<?php
require '../../PushAPI.inc';
require 'tick.inc';
try {
echo "Create\n";
$papi = PushAPIFactory::createHttp("localhost", 42002, "default");
echo "Ping\n";
$papi->ping();
$myPart1 = new Part(array('content' => '<html><body>Hello, world</body></html>', 'directives' => array('mimeHint' => 'text/html', 'encoding' => 'utf-8')));
$myDoc1 = new Document(array('uri' => 'http://example.com/doc.html', 'parts' => $myPart1, 'metas' => array('path' => 'Path/To/Example/1')));
$tsukemono = "漬物";
$tsukemonoh = "つけもの";
$myPart2 = new Part(array('content' => '<html>' . '<head><title>Hello World</title></head>' . '<body>Hello, world' . '<br />I like eating ' . $tsukemono . ' (' . $tsukemonoh . ')!' . '</body></html>', 'directives' => array('mimeHint' => 'text/html', 'encoding' => 'utf-8')));
$myDoc2 = new Document(array('uri' => 'http://example.com/doc2.html', 'parts' => $myPart2, 'metas' => array('path' => 'Path/To/Example/2')));
tickSet('serialization');
echo "Checking if document exists\n";
$stamp = $papi->getDocumentStatus('http://example.com/doc.html');
if ($stamp !== false) {
echo "Yes, its stamp is " . $stamp . "\n";
} else {
echo "No, it does not yet exist\n";
}
echo "Pushing documents\n";
$resp = $papi->addDocument(array($myDoc1, $myDoc2));
tickSet('pushing');
echo "Set checkpoint\n";
$serial = $papi->setCheckpoint(0, true);
$cp = $papi->enumerateCheckpointInfo();
foreach ($cp as $k => $v) {
echo "Checkpoint: name='" . $k . "' value='" . $v . "'\n";
{
$papi->deleteDocument(getUri($mid));
}
/* The Exalead "newsburst" server address and port */
$newsburst = 'reddmz005.prod.exalead.com';
$newsburstport = 42119;
$flushTime = 60;
/* Connect to PAPI */
echo "Create\n";
if (count($argv) != 4) {
$stderr = fopen('php://stderr', 'w');
fwrite($stderr, "Usage: " . $argv[0] . " papi_hostname papi_port papi_source\n");
fwrite($stderr, "Example: " . $argv[0] . " localhost 10002 default\n");
exit;
}
$papi = PushAPIFactory::createHttp($argv[1], intval($argv[2]), $argv[3]);
echo "Ping\n";
$papi->ping();
/* Let's have some burst */
$last = time();
$serial = false;
$flushRetry = 0;
$seen = 0;
$checkpoint = intval($papi->getCheckpoint());
for ($fp = false;;) {
/* (Re)Open a connection to newsburst */
if ($fp === false) {
echo "Connecting to newsburst: ";
for ($i = 0; $i < 60; $i++) {
$fp = @fsockopen($newsburst, $newsburstport, &$errno, &$errstr, 120);
if ($fp !== false && is_resource($fp)) {
$feedColl = $db->feed;
$docsPAPI = array();
$q = empty($_GET['q']) ? "all" : $_GET['q'];
$forceUpdate = empty($_GET['forceUpdate']) ? 0 : $_GET['forceUpdate'];
if ($q != '') {
if ($q == 'all') {
// we parse all valid feeds
$query = array('status' => 1);
} else {
$query = array('name' => $q, 'status' => 1);
}
$res = $feedColl->find($query);
$feeds = iterator_to_array($res);
//var_dump($feeds);
try {
$papi = PushAPIFactory::createHttp("ec2-54-246-84-102.eu-west-1.compute.amazonaws.com", 62002, "parserPAPI");
//echo "Ping\n";
//$papi->ping();
} catch (PushAPIFactory $e) {
echo "Error: " . $e->getMessage() . "\n";
}
function mapIt($i)
{
$i = convert_smart_quotes($i);
// clean rounded quotes from MSWord
return '/' . $i . '/';
}
foreach ($feeds as $feed) {
//var_dump($feed);
$file = $feed['name'] . ".xml";
echo 'Parsing feed : ' . $feed['name'] . '<br>';
