trying to implement this loop

eric1

but it doesn't work, the idea is to delete all links on my links page that doesn't have a link pointing back to my site from their page (linkback).

if ($action == "check_link" && $valid) {
    $query = mysql_query("SELECT $link_table.ID, Website, URL, linkback FROM $link_table WHERE linkback  != '' ");
    while (list($LID, $Website, $URL, $linkback) = mysql_fetch_array($query)) {

 if ($linkback !== "adminok"){
           $recurl = $linkback;
           $linktext = "http://www.sunnymead.org";
           $filepointer = fopen($recurl,"r");

           if($filepointer){
                while(!feof($filepointer)){
                        $buffer = fgets($filepointer, 4096);
                        $file .= $buffer;
                 }
                 fclose($filepointer);
                 echo "$linkback was accessed. ";
                 $statcode = 1;
            }
            else {
                 echo "Coudln't access $linkback. ";
                 $statcode = 0;
            }

          if ($statcode == 1) {
                  $string = $linktext;
                  $pattern = "<a href=\"\/\">";
                  $replacement = "\/";
                  $newlinktext = preg_replace($pattern, $replacement, $string);
                  preg_match("<a href=\"(.*)$newlinktext(.*)\">",$file,$matches);

                  if (count($matches) != 0) {
                       echo "Success, '$linktext' was found on '$recurl'.<br>";
                        $returnValue = 1;
                  }
                  else {
                       echo "Failure, '$linktext' was not found on '$recurl'.<br>";
                       $returnValue = 0;
                 }
         }
}

    if ($returnValue == 0){
    echo "<div>Removing $LID , $Website as they didn't have a '$linkback'.</div>";
    mysql_query("DELETE FROM $link_table WHERE ID = $LID");
}
   }
}

The while loop isn't working, all found items get the same parameters as the first one.
If I just put:

if ($action == "check_link" && $valid) {
    $query = mysql_query("SELECT $link_table.ID, Website, URL, linkback FROM $link_table WHERE linkback  != '' ");
    while (list($LID, $Website, $URL, $linkback) = mysql_fetch_array($query)) {
    echo "<div>$LID , $Website with '$linkback'.</div>";
   }
}

it does add every lines, please help, thanks.
Eric

paulnaj

Hi eric,

There's a few things that need more explanation ...

Why are doing the preg_replace() on $linktext, when $linktext is a constant?

And why are you putting (.*) either side of $newlinktext?

Anyway, in order to try and work out what you are trying to do I messed around with your code and simplified a bit. I hope you don't take offence!

I've stuck in some comments as well!

Have a look ...

<?
if ($action == "check_link" && $valid) { 

// constant - leave outside the loop
$linktext = "http://www.sunnymead.org";

// can leave out the adminok's in the query!
$query = mysql_query("
	SELECT ".$link_table.".ID, Website, URL, linkback 
	FROM  ".$link_table." 
	WHERE linkback != '' 
	AND linkback != 'adminok' 
");

while(list($LID, $Website, $URL, $linkback) = mysql_fetch_array($query)) {

	// Just get the whole file in one go in an array!!
	$file = file($linkback); 
	if($file === false){
		echo "Couldn't access '".$linkback."'.<br>";
		// Go directly to the next iteration of the while loop
		continue; 
	}
	// Glue the array elements together to make one long string
	$file = join('', $file); 
	preg_match("<a href=\"".$linktext."\">", $file, $matches); 
	if (count($matches) != 0) { 
		echo "Success, '".$linktext."' was found on '".$linkback."'.<br>"; 
	} else { 
		echo "Failure ...removing ".$LID." , ".$Website." as they didn't have a '".$linkback."'.<br>"; 
		mysql_query("DELETE FROM ".$link_table." WHERE ID = ".$LID); 
	}
}
}
?>

If it works, then great, but I imagine that the preg_match() is going to have to be changed.

Paul.

eric1

Hi Paul,

Thanks for your answer. Your codes are much leaner, I have a definite tendency to complicate things. No offence taken, I'm here to learn.

After regular visits from spamers adding about 100 links a week I felt I needed a new system. As I started learning php/mysql a couple of months ago I thought this could be a good test.

So the front page will have categories each opening one list, people will therefore be able to add their link from within the proper category (often a mistake they did). A start session at the top of the add link module should prevent multiple posting. A reciprocal exchange procedure with a live verification should also minimise abuses while giving me back what seems fair: a link. And finally generating a html page just for the purpose of search engine access. (reading lots of tutorials and looking at other code allowed me to make it work, it certainly isn't good coding but up to this point those stages works).

But this didn't insure that the linkback was still there once the visitor submitted his and as on my links'page I had 488 links, I had to go through all of them manually and now 308 are left, those don't have linkbacks (hence the adminok). This was very long and I didn't see myself doing it every week.

So I though to reuse the code used during the link posting in a loop. As listing every row worked well, I belived I could add this code to 'validate' each row while letting me know if the external site was reached or not ($linkback), if my link was among the text of the page ($linktext), and if not delete the row from my database.

I used the preg_replace, preg match to make sure the code would be a real link and allow wildcards so people could add style conditions etc... (I thought * allowed wildcards for the replacement string).

