Add sitemap.xml.php
Add sitemap.xml.php


Former-commit-id: d88af9d576a0744d366a5881a34b742975d70a45

<?php <?php
   
date_default_timezone_set("Australia/Sydney"); date_default_timezone_set("Australia/Sydney");
   
$basePath = ""; $basePath = "";
if (strstr($_SERVER['PHP_SELF'], "alaveteli/") if (strstr($_SERVER['PHP_SELF'], "alaveteli/")
|| strstr($_SERVER['PHP_SELF'], "admin/") || strstr($_SERVER['PHP_SELF'], "admin/")
|| strstr($_SERVER['PHP_SELF'], "lib/") || strstr($_SERVER['PHP_SELF'], "lib/")
|| strstr($_SERVER['PHP_SELF'], "include/")) || strstr($_SERVER['PHP_SELF'], "include/"))
$basePath = "../"; $basePath = "../";
   
include_once ('couchdb.inc.php'); include_once ('couchdb.inc.php');
include_once ('template.inc.php'); include_once ('template.inc.php');
require_once $basePath.'lib/Requests/library/Requests.php'; require_once $basePath.'lib/Requests/library/Requests.php';
   
Requests::register_autoloader(); Requests::register_autoloader();
# Convert a stdClass to an Array. http://www.php.net/manual/en/language.types.object.php#102735 # Convert a stdClass to an Array. http://www.php.net/manual/en/language.types.object.php#102735
   
function object_to_array(stdClass $Class) { function object_to_array(stdClass $Class) {
# Typecast to (array) automatically converts stdClass -> array. # Typecast to (array) automatically converts stdClass -> array.
$Class = (array) $Class; $Class = (array) $Class;
   
# Iterate through the former properties looking for any stdClass properties. # Iterate through the former properties looking for any stdClass properties.
# Recursively apply (array). # Recursively apply (array).
foreach ($Class as $key => $value) { foreach ($Class as $key => $value) {
if (is_object($value) && get_class($value) === 'stdClass') { if (is_object($value) && get_class($value) === 'stdClass') {
$Class[$key] = object_to_array($value); $Class[$key] = object_to_array($value);
} }
} }
return $Class; return $Class;
} }
   
# Convert an Array to stdClass. http://www.php.net/manual/en/language.types.object.php#102735 # Convert an Array to stdClass. http://www.php.net/manual/en/language.types.object.php#102735
   
function array_to_object(array $array) { function array_to_object(array $array) {
# Iterate through our array looking for array values. # Iterate through our array looking for array values.
# If found recurvisely call itself. # If found recurvisely call itself.
foreach ($array as $key => $value) { foreach ($array as $key => $value) {
if (is_array($value)) { if (is_array($value)) {
$array[$key] = array_to_object($value); $array[$key] = array_to_object($value);
} }
} }
   
# Typecast to (object) will automatically convert array -> stdClass # Typecast to (object) will automatically convert array -> stdClass
return (object) $array; return (object) $array;
} }
   
function dept_to_portfolio($deptName) { function dept_to_portfolio($deptName) {
return trim(str_replace("Department of", "", str_replace("Department of the", "Department of", $deptName))); return trim(str_replace("Department of", "", str_replace("Department of the", "Department of", $deptName)));
} }
function phrase_to_tag ($phrase) { function phrase_to_tag ($phrase) {
return str_replace(" ","_",str_replace("'","",str_replace(",","",strtolower($phrase)))); return str_replace(" ","_",str_replace("'","",str_replace(",","",strtolower($phrase))));
} }
  function local_url() {
  return "http://" . $_SERVER['HTTP_HOST'] . rtrim(dirname($_SERVER['PHP_SELF']), '/\\') . "/";
  }
function GetDomain($url) function GetDomain($url)
{ {
$nowww = ereg_replace('www\.','',$url); $nowww = ereg_replace('www\.','',$url);
$domain = parse_url($nowww); $domain = parse_url($nowww);
if(!empty($domain["host"])) if(!empty($domain["host"]))
{ {
return $domain["host"]; return $domain["host"];
} else } else
{ {
return $domain["path"]; return $domain["path"];
} }
} }
   
