projects / disclosr.git / commitdiff
search:
summary | shortlog | log | commit | commitdiff | tree
unified | plain
Testing facebook imports
Testing facebook imports


Former-commit-id: 26f7cc47bbd4c30bb9dab99a095d0363e52c3766

file:a/.gitmodules -> file:b/.gitmodules
[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"] [submodule "couchdb/settee"]
path = couchdb/settee path = couchdb/settee
url = https://github.com/inadarei/settee.git url = https://github.com/inadarei/settee.git
[submodule "lib/springy"] [submodule "lib/springy"]
path = lib/springy path = lib/springy
url = https://github.com/dhotson/springy.git url = https://github.com/dhotson/springy.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 "javascripts/flotr2"] [submodule "javascripts/flotr2"]
path = javascripts/flotr2 path = javascripts/flotr2
url = https://github.com/HumbleSoftware/Flotr2.git url = https://github.com/HumbleSoftware/Flotr2.git
  [submodule "lib/phpquery"]
  path = lib/phpquery
  url = https://github.com/TobiaszCudnik/phpquery.git
   
file:b/admin/importAustraliaGovAuGov2.php (new)
  <?php
 
  require_once '../include/common.inc.php';
 
  $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;
  }
 
  function extractCSVAccounts($url, $nameField, $accountField, $filter) {
  global $accounts, $nametoid;
  $request = Requests::get($url);
  $Data = str_getcsv($request->body, "\n"); //parse the rows
  $headers = Array();
  foreach ($Data as $num => $line) {
  $Row = str_getcsv($line, ",");
  if ($num == 0) {
 
  } else if ($num == 1) {
  $headers = $Row;
  //print_r($headers);
  } else {
  if (isset($Row[array_search($nameField, $headers)])) {
  $agencyName = $Row[array_search($nameField, $headers)];
  if (!$filter || $Row[array_search("State", $headers)] == "NAT") {
  if (!in_array(trim($agencyName), array_keys($nametoid))) {
  echo "$agencyName missing" . PHP_EOL;
  } else {
  // echo $Row[array_search($nameField, $headers)] . PHP_EOL;
  }
  }
  } else {
  //echo "error finding agency" . $line . PHP_EOL;
  }
  }
  }
  }
 
  // http://agimo.govspace.gov.au/page/gov2register/
  // twitter
  //extractCSVAccounts("https://docs.google.com/spreadsheet/pub?key=0Ap1exl80wB8OdHNKVmQ5RVlvQWpibDAxNHkzcU1nV2c&single=true&gid=0&output=csv", "Agency/Body/Event", "", true);
  // RSS
  // https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGJxandJREhLSGlWWUZfZ2xKOTNHZ0E&output=csv
  // facebook
  extractCSVAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGtjcW9vOXdyZ3pOV21vQU51VmhzQnc&single=true&gid=0&output=csv","Agency","Name");
 
  /*
  * http://australia.gov.au/news-and-media/media-release-rss-feeds
  * http://australia.gov.au/news-and-media/social-media/blogs
  * http://australia.gov.au/news-and-media/social-media/twitter
  * http://australia.gov.au/news-and-media/social-media/facebook
  * http://australia.gov.au/news-and-media/social-media/youtube
  * http://australia.gov.au/news-and-media/social-media/flickr
  * http://australia.gov.au/news-and-media/social-media/apps http://www.harmony.gov.au/get-involved/app-downloads.htm http://www.em.gov.au/Resources/Pages/Before-the-Storm-phone-game.aspx
  * http://australia.gov.au/news-and-media/social-media/podcasts
  */
  ?>
 
file:a/admin/importGov2RegisterRSSFacebookTwitter.php -> file:b/admin/importGov2RegisterRSSFacebookTwitter.php
<?php <?php
   
