projects / contractdashboard.git / commitdiff
search:
summary | shortlog | log | commit | commitdiff | tree
unified | plain
Merge branch 'master' of ssh://apples.lambdacomplex.org/git/contractdashboard
Merge branch 'master' of ssh://apples.lambdacomplex.org/git/contractdashboard

file:a/admin/partialdata/import.php -> file:b/admin/partialdata/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;
echo " ============== $fpath ============== <br>"; echo " ============== $fpath ============== <br>";
flush(); flush();
$row = 1; $row = 1;
$success = 0; $success = 0;
$dupes = 0; $dupes = 0;
$handle = fopen($fpath, "r"); $handle = fopen($fpath, "r");
//"t" mode string translates windows line breaks to unix //"t" mode string translates windows line breaks to unix
$datamapping0507 = array( $datamapping0507 = array(
"Agency" => "agencyName", "Agency" => "agencyName",
"CN ID" => "CNID", "CN ID" => "CNID",
"Publish Date" => "publishDate", "Publish Date" => "publishDate",
"Contract Start Date" => "contractStart", "Contract Start Date" => "contractStart",
"Contract End Date" => "contractEnd", "Contract End Date" => "contractEnd",
"Value (AUD)" => "value", "Value (AUD)" => "value",
"Title" => "description", "Title" => "description",
"Category" => "category", "Category" => "category",
"ATM ID" => "atmID", "ATM ID" => "atmID",
"Supplier Name" => "supplierName", "Supplier Name" => "supplierName",
"LastUpdated" => "amendDate", "LastUpdated" => "amendDate",
"" => "" "" => ""
); );
$headers; $headers;
$contractNoticeFields = array( $contractNoticeFields = array(
"importFile", "importFile",
"CNID", "CNID",
"description", "description",
"agencyName", "agencyName",
"publishDate", "publishDate",
"category", "category",
"contractStart", "contractStart",
"contractEnd", "contractEnd",
"value", "value",
"atmID", "atmID",
"supplierName", "supplierName",
"amendDate" "amendDate"
); );
if ($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);
} }
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($datamapping0507)) { if ($num > count($datamapping0507)) {
die("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>" . $fname . "data:" .$num. print_r($data ,true). "mapping:" . count($datamapping0507). print_r($datamapping0507 ,true)); die("<font color=red>Error in data import; data mapping fields out of bounds or changed</font><br>" . $fname . "data:" .$num. print_r($data ,true). "mapping:" . count($datamapping0507). print_r($datamapping0507 ,true));
} }
$contractNoticeInsert = Array(); $contractNoticeInsert = Array();
$contractNoticeInsert[] = $fpath; $contractNoticeInsert[] = $fpath;
$keys = array_keys($datamapping0507); $keys = array_keys($datamapping0507);
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(($datamapping0507[$headers[$c]]) , $contractNoticeFields)) { if (in_array(($datamapping0507[$headers[$c]]) , $contractNoticeFields)) {
if (($datamapping0507[$headers[$c]]) == "parentCN" || ($datamapping0507[$headers[$c]]) == "CNID") { if (($datamapping0507[$headers[$c]]) == "parentCN" || ($datamapping0507[$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]) && $data[$c] != "") die($data[$c] . " is not numeric"); if (!is_numeric($data[$c]) && $data[$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; $contractNoticeInsert[] = 0;
} }
} }
elseif (($datamapping0507[$headers[$c]]) == "supplierABN") { elseif (($datamapping0507[$headers[$c]]) == "supplierABN") {
if ($data[$c] > 0) { if ($data[$c] > 0) {
$contractNoticeInsert[] = $data[$c]; $contractNoticeInsert[] = $data[$c];
} }
else { else {
$contractNoticeInsert[] = null; $contractNoticeInsert[] = null;
} }
} }
elseif (($datamapping0507[$headers[$c]]) == "amendDate" || ($datamapping0507[$headers[$c]]) == "publishDate" || ($datamapping0507[$headers[$c]]) == "contractStart" || ($datamapping0507[$headers[$c]]) == "contractEnd") { elseif (($datamapping0507[$headers[$c]]) == "amendDate" || ($datamapping0507[$headers[$c]]) == "publishDate" || ($datamapping0507[$headers[$c]]) == "contractStart" || ($datamapping0507[$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);
} }
$colvalue = preg_replace( '/[^[:print:]]/', '',utf8_encode( $data[$c])); $colvalue = preg_replace( '/[^[:print:]]/', '',utf8_encode( $data[$c]));
   
