Remove troublesome scraper script
Remove troublesome scraper script

file:b/.gitignore (new)
  /admin/neo4jimporter/target/
  /admin/neo4jimporter/.settings/
  /nbproject/private/
<?php <?php
include_once("../lib/common.inc.php"); include_once("../lib/common.inc.php");
   
$query = 'update contractnotice set "parentCN" = null where "parentCN" = \'0\''; $query = 'update contractnotice set "parentCN" = null where "parentCN" = \'0\'';
$result = $conn->prepare($query); $result = $conn->prepare($query);
$result->execute(); $result->execute();
$query = 'update contractnotice set "childCN" = null where "childCN" = \'0\''; $query = 'update contractnotice set "childCN" = null where "childCN" = \'0\'';
$result = $conn->prepare($query); $result = $conn->prepare($query);
$result->execute(); $result->execute();
$query = 'select "CNID","parentCN" from contractnotice where "CNID" like \'%00_\' and "parentCN" is not null'; $query = 'select "CNID","parentCN" from contractnotice where "CNID" like \'%00_\' and "parentCN" is not null';
$result = $conn->prepare($query); $result = $conn->prepare($query);
$result->execute(); $result->execute();
foreach ($result->fetchAll() as $record) { foreach ($result->fetchAll() as $record) {
$oldCN = $record['CNID']; $oldCN = $record['CNID'];
$parentCN = substr($oldCN,0,-3); $parentCN = substr($oldCN,0,-3);
if ($parentCN == $record['parentCN']) { if ($parentCN == $record['parentCN']) {
$newCN = $parentCN . "-A". substr($oldCN,-1); $newCN = $parentCN . "-A". substr($oldCN,-1);
$updateresult = $conn->exec('UPDATE contractnotice SET "CNID" = $updateresult = $conn->exec('UPDATE contractnotice SET "CNID" =
\'' . $newCN . '\' where "CNID" = \'' . $oldCN . '\';'); \'' . $newCN . '\' where "CNID" = \'' . $oldCN . '\';');
echo "$oldCN => $newCN (from parent CN $parentCN) <br>\n"; echo "$oldCN => $newCN (from parent CN $parentCN) <br>\n";
} }
else { else {
echo "parent CN unexpected - $oldCN doesn't look like child of {$record['parentCN']} <br>\n"; // echo "parent CN unexpected - $oldCN doesn't look like child of {$record['parentCN']} <br>\n";
} }
} }
   
<?php  
date_default_timezone_set('Australia/Melbourne');  
$split = false;  
function format_bytes($size) {  
$units = array(' B', ' KB', ' MB', ' GB', ' TB');  
for ($i = 0; $size >= 1024 && $i < 4; $i++) $size /= 1024;  
return round($size, 2).$units[$i];  
}  
 
$days = 4;  
if (isset($_REQUEST['days'])) $days = $_REQUEST['days'];  
$startDate = strtotime("05-Jun-2008");  
if (isset($_REQUEST['startDate'])) $startDate = $_REQUEST['startDate'];  
 
function getFile($startDate, $days, $minVal, $maxVal) {  
global $split;  
$endDate = strtotime(date("Y-m-d", $startDate)." +".$days." days");  
$file = date("dMY",$startDate).'to'.date("dMY",$endDate).'val'.$minVal.'to'.$maxVal.'.xls';  
echo "Fetching $file ($days days) ($minVal < value < $maxVal )... ";  
$url = "https://www.tenders.gov.au/?event=public.advancedsearch.CNSONRedirect&type=cnEvent&atmType=archived%2Cclosed%2Cpublished%2Cproposed&agencyUUID=&agencyStatus=-1&portfolioUUID=&keyword=&KeywordTypeSearch=AllWord&CNID=&dateType=Publish+Date&dateStart=".date("d-M-Y",$startDate)."&dateEnd=".date("d-M-Y",$endDate)."&supplierName=&supplierABN=&valueFrom=".$minVal."&valueTo=".$maxVal."&ATMID=&AgencyRefId=&consultancy=&download=Download+results";  
echo "<!-- $url -->";  
$current = file_get_contents($url);  
if (strpos($current,"There are no results that match your selection.")> 0 ) {  
echo "<font color=red>Empty file!</font><br>";  
}  
if (strpos($current,"Your search returned more than 1000 results.") === false) {  
file_put_contents($file, $current);  
echo "$file saved<br>";  
echo format_bytes(filesize($file))."<br>";  
echo '<a href="?startDate='.$endDate.'&days='.$days.'">Load next '.($days).' days </a><br>';  
echo '<a href="?startDate='.$endDate.'&days='.($days*2).'">Load next '.($days*2).' days </a><br>';  
echo '<a href="?startDate='.$endDate.'&days='.$days.'&split=yes">Load next '.($days).' days with split</a><br>';  
flush();  
if (!isset($_REQUEST['split']) && !$split) {  
echo "Success so fetching next $days... <br>";  
getFile($endDate, $days, "" , "");  
}  
return true;  
} else {  
echo "<font color=red>Too many records!</font><br>";  
echo '<a href="?startDate='.$startDate.'&days='.floor($days/2).'">Load '.($days/2).' days instead?</a><br>';  
echo '<a href="?startDate='.$startDate.'&days='.$days.'&split=yes">Split instead?</a><br>';  
flush();  
if (!isset($_REQUEST['split']) && !$split) {  
echo "Failure so splitting ... <br>";  
doSplit($startDate, $days);  
}  
return false;  
}  
}  
function doSplit($startDate, $days) {  
global $split;  
$split = true;  
set_time_limit(20);  
getFile($startDate, $days, 0, 12000);  
getFile($startDate, $days, 12000, 16000);  
getFile($startDate, $days, 16000, 20000);  
getFile($startDate, $days, 20000, 30000);  
getFile($startDate, $days, 30000, 40000);  
// getFile($startDate, $days, 40000, 80000);  
getFile($startDate, $days, 40000, 60000);  
getFile($startDate, $days, 60000, 80000);  
// getFile($startDate, $days, 80000, 300000);  
getFile($startDate, $days, 80000, 150000);  
getFile($startDate, $days, 150000, 300000);  
getFile($startDate, $days, 300000, 999999999);  
}  
if (isset($_REQUEST['split'])) {  
doSplit($startDate, $days);  
} else {  
getFile($startDate, $days, "" , "");  
}  
?>  
 
