beginning semantic markup
beginning semantic markup


Former-commit-id: a5eba7337e3796c4f37731406127e5df85c440fa

[submodule "couchdb/couchdb-lucene"] [submodule "couchdb/couchdb-lucene"]
path = couchdb/couchdb-lucene path = couchdb/couchdb-lucene
url = https://github.com/rnewson/couchdb-lucene.git url = https://github.com/rnewson/couchdb-lucene.git
[submodule "couchdb/settee"]  
path = couchdb/settee  
url = https://github.com/inadarei/settee.git  
[submodule "lib/php-diff"] [submodule "lib/php-diff"]
path = lib/php-diff path = lib/php-diff
url = https://github.com/chrisboulton/php-diff.git url = https://github.com/chrisboulton/php-diff.git
[submodule "lib/Requests"] [submodule "lib/Requests"]
path = lib/Requests path = lib/Requests
url = https://github.com/rmccue/Requests.git url = https://github.com/rmccue/Requests.git
[submodule "js/flotr2"] [submodule "js/flotr2"]
path = js/flotr2 path = js/flotr2
url = https://github.com/HumbleSoftware/Flotr2.git url = https://github.com/HumbleSoftware/Flotr2.git
[submodule "lib/phpquery"] [submodule "lib/phpquery"]
path = lib/phpquery path = lib/phpquery
url = https://github.com/TobiaszCudnik/phpquery.git url = https://github.com/TobiaszCudnik/phpquery.git
[submodule "js/sigma"] [submodule "js/sigma"]
path = js/sigma path = js/sigma
url = https://github.com/jacomyal/sigma.js.git url = https://github.com/jacomyal/sigma.js.git
[submodule "js/bubbletree"] [submodule "js/bubbletree"]
path = js/bubbletree path = js/bubbletree
url = https://github.com/okfn/bubbletree.git url = https://github.com/okfn/bubbletree.git
[submodule "lib/querypath"] [submodule "lib/querypath"]
path = lib/querypath path = lib/querypath
url = https://github.com/technosophos/querypath.git url = https://github.com/technosophos/querypath.git
[submodule "lib/amon-php"] [submodule "lib/amon-php"]
path = lib/amon-php path = lib/amon-php
url = https://github.com/martinrusev/amon-php.git url = https://github.com/martinrusev/amon-php.git
[submodule "documents/lib/parsedatetime"] [submodule "documents/lib/parsedatetime"]
path = documents/lib/parsedatetime path = documents/lib/parsedatetime
url = git://github.com/bear/parsedatetime.git url = git://github.com/bear/parsedatetime.git
[submodule "lib/FeedWriter"] [submodule "lib/FeedWriter"]
path = lib/FeedWriter path = lib/FeedWriter
url = https://github.com/mibe/FeedWriter url = https://github.com/mibe/FeedWriter
   
<?php <?php
   
include_once("../include/common.inc.php"); include_once("../include/common.inc.php");
   
$format = "csv"; $format = "csv";
//$format = "json"; //$format = "json";
if (isset($_REQUEST['format'])) $format = $_REQUEST['format']; if (isset($_REQUEST['format']))
  $format = $_REQUEST['format'];
setlocale(LC_CTYPE, 'C'); setlocale(LC_CTYPE, 'C');
if ($format == "csv") { if ($format == "csv") {
$headers = Array("name"); $headers = Array("name");
} else { } else {
$headers = Array(); $headers = Array();
} }
   
$db = $server->get_db('disclosr-agencies'); $db = $server->get_db('disclosr-agencies');
try { try {
$rows = $db->get_view("app", "all", null, true)->rows; $rows = $db->get_view("app", "all", null, true)->rows;
   
$dataValues = Array(); $dataValues = Array();
foreach ($rows as $row) { foreach ($rows as $row) {
if (isset($row->value->statistics->employees)) { if (isset($row->value->statistics->employees)) {
   
$headers = array_unique(array_merge($headers, array_keys(object_to_array($row->value->statistics->employees)))); $headers = array_unique(array_merge($headers, array_keys(object_to_array($row->value->statistics->employees))));
   
} }
} }
} catch (SetteeRestClientException $e) { } catch (SetteeRestClientException $e) {
setteErrorHandler($e); setteErrorHandler($e);
} }
   
$fp = fopen('php://output', 'w'); $fp = fopen('php://output', 'w');
if ($fp && $db) { if ($fp && $db) {
if ($format == "csv") { if ($format == "csv") {
header('Content-Type: text/csv; charset=utf-8'); header('Content-Type: text/csv; charset=utf-8');
header('Content-Disposition: attachment; filename="export.employeestats.' . date("c") . '.csv"'); header('Content-Disposition: attachment; filename="export.employeestats.' . date("c") . '.csv"');
} }
header('Pragma: no-cache'); header('Pragma: no-cache');
header('Expires: 0'); header('Expires: 0');
if ($format == "csv") { if ($format == "csv") {
fputcsv($fp, $headers); fputcsv($fp, $headers);
} else if ($format == "json") { } else if ($format == "json") {
echo '{ echo '{
"labels" : ["' . implode('","', $headers) . '"],'.PHP_EOL; "labels" : ["' . implode('","', $headers) . '"],' . PHP_EOL;
} }
try { try {
$agencies = $db->get_view("app", "all", null, true)->rows; $agencies = $db->get_view("app", "all", null, true)->rows;
//print_r($agencies); //print_r($agencies);
$first = true; $first = true;
if ($format == "json") { if ($format == "json") {
echo '"data" : ['.PHP_EOL; echo '"data" : [' . PHP_EOL;
   
