From c48590a7927b9ab63797501d3e30fd2300d609be Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Wed, 11 Mar 2020 10:09:22 -0400
Subject: [PATCH] Initial failure tests

---
 lib/Exception.php             |  2 ++
 lib/Parser/XML/Feed.php       | 11 +++++++++--
 tests/cases/XML/failures.yaml | 11 +++++++++++
 3 files changed, 22 insertions(+), 2 deletions(-)
 create mode 100644 tests/cases/XML/failures.yaml

diff --git a/lib/Exception.php b/lib/Exception.php
index 4c4c520..85d65bf 100644
--- a/lib/Exception.php
+++ b/lib/Exception.php
@@ -11,7 +11,9 @@ abstract class Exception extends \Exception {
         // Parsing: 0x1100
         "notJSONType"                => [0x1111, "Document Content-Type is not either that of JSON Feed or generic JSON"],
         "notJSON"                    => [0x1112, "Document is not valid JSON"],
+        "notXML"                     => [0x1112, "Document is not well-formed XML"],
         "notJSONFeed"                => [0x1113, "Document is not a JSON Feed document"],
+        "notXMLFeed"                 => [0x1113, "Document is not a newsfeed"],
     ];
 
     public function __construct(string $symbol, \Exception $e = null) {
diff --git a/lib/Parser/XML/Feed.php b/lib/Parser/XML/Feed.php
index 079d735..fcd63bd 100644
--- a/lib/Parser/XML/Feed.php
+++ b/lib/Parser/XML/Feed.php
@@ -6,6 +6,7 @@
 declare(strict_types=1);
 namespace MensBeam\Lax\Parser\XML;
 
+use MensBeam\Lax\Parser\Exception;
 use MensBeam\Lax\Person\Collection as PersonCollection;
 use MensBeam\Lax\Category\Collection as CategoryCollection;
 use MensBeam\Lax\Feed as FeedStruct;
@@ -18,6 +19,8 @@ class Feed implements \MensBeam\Lax\Parser\Feed {
     use Primitives\Construct;
     use Primitives\Feed;
 
+    protected const LIBXML_OPTIONS = \LIBXML_BIGLINES | \LIBXML_COMPACT | \LIBXML_HTML_NODEFDTD | \LIBXML_NOCDATA | \LIBXML_NOENT | \LIBXML_NONET | \LIBXML_NOERROR | LIBXML_NOWARNING;
+
     /** @var string */
     protected $data;
     /** @var string */
@@ -41,7 +44,9 @@ class Feed implements \MensBeam\Lax\Parser\Feed {
     /** Performs initialization of the instance */
     protected function init(FeedStruct $feed): FeedStruct {
         $this->document = new \DOMDocument();
-        $this->document->loadXML($this->data, \LIBXML_BIGLINES | \LIBXML_COMPACT);
+        if (!$this->document->loadXML($this->data, self::LIBXML_OPTIONS)) {
+            throw new Exception("notXML");
+        }
         $this->document->documentURI = (string) $this->url;
         $this->xpath = new XPath($this->document);
         $this->subject = $this->document->documentElement;
@@ -61,13 +66,15 @@ class Feed implements \MensBeam\Lax\Parser\Feed {
                 $element = $this->fetchElement("rss1:item|rss0:item|rss1:image|rss0:image");
                 if ($element) {
                     $feed->version = ($element->namespaceURI === XPath::NS['rss1']) ? "1.0" : "0.90";
+                } else {
+                    throw new Exception("notXMLFeed");
                 }
             }
         } elseif ($ns === XPath::NS['atom'] && $name === "feed") {
             $feed->format = "atom";
             $feed->version = "1.0";
         } else {
-            throw new \Exception;
+            throw new Exception("notXMLFeed");
         }
         $feed->meta->url = $this->url;
         return $feed;
diff --git a/tests/cases/XML/failures.yaml b/tests/cases/XML/failures.yaml
new file mode 100644
index 0000000..6d8a9c0
--- /dev/null
+++ b/tests/cases/XML/failures.yaml
@@ -0,0 +1,11 @@
+Malformed XML:
+    input: >
+        <rss>
+            <channel>
+        </rss>
+    exception: notXML
+
+Root element of non-feed type:
+    input: >
+        <html/>
+    exception: notXMLFeed