pyquery scraper beginnings
Former-commit-id: 7f7aa89526b530d6444397489fd11eed7d17369f
--- a/couchdb/settee/src/classes/SetteeDatabase.class.php
+++ b/couchdb/settee/src/classes/SetteeDatabase.class.php
@@ -251,7 +251,7 @@
*
* @return void
*/
- function get_view($design_doc, $view_name, $key = null, $descending = false, $limit = false) {
+ function get_view($design_doc, $view_name, $key = null, $descending = false, $limit = false, $reduce=false) {
$id = "_design/" . urlencode($design_doc);
$view_name = urlencode($view_name);
$id .= "/_view/$view_name";
@@ -269,6 +269,11 @@
if ($descending) {
$data .= "&descending=true";
}
+ if ($reduce) {
+ $data .= "&reduce=true";
+ } else {
+ $data .= "&reduce=false";
+ }
if ($limit) {
$data .= "&limit=".$limit;
}
@@ -281,9 +286,11 @@
}
$full_uri = $this->dbname . "/" . $this->safe_urlencode($id);
+
$full_uri = str_replace("%253Fgroup%253D","?group=",$full_uri);
$full_uri = str_replace("%253Flimit%253D","?limit=",$full_uri);
$ret = $this->rest_client->http_get($full_uri, $data);
+ //$ret['decoded'] = str_replace("?k","&k",$ret['decoded']);
return $ret['decoded'];
}
--- a/documents/about.php
+++ b/documents/about.php
@@ -1,7 +1,7 @@
<?php
include('template.inc.php');
-include_header_documents("");
+include_header_documents("About");
include_once('../include/common.inc.php');
?>
<h1>About</h1>
--- /dev/null
+++ b/documents/agency.php
@@ -1,1 +1,41 @@
+<?php
+include('template.inc.php');
+include_once('../include/common.inc.php');
+$agenciesdb = $server->get_db('disclosr-agencies');
+$idtoname = Array();
+foreach ($agenciesdb->get_view("app", "byCanonicalName")->rows as $row) {
+ $idtoname[$row->id] = trim($row->value->name);
+}
+$foidocsdb = $server->get_db('disclosr-foidocuments');
+
+include_header_documents((isset($_REQUEST['id']) ? $idtoname[$_REQUEST['id']] : 'Entries by Agency'));
+$endkey = (isset($_REQUEST['end_key']) ? $_REQUEST['end_key'] : '9999-99-99');
+?>
+<div class="headline">Read all the information released by Australian Federal Government agencies under the FOI Act in one place!</div>
+<a style='float:right' href="rss.xml.php"><img src="img/feed-icon-14x14.png" alt="RSS Icon"/> All Agencies RSS Feed</a><br>
+<?php
+try {
+ if ($_REQUEST['id']) {
+ $rows = $foidocsdb->get_view("app", "byAgencyID", $_REQUEST['id'], false, false, false)->rows;
+ foreach ($rows as $row) {
+ //print_r($rows);
+ echo displayLogEntry($row, $idtoname);
+ if (!isset($startkey))
+ $startkey = $row->key;
+ $endkey = $row->key;
+ }
+ } else {
+ $rows = $foidocsdb->get_view("app", "byAgencyID?group=true", null, false, false, true)->rows;
+ if ($rows) {
+ foreach ($rows as $row) {
+ echo '<a href="agency.php?id=' . $row->key . '">' . $idtoname[$row->key] . " (" . $row->value . " records)</a> <br>\n";
+ }
+ }
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+echo "<a class='btn btn-large btn-primary' href='?end_key=$endkey' style='float:right;'>next page <i class='icon-circle-arrow-right icon-white'></i></a>";
+include_footer_documents();
+?>
--- a/documents/charts.php
+++ b/documents/charts.php
@@ -1,6 +1,6 @@
<?php
include('template.inc.php');
-include_header_documents("");
+include_header_documents("Charts");
include_once('../include/common.inc.php');
$agenciesdb = $server->get_db('disclosr-agencies');
