TRUNCATE mysql table in the while loop only executes ones - mysql

Question

I have a bunch of websites stored as strings in 3 mysql tables. My script puts them into arrays, empties the table, parse the strings, extracts all the links and sort them into 2 tables. Its broken in 3 identical modules which do the sorting. The above mentioned 3 mysql tables gets populate with new data every few seconds from external source (other php script).
The whole thing is looped with while to perform its operations every 90 seconds.
For some reason only on the first go of the loop the TRUNCATE part gets executed, on every next go it doesn't empty the table.
Before I get to my code, I apologize for depreciated mysql, this script will be only to used on local machine and I will update it when the time is right.
Here is my code:
$i=1;
$domain1 = 'example1.com';
$domain2 = 'example2.com';
$domain3 = 'example3.com';
$robots1 = array("url1",
"url2",
"url3");
$robots2 = array("url1",
"url2",
"url3");
$robots3 = array("url1",
"url2",
"url3");
require_once 'Normalizer.php';
$conn = mysql_connect('localhost:3306','user', 'pass', true );
mysql_select_db( 't1000', $conn );
do {
$query = 'SELECT * FROM dump1';
$result1=mysql_query( $query, $conn );
$strings1=array();
while ($row = mysql_fetch_assoc($result1)) {
array_push($strings1, $row["link"]);
}
$query = 'TRUNCATE TABLE dump1';
$delete=mysql_query( $query, $conn );
$query = 'SELECT * FROM dump2';
$result1=mysql_query( $query, $conn );
$strings2=array();
while ($row = mysql_fetch_assoc($result1)) {
array_push($strings2, $row["link"]);
}
$query = 'TRUNCATE TABLE dump2';
$delete=mysql_query( $query, $conn );
$query = 'SELECT * FROM dump3';
$result1=mysql_query( $query, $conn );
$strings3=array();
while ($row = mysql_fetch_assoc($result1)) {
array_push($strings3, $row["link"]);
}
$query = 'TRUNCATE TABLE dump3';
$delete=mysql_query( $query, $conn );
// Module 1 start
$ii=0;
$links = array();
$edofollow = array();
$enofollow = array();
$internal = array();
foreach ($strings1 as $value)
{
$input=$strings1[$ii];
$htm=stripcslashes($input);
$doc = new DOMDocument();
#$doc->loadHTML($htm);
$arr = $doc->getElementsByTagName("a"); // DOMNodeList Object
foreach($arr as $item) { // DOMElement Object
$href = $item->getAttribute("href");
$rel = $item->getAttribute("rel");
$text = trim(preg_replace("/[\r\n]+/", " ", $item->nodeValue));
$links[] = array(
'href' => $href,
'rel' => $rel,
'text' => $text
);
if (strpos($href, '://')!==false AND strpos($href, $domain1)==false AND $rel!=='nofollow')
{
$un = new URL\Normalizer();
$un->setUrl( $href );
$href= parse_url($un->normalize(), PHP_URL_HOST);
array_push($edofollow, $href);
}
else if (strpos($href, '://')!==false AND strpos($href, $domain1)==false AND $rel=='nofollow')
{
$un1 = new URL\Normalizer();
$un1->setUrl( $href );
array_push($enofollow, $un1->normalize());
}
else if (strpos($href,'://')==false or strpos($href,$domain1)!==false)
{
$un2 = new URL\Normalizer();
$un2->setUrl( $href );
$href1=$un2->normalize();
if (strpos($href1, 'TRANSCRIPTS')==false AND strpos($href1, '(')==false AND strpos($href1, ')')==false AND strpos($href1, '#')==false AND strpos($href1, 'javascript')==false AND strpos($href1, '?')==false AND strpos($href1, 'void')==false)
{
if($href1=='' or $href1=='/')
{}
else{
if (strpos($href1, '://')==false)
{$href1='http://'.$domain1.$href1;}
if (in_array($href1, $robots1)) { }
else {
array_push($internal, $href1);
}
}
}
}
}
$uedofollow = array_values(array_unique($edofollow));
foreach ($uedofollow as $value) {
$query=mysql_query("select * from dofollow where link='".$value."' ");
$duplicate=0;
if($query){
$duplicate=mysql_num_rows($query);
}
if($duplicate==0)
{
$sql='INSERT INTO dofollow (link) VALUES ("'.$value.'")';
mysql_query( $sql, $conn );
}
}
$uinternal = array_values(array_unique($internal));
foreach ($uinternal as $value2) {
$query=mysql_query("select * from joblist1 where link='".$value2."' ");
if ($query) {
$duplicate=0;
$duplicate=mysql_num_rows($query);
if($duplicate==0)
{
$sql='INSERT INTO joblist1 (link) VALUES ("'.$value2.'")';
mysql_query( $sql, $conn );
}
}
}
$ii=$ii+1;
}
// Module 1 ends
// Module 2 start
$links = array();
$edofollow = array();
$enofollow = array();
$internal = array();
$ii=0;
foreach ($strings2 as $value)
{
$input=$strings2[$ii];
$htm=stripcslashes($input);
$doc = new DOMDocument();
#$doc->loadHTML($htm);
$arr = $doc->getElementsByTagName("a"); // DOMNodeList Object