require_once '../include/common.inc.php'; require_once '../include/common.inc.php';
try { require($basePath.'lib/phpquery/phpQuery/phpQuery.php');
$server->create_db('disclosr-agencies');  
} catch (SetteeRestClientException $e) { $db = $server->get_db('disclosr-agencies');
setteErrorHandler($e); $rows = $db->get_view("app", "byName")->rows;
  $nametoid = Array();
  $accounts = Array();
  foreach ($rows as $row) {
  $nametoid[trim($row->key)] = $row->value;
} }
$db = $server->get_db('disclosr-agencies'); function extractHTMLAccounts($url, $accountType) {
createAgencyDesignDoc(); global $accounts, $nametoid;
  $request = Requests::get($url);
  $doc = phpQuery::newDocumentHTML($request->body);
  phpQuery::selectDocument($doc);
  foreach (pq('tr')->elements as $tr) {
  //echo $tr->nodeValue.PHP_EOL;
  $agency = "";
  $url = "";
  foreach ($tr->childNodes as $td) {
  $class = $td->getAttribute("class");
  //echo "cccc $class ".$td->nodeValue.PHP_EOL;
  if ($class == "s11" || $class == "s10" || $class == "s7") {
  $agency = $td->nodeValue;
  } else if ($class == "s6" || $class == "s9"){
  $url = $td->nodeValue;
  foreach($td->childNodes as $a) {
  $href = $a->getAttribute("href");
  if ($href != "") {
  $url = $href;
  }
  }
  }
  }
  if ($agency != "" && $url != "") {
  if (!in_array(trim($agency), array_keys($nametoid))) {
  echo trim($agency)." missing" . PHP_EOL;
  } else {
  // echo $agency." = ".$url.PHP_EOL;
  $accounts[$nametoid[trim($agency)]][$accountType][] = $url;
  }
   
  }
  }
   
  }
   
// twitter https://docs.google.com/spreadsheet/fm?id=tsJVd9EYoAjbl014y3qMgWg.03918275400592898296.8568379511161083736&hl=en&fmcmd=5&gid=0 function extractCSVAccounts($url, $accountType, $nameField, $accountField, $filter) {
// RSS https://docs.google.com/spreadsheet/fm?id=tbqjwIDHKHiVYF_glJ93GgA.03918275400592898296.8789688748524615194&authkey=CJDP-uQG&hl=en_GB&fmcmd=5&gid=0 global $accounts, $nametoid;
// facebook https://docs.google.com/spreadsheet/fm?id=tkcqoo9wrgzNWmoANuVhsBw.03918275400592898296.3040387705062056060&authkey=CKzl7r0I&hl=en_GB&fmcmd=5&gid=0 $request = Requests::get($url);
  $Data = str_getcsv($request->body, "\n"); //parse the rows
  $headers = Array();
  foreach ($Data as $num => $line) {
  $Row = str_getcsv($line, ",",'"');
  if ($num == 0) {
   
  } else if ($num == 1) {
  $headers = $Row;
  //print_r($headers);
  } else {
  if (isset($Row[array_search($nameField, $headers)])) {
  $agencyName = $Row[array_search($nameField, $headers)];
  if (!$filter || $Row[array_search("State", $headers)] == "NAT") {
  if (!in_array(trim($agencyName), array_keys($nametoid))) {
  echo trim($agencyName)." missing" . PHP_EOL;
  } else {
  // echo $Row[array_search($nameField, $headers)] . PHP_EOL;
  $accounts[$nametoid[trim($agencyName)]][$accountType][] = $Row[array_search($accountField, $headers)];
  }
  }
  } else {
  //echo "error finding agency" . $line . PHP_EOL;
  }
  }
  }
  }
   
/* // http://agimo.govspace.gov.au/page/gov2register/
* http://australia.gov.au/news-and-media/media-release-rss-feeds // twitter
* http://australia.gov.au/news-and-media/social-media/blogs extractCSVAccounts("https://docs.google.com/spreadsheet/pub?key=0Ap1exl80wB8OdHNKVmQ5RVlvQWpibDAxNHkzcU1nV2c&single=true&gid=0&output=csv", "Twitter", "Agency/Body/Event", "", true);
* http://australia.gov.au/news-and-media/social-media/twitter // RSS
* http://australia.gov.au/news-and-media/social-media/facebook extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGJxandJREhLSGlWWUZfZ2xKOTNHZ0E&output=html", "RSS");
* http://australia.gov.au/news-and-media/social-media/youtube // facebook
* http://australia.gov.au/news-and-media/social-media/flickr extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGtjcW9vOXdyZ3pOV21vQU51VmhzQnc&single=true&gid=0&output=html", "Facebook");
* http://australia.gov.au/news-and-media/social-media/apps http://www.harmony.gov.au/get-involved/app-downloads.htm http://www.em.gov.au/Resources/Pages/Before-the-Storm-phone-game.aspx  
* http://australia.gov.au/news-and-media/social-media/podcasts  
*/  
?> ?>
   
