projects / disclosr.git / commitdiff
search:
summary | shortlog | log | commit | commitdiff | tree
unified | plain
more scrapers
more scrapers


Former-commit-id: 012f6a5c1fa63f9ddaeb47cb13297668fa35b23e

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/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"]
  path = lib/FeedWriter
  url = https://github.com/mibe/FeedWriter
   
file:a/admin/refreshDesignDoc.php -> file:b/admin/refreshDesignDoc.php
<?php <?php
   
require_once '../include/common.inc.php'; require_once '../include/common.inc.php';
//function createFOIDocumentsDesignDoc() { //function createFOIDocumentsDesignDoc() {
   
$foidb = $server->get_db('disclosr-foidocuments'); $foidb = $server->get_db('disclosr-foidocuments');
$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->byDate->map = "function(doc) { emit(doc.date, doc); };"; $obj->views->byDate->map = "function(doc) { emit(doc.date, doc); };";
  $obj->views->byDate->reduce = "_count";
  $obj->views->byAgencyID->map = "function(doc) { emit(doc.agencyID, doc); };";
  $obj->views->byAgencyID->reduce = "_count";
   
// 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).
$foidb->save($obj, true); $foidb->save($obj, true);
   
   
function createDocumentsDesignDoc() { function createDocumentsDesignDoc() {
/* /*
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); };";
"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() {
$db = $server->get_db('disclosr-agencies'); $db = $server->get_db('disclosr-agencies');
$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);
} }
} }
for (name in doc.positions) { for (name in doc.positions) {
if (doc.positions[name] != "" && doc.positions[name] != doc.name) { if (doc.positions[name] != "" && doc.positions[name] != doc.name) {
emit(doc.positions[name], doc._id); emit(doc.positions[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) {