Add 'assoc' => true mode to it_url::get() and prepare to mark function static

2 files changed, 56 insertions, 43 deletions

diff --git a/it_url.class b/it_url.class
index 5bd796c..85471d4 100644
--- a/it_url.class
+++ b/it_url.class
@@ -72,7 +72,7 @@ function __construct($url = null)
 function is_reachable($timeout = 5)
 {
 	$url = new it_url($this->url);
-	@$url->get(['method' => 'HEAD', 'totaltimeout' => $timeout]);
+	@$url->_get(['method' => 'HEAD', 'totaltimeout' => $timeout]);
 
 	return $url->result >= 200 && $url->result < 400;
 }
@@ -93,6 +93,7 @@ static function _postprocess($data, $p)
  *
  * @param $p parameter array with the following keys
  * @param $p['url']           url to get, defaults to constructor URL
+ * @param $p['assoc']         Return [ 'data' => string, 'status' => int, 'cookies' => array, 'headers' => array, 'errstr' => string ] instead of just data
  * @param $p['headers']       optional associative array of HTTP headers to send
  * @param $p['safety']        DEPRECATED. 0 = ignore errors, 1 = errors, 2 = fatals
  * @param $p['it_error']      extra arguments for it_error or false to ignore errors
@@ -108,35 +109,39 @@ static function _postprocess($data, $p)
  * @param $p['retrysleep']    Number of seconds to wait before retry (additional to fetchsleep), fractions ok
  * @param $p['compression']   use compression (uses curl to do that)
  * @param $p['postprocess']   function called with content and $p which has it_error. returns content or null
- * @return contents of resulting page, considering redirects, excluding headers, or false on error
+ * @return Content of resulting page (considering redirects, excluding headers or false on error) or array (empty on error) if 'assoc' => true
  */
-function get($p=null, $timeout=5)
+function get($p = [], $timeout = null)
 {
-	if (!is_int($timeout))
-		it::error("Wrong value for second argument of it_url::get()!");
-	if (!is_array($p))
-		$p = array('url' => $p, 'timeout' => $timeout);
+	if (isset($this) && $this instanceof it_url)	# FIXME 202006 cschneid Compatibility until all application switched to 'assoc' => true or it_ur::get()
+		return $this->_get($p, $timeout);
+	return (new static)->_get($p, $timeout);
+}
+
+/**
+ * Non-static alias for get so we can make get() static
+ */
+function _get($p = [], $timeout = null)
+{
+	if (isset($timeout))
+		it::error("Deprecated second argument of it_url::get()!");
+	if (is_string($p))
+		$p = array('url' => $p, 'timeout' => 5);
 	$p += array('retries' => 1);
 
 	if (($filter = EDC('req')) && ($filter == 1 || strstr($p['url'], "/$filter.")))
 		ED($p['url'] ?: $this->url);
 
-	if (isset($this) && $this instanceof it_url)
-	{
-		$url = $this;
-		if ($p['url'])
-			$this->__construct($p['url']);
-	}
-	else	# called statically
-		$url = new it_url($p['url']);
+	if ($p['url'])
+		$this->__construct($p['url']);
 
-	$result = $url->request($p + ['followlocation' => true]);
+	$result = $this->request($p + ['followlocation' => true]);
 	$result = self::_postprocess($result, $p);
 
-	if (!$result && $p['retries'] > 0 && !it::match('^(4..|204)$', $url->result))
+	if (!$result && $p['retries'] > 0 && !it::match('^(4..|204)$', $this->result))
 	{
 		usleep($p['retrysleep']*1000000);
-		$result = $url->get(array('retries' => $p['retries'] - 1) + $p);
+		$result = $this->_get(array('retries' => $p['retries'] - 1) + $p);
 	}
 
 	if (($filter = EDC('res')) && strstr($p['url'], it::replace(array('1' => ":"), $filter)))
@@ -144,6 +149,9 @@ function get($p=null, $timeout=5)
 
 	usleep($p['fetchsleep'] * 1000000);
 
+	if ($p['assoc'])
+		$result = $result !== false ? [ 'status' => $this->result, 'data' => $this->data, 'headers' => $this->headers, 'cookies' => $this->cookies, 'errstr' => $this->errstr ] : [];
+
 	return $result;
 }
 
@@ -524,9 +532,9 @@ static function get_cache($p = array())
 