file:a/getAgency.php -> file:b/getAgency.php
<
<?php <?php
   
include_once('include/common.inc.php'); include_once('include/common.inc.php');
include_header(); include_header();
   
function displayValue($key, $value, $mode) { function displayValue($key, $value, $mode) {
global $db, $schemas; global $db, $schemas;
if ($mode == "view") { if ($mode == "view") {
   
echo "<tr>"; echo "<tr>";
   
echo "<td>" . $schemas['agency']["properties"][$key]['x-title'] . "<br><small>" . $schemas['agency']["properties"][$key]['description'] . "</small></td><td>"; echo "<td>" . $schemas['agency']["properties"][$key]['x-title'] . "<br><small>" . $schemas['agency']["properties"][$key]['description'] . "</small></td><td>";
if (is_array($value)) { if (is_array($value)) {
echo "<ol>"; echo "<ol>";
foreach ($value as $subkey => $subvalue) { foreach ($value as $subkey => $subvalue) {
if (isset($schemas['agency']["properties"][$key]['x-property'])) { if (isset($schemas['agency']["properties"][$key]['x-property'])) {
echo '<li property="' . $schemas['agency']["properties"][$key]['x-property'] . '">'; echo '<li property="' . $schemas['agency']["properties"][$key]['x-property'] . '">';
} else { } else {
echo "<li>"; echo "<li>";
} }
echo "$subvalue</li>"; echo "$subvalue</li>";
} }
echo "</ol></td></tr>"; echo "</ol></td></tr>";
} else { } else {
if (isset($schemas['agency']["properties"][$key]['x-property'])) { if (isset($schemas['agency']["properties"][$key]['x-property'])) {
echo '<span property="' . $schemas['agency']["properties"][$key]['x-property'] . '">'; echo '<span property="' . $schemas['agency']["properties"][$key]['x-property'] . '">';
} else { } else {
echo "<span>"; echo "<span>";
} }
if ((strpos($key, "URL") > 0 || $key == 'website') && $value != "") { if ((strpos($key, "URL") > 0 || $key == 'website') && $value != "") {
echo "<a href='$value'>view</a></span>"; echo "<a href='$value'>view</a></span>";
} else { } else {
echo "$value</span>"; echo "$value</span>";
} }
} }
echo "</td></tr>"; echo "</td></tr>";
} }
if ($mode == "edit") { if ($mode == "edit") {
if (is_array($value)) { if (is_array($value)) {
echo '<div class="row"> echo '<div class="row">
<div class="seven columns"> <div class="seven columns">
<fieldset> <fieldset>
<h5>' . $key . '</h5>'; <h5>' . $key . '</h5>';
foreach ($value as $subkey => $subvalue) { foreach ($value as $subkey => $subvalue) {
echo "<label>$subkey</label><input class='input-text' type='text' id='$key$subkey' name='$key" . '[' . $subkey . "]' value='$subvalue'/></tr>"; echo "<label>$subkey</label><input class='input-text' type='text' id='$key$subkey' name='$key" . '[' . $subkey . "]' value='$subvalue'/></tr>";
} }
echo "</fieldset> echo "</fieldset>
</div> </div>
</div>"; </div>";
} else { } else {
if (strpos($key, "_") === 0) { if (strpos($key, "_") === 0) {
echo"<input type='hidden' id='$key' name='$key' value='$value'/>"; echo"<input type='hidden' id='$key' name='$key' value='$value'/>";
} else if ($key == "parentOrg") { } else if ($key == "parentOrg") {
echo "<label for='$key'>$key</label><select id='$key' name='$key'><option value=''> Select... </option>"; echo "<label for='$key'>$key</label><select id='$key' name='$key'><option value=''> Select... </option>";
$rows = $db->get_view("app", "byDeptStateName")->rows; $rows = $db->get_view("app", "byDeptStateName")->rows;
//print_r($rows); //print_r($rows);
foreach ($rows as $row) { foreach ($rows as $row) {
echo "<option value='{$row->value}'" . (($row->value == $value) ? "SELECTED" : "") . " >" . str_replace("Department of ", "", $row->key) . "</option>"; echo "<option value='{$row->value}'" . (($row->value == $value) ? "SELECTED" : "") . " >" . str_replace("Department of ", "", $row->key) . "</option>";
} }
echo" </select>"; echo" </select>";
} else { } else {
echo "<label>$key</label><input class='input-text' type='text' id='$key' name='$key' value='$value'/>"; echo "<label>$key</label><input class='input-text' type='text' id='$key' name='$key' value='$value'/>";
if ((strpos($key, "URL") > 0 || $key == 'website') && $value != "") { if ((strpos($key, "URL") > 0 || $key == 'website') && $value != "") {
echo "<a href='$value'>view</a>"; echo "<a href='$value'>view</a>";
} }
if ($key == 'abn') { if ($key == 'abn') {
echo "<a href='http://www.abr.business.gov.au/SearchByAbn.aspx?SearchText=$value'>view abn</a>"; echo "<a href='http://www.abr.business.gov.au/SearchByAbn.aspx?SearchText=$value'>view abn</a>";
} }
} }
} }
} }
// //
} }
   
