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"); if (php_sapi_name() != "cli") {
   
function processFile($fpath,$tablename){  
global $conn;  
$row = 1;  
$handle = fopen($fpath, "r");  
//"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", "" => "");  
$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");  
$agencyFields = array("agencyName");  
$supplierFields = array("supplierName", "supplierAddress", "supplierCity", "supplierPostcode", "supplierCountry", "supplierABNExempt", "supplierABN");  
   
if ($tablename == "contractnotice") {  
$contractNoticeInsertQ = 'INSERT INTO contractnotice ("' . implode('" , "', $contractNoticeFields) . '") VALUES ( ';  
foreach ($contractNoticeFields as $key => $f) {  
$contractNoticeInsertQ .= ($key == 0 ? "" : ", ")."?";  
}  
$contractNoticeInsertQ .= ");";  
$contractNoticeInsertQ = $conn->prepare($contractNoticeInsertQ);  
   
   
} else if ($tablename == "supplierdetails") {  
$supplierInsertQ = 'INSERT INTO supplierdetails ("' . implode('" , "', $supplierFields) . '") VALUES ( ';  
foreach ($supplierFields as $key => $f) {  
$supplierInsertQ .= ($key == 0 ? "" : ", ")."?";  
}  
$supplierInsertQ .= ");";  
$supplierInsertQ = $conn->prepare($supplierInsertQ);  
   
} else if ($tablename == "agency") {  
$agencyInsertQ = 'INSERT INTO agency ("' . implode('" , "', $agencyFields) . '") VALUES ( ';  
foreach ($agencyFields as $key => $f) {  
$agencyInsertQ .= ($key == 0 ? "" : ", ")."?";  
}  
$agencyInsertQ .= ");";  
$agencyInsertQ = $conn->prepare($agencyInsertQ);  
}  
   
while (($data = fgetcsv($handle, 1000, "\t")) !== false) {  
$num = count($data);  
   
if ($row == 3) {  
$headers = $data;  
} elseif ($row > 3) {  
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));  
}  
$contractNoticeInsert = Array();  
$supplierInsert = Array();  
$agencyInsert = Array();  
$contractNoticeInsert[] = $fname;  
$keys = array_keys($datamapping0711);  
for ($c = 0; $c < $num; $c++) {  
$data[$c] = trim($data[$c], "=");  
$data[$c] = trim($data[$c], "\"");  
if ($tablename == "contractnotice") {  
if (in_array(($datamapping0711[$headers[$c]]), $contractNoticeFields)) {  
if (($datamapping0711[$headers[$c]]) == "parentCN" || ($datamapping0711[$headers[$c]]) == "CNID") {  
$data[$c] = substr($data[$c], 2); // take off the "CN" prefix  
$data[$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) {  
$contractNoticeInsert[] = $data[$c];  
} else {  
$contractNoticeInsert[] = 0;  
}  
} elseif (($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]));  
} else {  
if (strstr("\" =", $data[$c] > 0)) {  
die("Invalid Description field" . $contractNoticeInsert);  
}  
$contractNoticeInsert[] = $data[$c];  
}  
}  
   
} else if ($tablename == "supplierdetails") {  
if (in_array(($datamapping0711[$headers[$c]]), $supplierFields)) {  
$supplierInsert[] = $data[$c];  
}  
   
} else if ($tablename == "agency") {  
if (in_array(($datamapping0711[$headers[$c]]), $agencyFields)) {  
$agencyInsert[] = $data[$c];  
}  
}  
   
  include_once ("../lib/common.inc.php");
  //auth();
  $contractNoticeFields = array(
} "importFile",
flush(); "agencyName",
if ($tablename == "contractnotice") { "parentCN",
$contractNoticeInsertQ->execute($contractNoticeInsert); "CNID",
$errors = $conn->errorInfo(); "publishDate",
  "amendDate",
if ($errors[1] == 7 || $errors[1] ==0) "contractStart",
$success++; "contractEnd",
else "value",
echo $data[2] . " failed CN insert.<br>" . print_r($errors,true) "description",
. " <br> ". print_r($contractNoticeInsert,true) ."<br> $row <br><br>\n"; "agencyID",
  "category",
  "procurementMethod",
} else if ($tablename == "supplierdetails") { "atmID",
$supplierInsertQ->execute($supplierInsert); "SONID",
$errors = $conn->errorInfo(); "confidentialityContract",
  "confidentialityContractReason",
if ($errors[1] == 7 || $errors[1] ==0) "confidentialityOutputs",
$success++; "confidentialityOutputsReason",
else "consultancy",
echo $data[2] . " failed agency insert.<br>" . print_r($errors,true) "consultancyReason",
. " <br> ". print_r($agencyInsert,true) ."<br> $row <br><br>\n"; "amendmentReason",
  "supplierName",
  "supplierAddress",
} else if ($tablename == "agency") { "supplierCity",
$agencyInsertQ->execute($agencyInsert); "supplierPostcode",
$errors = $conn->errorInfo(); "supplierCountry",
  "supplierABNExempt",
if ($errors[1] == 7 || $errors[1] ==0) "supplierABN",
$success++; "contactBranch",
else "contactDivision",
echo $data[2] . " failed supplier insert.<br>" . print_r($errors,true) "contactPostcode"
. " <br> ". print_r($supplierInsert,true) ."<br> $row <br><br>\n"; );
   
