more ausbudget work
[disclosr.git] / admin / importAusbudget.php
blob:a/admin/importAusbudget.php -> blob:b/admin/importAusbudget.php
<?php <?php
   
include_once("../include/common.inc.php"); require_once '../include/common.inc.php';
require($basePath . 'lib/phpquery/phpQuery/phpQuery.php');  
   
setlocale(LC_CTYPE, 'C'); $db = $server->get_db('disclosr-agencies');
  $rows = $db->get_view("app", "byName")->rows;
  $nametoid = Array();
  $accounts = Array();
  foreach ($rows as $row) {
  $nametoid[trim($row->key)] = $row->value;
  }
  $nameField = "Agency";
  $elevenField = "2011-2012";
  $twelveField = "2012-2013";
  // $request = Requests::get($url);
  //echo $url;
  // $Data = str_getcsv($request->body, "\n"); //parse the rows
  $Data = str_getcsv(file_get_contents("ausbudget.csv"), "\n"); //parse the rows
  $headers = Array();
  foreach ($Data as $num => $line) {
  $Row = str_getcsv($line, ",");
  if ($num == 0) {
  $headers = $Row;
  print_r($headers);
  } else {
  if (isset($Row[array_search($nameField, $headers)])) {
  $agencyName = $Row[array_search($nameField, $headers)];
  if (!in_array(trim($agencyName), array_keys($nametoid))) {
  echo trim($agencyName) . " missing" . PHP_EOL;
  } else {
  //echo $Row[array_search($nameField, $headers)] . PHP_EOL;
  // print_r($Row);
  if (!is_array($accounts[$nametoid[trim($agencyName)]]["budget"]['2011-2012'])) $accounts[$nametoid[trim($agencyName)]]["budget"]['2011-2012'] = Array();
  if (!is_array($accounts[$nametoid[trim($agencyName)]]["budget"]['2012-2013'])) $accounts[$nametoid[trim($agencyName)]]["budget"]['2012-2013'] = Array();
   
function getTextFromNode($Node, $Text = "") { if (!isset($accounts[$nametoid[trim($agencyName)]]["budget"]['2011-2012']['value'])) $accounts[$nametoid[trim($agencyName)]]["budget"]['2011-2012']['value'] = 0;
if (!isset($Node->tagName) || $Node->tagName == null) { if (!isset($accounts[$nametoid[trim($agencyName)]]["budget"]['2012-2013']['value'])) $accounts[$nametoid[trim($agencyName)]]["budget"]['2012-2013']['value'] = 0;
return $Text . $Node->textContent; $accounts[$nametoid[trim($agencyName)]]["budget"]['2011-2012']['value'] += $Row[array_search($elevenField, $headers)];
  $accounts[$nametoid[trim($agencyName)]]["budget"]['2012-2013']['value'] += $Row[array_search($twelveField, $headers)];
   
  $accounts[$nametoid[trim($agencyName)]]["budget"]['2011-2012']['source'] = 'http://budgetaus.net/';
  $accounts[$nametoid[trim($agencyName)]]["budget"]['2012-2013']['source'] = 'http://budgetaus.net/';
  }
  } else {
  echo "error finding any agency" . $line . PHP_EOL;
  }
} }
$Node = $Node->firstChild;  
if (isset($Node) && $Node != null) {  
$Text = getTextFromNode($Node, $Text);  
}  
while (isset($Node->nextSibling) && $Node->nextSibling != null) {  
$Text = getTextFromNode($Node->nextSibling, $Text);  
$Node = $Node->nextSibling;  
}  
return $Text;  
} }
   
$dir = "./ausbudget/";  
$dhandle = opendir("./ausbudget/");  
// define an array to hold the files //print_r($accounts);
$files = array(); foreach ($accounts as $id => $allvalues) {
if ($dhandle) { echo $id . "<br>" . PHP_EOL;
// loop through all of the files $doc = object_to_array($db->get($id));
while (false !== ($fname = readdir($dhandle))) { // print_r($doc);
if (($fname != '.') && ($fname != '..')) {  
echo "$fname <br>"; foreach ($allvalues as $valueType => $values) {
$html = phpQuery::newDocumentHTML(file_get_contents($dir . $fname)); if (!isset($doc['statistics'][$valueType]) || !is_array($doc['statistics'][$valueType])) {
phpQuery::selectDocument($html);  
foreach (pq('table')->elements as $table) { $doc['statistics'][$valueType] = Array();
echo "loltable";  
//echo $table->ownerDocument->saveXML($table);  
foreach (pq('tr')->elements as $row) {  
echo "lolrow";  
$rowText = getTextFromNode($row);  
if (strpos($rowText,"Twitter") === false) {  
echo "<b>".pq("td:first",$row)->text()."</b><br>";  
echo pq("td",$row)->text() . "<br>";  
}  
}  
}  
} }
break; $doc['statistics'][$valueType] = array_unique(array_merge($doc['statistics'][$valueType], $values));
} }
   
  $db->save($doc);
} }
?> ?>