Start graph and import refactoring
[contractdashboard.git] / admin / import.php
blob:a/admin/import.php -> blob:b/admin/import.php
<?php <?php
include_once("../lib/common.inc.php"); include_once("../lib/common.inc.php");
   
function processFile($fpath,$tablename){ function processFile($fpath,$tablename){
global $conn; global $conn;
$row = 1; $row = 1;
$handle = fopen($fpath, "r"); $handle = fopen($fpath, "r");
//"t" mode string translates windows line breaks to unix //"t" mode string translates windows line breaks to unix
$datamapping0711 = array("Agency" => "agencyName", "Parent CN ID" => "parentCN", "CN ID" => "CNID", "Publish Date" => "publishDate", "Amendment Date" => "amendDate", "Status" => "", "StartDate" => "contractStart", "EndDate" => "contractEnd", "Value" => "value", "Description" => "description", "Agency Ref Id" => "agencyID", "Category" => "category", "Procurement Method" => "procurementMethod", "ATM ID" => "atmID", "SON ID" => "SONID", "Confidentiality - Contract" => "confidentialityContract", "Confidentiality - Contract Reason(s)" => "confidentialityContractReason", "Confidentiality - Outputs" => "confidentialityOutputs", "Confidentiality - Outputs Reason(s)" => "confidentialityOutputsReason", "Consultancy" => "consultancy", "Consultancy Reason(s)" => "consultancyReason", "Amendment Reason" => "amendmentReason", "Supplier Name" => "supplierName", "Supplier Address" => "supplierAddress", "Supplier City" => "supplierCity", "Supplier Postcode" => "supplierPostcode", "Supplier Country" => "supplierCountry", "Supplier ABNExempt" => "supplierABNExempt", "Supplier ABN" => "supplierABN", "Agency Branch" => "contactBranch", "Agency Divison" => "contactDivision", "Agency Postcode" => "contactPostcode", "" => ""); $datamapping0711 = array("Agency" => "agencyName", "Parent CN ID" => "parentCN", "CN ID" => "CNID", "Publish Date" => "publishDate", "Amendment Date" => "amendDate", "Status" => "", "StartDate" => "contractStart", "EndDate" => "contractEnd", "Value" => "value", "Description" => "description", "Agency Ref Id" => "agencyID", "Category" => "category", "Procurement Method" => "procurementMethod", "ATM ID" => "atmID", "SON ID" => "SONID", "Confidentiality - Contract" => "confidentialityContract", "Confidentiality - Contract Reason(s)" => "confidentialityContractReason", "Confidentiality - Outputs" => "confidentialityOutputs", "Confidentiality - Outputs Reason(s)" => "confidentialityOutputsReason", "Consultancy" => "consultancy", "Consultancy Reason(s)" => "consultancyReason", "Amendment Reason" => "amendmentReason", "Supplier Name" => "supplierName", "Supplier Address" => "supplierAddress", "Supplier City" => "supplierCity", "Supplier Postcode" => "supplierPostcode", "Supplier Country" => "supplierCountry", "Supplier ABNExempt" => "supplierABNExempt", "Supplier ABN" => "supplierABN", "Agency Branch" => "contactBranch", "Agency Divison" => "contactDivision", "Agency Postcode" => "contactPostcode", "" => "");
$headers; $headers;