} $contractNoticeInsertQ = 'INSERT INTO contractnotice ("' . implode('" , "', $contractNoticeFields) . '") VALUES ( ';
  foreach ($contractNoticeFields as $key => $f) {
  $contractNoticeInsertQ.= ($key == 0 ? "" : ", ") . "?";
  }
flush(); $contractNoticeInsertQ.= ");";
  $contractNoticeInsertQ = $conn->prepare($contractNoticeInsertQ);
//echo "<hr>\n"; function processFile($fpath) {
} global $conn, $contractNoticeFields, $contractNoticeInsertQ;
  $success = 0;
$row++; $row = 1;
} $handle = fopen($fpath, "r");
  //"t" mode string translates windows line breaks to unix
fclose($handle); //On the 15 April 2014 AusTender Contract Notice functionality changed to improve transparency. Two new fields have been introduced to capture the ‘Amendment Value’ and the ‘Amendment Start Date’. Where applicable, these fields will display on amendment records published after this date.
return $success; $datamapping0712 = array(
} "Agency" => "agencyName",
  "Parent CN ID" => "parentCN",
$path = 'data/'; "CN ID" => "CNID",
if ($_REQUEST["fname"] == "") { "Publish Date" => "publishDate",
echo "Get files from: https://www.tenders.gov.au/?event=public.reports.list<br>"; "Amendment Date" => "amendDate",
$dhandle = opendir($path); "Amendment Publish Date" => "amendDate",
// define an array to hold the files "Status" => "",
$files = array(); "StartDate" => "contractStart",
  "EndDate" => "contractEnd",
if ($dhandle) { "Value" => "value",
// loop through all of the files "Description" => "description",
while (false !== ($fname = readdir($dhandle))) { "Agency Ref Id" => "agencyID",
if (($fname != '.') && ($fname != '..')) { "Agency Ref. ID" => "agencyID",
echo "<a href=\"import.php?fname=$fname\">$fname</a>&nbsp;" . filesize($path . $fname) . "&nbsp;" . date("c",filemtime($path . $fname)) . "<br/>"; "Category" => "category",
} "Procurement Method" => "procurementMethod",
} "ATM ID" => "atmID",
} "SON ID" => "SONID",
} else { "Confidentiality - Contract" => "confidentialityContract",
$success = 0; "Confidentiality - Contract Reason(s)" => "confidentialityContractReason",
$fname = $_REQUEST["fname"]; "Confidentiality - Outputs" => "confidentialityOutputs",
echo " ============== $fname ============== <br>"; "Confidentiality - Outputs Reason(s)" => "confidentialityOutputsReason",
flush(); "Consultancy" => "consultancy",
$success += processFile($path . $fname,"contractnotice"); "Consultancy Reason(s)" => "consultancyReason",
$success += processFile($path . $fname,"agency"); "Amendment Reason" => "amendmentReason",
$success += processFile($path . $fname,"agency"); "Supplier Name" => "supplierName",
echo "<br> $success records successfully created"; "Supplier Address" => "supplierAddress",
flush(); "Supplier City" => "supplierCity",
} "Supplier Postcode" => "supplierPostcode",
// run post import data processing "Supplier Country" => "supplierCountry",
// cn "Supplier ABNExempt" => "supplierABNExempt",
include("linkAmendments.php"); "Supplier ABN" => "supplierABN",
include("updateUNSPSC.php"); "Agency Branch" => "contactBranch",
  "Agency Divison" => "contactDivision",
  "Agency Postcode" => "contactPostcode",
  "" => ""
  );
   
