projects / contractdashboard.git / blobdiff
search:
summary | shortlog | log | commit | commitdiff | tree
unified | plain
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",
  "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"
  );
   
  $contractNoticeInsertQ = 'INSERT INTO contractnotice ("' . implode('" , "', $contractNoticeFields) . '") VALUES ( ';
  foreach ($contractNoticeFields as $key => $f) {
  $contractNoticeInsertQ.= ($key == 0 ? "" : ", ") . "?";
  }
  $contractNoticeInsertQ.= ");";
  $contractNoticeInsertQ = $conn->prepare($contractNoticeInsertQ);
   
  function processFile($fpath) {
  global $conn, $contractNoticeFields, $contractNoticeInsertQ;
  $row = 1;
  $handle = fopen($fpath, "r");
  //"t" mode string translates windows line breaks to unix
  $datamapping0712 = 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",
  "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",
  "" => ""
  );
  $headers;
}  
flush(); while (($data = fgetcsv($handle, 1000, "\t")) !== false) {
if ($tablename == "contractnotice") { $num = count($data);
$contractNoticeInsertQ->execute($contractNoticeInsert); if ($row == 3) {
$errors = $conn->errorInfo(); $headers = $data;
  } elseif ($row > 3) {
if ($errors[1] == 7 || $errors[1] ==0) if ($num > count($datamapping0712)) {
$success++; die("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>" . $fname . print_r($data));
else }
echo $data[2] . " failed CN insert.<br>" . print_r($errors,true) $contractNoticeInsert = Array();
. " <br> ". print_r($contractNoticeInsert,true) ."<br> $row <br><br>\n"; $supplierInsert = Array();
  $agencyInsert = Array();
  $contractNoticeInsert[] = $fpath;
} else if ($tablename == "supplierdetails") { $keys = array_keys($datamapping0712);
$supplierInsertQ->execute($supplierInsert); for ($c = 0; $c < $num; $c++) {
$errors = $conn->errorInfo(); $data[$c] = trim($data[$c], "=");
  $data[$c] = trim($data[$c], "\"");
if ($errors[1] == 7 || $errors[1] ==0) if (in_array(($datamapping0712[$headers[$c]]), $contractNoticeFields)) {
$success++; if (($datamapping0712[$headers[$c]]) == "parentCN" || ($datamapping0712[$headers[$c]]) == "CNID") {
else $data[$c] = substr($data[$c], 2); // take off the "CN" prefix
echo $data[2] . " failed agency insert.<br>" . print_r($errors,true) if ($data[$c] > 0 && $data[$c] != '0') {
. " <br> ". print_r($agencyInsert,true) ."<br> $row <br><br>\n"; $contractNoticeInsert[] = $data[$c];
  } else {
  $contractNoticeInsert[] = null;
} else if ($tablename == "agency") { }
$agencyInsertQ->execute($agencyInsert); } elseif (($datamapping0712[$headers[$c]]) == "supplierABN") {
$errors = $conn->errorInfo(); if ($data[$c] > 0 && $data[$c] != '0') {
  $contractNoticeInsert[] = $data[$c];
if ($errors[1] == 7 || $errors[1] ==0) } else {
$success++; $contractNoticeInsert[] = null;
else }
echo $data[2] . " failed supplier insert.<br>" . print_r($errors,true) } elseif (($datamapping0712[$headers[$c]]) == "amendDate" || ($datamapping0712[$headers[$c]]) == "publishDate" || ($datamapping0712[$headers[$c]]) == "contractStart" || ($datamapping0712[$headers[$c]]) == "contractEnd") {
. " <br> ". print_r($supplierInsert,true) ."<br> $row <br><br>\n"; $contractNoticeInsert[] = date('Y-m-d H:i:s', strtotime($data[$c]));
  } else {
} if (strstr("\" =", $data[$c] > 0)) {
  die("Invalid Description field" . $contractNoticeInsert);
  }
  $colvalue = preg_replace('/[^[:print:]]/', '', utf8_encode($data[$c]));
flush();  
  $contractNoticeInsert[] = $colvalue;
//echo "<hr>\n"; }
} }
  }
$row++; flush();
} $contractNoticeInsertQ->execute($contractNoticeInsert);
  $errors = $conn->errorInfo();
fclose($handle); if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) {
return $success;  
} } elseif ($errors[1] == 0) {
  $success++;
$path = 'data/'; } else {
if ($_REQUEST["fname"] == "") { foreach ($contractNoticeFields as $key => $cnf) {
echo "Get files from: https://www.tenders.gov.au/?event=public.reports.list<br>"; echo var_dump($contractNoticeInsert[$key]) . $cnf . "<br>";
$dhandle = opendir($path); }
// define an array to hold the files echo $data[2] . " failed CN insert.<br>" . print_r($errors, true) . "<br> row $row <br><br>\n";
$files = array(); }
   
if ($dhandle) { flush();
// loop through all of the files //echo "<hr>\n";
while (false !== ($fname = readdir($dhandle))) { }
if (($fname != '.') && ($fname != '..')) { $row++;
echo "<a href=\"import.php?fname=$fname\">$fname</a>&nbsp;" . filesize($path . $fname) . "&nbsp;" . date("c",filemtime($path . $fname)) . "<br/>"; }
} fclose($handle);
} $contractNoticeInsertQ->closeCursor();
}  
} else { return $success;
$success = 0; }
$fname = $_REQUEST["fname"];  
echo " ============== $fname ============== <br>"; $path = 'data/';
flush(); if ($_REQUEST["fname"] == "") {
$success += processFile($path . $fname,"contractnotice"); echo "Get files from: https://www.tenders.gov.au/?event=public.reports.list<br>";
$success += processFile($path . $fname,"agency"); $dhandle = opendir($path);
$success += processFile($path . $fname,"agency"); // define an array to hold the files
echo "<br> $success records successfully created"; $files = array();
flush(); if ($dhandle) {
} // loop through all of the files
// run post import data processing while (false !== ($fname = readdir($dhandle))) {
// cn if (($fname != '.') && ($fname != '..')) {
include("linkAmendments.php"); $files[date("c", filemtime($path . $fname)) . md5($fname)] = $fname;
include("updateUNSPSC.php"); }
  }
  }
  ksort($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");
  }
  }
?> ?>
   
Unnamed repository; edit this file to name it for gitweb.
GitPHP by Chris Han