$contractNoticeFields = array("importFile", "agencyName", "parentCN", "CNID", "publishDate", "amendDate", "contractStart", "contractEnd", "value", "description", "agencyID", "category", "procurementMethod", "atmID", "SONID", "confidentialityContract", "confidentialityContractReason", "confidentialityOutputs", "confidentialityOutputsReason", "consultancy", "consultancyReason", "amendmentReason", "supplierName", "supplierAddress", "supplierCity", "supplierPostcode", "supplierCountry", "supplierABNExempt", "supplierABN", "contactBranch", "contactDivision", "contactPostcode"); $contractNoticeFields = array("importFile", "agencyName", "parentCN", "CNID", "publishDate", "amendDate", "contractStart", "contractEnd", "value", "description", "agencyID", "category", "procurementMethod", "atmID", "SONID", "confidentialityContract", "confidentialityContractReason", "confidentialityOutputs", "confidentialityOutputsReason", "consultancy", "consultancyReason", "amendmentReason", "supplierName", "supplierAddress", "supplierCity", "supplierPostcode", "supplierCountry", "supplierABNExempt", "supplierABN", "contactBranch", "contactDivision", "contactPostcode");
$agencyFields = array("agencyName"); $agencyFields = array("agencyName");
$supplierFields = array("supplierName", "supplierAddress", "supplierCity", "supplierPostcode", "supplierCountry", "supplierABNExempt", "supplierABN"); $supplierFields = array("supplierName", "supplierAddress", "supplierCity", "supplierPostcode", "supplierCountry", "supplierABNExempt", "supplierABN");
if ($tablename == "contractnotice") { if ($tablename == "contractnotice") {
$contractNoticeInsertQ = 'INSERT INTO contractnotice ("' . implode('" , "', $contractNoticeFields) . '") VALUES ( '; $contractNoticeInsertQ = 'INSERT INTO contractnotice ("' . implode('" , "', $contractNoticeFields) . '") VALUES ( ';
foreach ($contractNoticeFields as $key => $f) { foreach ($contractNoticeFields as $key => $f) {
$contractNoticeInsertQ .= ($key == 0 ? "" : ", ")."?"; $contractNoticeInsertQ .= ($key == 0 ? "" : ", ")."?";
} }
$contractNoticeInsertQ .= ");"; $contractNoticeInsertQ .= ");";
$contractNoticeInsertQ = $conn->prepare($contractNoticeInsertQ); $contractNoticeInsertQ = $conn->prepare($contractNoticeInsertQ);
} else if ($tablename == "supplierdetails") { } else if ($tablename == "supplierdetails") {
$supplierInsertQ = 'INSERT INTO supplierdetails ("' . implode('" , "', $supplierFields) . '") VALUES ( '; $supplierInsertQ = 'INSERT INTO supplierdetails ("' . implode('" , "', $supplierFields) . '") VALUES ( ';
foreach ($supplierFields as $key => $f) { foreach ($supplierFields as $key => $f) {
$supplierInsertQ .= ($key == 0 ? "" : ", ")."?"; $supplierInsertQ .= ($key == 0 ? "" : ", ")."?";
} }
$supplierInsertQ .= ");"; $supplierInsertQ .= ");";
$supplierInsertQ = $conn->prepare($supplierInsertQ); $supplierInsertQ = $conn->prepare($supplierInsertQ);
} else if ($tablename == "agency") { } else if ($tablename == "agency") {
$agencyInsertQ = 'INSERT INTO agency ("' . implode('" , "', $agencyFields) . '") VALUES ( '; $agencyInsertQ = 'INSERT INTO agency ("' . implode('" , "', $agencyFields) . '") VALUES ( ';
foreach ($agencyFields as $key => $f) { foreach ($agencyFields as $key => $f) {
$agencyInsertQ .= ($key == 0 ? "" : ", ")."?"; $agencyInsertQ .= ($key == 0 ? "" : ", ")."?";
} }
$agencyInsertQ .= ");"; $agencyInsertQ .= ");";
$agencyInsertQ = $conn->prepare($agencyInsertQ); $agencyInsertQ = $conn->prepare($agencyInsertQ);
} }
   
while (($data = fgetcsv($handle, 1000, "\t")) !== false) { while (($data = fgetcsv($handle, 1000, "\t")) !== false) {
$num = count($data); $num = count($data);
if ($row == 3) { if ($row == 3) {
$headers = $data; $headers = $data;
} elseif ($row > 3) { } elseif ($row > 3) {
if ($num > count($datamapping0711)) { if ($num > count($datamapping0711)) {
die("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>" . $fname . print_r($data)); die("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>" . $fname . print_r($data));
