beginnings of document scrapers mark 2
Former-commit-id: fc862f3bada775436406e0ee8adcd4e7d04ac8f6
--- a/alaveteli/exportAgencies.csv.php
+++ b/alaveteli/exportAgencies.csv.php
@@ -52,18 +52,23 @@
$row["request_email"] = (isset($agency->value->foiEmail) ? $agency->value->foiEmail : "");
$row["short_name"] = (isset($agency->value->shortName) ? $agency->value->shortName : "");
$row["notes"] = (isset($agency->value->description) ? $agency->value->description : "");
-
+
$otherBodies = Array();
if (isset($agency->value->foiBodies)) {
$otherBodies = array_merge($otherBodies, $agency->value->foiBodies);
}
if (isset($agency->value->positions)) {
- $otherBodies = array_merge($otherBodies, $agency->value->positions);
+ $positions = Array();
+ foreach ($agency->value->positions as $position) {
+ $positions[] = "Office of the ".$position;
+ }
+ $otherBodies = array_merge($otherBodies, $positions);
}
+ sort($otherBodies);
if (count($otherBodies) > 0) {
- $row["notes"] .= "<br/> This department also responds to requests for information held by ".implode(",",$otherBodies);
+ $row["notes"] .= "<br/> This department also responds to requests for information held by " . implode(", ", $otherBodies);
}
-
+
$row["publication_scheme"] = (isset($agency->value->infoPublicationSchemeURL) ? $agency->value->infoPublicationSchemeURL : "");
$row["home_page"] = (isset($agency->value->website) ? $agency->value->website : "");
if ($agency->value->orgType == "FMA-DepartmentOfState") {
@@ -74,9 +79,6 @@
$row["tag_string"] .= " " . $agency->value->orgType;
$row["tag_string"] .= " federal";
fputcsv($fp, array_values($row));
-
-
-
}
}
} catch (SetteeRestClientException $e) {
--- /dev/null
+++ b/documents/google676a414ad086cefb.html
@@ -1,1 +1,2 @@
+google-site-verification: google676a414ad086cefb.html
--- /dev/null
+++ b/documents/index.php
@@ -1,1 +1,15 @@
+australian disclosure logs
+are you looking for more information about:
+contracts
+gov orgs
+lobbyists
+
+1/1/11 title (Dept dfggdfgdf)
+description:
+source link:
+documents:
+ #1 title link
+
+
+rss feed here
--- /dev/null
+++ b/documents/rss.xml.php
@@ -1,1 +1,30 @@
+<?php
+// Agency X updated Y, new files, diff of plain text/link text,
+// feed for just one agency or all
+// This is a minimum example of using the Universal Feed Generator Class
+include("lib/FeedWriter.php");
+//Creating an instance of FeedWriter class.
+$TestFeed = new FeedWriter(RSS2);
+//Setting the channel elements
+//Use wrapper functions for common channelelements
+$TestFeed->setTitle('Last Modified - All');
+$TestFeed->setLink('http://disclosr.lambdacomplex.org/rss.xml.php');
+$TestFeed->setDescription('This is test of creating a RSS 2.0 feed Universal Feed Writer');
+//Retriving informations from database
+$rows = $db->get_view("app", "byLastModified")->rows;
+//print_r($rows);
+foreach ($rows as $row) {
+ //Create an empty FeedItem
+ $newItem = $TestFeed->createNewItem();
+ //Add elements to the feed item
+ $newItem->setTitle($row['name']);
+ $newItem->setLink($row['id']);
+ $newItem->setDate(date("c", $row['metadata']['lastModified']));
+ $newItem->setDescription($row['name']);
+ //Now add the feed item
+ $TestFeed->addItem($newItem);
+}
+//OK. Everything is done. Now genarate the feed.
