/** ************************************************
* Trims whitespace off the url, then removes the /
* off the right side of the url if it exists
* @param:
* @return: null
* @throws:
** ***********************************************/
public function setURL($url)
{
if (is_string($url) && WebUtils::isValidURL($url)) {
$url = rtrim(trim($url), '/');
$this->sURL = $url;
} else {
throw new Exception('LogParser.setURL accepts valid URLs only.');
}
}
/** ************************************************
* This function tries to travers the given url's path
* but will not work if the realtive path is not traversable
* with the given url.
* @param: $url string; a url
* @param: $relPath string; a relative path
* @return: string; the url after it was traversed by
* relPath. Or false if there was some error/problem
* TODO: Do a better job of adding / in the return string
***************************************************/
public static function traversRelPath($sUrl, $relPath)
{
if (!WebUtils::isValidURL($sUrl)) {
return false;
}
$baseP = WebUtils::getCurrentPathURL($sUrl);
$baseU = WebUtils::getBaseURL($sUrl);
$path = WebUtils::getPathOnly($sUrl);
$path = explode('/', $path);
$relPath = explode('/', $relPath);
//echo '<pre>Path '. var_export($path, true) . '</pre><br/>';
//echo '<pre>rel '. var_export($relPath, true) . '</pre><br/>';
foreach ($relPath as $r) {
if ($r != '.' && $r != '..') {
array_push($path, array_shift($relPath));
continue;
} else {
if ($r == '.') {
array_shift($relPath);
continue;
} else {
if ($r == '..') {
array_shift($relPath);
if (array_pop($path) == null) {
return false;
}
}
}
}
}
return rtrim($baseU . implode('/', $path) . implode('/', $relPath), '/');
}
/** ************************************************
* Put a valid URL on the stack. Use the url as a key
* to eliminate duplicates. It also makes sure the URL
* is within $this->aDomainLimits.
* @param: $sLink string; A url
* @return: null
* @throws: Exception
* DEPRICATED use pushTraversed
***************************************************/
public function pushToHash($sLink)
{
if (is_string($sLink) && WebUtils::isValidURL($sLink)) {
if ($this->isWithInDomain($sLink)) {
$this->aLinkStack[$sLink] = $sLink;
echo '<span style="margin-left:20px;">Starting with: </span>' . $sLink . '<br/>';
}
} else {
throw new Exception('Spiderman.pushToHash only accepts valid URL strings \'' . $sLink . '\'');
}
}
/** ************************************************
* Turns all relative URLs into absolute URLs
* @return: null
* Notes:
* Different forms of urls
* /dir/
* /dir
* /dir/index.php
* dir/index.php
* dir/index.php?var=v
* index.php
* /index.php
* ./dir
* ../../dir
* <a href="javascript:;" class="prev"></a>
***************************************************/
private function processURLs()
{
$tmpArray = array();
foreach ($this->aAnchors as $url) {
$url = trim($url);
echo $this->sSay('I\'m going to process url: ', $url);
if (WebUtils::isValidURL($url)) {
$url = rtrim($url, '/');
$tmpArray[] = $url;
//echo $this->sSay('After Processing url: ', $url);
} else {
if (strlen($url) <= 0) {
continue;
}
if ($this->containsJS($url)) {
continue;
}
if ($this->containsHash($url)) {
continue;
}
//TODO: Not sure about this might not need it. URLs can contain hashes but link to a differnt page?
if ($this->containsMailTo($url)) {
continue;
}
if (strpos($url, '/') === 0) {
//means root.
$url = rtrim($url, '/');
$tmpArray[] = $this->sBaseURL . $url;
} else {
if (strpos($url, './') === 0 || strpos($url, '../') === 0) {
//realative path try and travers
$url = rtrim($url, '/');
echo $this->sSay('I\'ve found a relative path ', $url);
$rtn = WebUtils::traversRelPath($this->sURL, $url);
echo $this->sSay('I\'ve tried to travers it and arrived at ', $rtn);
if (WebUtils::isValidURL($rtn)) {
$tmpArray[] = $rtn;
} else {
echo $this->sSay('Relative url parse failed. url is', $rtn);
}
} else {
//who knows just add it;
$url = rtrim($url, '/');
echo $this->sSay('Not sure what to do with this url: ', $url);
$tmpArray[] = $this->sCurrentPathURL . '/' . $url;
echo $this->sSay('So I\'ll add on the current path: ', $this->sCurrentPathURL . '/' . $url);
}
}
//echo $this->sSay('After Processing url: ', $tmpArray[count($tmpArray)-1]);
}
}
$this->aAnchors = null;
$this->aAnchors = $tmpArray;
}
