/**
* Tests Remote::status
*
* @test
* @dataProvider provider_status
* @param boolean $input Input for File::mime
* @param boolean $expected Output for File::mime
*/
public function test_status($input, $expected)
{
if ( ! $this->hasInternet())
$this->markTestSkipped('An internet connection is required for this test');
$this->assertSame($expected, Remote::status($input));
}
private function rss_feed()
{
// RSS Feed parameters:
$api_id = 2;
// api_id = 2 for RSS Feeds
// Perform gathering for each active RSS feed URL for given project_id
$rss_feeds = $this->model_gather->get_active_rss_feeds($this->project_id);
foreach ($rss_feeds as $rss_feed) {
$total_results_gathered = 0;
// If RSS feed was set searchable generate search query GET string
$keyword_str = "";
if ($rss_feed['searchable']) {
$num_keywords = count($this->keywords_phrases);
$i = 0;
foreach ($this->keywords_phrases as $keyword_phrase) {
$i++;
$word_split = explode(" ", $keyword_phrase['keyword_phrase']);
if (count($word_split) > 1) {
// Is phrase (more than 1 word)
// Check if searching "exact phrase" -> if so add quotes
if ($keyword_phrase['exact_phrase']) {
$keyword_str .= '"' . urlencode($keyword_phrase['keyword_phrase']) . '"';
} else {
$keyword_str .= '(' . urlencode($keyword_phrase['keyword_phrase']) . ')';
}
} else {
// Is single keyword
$keyword_str .= urlencode($keyword_phrase['keyword_phrase']);
}
if ($i < $num_keywords) {
$keyword_str .= '+OR+';
}
}
}
$connection_retries = 0;
$this->api_connect_error = 0;
while (TRUE) {
// Compile request URL
$request_url = $rss_feed['url'] . $keyword_str;
print "Query: {$request_url}\n";
// Check for connection errors
try {
$status_code = Remote::status($request_url);
} catch (Exception $e) {
$this->api_connect_error = "Error connecting to {$request_url}. Cannot locate host.";
}
if (!$this->api_connect_error and ($status_code < 200 or $status_code > 299)) {
$this->api_connect_error = "Error connecting to {$request_url}. Status code: {$status_code}";
}
if (!$this->api_connect_error) {
$rss_output = Feed::parse($request_url);
$num_results = count($rss_output);
if ($num_results > 0) {
// Loop through each result, parse, and store data
foreach ($rss_output as $item) {
$title = array_key_exists('title', $item) ? $item['title'] : '';
$text = array_key_exists('description', $item) ? $item['description'] : '';
$date_published_timestamp = array_key_exists('pubDate', $item) ? strtotime($item['pubDate']) : 0;
// Append title to text & strip all HTML tags except <br>'s
$text = "Title: {$title}<br>{$text}";
$text = strip_tags($text, "<br>");
// Determine unique identifier, if no URL -> use guid -> if no GUID give error
if (array_key_exists('link', $item) and $item['link'] != "") {
$url = $item['link'];
} else {
if (array_key_exists('guid', $item)) {
$url = $item['guid'];
} else {
print "Error: item has no URL or GUID. Cannot use.\n";
continue;
}
}
// Add each result to database
$require_keywords = $rss_feed['searchable'] ? 0 : 1;
// Only require keywords if RSS feed is NOT searchable
$total_results_gathered += $this->add_metadata($url, $text, $require_keywords, array('project_id' => $this->project_id, 'api_id' => $api_id, 'date_published' => $date_published_timestamp, 'date_retrieved' => time()));
}
}
break;
} else {
// Retry connecting to API
$connection_retries++;
// Connection error (only for non-searchable RSS feeds, it is assumed an error has occured if no item comes through the feed after multiple connection attempts)
if ($connection_retries > $this->connection_retries) {
if (!$rss_feed['searchable']) {
$this->model_gather->insert_gather_log(array('project_id' => $this->project_id, 'search_query' => $this->api_connect_error, 'date' => time(), 'results_gathered' => 0, 'error' => 1));
$this->api_connect_error = 1;
}
break;
}
}
}
// Add entry to gather log (as long as no errors occurred)
if (!$this->api_connect_error) {
$this->model_gather->insert_gather_log(array('project_id' => $this->project_id, 'search_query' => $request_url, 'date' => time(), 'results_gathered' => $total_results_gathered));
}
}
}
/**
* Tests Remote::status
*
* @test
* @dataProvider providerStatus
* @param boolean $input Input for File::mime
* @param boolean $expected Output for File::mime
*/
function testStatus($input, $expected)
{
$this->assertSame($expected, Remote::status($input));
}