@@ -15,29 +15,28 @@
<h1><a href="about.php">Charts</a></h1>
<h4 class="subheader">Lorem ipsum.</h4>
</div>
-<div id="employees" style="width:1000px;height:900px;"></div>
+<div id="bydate" style="width:1000px;height:300px;"></div>
+<div id="byagency" style="width:1200px;height:300px;"></div>
<script id="source">
window.onload = function() {
$(document).ready(function() {
var
d1 = [],
- start = new Date("2009/01/01 01:00").getTime(),
- options,
- graph,
- i, x, o;
+ options1,
+ o1;
<?php
try {
- $rows = $foidocsdb->get_view("app", "byDate?group=true", null, true)->rows;
+ $rows = $foidocsdb->get_view("app", "byDateMonthYear?group=true",null, false,false,true)->rows;
$dataValues = Array();
foreach ($rows as $row) {
- $dataValues[$row->value] = $row->key;
+ $dataValues[$row->key] = $row->value;
}
$i = 0;
ksort($dataValues);
- foreach ($dataValues as $value => $key) {
+ foreach ($dataValues as $key => $value) {
$date = date_create_from_format('Y-m-d', $key);
if (date_format($date, 'U') != "") {
echo " d1.push([".date_format($date, 'U')."000, $value]);" . PHP_EOL;
@@ -52,7 +51,7 @@
- options = {
+ options1 = {
xaxis : {
mode : 'time',
labelsAngle : 45
@@ -68,19 +67,19 @@
function drawGraph (opts) {
// Clone the options, so the 'options' variable always keeps intact.
- o = Flotr._.extend(Flotr._.clone(options), opts || {});
+ o1 = Flotr._.extend(Flotr._.clone(options1), opts || {});
// Return a new graph.
return Flotr.draw(
- document.getElementById("employees"),
+ document.getElementById("bydate"),
[ d1 ],
- o
+ o1
);
}
graph = drawGraph();
- Flotr.EventAdapter.observe(container, 'flotr:select', function(area){
+ Flotr.EventAdapter.observe(document.getElementById("bydate"), 'flotr:select', function(area){
// Draw selected area
graph = drawGraph({
xaxis : { min : area.x1, max : area.x2, mode : 'time', labelsAngle : 45 },
@@ -89,10 +88,74 @@
});
// When graph is clicked, draw the graph with default area.
- Flotr.EventAdapter.observe(container, 'flotr:click', function () { graph = drawGraph(); });
+ Flotr.EventAdapter.observe(document.getElementById("bydate"), 'flotr:click', function () { graph = drawGraph(); });
});
};
+
+var d2 = [];
+var agencylabels = [];
+function agencytrackformatter(obj) {
+
+ return agencylabels[Math.floor(obj.x)] +" = "+obj.y;
+
+ }
+ function agencytickformatter(val, axis) {
+ if (agencylabels[Math.floor(val)]) {
+ return '<p style="margin-top:8em;-webkit-transform:rotate(-90deg);">'+(agencylabels[Math.floor(val)])+"</b>";
+
+ } else {
+ return "";
+ }
+ }
+<?php
+ try {
+ $rows = $foidocsdb->get_view("app", "byAgencyID?group=true",null, false,false,true)->rows;
+
+
+ $dataValues = Array();
+ $i = 0;
+ foreach ($rows as $row) {
+ echo " d2.push([".$i.", $row->value]);" . PHP_EOL;
+ echo " agencylabels.push(['".str_replace("'","",$idtoname[$row->key])."']);" . PHP_EOL;
+
+ $i++;
+ }
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+ ?>
+ // Draw the graph
+ Flotr.draw(
+ document.getElementById("byagency"),
+ [d2],
+ {
+ bars : {
+ show : true,
+ horizontal : false,
+ shadowSize : 0,
+ barWidth : 0.5
+ },
+mouse : {
+ track : true,
+ relative : true,
+ trackFormatter: agencytrackformatter
+ },
+ yaxis : {
+ min : 0,
+ autoscaleMargin : 1
+ },
+ xaxis: {
+ minorTickFreq: 1,
+ noTicks: agencylabels.length,
+ showMinorLabels: true,