+$TestFeed->genarateFeed();
+?>
--- /dev/null
+++ b/documents/runScrapers.php
--- /dev/null
+++ b/documents/scrapers/3cd40b1240e987cbcd3f0e67054ce259.py
@@ -1,1 +1,1 @@
-
+http://www.apvma.gov.au/rss/disclosure.rss
--- /dev/null
+++ b/documents/scrapers/8c9421f852c441910bf1d93a57b31d64.py
@@ -1,1 +1,1 @@
-
+http://www.daff.gov.au/about/foi/ips/disclosure_log
--- /dev/null
+++ b/documents/scrapers/be9996f0ac58f71f23d074e82d44ead3.py
@@ -1,1 +1,2 @@
+http://foi.deewr.gov.au/disclosure-log/rss
--- /dev/null
+++ b/documents/scrapers/e2a845e55bc9986e6c75c5ad2c508b8d.py
@@ -1,1 +1,1 @@
-
+www.finance.gov.au/foi/disclosure-log/foi-rss.xml
--- /dev/null
+++ b/documents/scrapers/rtk.py
@@ -1,1 +1,1 @@
-
+http://www.righttoknow.org.au/feed/search/%20(latest_status:successful%20OR%20latest_status:partially_successful)
--- /dev/null
+++ b/documents/search.php
@@ -1,1 +1,25 @@
+<?php
+include_once('include/common.inc.php');
+include_header('Search');
+?>
+<div class="foundation-header">
+ <h1><a href="search.php">Search</a></h1>
+</div>
+<form>
+ <input type="text" name="q" value="<?php if (isset($_REQUEST['q']))echo $_REQUEST['q'];?>"/>
+ <input type="submit"/>
+</form>
+<?php
+if (isset($_REQUEST['q'])) {
+ $request = Requests::get($serverAddr."disclosr-documents/_fti/_design/lucene/by_all?include_docs=true&q=".$_REQUEST['q']);
+ $results = json_decode($request->body);
+ $db = $server->get_db('disclosr-documents');
+ foreach ($results->rows as $result) {
+ //print_r($result);
+ //$row = $db->get($result->id);
+ echo $result->doc->_id." ".$result->doc->url."<br>".PHP_EOL;
+ }
+}
+include_footer();
+?>
--- a/getAgency.php
+++ b/getAgency.php
@@ -9,7 +9,11 @@
if (strpos($key, "_") === 0 || $key== "metadata") return;
echo "<tr>";
- echo "<td>" . $schemas['agency']["properties"][$key]['x-title'] . "<br><small>" . $schemas['agency']["properties"][$key]['description'] . "</small></td><td>";
+ echo "<td>";
+ if (isset($schemas['agency']["properties"][$key])) {
+ echo $schemas['agency']["properties"][$key]['x-title'] . "<br><small>" . $schemas['agency']["properties"][$key]['description']."</small>";
+ }
+ echo "</td><td>";
if (is_array($value)) {
echo "<ol>";
foreach ($value as $subkey => $subvalue) {
--- a/include/common.inc.php
+++ b/include/common.inc.php
@@ -14,10 +14,10 @@
require_once $basePath.'lib/Requests/library/Requests.php';
Requests::register_autoloader();
-require $basePath."/lib/amon-php/amon.php";
-Amon::config(array('address'=> 'http://127.0.0.1:2465',
+require $basePath."lib/amon-php/amon.php";
+Amon::config(array('address'=> 'http://127.0.0.1:2464',
'protocol' => 'http',
- 'secret_key' => "g99127n3lkzigg8ob2rllth97d1pb4sj"));
+ 'secret_key' => "I2LJ6dOMmlnXgVAkTPFXd5M3ejkga8Gd2FbBt6iqZdw"));
Amon::setup_exception_handler();
# Convert a stdClass to an Array. http://www.php.net/manual/en/language.types.object.php#102735
--- /dev/null
+++ b/js/bubbletree
--- /dev/null
+++ b/js/flotr2
--- /dev/null
+++ b/js/sigma
--- a/rss.xml.php
+++ /dev/null
@@ -1,30 +1,1 @@
-<?php
-// Agency X updated Y, new files, diff of plain text/link text,
-// feed for just one agency or all
-// This is a minimum example of using the Universal Feed Generator Class
-include("lib/FeedWriter.php");
-//Creating an instance of FeedWriter class.
-$TestFeed = new FeedWriter(RSS2);
-//Setting the channel elements
-//Use wrapper functions for common channelelements
-$TestFeed->setTitle('Last Modified - All');
-$TestFeed->setLink('http://disclosr.lambdacomplex.org/rss.xml.php');
-$TestFeed->setDescription('This is test of creating a RSS 2.0 feed Universal Feed Writer');
-//Retriving informations from database
-$rows = $db->get_view("app", "byLastModified")->rows;
-//print_r($rows);
-foreach ($rows as $row) {
- //Create an empty FeedItem
- $newItem = $TestFeed->createNewItem();
- //Add elements to the feed item
- $newItem->setTitle($row['name']);
- $newItem->setLink($row['id']);
- $newItem->setDate(date("c", $row['metadata']['lastModified']));
- $newItem->setDescription($row['name']);
- //Now add the feed item
- $TestFeed->addItem($newItem);
-}
-//OK. Everything is done. Now genarate the feed.
-$TestFeed->genarateFeed();
-?>
--- a/search.php
+++ /dev/null
@@ -1,25 +1,1 @@
-<?php
-include_once('include/common.inc.php');
-include_header('Search');
-?>
-<div class="foundation-header">
- <h1><a href="search.php">Search</a></h1>
-</div>
-<form>
- <input type="text" name="q" value="<?php if (isset($_REQUEST['q']))echo $_REQUEST['q'];?>"/>
- <input type="submit"/>
-</form>
-<?php
-if (isset($_REQUEST['q'])) {
- $request = Requests::get($serverAddr."disclosr-documents/_fti/_design/lucene/by_all?include_docs=true&q=".$_REQUEST['q']);
- $results = json_decode($request->body);
- $db = $server->get_db('disclosr-documents');
- foreach ($results->rows as $result) {
- //print_r($result);
- //$row = $db->get($result->id);
- echo $result->doc->_id." ".$result->doc->url."<br>".PHP_EOL;
- }
-}
-include_footer();
-?>