<?php <?php
error_reporting(E_ALL); error_reporting(E_ALL);
include_once("../lib/common.inc.php"); include_once("../lib/common.inc.php");
$unspscresult = $conn->prepare('select * from "UNSPSCcategories";'); $unspscresult = $conn->prepare('select * from "UNSPSCcategories";');
$unspscresult->execute(); $unspscresult->execute();
foreach ($unspscresult->fetchAll() as $row) { foreach ($unspscresult->fetchAll() as $row) {
$unspsc[$row['Title']] = $row['UNSPSC']; $unspsc[$row['Title']] = $row['UNSPSC'];
// some Australian spellings // some Australian spellings
$isiz = str_replace("iz", "is", $row['Title']); $isiz = str_replace("iz", "is", $row['Title']);
$unspsc[$isiz] = $row['UNSPSC']; $unspsc[$isiz] = $row['UNSPSC'];
$filfill = str_replace("fill", "fil", $row['Title']); $filfill = str_replace("fill", "fil", $row['Title']);
$unspsc[$filfill] = $row['UNSPSC']; $unspsc[$filfill] = $row['UNSPSC'];
$defence = str_replace("efense", "efence", $row['Title']); $defence = str_replace("efense", "efence", $row['Title']);
$unspsc[$defence] = $row['UNSPSC']; $unspsc[$defence] = $row['UNSPSC'];
$armor = str_replace("rmored", "rmoured", $row['Title']); $armor = str_replace("rmored", "rmoured", $row['Title']);
$unspsc[$armor] = $row['UNSPSC']; $unspsc[$armor] = $row['UNSPSC'];
$erre = str_replace("er", "re", $row['Title']); $erre = str_replace("er", "re", $row['Title']);
$unspsc[$erre] = $row['UNSPSC']; $unspsc[$erre] = $row['UNSPSC'];
  $center = str_replace("center", "centre", $row['Title']);
  $unspsc[$center] = $row['UNSPSC'];
  $accessory = str_replace("accesor", "accessor", $row['Title']);
  $unspsc[$accessory] = $row['UNSPSC'];
$lyslyz = str_replace("lyz", "lys", $row['Title']); $lyslyz = str_replace("lyz", "lys", $row['Title']);
$unspsc[$lyslyz] = $row['UNSPSC']; $unspsc[$lyslyz] = $row['UNSPSC'];
  $tire = str_replace("ire", "yre", $row['Title']);
  $unspsc[$tire] = $row['UNSPSC'];
   
  $pe = str_replace("pe", "pae", $row['Title']);
  $unspsc[$pe] = $row['UNSPSC'];
  $ane = str_replace("ane", "anae", $row['Title']);
  $unspsc[$ane] = $row['UNSPSC'];
  $airo = str_replace("airplane", "aeroplane", $row['Title']);
  $unspsc[$airo] = $row['UNSPSC'];
// some divergence from standard // some divergence from standard
  $forensicit = str_replace("Information technology consultation services", "Forensic IT Services", $row['Title']);
  $unspsc[$forensicit] = $row['UNSPSC'];
  $powercable = str_replace( "Power cable", "Power cable installation and supply", $row['Title']);
  $unspsc[$powercable] = $row['UNSPSC'];
$tobacco = str_replace("Food Beverage and Tobacco Products", "Food and Beverage Products", $row['Title']); $tobacco = str_replace("Food Beverage and Tobacco Products", "Food and Beverage Products", $row['Title']);
$unspsc[$tobacco] = $row['UNSPSC']; $unspsc[$tobacco] = $row['UNSPSC'];
$architect = str_replace("Building and Construction and Maintenance Services", "Architectural services", $row['Title']); $architect = str_replace("Building and Construction and Maintenance Services", "Architectural services", $row['Title']);
$unspsc[$architect] = $row['UNSPSC']; $unspsc[$architect] = $row['UNSPSC'];
// some just plain wrong // some just plain wrong
$noOilRigs = str_replace("Building and Construction and Maintenance Services", "Management and provision of all facilities engineering modification and maintenance services for a site or platform", $row['Title']); $noOilRigs = str_replace("Building and Construction and Maintenance Services", "Management and provision of all facilities engineering modification and maintenance services for a site or platform", $row['Title']);
$unspsc[$noOilRigs] = $row['UNSPSC']; $unspsc[$noOilRigs] = $row['UNSPSC'];
} }
$query = 'SELECT "CNID","category","value" $query = 'SELECT "CNID","category","value"
FROM contractnotice FROM contractnotice
WHERE "categoryUNSPSC" IS NULL OR "categoryUNSPSC" = 0'; WHERE "categoryUNSPSC" IS NULL OR "categoryUNSPSC" = 0';