   
  while (($data = fgetcsv($handle, 1200, "\t")) !== false) {
  $num = count($data);
  if ($row == 3) {
  $headers = $data;
  } elseif ($row > 3) {
  if ($num > count($datamapping0712)) {
  die("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>" . $fname . print_r($data));
  }
  $contractNoticeInsert = Array();
  $supplierInsert = Array();
  $agencyInsert = Array();
  $contractNoticeInsert[] = $fpath;
  $keys = array_keys($datamapping0712);
  for ($c = 0; $c < $num; $c++) {
  $data[$c] = trim($data[$c], "=");
  $data[$c] = trim($data[$c], "\"");
  if (in_array(($datamapping0712[$headers[$c]]), $contractNoticeFields)) {
  if (($datamapping0712[$headers[$c]]) == "parentCN" || ($datamapping0712[$headers[$c]]) == "CNID") {
  $data[$c] = substr($data[$c], 2); // take off the "CN" prefix
  if ($data[$c] > 0 && $data[$c] != '0' && $data[$c] != '') {
  $contractNoticeInsert[] = $data[$c];
  } else {
  $contractNoticeInsert[] = null;
  }
  } elseif (($datamapping0712[$headers[$c]]) == "supplierABN" || ($datamapping0712[$headers[$c]]) == "SONID") {
  if ($data[$c] > 0 && $data[$c] != '0' && $data[$c] != '') {
  $contractNoticeInsert[] = $data[$c];
  } else {
  $contractNoticeInsert[] = null;
  }
  } elseif (($datamapping0712[$headers[$c]]) == "amendDate" || ($datamapping0712[$headers[$c]]) == "publishDate" || ($datamapping0712[$headers[$c]]) == "contractStart" || ($datamapping0712[$headers[$c]]) == "contractEnd") {
  $contractNoticeInsert[] = date('Y-m-d H:i:s', strtotime($data[$c]));
  } else {
  if (strstr("\" =", $data[$c] > 0)) {
  die("Invalid Description field" . $contractNoticeInsert);
  }
  $colvalue = preg_replace('/[^[:print:]]/', '', utf8_encode($data[$c]));
   
  $contractNoticeInsert[] = $colvalue;
  }
  }
  }
  flush();
  $contractNoticeInsertQ->execute($contractNoticeInsert);
  $errors = $conn->errorInfo();
  if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) {
   
  } elseif ($errors[1] <= 0) {
  //print_r($errors);
  $success = 1;
  } else {
  foreach ($contractNoticeFields as $key => $cnf) {
  echo var_dump($contractNoticeInsert[$key]) . $cnf . "<br>";
  }
  echo $data[2] . " failed CN insert.<br>" . print_r($errors, true) . "<br> row $row <br><br>\n";
  }
   
  flush();
  //echo "<hr>\n";
  }
  $row++;
  }
  fclose($handle);
  $contractNoticeInsertQ->closeCursor();
   
  return $success;
  }
   
  $path = 'data/';
  if ($_REQUEST["fname"] == "") {
  echo "Get files from: https://www.tenders.gov.au/?event=public.reports.list<br>";
  $dhandle = opendir($path);
  // define an array to hold the files
  $files = array();
  if ($dhandle) {
  // loop through all of the files
  while (false !== ($fname = readdir($dhandle))) {
  if (($fname != '.') && ($fname != '..')) {
  $files[date("c", filemtime($path . $fname)) . md5($fname)] = $fname;
  }
  }
  }
  ksort($files);
  foreach ($files as $date => $fname) {
  echo "<a href=\"import.php?fname=$fname\">$fname</a>&nbsp;" . filesize($path . $fname) . "&nbsp;" . $date . "<br/>";
  }
  } else {
  $success = 0;
  $fname = $_REQUEST["fname"];
  echo " ============== $fname ============== <br>";
  flush();
  $success+= processFile($path . $fname, "contractnotice");
  $success+= processFile($path . $fname, "agency");
  $success+= processFile($path . $fname, "supplier");
  echo "<br> $success records successfully created";
   
  flush();
  // run post import data processing
  //
  if ($success > 0) {
  $conn->exec("update datasets set \"lastUpdated\" = NOW() where title = 'Contract Notices'");
  echo "link amend<br>";
  include ("linkAmendments.php");
  echo "update UNSPSC<br>";
  include ("updateUNSPSC.php");
  }
  // cn
   
// agency // agency
include("setAgencyStatus.php"); //include ("setAgencyStatus.php");
include("setAgencyURLABN.php"); //include ("setAgencyURLABN.php");
  }
  }
?> ?>