function addDefaultFields($row) { function addDefaultFields($row) {
global $schemas; global $schemas;
$defaultFields = array_keys($schemas['agency']['properties']); $defaultFields = array_keys($schemas['agency']['properties']);
foreach ($defaultFields as $defaultField) { foreach ($defaultFields as $defaultField) {
if (!isset($row[$defaultField])) { if (!isset($row[$defaultField])) {
if ($schemas['agency']['properties'][$defaultField]['type'] == "string") { if ($schemas['agency']['properties'][$defaultField]['type'] == "string") {
  $row[$defaultField] = "";
$row[$defaultField] = "";  
   
} }
if ($schemas['agency']['properties'][$defaultField]['type'] == "array") { if ($schemas['agency']['properties'][$defaultField]['type'] == "array") {
   
$row[$defaultField] = Array(""); $row[$defaultField] = Array("");
  }
  } else if ($schemas['agency']['properties'][$defaultField]['type'] == "array") {
  if (is_array($row[$defaultField])) {
  $row[$defaultField][] = "";
  } else {
  $value = $row[$defaultField];
  $row[$defaultField] = Array($value);
} }
} }
} }
return $row; return $row;
} }
   
$db = $server->get_db('disclosr-agencies'); $db = $server->get_db('disclosr-agencies');
   
if (isset($_REQUEST['id'])) { if (isset($_REQUEST['id'])) {
//get an agency record as json/html, search by name/abn/id //get an agency record as json/html, search by name/abn/id
// by name = startkey="Ham"&endkey="Ham\ufff0" // by name = startkey="Ham"&endkey="Ham\ufff0"
// edit? // edit?
   
$row = $db->get($_REQUEST['id']); $row = $db->get($_REQUEST['id']);
//print_r($row); //print_r($row);
if (sizeof($_POST) > 0) { if (sizeof($_POST) > 0) {
//print_r($_POST); //print_r($_POST);
foreach ($_POST as $postkey => $postvalue) { foreach ($_POST as $postkey => $postvalue) {
if ($postvalue == "") { if ($postvalue == "") {
unset($_POST[$postkey]); unset($_POST[$postkey]);
} }
if (is_array($postvalue) && count($postvalue) == 1 && $postvalue[0] == "") { if (is_array($postvalue)) {
unset($_POST[$postkey]); if (count($postvalue) == 1 && $postvalue[0] == "") {
  unset($_POST[$postkey]);
  } else {
  foreach ($_POST[$postkey] as $key => &$value) {
  if ($value == "") {
  unset($_POST[$postkey][$key]);
  }
  }
  }
} }
} }
if (isset($_POST['_id']) && $db->get_rev($_POST['_id']) == $_POST['_rev']) { if (isset($_POST['_id']) && $db->get_rev($_POST['_id']) == $_POST['_rev']) {
echo "Edited version was latest version, continue saving"; echo "Edited version was latest version, continue saving";
$newdoc = $_POST; $newdoc = $_POST;
$newdoc['metadata']['lastModified'] = time(); $newdoc['metadata']['lastModified'] = time();
$row = $db->save($newdoc); $row = $db->save($newdoc);
} else { } else {
echo "ALERT doc revised by someone else while editing. Document not saved."; echo "ALERT doc revised by someone else while editing. Document not saved.";
} }
} }
   
$mode = "edit"; $mode = "edit";
if ($mode == "edit") { if ($mode == "edit") {
$row = addDefaultFields(object_to_array($row)); $row = addDefaultFields(object_to_array($row));
} else { } else {
$row = object_to_array($row); $row = object_to_array($row);
} }