} }
foreach ($agencies as $agency) { foreach ($agencies as $agency) {
   
if (isset($agency->value->statistics->employees)) { if (isset($agency->value->statistics->employees)) {
$row = Array(); $row = Array();
$agencyEmployeesArray = object_to_array($agency->value->statistics->employees); $agencyEmployeesArray = object_to_array($agency->value->statistics->employees);
foreach ($headers as $i => $fieldName) { foreach ($headers as $i => $fieldName) {
  if ($format == "csv") {
  if (isset($agencyEmployeesArray[$fieldName])) {
  $row[] = $agencyEmployeesArray[$fieldName]["value"] ;
  } else if ($i == 0) {
  $row[] = $agency->value->name;
  } else {
  $row[] = 0;
  }
  } else if ($format == "json") {
if (isset($agencyEmployeesArray[$fieldName])) { if (isset($agencyEmployeesArray[$fieldName])) {
$row[] = '['.$i.','.$agencyEmployeesArray[$fieldName]["value"].']'; $row[] = '[' . $i . ',' . $agencyEmployeesArray[$fieldName]["value"] . ']';
} else { } else {
$row[] = '['.$i.',0]'; $row[] = '[' . $i . ',0]';
} }
  }
} }
if ($format == "csv") { if ($format == "csv") {
fputcsv($fp, array_values($row)); fputcsv($fp, array_values($row));
} else if ($format == "json") { } else if ($format == "json") {
if (!$first) echo ","; if (!$first)
echo '{"data" : [' . implode(",", array_values($row)) . '], "label": "'.$agency->value->name.'", "lines" : { "show" : true }, "points" : { "show" : true }}'.PHP_EOL; echo ",";
  echo '{"data" : [' . implode(",", array_values($row)) . '], "label": "' . $agency->value->name . '", "lines" : { "show" : true }, "points" : { "show" : true }}' . PHP_EOL;
$first = false; $first = false;
} }
} }
} }
if ($format == "json") { if ($format == "json") {
echo '] echo ']
}'.PHP_EOL; }' . PHP_EOL;
   
} }
} catch (SetteeRestClientException $e) { } catch (SetteeRestClientException $e) {
setteErrorHandler($e); setteErrorHandler($e);
} }
   
die; die;
} }
?> ?>
   
<?php <?php
   
require_once '../include/common.inc.php'; require_once '../include/common.inc.php';
   
$db = $server->get_db('disclosr-agencies'); $db = $server->get_db('disclosr-agencies');
$rows = $db->get_view("app", "byName")->rows; $rows = $db->get_view("app", "byName")->rows;
$nametoid = Array(); $nametoid = Array();
$sums = Array(); $sums = Array();
foreach ($rows as $row) { foreach ($rows as $row) {
$nametoid[trim($row->key)] = $row->value; $nametoid[trim($row->key)] = $row->value;
} }
$employeeCSVs = Array("2002-2003" => "0203apsemployees.csv", $employeeCSVs = Array("2002-2003" => "0203apsemployees.csv",
"2003-2004" => "0304apsemployees.csv", "2003-2004" => "0304apsemployees.csv",
"2004-2005" => "0405apsemployees.csv", "2004-2005" => "0405apsemployees.csv",
"2005-2006" => "0506apsemployees.csv", "2005-2006" => "0506apsemployees.csv",
"2006-2007" => "0607apsemployees.csv", "2006-2007" => "0607apsemployees.csv",
"2007-2008" => "0708apsemployees.csv", "2007-2008" => "0708apsemployees.csv",
"2008-2009" => "0809apsemployees.csv", "2008-2009" => "0809apsemployees.csv",
"2009-2010" => "0910apsemployees.csv", "2009-2010" => "0910apsemployees.csv",
"2010-2011" => "1011apsemployees.csv" "2010-2011" => "1011apsemployees.csv"
); );
foreach ($employeeCSVs as $timePeriod => $employeeCSV) { foreach ($employeeCSVs as $timePeriod => $employeeCSV) {
echo $employeeCSV . "<br>" . PHP_EOL; echo $employeeCSV . "<br>" . PHP_EOL;
$row = 1; $row = 1;
if (($handle = fopen($employeeCSV, "r")) !== FALSE) { if (($handle = fopen($employeeCSV, "r")) !== FALSE) {
while (($data = fgetcsv($handle, 1000, ",")) !== FALSE) { while (($data = fgetcsv($handle, 1000, ",")) !== FALSE) {
//print_r($data); //print_r($data);
$name = trim($data[0]); $name = trim($data[0]);
if (isset($nametoid[$name])) { if (isset($nametoid[$name])) {
$id = $nametoid[$name]; $id = $nametoid[$name];
//echo $id . "<br>" . PHP_EOL; //echo $id . "<br>" . PHP_EOL;
@$sums[$id][$timePeriod] += $data[1]; @$sums[$id][$timePeriod] += $data[1];
} else { } else {
echo "<br>ERROR NAME MISSING FROM ID LIST<br><bR>" . PHP_EOL; echo "<br>ERROR NAME MISSING FROM ID LIST<br><bR>" . PHP_EOL;
   