$contractNoticeInsert[] = $colvalue; $contractNoticeInsert[] = $colvalue;
} }
} }
} }
} }
flush(); flush();
if ($tablename == "contractnotice") { if ($tablename == "contractnotice") {
$contractNoticeInsertQ->execute($contractNoticeInsert); $contractNoticeInsertQ->execute($contractNoticeInsert);
$errors = $conn->errorInfo(); $errors = $conn->errorInfo();
if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) { if ($errors[1] == 7 && strpos($errors[2], "duplicate key")) {
$dupes++; $dupes++;
} }
elseif ($errors[1] == 0) { elseif ($errors[1] == 0) {
$success++; $success++;
} }
else { else {
foreach ($contractNoticeFields as $key => $cnf) { foreach ($contractNoticeFields as $key => $cnf) {
echo var_dump($contractNoticeInsert[$key]) . $cnf . "<br>"; echo var_dump($contractNoticeInsert[$key]) . $cnf . "<br>";
} }
echo $data[2] . " failed CN insert.<br>" . print_r($errors, true) . "<br> row $row <br><br>\n"; echo $data[2] . " failed CN insert.<br>" . print_r($errors, true) . "<br> row $row <br><br>\n";
} }
} }
flush(); flush();
//echo "<hr>\n"; //echo "<hr>\n";
} }
$row++; $row++;
} }
fclose($handle); fclose($handle);
echo " $dupes duplicate records<br>"; echo " $dupes duplicate records<br>";
echo " $success records successfully created<br>"; echo " $success records successfully created<br>";
flush(); flush();
return $success; return $success;
} }
$path = './'; $path = './';
if ($_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 != '.') && ($fname != '..') && (!isset($_REQUEST["filter"]) || strpos($fname,$_REQUEST["filter"]) != false)) {
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 . $fname)) . "<br/>";
processFile($path . $fname, "contractnotice"); processFile($path . $fname, "contractnotice");
} }
} }
} }
} }
else { else {
$success = 0; $success = 0;
$fname = $_REQUEST["fname"]; $fname = $_REQUEST["fname"];
$success+= processFile($path . $fname, "contractnotice"); $success+= processFile($path . $fname, "contractnotice");
   
} }
?> ?>
   
file:b/admin/partialdata/importamendments.php (new)
  <?php
  include_once ("../../lib/common.inc.php");
 
  // http://www.lastcraft.com/browser_documentation.php
  // http://code.google.com/p/phpquery/
  require('phpQuery-onefile.php');
  function getURL($url) {
  //return file_get_contents($url);
  $ch = curl_init($url);
  curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
  curl_setopt($ch, CURLOPT_HEADER, 0);
  curl_setopt($ch, CURLOPT_TIMEOUT, 45);
  $page = curl_exec($ch);
  if (curl_errno($ch)) {
  echo "<font color=red> Database temporarily unavailable: ";
  echo curl_errno($ch) . " " . curl_error($ch);
 
  echo $url;
  echo "</font><br>";
  }
  curl_close($ch);
  return $page;
  }
 
 
  function getTextFromTHNode($Node, $Text = "") {
  if ($Node->tagName == null)
  return $Text.$Node->textContent;
 
  if ($Node->tagName != "td") {
  $Node = $Node->firstChild;
  if ($Node != null)
  $Text = getTextFromTHNode($Node, $Text);
 
  while($Node->nextSibling != null) {
  $Text = getTextFromTHNode($Node->nextSibling, $Text);
  $Node = $Node->nextSibling;
  }
  }
  return $Text;
  }
 
  function getTextFromNode($Node, $Text = "") {
  if ($Node->tagName == null)
  return $Text.$Node->textContent;
  if ($Node->tagName != "th" && $Node->tagName != "span") {
  $Node = $Node->firstChild;
  if ($Node != null)
  $Text = getTextFromNode($Node, $Text);
 
  while($Node->nextSibling != null) {
  $Text = getTextFromNode($Node->nextSibling, $Text);
  $Node = $Node->nextSibling;
  }
  }
  return $Text;
  }
  function dom_to_array($root)
  {
  $result = array();
 
  if ($root->hasAttributes())
  {
  $attrs = $root->attributes;
 
  foreach ($attrs as $i => $attr)
  $result[$attr->name] = $attr->value;
  }
 
  $children = $root->childNodes;
  if ($root->childNodes) {
  if ($children->length == 1)
  {
  $child = $children->item(0);
 
  if ($child->nodeType == XML_TEXT_NODE)
  {
  $result['_value'] = $child->nodeValue;
 
  if (count($result) == 1)
  return $result['_value'];
  else
  return $result;
  }
  }
 
  $group = array();
 
  for($i = 0; $i < $children->length; $i++)
  {
  $child = $children->item($i);
 
  if (!isset($result[$child->nodeName]))
  $result[$child->nodeName] = dom_to_array($child);
  else
  {
  if (!isset($group[$child->nodeName]))
  {
  $tmp = $result[$child->nodeName];
  $result[$child->nodeName] = array($tmp);
  $group[$child->nodeName] = 1;
  }
 