+ tickFormatter: agencytickformatter
+ },
+ legend: {
+ show: false
+ }
+ }
+ );
</script>
<?php
--- /dev/null
+++ b/documents/date.php
@@ -1,1 +1,34 @@
+<?php
+include('template.inc.php');
+include_header_documents("Entries by Date");
+include_once('../include/common.inc.php');
+$endkey = (isset($_REQUEST['end_key']) ? $_REQUEST['end_key'] : '9999-99-99');
+?>
+<div class="headline">Read all the information released by Australian Federal Government agencies under the FOI Act in one place!</div>
+<a style='float:right' href="rss.xml.php"><img src="img/feed-icon-14x14.png" alt="RSS Icon"/> All Agencies RSS Feed</a><br>
+<?php
+/*$agenciesdb = $server->get_db('disclosr-agencies');
+
+$idtoname = Array();
+foreach ($agenciesdb->get_view("app", "byCanonicalName")->rows as $row) {
+ $idtoname[$row->id] = trim($row->value->name);
+}
+$foidocsdb = $server->get_db('disclosr-foidocuments');
+try {
+ $rows = $foidocsdb->get_view("app", "byDate", Array($endkey, '0000-00-00'), true, 20)->rows;
+ if ($rows) {
+ foreach ($rows as $key => $row) {
+ echo displayLogEntry($row, $idtoname);
+ if (!isset($startkey)) $startkey = $row->key;
+ $endkey = $row->key;
+ }
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+echo "<a class='btn btn-large btn-primary' href='?end_key=$endkey' style='float:right;'>next page <i class='icon-circle-arrow-right icon-white'></i></a>";
+*/
+include_footer_documents();
+?>
+
--- a/documents/disclogsList.php
+++ b/documents/disclogsList.php
@@ -1,7 +1,7 @@
<?php
include('template.inc.php');
-include_header_documents("");
+include_header_documents("List of Disclosure Logs");
include_once('../include/common.inc.php');
echo "<table>
Binary files /dev/null and b/documents/img/feed-icon-14x14.png differ
--- a/documents/index.php
+++ b/documents/index.php
@@ -3,10 +3,11 @@
include('template.inc.php');
include_header_documents("");
include_once('../include/common.inc.php');
-$startkey = (isset($_REQUEST['start_key']) ? $_REQUEST['start_key'] : '9999-99-99');
+$endkey = (isset($_REQUEST['end_key']) ? $_REQUEST['end_key'] : '9999-99-99');
?>
+<div class="headline">Read all the information released by Australian Federal Government agencies under the FOI Act in one place!</div>
+<a style='float:right' href="rss.xml.php"><img src="img/feed-icon-14x14.png" alt="RSS Icon"/> All Agencies RSS Feed</a><br>
<?php
-
$agenciesdb = $server->get_db('disclosr-agencies');
$idtoname = Array();
@@ -15,17 +16,18 @@
}
$foidocsdb = $server->get_db('disclosr-foidocuments');
try {
- $rows = $foidocsdb->get_view("app", "byDate", Array($startkey, '0000-00-00'), true, 20)->rows;
+ $rows = $foidocsdb->get_view("app", "byDate", Array($endkey, '0000-00-00'), true, 20)->rows;
if ($rows) {
foreach ($rows as $key => $row) {
echo displayLogEntry($row, $idtoname);
+ if (!isset($startkey)) $startkey = $row->key;
$endkey = $row->key;
}
}
} catch (SetteeRestClientException $e) {
setteErrorHandler($e);
}
-echo "<a href='?start_key=$endkey'>next page</a>";
+echo "<a class='btn btn-large btn-primary' href='?end_key=$endkey' style='float:right;'>next page <i class='icon-circle-arrow-right icon-white'></i></a>";
include_footer_documents();
?>
--- a/documents/rss.xml.php
+++ b/documents/rss.xml.php
@@ -8,21 +8,28 @@
//Creating an instance of FeedWriter class.