Yes you are right the linktext doesn't need to be in the loop.

Yes getting all the file at once and searching for the match on a long string works, I used the buffer section because I didn't now the length of the file I was going to read and I've read that sometimes a loss of a packet might terminate the connection therefore I wouldn't get all the file.

Yes, not changing the preg_match didn't work as bad or good links were deleted while the couldn't access remained. But now it works. Would you know if it is possible to distinguish the reason the file wasn't accessed (either if it didn't exist or if the server couldn't be accessed)?

Thanks again for your help and advises.
eric1

eric1

Hi Paul,

I need another advise as I found a hole in my system.

As is, one can submit a link and use my url for linkback, I could state that the linkback != my URL but then one could put a link found on google and it will work.

How can I make sure the linkback originates from the submiter's server or site?

Thanks in advance,
eric

paulnaj

Hi eric,

Been thinking about your problem and have got a couple of ideas ...

1) I don't know a test to see whether the server is inaccessible ( offline, etc) or if the file does not exist. All of the filesystem functions (file, fopen, etc) just return FALSE if the function fails.
But, what you could do is put such links on 'standby' ... that is, don't delete them from your database straightaway, but also, don't allow the links onto your until the file is accessed.
You could put a timestamp field in the table for the 'creation date' and then maybe delete 'standby' links after a month, or whatever.

2) About searching the page for the link back to your site.

First of all, you could insist that the HTML is exactly as you specify. For example you could insist that the link is exactly ...

<a href="http://www.sunnymead.org">Sunnymead</a>

... and then do a very straightforward search using strstr() or strpos() and not have to worry about regular expressions.

If you wanted to allow a bit flexibility and let them use other attributes (class, border, target, etc) then you are going to need a regex. I could have a think about what that should look like if you want.

Paul. 🙂

PS. I didn't quite understand what you were trying to say in your last post ... could you try and explain it again.

Having said that, maybe the way to think about it is to assume that every link is malicious!! Then think up as many tests that would filter as many of them as you can. Then check by hand only those that remain.

eric1

Hi Paul,

For checking the state of the access I modified the echo to offer a link as such:

        if($file === false){
            echo "<span style=\"color: #008000;\">Couldn't access ".$linkback." for <a href=\"".$linkback."\" target=\"_blank\">".$Website."</a>.(".$LID.").</span><br>";
            // Go directly to the next iteration of the while loop
            continue; 
        }

while testing I thought that once the system is in function there couldn't be so many unaccessible pages so this does list unaccessed page with its ID, I simply click the href, if the page opens I search for the linkback, if not I take note of the LID and manually delete it as the loop doesn't delete it from the database.
It is simpler than checking the status code for a 200, 302 or 404. I could still make a cron to automate it but I would have anyhow to generate this list and go through it so as well directly validating it myself, I do think that the number of submission will greatly go down.

My last post shouldn't' t really be here but I found the problem while testing the loop.
When a visitor submit his site I validate the fields of the form with this:

if ($action == "submit"){
      if ($name == "") {myerror("Please enter your name to continue.");}
        elseif ($email == "") {myerror("Please enter your email to continue.");}
        elseif (!(preg_match("/([\w\-]+\@[\w\-]+\.[\w\-]+)/",$email))) {myerror("Please enter a valid e-mail address!");}
        elseif ($linkback == "http://" || $linkback == "") {myerror("Please enter the url where a reciprocal link to our site is placed!");}
        elseif ($linkback != "adminok" && !(preg_match("/(http:\/\/+[\w\-]+\.[\w\-]+)/i",$linkback))) {myerror("The reciprocal page URL is not valid!");}
        elseif ($URL == "http://" || $URL == "") {myerror("Please enter the url to your site.");}
        elseif (!(preg_match("/(http:\/\/+[\w\-]+\.[\w\-]+)/i",$URL))) {myerror("The site URL is not valid!");}
        elseif ($CID == "") {myerror("Please select a category for your site.");}
        elseif ($Website == "") {myerror("Please enter a title for your site.");}
        elseif (strlen($Website)>100) {myerror("The Description is too long! (length: " . strlen($Website) . ") The Title of your website is limited to 100 chars.");}
        elseif ($Description == "") {myerror("Please enter description of your site!");}
        elseif (strlen($Description)>200) {myerror("The Description is too long! (length: " . strlen($Description) . ") The Description of your website is limited to 200 chars.");}
        else
                {

As is, one can enter my url as linkback and the system will approve it, with a != myURL it wouldn't stop someone to make a search in google and post any of the results as a linkback (I tested it and it works).

What I was thinking to do is similar to the condition I use to start a session:
I use a request for a hidden variable in the submit form and depending of the initial query string it opens the session (this seems to stop using the back button to repost multiple times the same link).

So (and I don't know how to do this) if I inverse the URL & linkback I could check if their URL is contained within the linkback and validate their URL through a request hidden field using a query like $ENV{'HTTP_REFERER'} (but the manual states that HTTP_REFERER cannot really be trusted). The point is the linkback has to originate from the submitter's website/server... and not from another previously submitted site just to be fair to those that really want their site on my link page.

Thanks Paul, regards,
eric