<?php <?php
   
include $basePath . "schemas/schemas.inc.php"; include $basePath . "schemas/schemas.inc.php";
   
require ($basePath . 'couchdb/settee/src/settee.php'); require ($basePath . 'couchdb/settee/src/settee.php');
   
function createDocumentsDesignDoc() { function createDocumentsDesignDoc() {
/* "views": { /* "views": {
"web_server": { "web_server": {
"map": "function(doc) {\n emit(doc.web_server, 1);\n}", "map": "function(doc) {\n emit(doc.web_server, 1);\n}",
"reduce": "function (key, values, rereduce) {\n return sum(values);\n}" "reduce": "function (key, values, rereduce) {\n return sum(values);\n}"
}, },
"byAgency": { "byAgency": {
"map": "function(doc) {\n emit(doc.agencyID, 1);\n}", "map": "function(doc) {\n emit(doc.agencyID, 1);\n}",
"reduce": "function (key, values, rereduce) {\n return sum(values);\n}" "reduce": "function (key, values, rereduce) {\n return sum(values);\n}"
}, },
"byURL": { "byURL": {
"map": "function(doc) {\n emit(doc.url, doc);\n}" "map": "function(doc) {\n emit(doc.url, doc);\n}"
}, },
"agency": { "agency": {
"map": "function(doc) {\n emit(doc.agencyID, doc);\n}" "map": "function(doc) {\n emit(doc.agencyID, doc);\n}"
}, },
"byWebServer": { "byWebServer": {
"map": "function(doc) {\n emit(doc.web_server, doc);\n}" "map": "function(doc) {\n emit(doc.web_server, doc);\n}"
}, },
"getValidationRequired": { "getValidationRequired": {
"map": "function(doc) {\nif (doc.mime_type == \"text/html\" \n&& typeof(doc.validation) == \"undefined\") {\n emit(doc._id, doc._attachments);\n}\n}" "map": "function(doc) {\nif (doc.mime_type == \"text/html\" \n&& typeof(doc.validation) == \"undefined\") {\n emit(doc._id, doc._attachments);\n}\n}"
} }
} */ } */
} }
   
function createAgencyDesignDoc() { function createAgencyDesignDoc() {
global $db; global $db;
$obj = new stdClass(); $obj = new stdClass();
$obj->_id = "_design/" . urlencode("app"); $obj->_id = "_design/" . urlencode("app");
$obj->language = "javascript"; $obj->language = "javascript";
$obj->views->all->map = "function(doc) { emit(doc._id, doc); };"; $obj->views->all->map = "function(doc) { emit(doc._id, doc); };";
$obj->views->byABN->map = "function(doc) { emit(doc.abn, doc); };"; $obj->views->byABN->map = "function(doc) { emit(doc.abn, doc); };";
$obj->views->byCanonicalName->map = "function(doc) { $obj->views->byCanonicalName->map = "function(doc) {
if (doc.parentOrg || doc.orgType == 'FMA-DepartmentOfState') { if (doc.parentOrg || doc.orgType == 'FMA-DepartmentOfState') {
emit(doc.name, doc); emit(doc.name, doc);
} }
};"; };";
$obj->views->byDeptStateName->map = "function(doc) { $obj->views->byDeptStateName->map = "function(doc) {
if (doc.orgType == 'FMA-DepartmentOfState') { if (doc.orgType == 'FMA-DepartmentOfState') {
emit(doc.name, doc._id); emit(doc.name, doc._id);
} }
};"; };";
$obj->views->parentOrgs->map = "function(doc) { $obj->views->parentOrgs->map = "function(doc) {
if (doc.parentOrg) { if (doc.parentOrg) {
emit(doc._id, doc.parentOrg); emit(doc._id, doc.parentOrg);
} }
};"; };";
$obj->views->byName->map = 'function(doc) { $obj->views->byName->map = 'function(doc) {
if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") { if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") {
emit(doc.name, doc._id); emit(doc.name, doc._id);
if (typeof(doc.shortName) != "undefined" && doc.shortName != doc.name) { if (typeof(doc.shortName) != "undefined" && doc.shortName != doc.name) {
emit(doc.shortName, doc._id); emit(doc.shortName, doc._id);
} }
for (name in doc.otherNames) { for (name in doc.otherNames) {
if (doc.otherNames[name] != "" && doc.otherNames[name] != doc.name) { if (doc.otherNames[name] != "" && doc.otherNames[name] != doc.name) {
emit(doc.otherNames[name], doc._id); emit(doc.otherNames[name], doc._id);
} }
} }
for (name in doc.foiBodies) { for (name in doc.foiBodies) {
if (doc.foiBodies[name] != "" && doc.foiBodies[name] != doc.name) { if (doc.foiBodies[name] != "" && doc.foiBodies[name] != doc.name) {
emit(doc.foiBodies[name], doc._id); emit(doc.foiBodies[name], doc._id);
} }
} }
} }
};'; };';
   