$TestFeed = new RSS2FeedWriter();
//Setting the channel elements
-//Use wrapper functions for common channelelements
-$TestFeed->setTitle('disclosurelo.gs Newest Entries - All');
-$TestFeed->setLink('http://disclosurelo.gs/rss.xml.php');
-$TestFeed->setDescription('disclosurelo.gs Newest Entries - All Agencies');
-$TestFeed->setChannelElement('language', 'en-us');
-$TestFeed->setChannelElement('pubDate', date(DATE_RSS, time()));
-
-//Retriving informations from database
+////Retriving informations from database
$idtoname = Array();
$agenciesdb = $server->get_db('disclosr-agencies');
foreach ($agenciesdb->get_view("app", "byCanonicalName")->rows as $row) {
$idtoname[$row->id] = trim($row->value->name);
}
$foidocsdb = $server->get_db('disclosr-foidocuments');
-$rows = $foidocsdb->get_view("app", "byDate", Array('9999-99-99', '0000-00-00', 50), true)->rows;
+if (isset($_REQUEST['id'])) {
+ $rows = $foidocsdb->get_view("app", "byAgencyID", $_REQUEST['id'], false, false, false)->rows;
+ $title = $idtoname[$_REQUEST['id']];
+} else {
+ $rows = $foidocsdb->get_view("app", "byDate", Array('9999-99-99', '0000-00-00', 50), true)->rows;
+ $title = 'All Agencies';
+}
+//Use wrapper functions for common channelelements
+$TestFeed->setTitle('disclosurelo.gs Newest Entries - '.$title);
+$TestFeed->setLink('http://disclosurelo.gs/rss.xml.php'.(isset($_REQUEST['id'])? '?id='.$_REQUEST['id'] : ''));
+$TestFeed->setDescription('disclosurelo.gs Newest Entries - '.$title);
+$TestFeed->setChannelElement('language', 'en-us');
+$TestFeed->setChannelElement('pubDate', date(DATE_RSS, time()));
+
+
//print_r($rows);
foreach ($rows as $row) {
//Create an empty FeedItem
--- /dev/null
+++ b/documents/scrapers/7c6adc1d41cf029bf1a0959e5156477a.py
@@ -1,1 +1,42 @@
+import sys
+import os
+sys.path.insert(0, os.path.join(os.path.dirname(__file__) or '.', '../'))
+import genericScrapers
+import scrape
+from datetime import date
+from pyquery import PyQuery as pq
+from lxml import etree
+import urllib
+class ACMADisclogScraper(genericScrapers.GenericDisclogScraper):
+
+ def doScrape(self):
+ foidocsdb = scrape.couch['disclosr-foidocuments']
+ (url, mime_type, content) = scrape.fetchURL(scrape.docsdb,
+ self.getURL(), "foidocuments", self.getAgencyID())
+
+ d = pq(content)
+ d.make_links_absolute()
+ d.table.filter('.ncTAF_DataTABLE')
+ print [i.text() for i in d.items('span')]
+ description = ""
+ dochash = scrape.mkhash(description)
+ doc = foidocsdb.get(dochash)
+ if doc is None:
+ print "saving " + dochash
+ edate = date.today().strftime("%Y-%m-%d")
+ doc = {'_id': dochash, 'agencyID': self.getAgencyID()
+ , 'url': self.getURL(), 'docID': dochash,
+ "date": edate, "title": "Disclosure Log Updated", "description": description}
+ #foidocsdb.save(doc)
+ else:
+ print "already saved"
+
+
+if __name__ == '__main__':
+ print 'Subclass:', issubclass(ACMADisclogScraper,
+ genericScrapers.GenericDisclogScraper)
+ print 'Instance:', isinstance(ACMADisclogScraper(),