} }
$contractNoticeInsert = Array(); $contractNoticeInsert = Array();
$supplierInsert = Array(); $supplierInsert = Array();
$agencyInsert = Array(); $agencyInsert = Array();
$contractNoticeInsert[] = $fname; $contractNoticeInsert[] = $fname;
$keys = array_keys($datamapping0711); $keys = array_keys($datamapping0711);
for ($c = 0; $c < $num; $c++) { for ($c = 0; $c < $num; $c++) {
$data[$c] = trim($data[$c], "="); $data[$c] = trim($data[$c], "=");
$data[$c] = trim($data[$c], "\""); $data[$c] = trim($data[$c], "\"");
if ($tablename == "contractnotice") { if ($tablename == "contractnotice") {
if (in_array(($datamapping0711[$headers[$c]]), $contractNoticeFields)) { if (in_array(($datamapping0711[$headers[$c]]), $contractNoticeFields)) {
if (($datamapping0711[$headers[$c]]) == "parentCN" || ($datamapping0711[$headers[$c]]) == "CNID") { if (($datamapping0711[$headers[$c]]) == "parentCN" || ($datamapping0711[$headers[$c]]) == "CNID") {
$data[$c] = substr($data[$c], 2); // take off the "CN" prefix $data[$c] = substr($data[$c], 2); // take off the "CN" prefix
$data[$c] = str_replace("-A","00",$data[$c]); // make amendments really big numbers $data[$c] = str_replace("-A","00",$data[$c]); // make amendments really big numbers
  if (!is_numeric($data[$c])) die ($data[$c]." is not numeric");
if ($data[$c] > 0) { if ($data[$c] > 0) {
$contractNoticeInsert[] = $data[$c]; $contractNoticeInsert[] = $data[$c];
} else { } else {
$contractNoticeInsert[] = 0; $contractNoticeInsert[] = 0;
} }
} elseif (($datamapping0711[$headers[$c]]) == "amendDate" || ($datamapping0711[$headers[$c]]) == "publishDate" || ($datamapping0711[$headers[$c]]) == "contractStart" || ($datamapping0711[$headers[$c]]) == "contractEnd") { } elseif (($datamapping0711[$headers[$c]]) == "amendDate" || ($datamapping0711[$headers[$c]]) == "publishDate" || ($datamapping0711[$headers[$c]]) == "contractStart" || ($datamapping0711[$headers[$c]]) == "contractEnd") {
$contractNoticeInsert[] = date('Y-m-d H:i:s', strtotime($data[$c])); $contractNoticeInsert[] = date('Y-m-d H:i:s', strtotime($data[$c]));
} else { } else {
if (strstr("\" =", $data[$c] > 0)) { if (strstr("\" =", $data[$c] > 0)) {
die("Invalid Description field" . $contractNoticeInsert); die("Invalid Description field" . $contractNoticeInsert);
} }
$contractNoticeInsert[] = $data[$c]; $contractNoticeInsert[] = $data[$c];
} }
} }
} else if ($tablename == "supplierdetails") { } else if ($tablename == "supplierdetails") {
if (in_array(($datamapping0711[$headers[$c]]), $supplierFields)) { if (in_array(($datamapping0711[$headers[$c]]), $supplierFields)) {
$supplierInsert[] = $data[$c]; $supplierInsert[] = $data[$c];
} }
} else if ($tablename == "agency") { } else if ($tablename == "agency") {
if (in_array(($datamapping0711[$headers[$c]]), $agencyFields)) { if (in_array(($datamapping0711[$headers[$c]]), $agencyFields)) {
$agencyInsert[] = $data[$c]; $agencyInsert[] = $data[$c];
} }
} }
} }
flush(); flush();
if ($tablename == "contractnotice") { if ($tablename == "contractnotice") {
$contractNoticeInsertQ->execute($contractNoticeInsert); $contractNoticeInsertQ->execute($contractNoticeInsert);
$errors = $conn->errorInfo(); $errors = $conn->errorInfo();
if ($errors[1] == 7 || $errors[1] ==0) if ($errors[1] == 7 || $errors[1] ==0)
$success++; $success++;
else else
echo $data[2] . " failed CN insert.<br>" . print_r($errors,true) echo $data[2] . " failed CN insert.<br>" . print_r($errors,true)