 			EDC('getcache', "new", $filemtime, $p['url'], $path);
 			$url = new it_url;
-			if ($success = $url->get($p + ['checkonly' => true, 'filemtime' => EDC('nocache') ? null : $filemtime])) # => true means not modified (no new data fetched)
+			if ($success = $url->_get($p + ['checkonly' => true, 'filemtime' => EDC('nocache') ? null : $filemtime])) # => true means not modified (no new data fetched)
 			{
-				$newfile = it_url::_atomicwrite($path, $success);
+				$newfile = it_url::_atomicwrite($path, $p['assoc'] ? serialize($success) : $success);
 				if ($p['returnheaders'])
 					it::file_put_contents("$path.headers", '<?php return ' . var_export($url->headers, true) . ";\n");
 			}
@@ -618,11 +626,28 @@ static function get_cache($p = array())
 
 /**
  * Fetch a file, cache it and return contents
- * @param see it_url::get_cache()
+ * @param @see it_url::get_cache()
+ * @param $p['assoc']         Return [ 'data' => string, 'status' => int, 'cookies' => array, 'headers' => array, 'errstr' => string ] instead of just data
+ * @return @see it_url::get()
  */
-function get_cache_contents($p)
+static function get_cache_contents($p)
 {
-	return ($fn = self::get_cache($p)) ? self::_postprocess(it::file_get_contents($fn), $p) : it::error((array)$p['it_error'] + ['title' => $p['safety'] == 0 ? false : "failed getting " . it_url::absolute($p['url']), 'body' => $p]);
+	if ($fn = self::get_cache($p))
+	{
+		$result = it::file_get_contents($fn);
+		if ($p['assoc'])
+		{
+			$response = unserialize($result);
+			$response['data'] = self::_postprocess($response['data'], $p);
+			$result = $response;
+		}
+		else
+			$result = self::_postprocess($result, $p);
+	}
+	else
+		$result = it::error((array)$p['it_error'] + ['title' => $p['safety'] == 0 ? false : "failed getting " . it_url::absolute($p['url']), 'body' => $p]);
+
+	return $result;
 }
 
 /**
@@ -671,7 +696,7 @@ static function _unlock($path, $lock)
 /**
  * Wait for file which is currently locked
  * @param $path File to wait for
- * @param $p Wait parameters, see @get_cache
+ * @param $p Wait parameters, @see get_cache()
  * @return Whether lock was released within timeout and file is still there
  */
 static function _waitforlockedfile($path, $p)
diff --git a/test/it_url.t b/test/it_url.t
index edfbdb2..5bac758 100755
--- a/test/it_url.t
+++ b/test/it_url.t
@@ -119,41 +119,29 @@ is(
 
 $url = new it_url('http://www.gna.ch/');
 is((bool)$url->is_reachable(), true, "is_reachable($url->url)");
-$page = $url->get();
-ok(
-	strpos($page, '</html>'), # UTF8SAFE
-	'$url->get with url in constructor'
-);
 
 $url = new it_url('http://www.search.ch/not_found');
 is((bool)$url->is_reachable(), false, "is_reachable($url->url)");
 
 $url = new it_url('http://bogus.url');
 is((bool)$url->is_reachable(), false, "is_reachable($url->url)");
-$page = $url->get('http://www.gna.ch/');
-ok(
-	strpos($page, '</html>'), # UTF8SAFE
-	'$url->get(url) with url as string arg'
-);
 
-$url = new it_url('http://bogus.url');
-$page = $url->get(['url' => 'http://www.gna.ch/']);
+$response = it_url::get(['url' => 'http://www.gna.ch/', 'assoc' => true]);
 ok(
-	strpos($page, '</html>'), # UTF8SAFE
-	'$url->get(\'url\' => url) with url as named arg'
+	strpos($response['data'], '</html>'), # UTF8SAFE
+	'Get with url as named arg'
 );
 is(
-	$url->result,
+	$response['status'],
 	'200',
-	'$url->result = 200'
+	'Reponse status = 200'
 );
 is(
-	$url->headers['Server'],
+	$response['headers']['Server'],
 	'Apache',
-	'$url->headers correctly set'
+	'Response headers correctly set'
 );
 
-unset($url, $page);
 $page = it_url::get('http://www.gna.ch/');
 ok(
 	strpos($page, '</html>'), # UTF8SAFE