$obj->views->foiEmails->map = "function(doc) { $obj->views->foiEmails->map = "function(doc) {
emit(doc._id, doc.foiEmail); emit(doc._id, doc.foiEmail);
};"; };";
   
$obj->views->byLastModified->map = "function(doc) { emit(doc.metadata.lastModified, doc); }"; $obj->views->byLastModified->map = "function(doc) { emit(doc.metadata.lastModified, doc); }";
$obj->views->getActive->map = 'function(doc) { if (doc.status == "active") { emit(doc._id, doc); } };'; $obj->views->getActive->map = 'function(doc) { if (doc.status == "active") { emit(doc._id, doc); } };';
$obj->views->getSuspended->map = 'function(doc) { if (doc.status == "suspended") { emit(doc._id, doc); } };'; $obj->views->getSuspended->map = 'function(doc) { if (doc.status == "suspended") { emit(doc._id, doc); } };';
$obj->views->getScrapeRequired->map = "function(doc) { $obj->views->getScrapeRequired->map = "function(doc) {
   
var lastScrape = Date.parse(doc.metadata.lastScraped); var lastScrape = Date.parse(doc.metadata.lastScraped);
   
var today = new Date(); var today = new Date();
   
if (!lastScrape || lastScrape.getTime() + 1000 != today.getTime()) { if (!lastScrape || lastScrape.getTime() + 1000 != today.getTime()) {
emit(doc._id, doc); emit(doc._id, doc);
} }
   
};"; };";
$obj->views->showNamesABNs->map = "function(doc) { emit(doc._id, {name: doc.name, abn: doc.abn}); };"; $obj->views->showNamesABNs->map = "function(doc) { emit(doc._id, {name: doc.name, abn: doc.abn}); };";
$obj->views->getConflicts->map = "function(doc) { $obj->views->getConflicts->map = "function(doc) {
if (doc._conflicts) { if (doc._conflicts) {
emit(null, [doc._rev].concat(doc._conflicts)); emit(null, [doc._rev].concat(doc._conflicts));
} }
}"; }";
// http://stackoverflow.com/questions/646628/javascript-startswith // http://stackoverflow.com/questions/646628/javascript-startswith
$obj->views->score->map = 'if(!String.prototype.startsWith){ $obj->views->score->map = 'if(!String.prototype.startsWith){
String.prototype.startsWith = function (str) { String.prototype.startsWith = function (str) {
return !this.indexOf(str); return !this.indexOf(str);
} }
} }
   