  $result[$child->nodeName][] = dom_to_array($child);
  }
  }
  }
 
  return $result;
  }
 
  function importCN($cnid) {
  global $conn;
  $CN = str_replace("-A", "00",$cnid);
  // check if already complete
  $query = 'Select "parentCN" from contractnotice
  where "CNID" = :CNID';
  $query = $conn->prepare($query);
  $query->bindParam(":CNID", $CN);
  $query->execute();
  $r = $query->fetch(PDO::FETCH_ASSOC);
  if ($r['parentCN'] == NULL) {
  $site = "https://www.tenders.gov.au/";
  $searchResult = phpQuery::newDocument(getURL("https://www.tenders.gov.au/?event=public.advancedsearch.keyword&keyword=CN".$cnid));
  //echo "https://www.tenders.gov.au/?event=public.advancedsearch.keyword&keyword=CN".$cnid;
  $url = "";
  foreach(pq('a') as $a) {
  if (strpos($a->getAttribute("href"),"public.cn.view") >0 ) {
  //echo $a->getAttribute("href");
  $url = $a->getAttribute("href");
  break;
  }
  }
  $cn = phpQuery::newDocument(getURL($site.$url));
  $datamapping0711 = array(
  "Agency" => "agencyName",
  "Parent CN" => "parentCN",
  "CN ID" => "CNID",
  "Publish Date" => "publishDate",
  "Amendment Date" => "amendDate",
  "Status" => "",
  "StartDate" => "contractStart",
  "EndDate" => "contractEnd",
  "Contract Value (AUD)" => "value",
  "Description" => "description",
  "Agency Reference ID" => "agencyID",
  "Category" => "category",
  "Procurement Method" => "procurementMethod",
  "ATM ID" => "atmID",
  "SON ID" => "SONID",
  "Confidentiality - Contract" => "confidentialityContract",
  "Confidentiality Reason(s) - Contract" => "confidentialityContractReason",
  "Confidentiality - Outputs" => "confidentialityOutputs",
  "Confidentiality Reason(s) - Outputs" => "confidentialityOutputsReason",
  "Consultancy" => "consultancy",
  "Consultancy Reason(s)" => "consultancyReason",
  "Amendment Reason" => "amendmentReason",
  "Name" => "supplierName",
  "Postal Address" => "supplierAddress",
  "Town/City" => "supplierCity",
  "Postcode" => "supplierPostcode",
  "Country" => "supplierCountry",
  "ABN Exempt" => "supplierABNExempt",
  "ABN" => "supplierABN",
  "Branch" => "contactBranch",
  "Division" => "contactDivision",
  "Office Postcode" => "contactPostcode"
  );
  $cnFields = Array();
  foreach(pq('tr') as $tr) {
  $tra = dom_to_array($tr);
  if (is_array($tra['th'])) {
  $fieldName = trim(getTextFromTHNode($tr));
  } else {
  $fieldName = trim(str_replace("/th>","",$tra['th']));
  }
  $fieldValue = trim(print_r($tra['td'],true));
  if ($fieldName == "State/Territory" || $fieldName == "Contact Name"
  || $fieldName == "Contact Phone" || $fieldName == "Contact Email"
  ||$fieldName == "Amendments") {
  // do nothing
  } else if ($fieldName == "Contract Period") {
  $contractPeriod = explode("to",$fieldValue);
  $cnFields["contractStart"] = trim($contractPeriod[0]);
  $cnFields["contractEnd"] = trim($contractPeriod[1]);
  } else {
  $fieldName = $datamapping0711[$fieldName];
 
  if ($fieldName == "parentCN" || $fieldName == "CNID") {
  if (is_array($tra['td'])) {
  $fieldValue = trim(getTextFromNode($tr));
  }
  $fieldValue = substr($fieldValue, 2); // take off the "CN" prefix
  $fieldValue = str_replace("-A", "00", $fieldValue); // make amendments really big numbers
  } elseif ($fieldName == "description") {
 
  if (is_array($tra['td'])) $fieldValue = print_r($tra['td']['p'],true);
 
  } elseif ($fieldName == "value" || $fieldName == "supplierABN") {
  if (is_array($tra['td'])) {
  $fieldValue = trim(getTextFromNode($tr));
  }
  $fieldValue = str_replace(Array("$",","," "), "", $fieldValue);
  //if (!is_numeric($fieldValue)) $fieldValue = 0;
  if ($fieldValue == "Exempt") $fieldValue = NULL;
  } elseif ($fieldName == "amendDate" || $fieldName == "publishDate" || $fieldName == "contractStart" || $fieldName == "contractEnd") {
  $fieldValue = date('Y-m-d H:i:s', strtotime($fieldValue));
  } elseif (is_array($tra['td'])) {
  $fieldValue = trim(getTextFromNode($tr));
  }
  echo $fieldName. " = " .$fieldValue."<br>\n";
  $cnFields[$fieldName] = $fieldValue;
  }
  }
  $cnFields["importFile"] = $url;
  $contractNoticeInsertQ = 'INSERT INTO contractnotice ("' . implode('" , "', array_keys($cnFields)) . '") VALUES ( ';
  for($key = 0; $key < sizeof($cnFields); $key++) {
  $contractNoticeInsertQ.= ($key == 0 ? "" : ", ") . "?";
  }
  $contractNoticeInsertQ.= ");";
  //echo $contractNoticeInsertQ;
  $contractNoticeInsertQ = $conn->prepare($contractNoticeInsertQ);
  $contractNoticeInsertQ->execute(array_values($cnFields));
  $errors = $conn->errorInfo();
  if ($errors[1] == 7 && strpos($errors[2