Speeding up a function

[Cleverbum]

I was hoping to parse a webpage and extract all link and image URLs
from it and enter the new ones into mySQL tables, below is my code to
do it, I've tried to optimise it as much as I can but it still takes
too long to execute (server timeouts on a server which I can not
control) I was wondering if there was some way to compile the code or
if anyone can spot something which could be better written.
Thanks in advance,
Martin

<?
$start=microtime_float();
$uid=$_GET['uid'];
$restriction= "dontstayin";
include("loadstuff.php"); //just contains a function which echos a tick
or cross
include("userpass.php"); //contains database access details.
mysql_connect($host,$user,$password);
@mysql_select_db("$database") or die(cross());
$result = mysql_query("SELECT * FROM mark_toparse WHERE uid='$uid'");
//mysql_close();
$contents = mysql_result($result,0,"fcontents");
$fullorig = mysql_result($result,0,"originalpath");
$origpath = substr($origpath,0,strrpos($fullorig,"/")+1);
$lines = explode(">",$contents);
for($i=0;$i<count($lines);$i++){
$imgsrc = stristr($lines[$i],"<img");
if($imgsrc!=false){
$imgsrc = str_replace("'","",$imgsrc);
$f = strpos($imgsrc,"\"",strpos($imgsrc,"src"));
$l = strpos($imgsrc,"\"",$f+1);
$url = substr($imgsrc,$f+1,$l-$f-1);
if(strncasecmp($url,"http:",5)!=0){
$url = $origpath . $url;
$url= str_replace(":/","://",str_replace("//","/",$url));
}
$resone = mysql_query("SELECT count(*) FROM mark_images WHERE
url='$url'");
$a = (mysql_fetch_array($resone,MYSQL_NUM));
if($a[0]==0){
mysql_query("INSERT INTO mark_images VALUES('','$url')");
}
}else{
$link = stristr($lines[$i],"href=");
if($link!=false){
$link = str_replace("'","",$link);
$f = strpos($link,"\"");
$l = strpos($link,"\"",$f+1);
$url= substr($link,$f+1,$l-$f-1);
if(strncasecmp($url,"http:",5)!=0){
if(strncasecmp($url,"mailto:",7)==0||strncasecmp($ url,"ftp:",4)==0||strncasecmp($url,"msnim:",6)==0) {
//ignore ftp, mailto and msn links
}else{
$url = $origpath . $url;
$url= str_replace(":/","://",str_replace("//","/",$url));
if(eregi("\.jp[eg2]{1,2}$",$url)){
$resone = mysql_query("SELECT count(*) FROM mark_images WHERE
url='$url'");
$a = (mysql_fetch_array($resone,MYSQL_NUM));
if($a[0]==0){
mysql_query("INSERT INTO mark_images VALUES('','$url')");
}
}else{
$resone = mysql_query("SELECT count(*) FROM mark_linktodl WHERE
url='$url'");
$a = (mysql_fetch_array($resone,MYSQL_NUM));
if($a[0]==0){
mysql_query("INSERT INTO mark_linktodl VALUES('','$url')");
}
}
}
}elseif(eregi($restriction,$url)){
if(eregi("\.jp[eg2]{1,2}$",$url)){
$resone = mysql_query("SELECT count(*) FROM mark_images WHERE
url='$url'");
$a = (mysql_fetch_array($resone,MYSQL_NUM));
if($a[0]==0){
mysql_query("INSERT INTO mark_images VALUES('','$url')");
}
}else{
$resone = mysql_query("SELECT count(*) FROM mark_linktodl WHERE
url='$url'");
$a = (mysql_fetch_array($resone,MYSQL_NUM));
if($a[0]==0){
mysql_query("INSERT INTO mark_linktodl VALUES('','$url')");
}
}
}
}
}

}
//mysql_query("DELETE FROM mark_toparse WHERE
originalpath='$fullorig'");
mysql_query("INSERT INTO mark_parsed
VALUES('','$fullorig','".md5($contents)."')");
mysql_close();
$finish=microtime_float();

if(strcmp($_GET['debug'],"t")==0){
$tim=$finish-$start;
include("error_image.php");
echo imagepng(errorimage("Analysis: $tim s"));
}else{
echo tick();
}

function microtime_float()
{
list($usec, $sec) = explode(" ", microtime());
return ((float)$usec + (float)$sec);
}

?>

[Rik]

Cl*******@hotmail.com wrote:

I was hoping to parse a webpage and extract all link and image URLs
from it and enter the new ones into mySQL tables, below is my code to
do it, I've tried to optimise it as much as I can but it still takes
too long to execute (server timeouts on a server which I can not
control) I was wondering if there was some way to compile the code or
if anyone can spot something which could be better written.
Thanks in advance,
Martin

Well, it would be simpler for us if you could describe what it is exactly
what you're trying to do, instead of letting us decypher it.

1. You can use a WHERE REGEXP to have a small result to check from the
database.
2. In the case, I really advise preg_replace() to change the img src
instead of the exploding, looping strpos, str_replace etc. This can be a
one-liner.
3. MySQL has a handy REPLACE INTO, as long as you have a correct key, it
would mean that no checking on already existing rows is required.
--
Rik Wasmus

[Chung Leong]

Rik wrote:

>
Well, it would be simpler for us if you could describe what it is exactly
what you're trying to do, instead of letting us decypher it.

1. You can use a WHERE REGEXP to have a small result to check from the
database.
2. In the case, I really advise preg_replace() to change the img src
instead of the exploding, looping strpos, str_replace etc. This can be a
one-liner.
3. MySQL has a handy REPLACE INTO, as long as you have a correct key, it
would mean that no checking on already existing rows is required.
--
Rik Wasmus

Combining all the insert/replace operation into one statement would
help as well. Instead of doing a query immediate, store the links in
different arrays like this:

$images[$url] = true;

That collapses duplicate links in the page being parsed. Then it's just
a matter of looping through the arrays to build sql statements that
would create the records in one fell swoop.

[Chung Leong]

Rik wrote:

>
Well, it would be simpler for us if you could describe what it is exactly
what you're trying to do, instead of letting us decypher it.

1. You can use a WHERE REGEXP to have a small result to check from the
database.
2. In the case, I really advise preg_replace() to change the img src
instead of the exploding, looping strpos, str_replace etc. This can be a
one-liner.
3. MySQL has a handy REPLACE INTO, as long as you have a correct key, it
would mean that no checking on already existing rows is required.
--
Rik Wasmus

Combining all the insert/replace operation into one statement would
help as well. Instead of doing a query immediate, store the links in
different arrays like this:

$images[$url] = true;

That collapses duplicate links in the page being parsed. Then it's just
a matter of looping through the arrays to build sql statements that
would create the records in one fell swoop.