tweak year/agency modifer for queries
[contractdashboard.git] / admin / import.php
blob:a/admin/import.php -> blob:b/admin/import.php
--- a/admin/import.php
+++ b/admin/import.php
@@ -1,120 +1,270 @@
 <?php
-include_once("../lib/common.inc.php");
+include_once ("../lib/common.inc.php");
+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[] = $fpath;
+			$keys = array_keys($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]) && $data[$c] != "") die($data[$c] . " is not numeric");
+							if ($data[$c] > 0) {
+								$contractNoticeInsert[] = $data[$c];
+							}
+							else {
+								$contractNoticeInsert[] = 0;
+							}
+						}
+						elseif (($datamapping0711[$headers[$c]]) == "supplierABN") {
+							if ($data[$c] > 0) {
+								$contractNoticeInsert[] = $data[$c];
+							}
+							else {
+								$contractNoticeInsert[] = null;
+							}
+						}
+						elseif (($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" . $contractNoticeInsert);
+							}
+$colvalue =  preg_replace( '/[^[:print:]]/', '',utf8_encode( $data[$c]));
+
+							$contractNoticeInsert[] = $colvalue;
+						}
+					}
+				}
+				else if ($tablename == "supplierdetails") {
+					if (in_array(($datamapping0711[$headers[$c]]) , $supplierFields)) {
+						if (($datamapping0711[$headers[$c]]) == "supplierABN") {
+							if ($data[$c] > 0) {
+								$contractNoticeInsert[] = $data[$c];
+							}
+							else {
+								$contractNoticeInsert[] = 0;
+							}
+						}
+						else {
+							$supplierInsert[] = $data[$c];
+						}
+					}
+				}
+				else if ($tablename == "agency") {
+					if (in_array(($datamapping0711[$headers[$c]]) , $agencyFields)) {
+						$agencyInsert[] = $data[$c];
+					}
+				}
+			}
+			flush();
+			if ($tablename == "contractnotice") {
+				$contractNoticeInsertQ->execute($contractNoticeInsert);
+				$errors = $conn->errorInfo();
+				if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) {
+				}
+				elseif ($errors[1] == 0) {
+					$success++;
+				}
+				else {
+					foreach ($contractNoticeFields as $key => $cnf) {
+						echo var_dump($contractNoticeInsert[$key]) . $cnf . "<br>";
+					}
+					echo $data[2] . " failed CN insert.<br>" . print_r($errors, true) . "<br> row $row <br><br>\n";
+				}
+			}
+			else if ($tablename == "supplierdetails") {
+				$supplierInsertQ->execute($supplierInsert);
+				$errors = $conn->errorInfo();
+				if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) {
+				}
+				elseif ($errors[1] == 0) {
+					$success++;
+				}
+				else {
+					echo $data[2] . " failed supplier insert.<br>" . print_r($errors, true) . " <br> " . print_r($supplierInsert, true) . "<br> $row <br><br>\n";
+				}
+			}
+			else if ($tablename == "agency") {
+				$agencyInsertQ->execute($agencyInsert);
+				$errors = $conn->errorInfo();
+				if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) {
+				}
+				elseif ($errors[1] == 0) {
+					$success++;
+				}
+				else {
+					echo $data[2] . " failed agency insert.<br>" . print_r($errors, true) . " <br> " . print_r($agencyInsert, true) . "<br> $row <br><br>\n";
+				}
+			}
+			flush();
+			//echo "<hr>\n";
+			
+		}
+		$row++;
+	}
+	fclose($handle);
+	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 != '..')) {
-	  echo "<a href=\"import.php?fname=$fname\">$fname</a>".filesize($path.$fname)."<br/>";
-	  }
+	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 != '..')) {
+				echo "<a href=\"import.php?fname=$fname\">$fname</a>&nbsp;" . filesize($path . $fname) . "&nbsp;" . date("c", filemtime($path . $fname)) . "<br/>";
+			}
+		}
 	}
 }
-
-} else {
-$success =0;
-$fname = $_REQUEST["fname"];
-echo " ============== $fname  ============== <br>";