. " <br> ". print_r($contractNoticeInsert,true) ."<br> $row <br><br>\n"; . " <br> ". print_r($contractNoticeInsert,true) ."<br> $row <br><br>\n";
} else if ($tablename == "supplierdetails") { } else if ($tablename == "supplierdetails") {
$supplierInsertQ->execute($supplierInsert); $supplierInsertQ->execute($supplierInsert);
$errors = $conn->errorInfo(); $errors = $conn->errorInfo();
if ($errors[1] == 7 || $errors[1] ==0) if ($errors[1] == 7 || $errors[1] ==0)
$success++; $success++;
else else
echo $data[2] . " failed agency insert.<br>" . print_r($errors,true) echo $data[2] . " failed agency insert.<br>" . print_r($errors,true)
. " <br> ". print_r($agencyInsert,true) ."<br> $row <br><br>\n"; . " <br> ". print_r($agencyInsert,true) ."<br> $row <br><br>\n";
   
} else if ($tablename == "agency") { } else if ($tablename == "agency") {
$agencyInsertQ->execute($agencyInsert); $agencyInsertQ->execute($agencyInsert);
$errors = $conn->errorInfo(); $errors = $conn->errorInfo();
if ($errors[1] == 7 || $errors[1] ==0) if ($errors[1] == 7 || $errors[1] ==0)
$success++; $success++;
else else
echo $data[2] . " failed supplier insert.<br>" . print_r($errors,true) echo $data[2] . " failed supplier insert.<br>" . print_r($errors,true)
. " <br> ". print_r($supplierInsert,true) ."<br> $row <br><br>\n"; . " <br> ". print_r($supplierInsert,true) ."<br> $row <br><br>\n";
} }
flush(); flush();
//echo "<hr>\n"; //echo "<hr>\n";
} }
$row++; $row++;
} }
fclose($handle); fclose($handle);
return $success; return $success;
} }
   
$path = 'data/'; $path = 'data/';
if ($_REQUEST["fname"] == "") { if ($_REQUEST["fname"] == "") {
echo "Get files from: https://www.tenders.gov.au/?event=public.reports.list<br>"; echo "Get files from: https://www.tenders.gov.au/?event=public.reports.list<br>";
$dhandle = opendir($path); $dhandle = opendir($path);
// define an array to hold the files // define an array to hold the files
$files = array(); $files = array();
if ($dhandle) { if ($dhandle) {
// loop through all of the files // loop through all of the files
while (false !== ($fname = readdir($dhandle))) { while (false !== ($fname = readdir($dhandle))) {
if (($fname != '.') && ($fname != '..')) { if (($fname != '.') && ($fname != '..')) {
echo "<a href=\"import.php?fname=$fname\">$fname</a>&nbsp;" . filesize($path . $fname) . "&nbsp;" . date("c",filemtime($path . $fname)) . "<br/>"; echo "<a href=\"import.php?fname=$fname\">$fname</a>&nbsp;" . filesize($path . $fname) . "&nbsp;" . date("c",filemtime($path . $fname)) . "<br/>";
} }
} }
} }
} else { } else {
$success = 0; $success = 0;
$fname = $_REQUEST["fname"]; $fname = $_REQUEST["fname"];
echo " ============== $fname ============== <br>"; echo " ============== $fname ============== <br>";
flush(); flush();
$success += processFile($path . $fname,"contractnotice"); $success += processFile($path . $fname,"contractnotice");
$success += processFile($path . $fname,"agency"); $success += processFile($path . $fname,"agency");
$success += processFile($path . $fname,"agency"); $success += processFile($path . $fname,"agency");
echo "<br> $success records successfully created"; echo "<br> $success records successfully created";
flush(); flush();
} }
  // run post import data processing
  // cn
  include("linkAmendments.php");
  include("updateUNSPSC.php");
  // agency
  include("setAgencyStatus.php");
  include("setAgencyURLABN.php");
   
?> ?>