foreach($arr as $item) { // DOMElement Object
$href = $item->getAttribute("href");
$rel = $item->getAttribute("rel");
$text = trim(preg_replace("/[\r\n]+/", " ", $item->nodeValue));
$links[] = array(
'href' => $href,
'rel' => $rel,
'text' => $text
);
if (strpos($href, '://')!==false AND strpos($href, $domain2)==false AND $rel!=='nofollow')
{
$un = new URL\Normalizer();
$un->setUrl( $href );
$href= parse_url($un->normalize(), PHP_URL_HOST);
array_push($edofollow, $href);
}
else if (strpos($href, '://')!==false AND strpos($href, $domain2)==false AND $rel=='nofollow')
{
$un1 = new URL\Normalizer();
$un1->setUrl( $href );
array_push($enofollow, $un1->normalize());
}
else if (strpos($href,'://')==false or strpos($href,$domain2)!==false)
{
$un2 = new URL\Normalizer();
$un2->setUrl( $href );
$href1=$un2->normalize();
if (strpos($href1, 'TRANSCRIPTS')==false AND strpos($href1, '(')==false AND strpos($href1, ')')==false AND strpos($href1, '#')==false AND strpos($href1, 'javascript')==false AND strpos($href1, '?')==false AND strpos($href1, 'void')==false)
{
if($href1=='' or $href1=='/')
{}
else{
if (strpos($href1, '://')==false)
{$href1='http://'.$domain2.$href1;}
if (in_array($href1, $robots2)) { }
else {
array_push($internal, $href1);
}
}
}
}
}
$uedofollow = array_values(array_unique($edofollow));
foreach ($uedofollow as $value) {
$query=mysql_query("select * from dofollow where link='".$value."' ");
$duplicate=0;
if($query){
$duplicate=mysql_num_rows($query);
}
if($duplicate==0)
{
$sql='INSERT INTO dofollow (link) VALUES ("'.$value.'")';
mysql_query( $sql, $conn );
}
}
$uinternal = array_values(array_unique($internal));
foreach ($uinternal as $value2) {
$query=mysql_query("select * from joblist2 where link='".$value2."' ");
if ($query) {
$duplicate=0;
$duplicate=mysql_num_rows($query);
if($duplicate==0)
{
$sql='INSERT INTO joblist2 (link) VALUES ("'.$value2.'")';
mysql_query( $sql, $conn );
}
}
}
$ii=$ii+1;
}
// Module 2 Ends
// Module 3 start
$links = array();
$edofollow = array();
$enofollow = array();
$internal = array();
$ii=0;
foreach ($strings3 as $value)
{
$input=$strings3[$ii];
$htm=stripcslashes($input);
$doc = new DOMDocument();
#$doc->loadHTML($htm);
$arr = $doc->getElementsByTagName("a"); // DOMNodeList Object
foreach($arr as $item) { // DOMElement Object
$href = $item->getAttribute("href");
$rel = $item->getAttribute("rel");
$text = trim(preg_replace("/[\r\n]+/", " ", $item->nodeValue));
$links[] = array(
'href' => $href,
'rel' => $rel,
'text' => $text
);
if (strpos($href, '://')!==false AND strpos($href, $domain3)==false AND $rel!=='nofollow')
{
$un = new URL\Normalizer();
$un->setUrl( $href );
$href= parse_url($un->normalize(), PHP_URL_HOST);
array_push($edofollow, $href);
}
else if (strpos($href, '://')!==false AND strpos($href, $domain3)==false AND $rel=='nofollow')
{
$un1 = new URL\Normalizer();
$un1->setUrl( $href );
array_push($enofollow, $un1->normalize());
}
else if (strpos($href,'://')==false or strpos($href,$domain3)!==false)
{
$un2 = new URL\Normalizer();
$un2->setUrl( $href );
$href1=$un2->normalize();
if (strpos($href1, 'TRANSCRIPTS')==false AND strpos($href1, '(')==false AND strpos($href1, ')')==false AND strpos($href1, '#')==false AND strpos($href1, 'javascript')==false AND strpos($href1, '?')==false AND strpos($href1, 'void')==false)
{
if($href1=='' or $href1=='/')
{}
else{
if (strpos($href1, '://')==false)
{$href1='http://'.$domain3.$href1;}
if (in_array($href1, $robots3)) { }
else {
array_push($internal, $href1);
}
}
}
}
}
$uedofollow = array_values(array_unique($edofollow));
foreach ($uedofollow as $value) {
$query=mysql_query("select * from dofollow where link='".$value."' ");
$duplicate=0;
if($query){
$duplicate=mysql_num_rows($query);
}
if($duplicate==0)
{
$sql='INSERT INTO dofollow (link) VALUES ("'.$value.'")';
mysql_query( $sql, $conn );
}
}
$uinternal = array_values(array_unique($internal));
foreach ($uinternal as $value2) {
$query=mysql_query("select * from joblist3 where link='".$value2."' ");
if ($query) {
$duplicate=0;
$duplicate=mysql_num_rows($query);
if($duplicate==0)
{
$sql='INSERT INTO joblist3 (link) VALUES ("'.$value2.'")';
mysql_query( $sql, $conn );
}
}
}
$ii=$ii+1;
}
// Module 3 ends
sleep(90);
$i=$i++;
} while($i<=50000);
I was trying to troubleshoot it for days now, mixing things around but no luck...
How do I get it to empty the table each time?
I was also thinking to get rid of the loop and just run the script with cron, but I think this approach is counter productive :(