handle failed image urls better

mailstream/mailstream.php

@@ -239,6 +239,11 @@ function mailstream_do_images(array &$item, array &$attachments)
 		$cookiejar = tempnam(System::getTempPath(), 'cookiejar-mailstream-');
 		try {
 			$curlResult = DI::httpClient()->fetchFull($url, HttpClientAccept::DEFAULT, 0, $cookiejar);
+			if (!$curlResult->isSuccess()) {
+				Logger::debug('mailstream: fetch image url failed', [
+					'url' => $url, 'item_id' => $item['id'], 'return_code' => $curlResult->getReturnCode()]);
+				continue;
+			}
 		} catch (InvalidArgumentException $e) {
 			Logger::error('mailstream_do_images exception fetching url', ['url' => $url, 'item_id' => $item['id']]);
 			continue;

retriever/retriever.php

@@ -660,10 +660,14 @@ function retriever_extract(DOMDocument $doc, array $retriever) {
  * @return DOMDocument New DOM document with global URLs
  */
 function retriever_globalise_urls(DOMDocument $doc, array $resource) {
-	$components = parse_url($resource['redirect-url']);
+	$url = $resource['redirect-url'];
-        if (!array_key_exists('scheme', $components) || !array_key_exists('host', $components) || !array_key_exists('path', $components)) {
+	if ($url == "") {
+		$url = $resource['url'];
+	}
+	$components = parse_url($url);
+	if (!array_key_exists('scheme', $components) || !array_key_exists('host', $components) || !array_key_exists('path', $components)) {
 		return $doc;
-        }
+	}
 	$rooturl = $components['scheme'] . "://" . $components['host'];
 	$dirurl = $rooturl . dirname($components['path']) . "/";
 	$params = array('$dirurl' => $dirurl, '$rooturl' => $rooturl);
@@ -829,7 +833,7 @@ function retriever_transform_images(array &$item, array $resource) {
 		Logger::error('retriever_transform_images: unable to store photo ' . $resource['url'] . ' error: ' . $e->getMessage());
 		return;
 	}
-	$new_url = DI::baseUrl() . '/photo/' . $rid . '-0.' . $image->getExt();
+	$new_url = DI::baseUrl() . '/photo/' . $rid . '-0' . $image->getExt();
 	if (!strlen($new_url)) {
 		Logger::warning('retriever_transform_images: no replacement URL for image ' . $resource['url']);
 		return;

retriever/templates/fix-urls.tpl

@@ -22,5 +22,10 @@
       <xsl:value-of select="concat('{{$rooturl}}',.)"/>
     </xsl:attribute>
   </xsl:template>
+  <xsl:template match="*/@src[not(contains(.,':'))]">
+    <xsl:attribute name="src">
+      <xsl:value-of select="concat('{{$dirurl}}',.)"/>
+    </xsl:attribute>
+  </xsl:template>
 
 </xsl:stylesheet>