-$row = 1;
-$handle = fopen($path.$fname, "r"); 
-//"t" mode string translates windows line breaks to unix
-$datamapping0709 = 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("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" );
-
-while (($data = fgetcsv($handle, 1000, "\t")) !== FALSE) {
-    $num = count($data);
-
-	if ($row == 3)
-	{
-	$headers = $data;
-	}
-	else if ($row > 3)
-	{
-			if ($num > count($datamapping0709))
-		{
-		die ("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>".$fname.print_r($data));
-		}
-	$firstCN= false;
-    $firstSupplier= true;
-    $firstAgency= true;
-    $contractNoticeInsert = "INSERT INTO contractnotice (importFile, ".implode(" , ",$contractNoticeFields). ") VALUES ( \"$fname\" ";
-    $supplierInsert = "INSERT INTO supplierdetails (".implode(" , ",$supplierFields). ") VALUES (";
-    $agencyInsert = "INSERT INTO agency (".implode(" , ",$agencyFields). ") VALUES (";
-	
-	$keys = array_keys($datamapping0709);
-    for ($c=0; $c < $num; $c++) {
-	$data[$c] = trim($data[$c],"=");
-	$data[$c] = trim($data[$c], "\"");
-	
-	if (in_array(($datamapping0709[$headers[$c]]),$contractNoticeFields))
-	{
-	if (($datamapping0709[$headers[$c]]) == "parentCN" || ($datamapping0709[$headers[$c]]) == "CNID")
-	{
-		$data[$c] = substr($data[$c],2);
-		if ($data[$c] >0)
-		{
-		$contractNoticeInsert .= ($firstCN ? "" : ","). mysql_real_escape_string($data[$c]);
-		} else
-		{
-		$contractNoticeInsert .= ($firstCN ? "" : ",")."''";
-		}
-	} else 	if (($datamapping0709[$headers[$c]]) == "amendDate" || ($datamapping0709[$headers[$c]]) == "publishDate" || ($datamapping0709[$headers[$c]]) == "contractStart" || ($datamapping0709[$headers[$c]]) == "contractEnd")
-	{
-		$contractNoticeInsert .= ($firstCN ? "" : ","). "'".date( 'Y-m-d H:i:s', strtotime($data[$c])). "'";
-	} else
-	{
-		if (strstr("\" =",$data[$c] > 0))
-			{
-				die ("Invalid Description field" . $contractNoticeInsert); 
-			}
-        $contractNoticeInsert .= ($firstCN ? "" : ",") .  "'". mysql_real_escape_string($data[$c])."'";
-		
-		}
-		$firstCN = false;
-    }
-		if (in_array(($datamapping0709[$headers[$c]]),$supplierFields))
-	{
-        $supplierInsert .= ($firstSupplier ? "" : ",") .  "'". mysql_real_escape_string($data[$c])."'";
-		$firstSupplier = false;
-    }
-			if (in_array(($datamapping0709[$headers[$c]]),$agencyFields))
-	{
-        $agencyInsert .= ($firstAgency ? "" : ",") .  "'". mysql_real_escape_string($data[$c])."'";
-		$firstAgency = false;
-    }
-	
-	}
-	$contractNoticeInsert .= ");";
-	$supplierInsert .= ");";
-	$agencyInsert .= ");";
-	
-$result = mysql_query($contractNoticeInsert);
-if ($result) $success++;
-else if (strpos(mysql_error(),"Duplicate entry") === false) echo $data[2] . " failed CN insert.<br>". mysql_error() ." <br>  $contractNoticeInsert <br><br>\n";
-	
-	$result = mysql_query($agencyInsert);
-if ($result) $success++;
-else if (strpos(mysql_error(),"Duplicate entry") === false) echo $data[2] . " failed agency insert.<br>". mysql_error() ." <br> $agencyInsert <br><br>\n";
-	
-	$result = mysql_query($supplierInsert);
-if ($result) $success++;
-else if (strpos(mysql_error(),"Duplicate entry") === false) echo $data[2] . " failed supplier insert.<br>". mysql_error() ." <br> $supplierInsert <br><br>\n";
-
-//echo "<hr>\n";
-	}
-	
-    $row++;
+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();
 }
-echo "<br> $success records successfully created";
-fclose($handle);
-}
+// run post import data processing
+// cn
+include ("linkAmendments.php");
+include ("updateUNSPSC.php");
+// agency
+//include ("setAgencyStatus.php");
+//include ("setAgencyURLABN.php");
 ?>