+ genericScrapers.GenericDisclogScraper)
+ ACMADisclogScraper().doScrape()
+
--- a/documents/sitemap.xml.php
+++ b/documents/sitemap.xml.php
@@ -10,10 +10,18 @@
if (strpos($file, ".php") !== false && $file != "index.php" && $file != "sitemap.xml.php")
echo " <url><loc>" . local_url() . "$file</loc><priority>0.6</priority></url>\n";
}
-
-$db = $server->get_db('disclosr-foidocuments');
+$agenciesdb = $server->get_db('disclosr-agencies');
try {
- $rows = $db->get_view("app", "all")->rows;
+ $rows = $agenciesdb->get_view("app", "byCanonicalName")->rows;
+ foreach ($rows as $row) {
+ echo '<url><loc>' . local_url() . 'agency.php?id=' . $row->value->_id . "</loc><priority>0.3</priority></url>\n";
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+$foidocsdb = $server->get_db('disclosr-foidocuments');
+try {
+ $rows = $foidocsdb->get_view("app", "all")->rows;
foreach ($rows as $row) {
echo '<url><loc>' . local_url() . 'view.php?id=' . $row->value->_id . "</loc><priority>0.3</priority></url>\n";
}
--- a/documents/template.inc.php
+++ b/documents/template.inc.php
@@ -68,7 +68,8 @@
</p>
<ul class="nav">
- <li><a href="index.php">Home</a></li>
+ <li><a href="agency.php">By Agency</a></li>
+ <li><a href="date.php">By Date</a></li>
<li><a href="disclogsList.php">List of Disclosure Logs</a></li>
<li><a href="about.php">About</a></li>
@@ -89,7 +90,9 @@
<footer>
<p>Not affiliated with or endorsed by any government agency.</p>
</footer>
- <script type="text/javascript">
+ <?php
+ if ($ENV != "DEV") {
+ echo "<script type='text/javascript'>
var _gaq = _gaq || [];
_gaq.push(['_setAccount', 'UA-12341040-4']);
@@ -106,7 +109,9 @@
s.parentNode.insertBefore(ga, s);
})();
- </script>
+ </script>";
+ }
+ ?>
<!-- Le javascript
================================================== -->
<!-- Placed at the end of the document so the pages load faster -->
@@ -163,13 +168,13 @@
if (isset($row->value->links)) {
$result .= '<h3>Links/Documents</h3><ul itemprop="associatedMedia">';
foreach ($row->value->links as $link) {
- $result .= '<li itemscope itemtype="http://schema.org/MediaObject"><a href=' . $link . ' itemprop="url contentURL">' . urlencode($link) . "</a></li>";
+ $result .= '<li itemscope itemtype="http://schema.org/MediaObject"><a href=' . htmlspecialchars ($link) . ' itemprop="url contentURL">' . htmlspecialchars ( $link) . "</a></li>";
}
$result .= "</ul>";
}
$result .= "<small><A itemprop='url' href='" . $row->value->url . "'>View original source...</a> ID: " . strip_tags($row->value->docID) . "</small>";
- $result .= "</div>";
+ $result .= "</div>\n";
return $result;
}
--- a/documents/view.php
+++ b/documents/view.php
@@ -1,6 +1,5 @@
<?php
include('template.inc.php');
-include_header_documents("");
include_once('../include/common.inc.php');
?>
<?php
@@ -17,6 +16,8 @@
try {
$obj = new stdClass();
$obj->value = $foidocsdb->get($_REQUEST['id']);
+ include_header_documents($obj->value->title);
+
echo displayLogEntry($obj,$idtoname);
} catch (SetteeRestClientException $e) {