die(); die();
} }
} }
fclose($handle); fclose($handle);
} }
} }
foreach ($sums as $id => $sum) { foreach ($sums as $id => $sum) {
echo $id . "<br>" . PHP_EOL; echo $id . "<br>" . PHP_EOL;
$doc = $db->get($id); $doc = $db->get($id);
echo $doc->name . "<br>" . PHP_EOL; echo $doc->name . "<br>" . PHP_EOL;
// print_r($doc); // print_r($doc);
$changed = false; $changed = false;
if (!isset($doc->statistics)) { if (!isset($doc->statistics)) {
$changed = true; $changed = true;
$doc->statistics = Array(); $doc->statistics = new stdClass();
  }
  if (!isset($doc->statistics->employees)) {
  $changed = true;
  $doc->statistics->employees = new stdClass();
} }
foreach ($sum as $timePeriod => $value) { foreach ($sum as $timePeriod => $value) {
if (!isset($doc->statistics->employees->$timePeriod->value) if (!isset($doc->statistics->employees->$timePeriod->value)
|| $doc->statistics->employees->$timePeriod->value != $value) { || $doc->statistics->employees->$timePeriod->value != $value) {
$changed = true; $changed = true;
$doc->statistics["employees"][$timePeriod] = Array("value" => $value, "source" => "http://apsc.gov.au/stateoftheservice/"); $doc->statistics->employees->$timePeriod = Array("value" => $value, "source" => "http://apsc.gov.au/stateoftheservice/");
} }
} }
if ($changed) { if ($changed) {
$db->save($doc); $db->save($doc);
} else { } else {
echo "not changed" . "<br>" . PHP_EOL; echo "not changed" . "<br>" . PHP_EOL;
} }
} }
// employees: timeperiod, source = apsc state of service, value // employees: timeperiod, source = apsc state of service, value
?> ?>
   
  <?php
 
  require_once '../include/common.inc.php';
  require($basePath . 'lib/phpquery/phpQuery/phpQuery.php');
  $db = $server->get_db('disclosr-agencies');
  $rows = $db->get_view("app", "byName")->rows;
  $nametoid = Array();
  $sums = Array();
  $functions = Array();
  foreach ($rows as $row) {
  $nametoid[trim($row->key)] = $row->value;
  }
 
 
  $request = Requests::get("http://www.apsc.gov.au/publications-and-media/parliamentary/state-of-the-service/new-sosr/appendix-2-aps-agencies");
  $doc = phpQuery::newDocumentHTML($request->body);
  phpQuery::selectDocument($doc);
  foreach (pq('tr')->elements as $tr) {
  //echo $tr->nodeValue.PHP_EOL;
  $agency = "";
  $employees = "";
  $function = "";
  $i = 0;
  foreach ($tr->childNodes as $td) {
  //echo $td->nodeValue." $i <br>";
  if ($i == 0)
  $agency = $td->nodeValue;
  if ($i == 2) {
  $employees = trim(str_replace(",", "", $td->nodeValue));
  }
  if ($i == 4) {
  $function = $td->nodeValue;
  }
  $i++;
  }
  if ($agency != "" && $employees != "" && $function != "") {
  $name = trim(str_replace('2','',$agency));
  //echo "$name<br><bR>" . PHP_EOL;
  if (isset($nametoid[$name])) {
  $id = $nametoid[$name];
  //echo $id . "<br>" . PHP_EOL;
  @$sums[$id]["2011-2012"] += $employees;
  $functions[$id] = $function;
  } else if ($agency != "Agency"){
  echo "<br>ERROR NAME '$agency' MISSING FROM ID LIST<br><bR>" . PHP_EOL;
 
  die();
  }
  } else {
  echo "skipped $agency";
  }
  }
  //print_r($sums);
  foreach ($sums as $id => $sum) {
  echo $id . "<br>" . PHP_EOL;
  $doc = $db->get($id);
  echo $doc->name . "<br>" . PHP_EOL;
  // print_r($doc);
  $changed = false;
  if (!isset($doc->statistics)) {
  $changed = true;
  $doc->statistics = new stdClass();
  }
  if (!isset($doc->statistics->employees)) {
  $changed = true;
  $doc->statistics->employees = new stdClass();
  }
  foreach ($sum as $timePeriod => $value) {
  if (!isset($doc->statistics->employees->$timePeriod->value)
  || $doc->statistics->employees->$timePeriod->value != $value) {
  $changed = true;
  $doc->statistics->employees->$timePeriod = Array("value" => $value, "source" => "http://apsc.gov.au/stateoftheservice/");
  $doc->employees = $value;
  $doc->functionClassification = $functions[$id];
  }
  }