function(doc) { function(doc) {
count = 0; count = 0;
if (doc["status"] != "suspended") { if (doc["status"] != "suspended") {
for(var propName in doc) { for(var propName in doc) {
if(typeof(doc[propName]) != "undefined" && doc[propName] != "") { if(typeof(doc[propName]) != "undefined" && doc[propName] != "") {
count++; count++;
} }
} }
portfolio = doc.parentOrg; portfolio = doc.parentOrg;
if (doc.orgType == "FMA-DepartmentOfState") { if (doc.orgType == "FMA-DepartmentOfState") {
portfolio = doc._id; portfolio = doc._id;
} }
if (doc.orgType == "Court-Commonwealth" || doc.orgType == "FMA-DepartmentOfParliament") { if (doc.orgType == "Court-Commonwealth" || doc.orgType == "FMA-DepartmentOfParliament") {
portfolio = doc.orgType; portfolio = doc.orgType;
} }
emit(count+doc._id, {id:doc._id, name: doc.name, score:count, orgType: doc.orgType, portfolio:portfolio}); emit(count+doc._id, {id:doc._id, name: doc.name, score:count, orgType: doc.orgType, portfolio:portfolio});
} }
}'; }';
$obj->views->scoreHas->map = 'if(!String.prototype.startsWith){ $obj->views->scoreHas->map = 'if(!String.prototype.startsWith){
String.prototype.startsWith = function (str) { String.prototype.startsWith = function (str) {
return !this.indexOf(str); return !this.indexOf(str);
} }
} }
if(!String.prototype.endsWith){ if(!String.prototype.endsWith){
String.prototype.endsWith = function(suffix) { String.prototype.endsWith = function(suffix) {
    return this.indexOf(suffix, this.length - suffix.length) !== -1;     return this.indexOf(suffix, this.length - suffix.length) !== -1;
}; };
} }
function(doc) { function(doc) {
if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") { if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") {
for(var propName in doc) { for(var propName in doc) {
if(typeof(doc[propName]) != "undefined" && (propName.startsWith("has") || propName.endsWith("URL"))) { if(typeof(doc[propName]) != "undefined" && (propName.startsWith("has") || propName.endsWith("URL"))) {
emit(propName, 1); emit(propName, 1);
} }
} }
emit("total", 1); emit("total", 1);
} }
}'; }';
$obj->views->scoreHas->reduce = 'function (key, values, rereduce) { $obj->views->scoreHas->reduce = 'function (key, values, rereduce) {
return sum(values); return sum(values);
}'; }';
$obj->views->fieldNames->map = ' $obj->views->fieldNames->map = '
function(doc) { function(doc) {
for(var propName in doc) { for(var propName in doc) {
emit(propName, doc._id); emit(propName, doc._id);
} }
}'; }';
$obj->views->fieldNames->reduce = 'function (key, values, rereduce) { $obj->views->fieldNames->reduce = 'function (key, values, rereduce) {
return values.length; return values.length;
}'; }';
// allow safe updates (even if slightly slower due to extra: rev-detection check). // allow safe updates (even if slightly slower due to extra: rev-detection check).
return $db->save($obj, true); return $db->save($obj, true);
} }
   
if (php_uname('n') == "vanille") { if (php_uname('n') == "vanille") {
$serverAddr = 'http://192.168.178.21:5984/'; $serverAddr = 'http://192.168.178.21:5984/';
} else } else
if (php_uname('n') == "KYUUBEY") { if (php_uname('n') == "KYUUBEY") {
   
$serverAddr = 'http://127.0.0.1:5984/'; $serverAddr = 'http://192.168.1.148:5984/';
} else { } else {
$serverAddr = 'http://127.0.0.1:5984/'; $serverAddr = 'http://127.0.0.1:5984/';
} }
$server = new SetteeServer($serverAddr); $server = new SetteeServer($serverAddr);
   
function setteErrorHandler($e) { function setteErrorHandler($e) {
echo $e->getMessage() . "<br>" . PHP_EOL; echo $e->getMessage() . "<br>" . PHP_EOL;
} }
   
file:b/sitemap.xml.php (new)
  <?php
 
  include ('include/common.inc.php');
  $last_updated = date('Y-m-d', @filemtime('cbrfeed.zip'));
  header("Content-Type: text/xml");
  echo "<?xml version='1.0' encoding='UTF-8'?>";
  echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
  echo " <url><loc>" . local_url() . "index.php</loc><priority>1.0</priority></url>\n";
  foreach (scandir("./") as $file) {
  if (strpos($file, ".php") !== false && $file != "index.php" && $file != "sitemap.xml.php")
  echo " <url><loc>" . local_url() . "$file</loc><priority>0.3</priority></url>\n";
  }
  echo '</urlset>';
 
  $db = $server->get_db('disclosr-agencies');
  try {
  $rows = $db->get_view("app", "byCanonicalName")->rows;
  foreach ($rows as $row) {
  echo '<url><loc>' . local_url() . 'getAgency.php?id=' . $row->value->_id . "</loc><priority>0.6</priority></url>\n";
  }
  } catch (SetteeRestClientException $e) {
  setteErrorHandler($e);
  }
  ?>