Fix | Delete

if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }

[2337] Fix | Delete

[2338] Fix | Delete

foreach($this->noise as $noiseElement) {

[2339] Fix | Delete

if (strpos($noiseElement, $text) !== false) {

[2340] Fix | Delete

return $noiseElement;

[2341] Fix | Delete

}

[2342] Fix | Delete

}

[2343] Fix | Delete

}

[2344] Fix | Delete

[2345] Fix | Delete

function __toString()

[2346] Fix | Delete

{

[2347] Fix | Delete

return $this->root->innertext();

[2348] Fix | Delete

}

[2349] Fix | Delete

[2350] Fix | Delete

function __get($name)

[2351] Fix | Delete

{

[2352] Fix | Delete

switch ($name) {

[2353] Fix | Delete

case 'outertext':

[2354] Fix | Delete

return $this->root->innertext();

[2355] Fix | Delete

case 'innertext':

[2356] Fix | Delete

return $this->root->innertext();

[2357] Fix | Delete

case 'plaintext':

[2358] Fix | Delete

return $this->root->text();

[2359] Fix | Delete

case 'charset':

[2360] Fix | Delete

return $this->_charset;

[2361] Fix | Delete

case 'target_charset':

[2362] Fix | Delete

return $this->_target_charset;

[2363] Fix | Delete

}

[2364] Fix | Delete

}

[2365] Fix | Delete

[2366] Fix | Delete

function childNodes($idx = -1)

[2367] Fix | Delete

{

[2368] Fix | Delete

return $this->root->childNodes($idx);

[2369] Fix | Delete

}

[2370] Fix | Delete

[2371] Fix | Delete

function firstChild()

[2372] Fix | Delete

{

[2373] Fix | Delete

return $this->root->first_child();

[2374] Fix | Delete

}

[2375] Fix | Delete

[2376] Fix | Delete

function lastChild()

[2377] Fix | Delete

{

[2378] Fix | Delete

return $this->root->last_child();

[2379] Fix | Delete

}

[2380] Fix | Delete

[2381] Fix | Delete

function createElement($name, $value = null)

[2382] Fix | Delete

{

[2383] Fix | Delete

return @str_get_html("<$name>$value</$name>")->firstChild();

[2384] Fix | Delete

}

[2385] Fix | Delete

[2386] Fix | Delete

function createTextNode($value)

[2387] Fix | Delete

{

[2388] Fix | Delete

return @end(str_get_html($value)->nodes);

[2389] Fix | Delete

}

[2390] Fix | Delete

[2391] Fix | Delete

function getElementById($id)

[2392] Fix | Delete

{

[2393] Fix | Delete

return $this->find("#$id", 0);

[2394] Fix | Delete

}

[2395] Fix | Delete

[2396] Fix | Delete

function getElementsById($id, $idx = null)

[2397] Fix | Delete

{

[2398] Fix | Delete

return $this->find("#$id", $idx);

[2399] Fix | Delete

}

[2400] Fix | Delete

[2401] Fix | Delete

function getElementByTagName($name)

[2402] Fix | Delete

{

[2403] Fix | Delete

return $this->find($name, 0);

[2404] Fix | Delete

}

[2405] Fix | Delete

[2406] Fix | Delete

function getElementsByTagName($name, $idx = -1)

[2407] Fix | Delete

{

[2408] Fix | Delete

return $this->find($name, $idx);

[2409] Fix | Delete

}

[2410] Fix | Delete

[2411] Fix | Delete

function loadFile()

[2412] Fix | Delete

{

[2413] Fix | Delete

$args = func_get_args();

[2414] Fix | Delete

$this->load_file($args);

[2415] Fix | Delete

}

[2416] Fix | Delete

}

[2417] Fix | Delete

[2418] Fix | Delete

1 2 3 45