From cdd9f4dfbeb26faa4f077d4fb091a0219e36e764 Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Thu, 2 May 2019 21:54:49 -0400
Subject: [PATCH] More OPML parser tests

---
 lib/ImportExport/OPML.php                   |  4 ++--
 tests/cases/ImportExport/TestOPML.php       |  6 +++++-
 tests/docroot/Import/OPML/BrokenOPML.1.opml |  1 +
 tests/docroot/Import/OPML/BrokenOPML.2.opml |  1 +
 tests/docroot/Import/OPML/BrokenOPML.3.opml |  1 +
 tests/docroot/Import/OPML/BrokenOPML.4.opml |  5 +++++
 tests/docroot/Import/OPML/BrokenXML.opml    |  1 +
 tests/docroot/Import/OPML/Empty.1.opml      |  4 ++++
 tests/docroot/Import/OPML/Empty.2.opml      |  9 +++++++++
 tests/docroot/Import/OPML/Empty.3.opml      | 11 +++++++++++
 10 files changed, 40 insertions(+), 3 deletions(-)
 create mode 100644 tests/docroot/Import/OPML/BrokenOPML.4.opml
 create mode 100644 tests/docroot/Import/OPML/Empty.1.opml
 create mode 100644 tests/docroot/Import/OPML/Empty.2.opml
 create mode 100644 tests/docroot/Import/OPML/Empty.3.opml

diff --git a/lib/ImportExport/OPML.php b/lib/ImportExport/OPML.php
index b8711bdd..ca02f2d4 100644
--- a/lib/ImportExport/OPML.php
+++ b/lib/ImportExport/OPML.php
@@ -145,8 +145,8 @@ class OPML {
             $err = libxml_get_last_error();
             throw new Exception("invalidSyntax", ['line' => $err->line, 'column' => $err->column]);
         }
-        $body = $d->getElementsByTagName("body");
-        if ($d->documentElement->nodeName !== "opml" || !$body->length || !$body->item(0)->parentNode->isSameNode($d->documentElement)) {
+        $body = (new \DOMXPath($d))->query("/opml/body");
+        if ($body->length != 1) {
             // not a valid OPML document
             throw new Exception("invalidSemantics", ['type' => "OPML"]);
         }
diff --git a/tests/cases/ImportExport/TestOPML.php b/tests/cases/ImportExport/TestOPML.php
index a17cdaff..33904447 100644
--- a/tests/cases/ImportExport/TestOPML.php
+++ b/tests/cases/ImportExport/TestOPML.php
@@ -114,7 +114,7 @@ OPML_EXPORT_SERIALIZATION;
             $this->assertException($exp);
             $parser->parse($data, $flat);
         } else {
-            $this->assertSame($exp, $parse->parse($data, $flat));
+            $this->assertSame($exp, $parser->parse($data, $flat));
         }
     }
 
@@ -124,6 +124,10 @@ OPML_EXPORT_SERIALIZATION;
             ["BrokenOPML.1.opml", false, new Exception("invalidSemantics")],
             ["BrokenOPML.2.opml", false, new Exception("invalidSemantics")],
             ["BrokenOPML.3.opml", false, new Exception("invalidSemantics")],
+            ["BrokenOPML.4.opml", false, new Exception("invalidSemantics")],
+            ["Empty.1.opml", false, [[], []]],
+            ["Empty.2.opml", false, [[], []]],
+            ["Empty.3.opml", false, [[], []]],
         ];
     }
 }
diff --git a/tests/docroot/Import/OPML/BrokenOPML.1.opml b/tests/docroot/Import/OPML/BrokenOPML.1.opml
index 1f551eab..a626ae06 100644
--- a/tests/docroot/Import/OPML/BrokenOPML.1.opml
+++ b/tests/docroot/Import/OPML/BrokenOPML.1.opml
@@ -1 +1,2 @@
 <html/>
+<!-- Not an OPML document -->
diff --git a/tests/docroot/Import/OPML/BrokenOPML.2.opml b/tests/docroot/Import/OPML/BrokenOPML.2.opml
index a6c08015..ac70153f 100644
--- a/tests/docroot/Import/OPML/BrokenOPML.2.opml
+++ b/tests/docroot/Import/OPML/BrokenOPML.2.opml
@@ -1 +1,2 @@
 <opml/>
+<!-- Not body element -->
diff --git a/tests/docroot/Import/OPML/BrokenOPML.3.opml b/tests/docroot/Import/OPML/BrokenOPML.3.opml
index 466ca0ca..b087a1b6 100644
--- a/tests/docroot/Import/OPML/BrokenOPML.3.opml
+++ b/tests/docroot/Import/OPML/BrokenOPML.3.opml
@@ -3,3 +3,4 @@
         <body/>
     </head>
 </opml>
+<!-- No body as child of root -->
diff --git a/tests/docroot/Import/OPML/BrokenOPML.4.opml b/tests/docroot/Import/OPML/BrokenOPML.4.opml
new file mode 100644
index 00000000..544e4c36
--- /dev/null
+++ b/tests/docroot/Import/OPML/BrokenOPML.4.opml
@@ -0,0 +1,5 @@
+<opml>
+    <body/>
+    <body/>
+</opml>
+<!-- Only one body is allowed -->
diff --git a/tests/docroot/Import/OPML/BrokenXML.opml b/tests/docroot/Import/OPML/BrokenXML.opml
index 95028ac2..0cbc6fe2 100644
--- a/tests/docroot/Import/OPML/BrokenXML.opml
+++ b/tests/docroot/Import/OPML/BrokenXML.opml
@@ -1 +1,2 @@
 <opml>
+<!-- Not well-formed XML -->
diff --git a/tests/docroot/Import/OPML/Empty.1.opml b/tests/docroot/Import/OPML/Empty.1.opml
new file mode 100644
index 00000000..4999faaf
--- /dev/null
+++ b/tests/docroot/Import/OPML/Empty.1.opml
@@ -0,0 +1,4 @@
+<opml>
+    <body/>
+</opml>
+<!-- Empty body is not an error -->
diff --git a/tests/docroot/Import/OPML/Empty.2.opml b/tests/docroot/Import/OPML/Empty.2.opml
new file mode 100644
index 00000000..6dcd03f2
--- /dev/null
+++ b/tests/docroot/Import/OPML/Empty.2.opml
@@ -0,0 +1,9 @@
+<opml>
+    <head>
+        <body>
+            <outline text="I should be ignored!"/>
+        </body>
+    </head>
+    <body/>
+</opml>
+<!-- Only body in root counts -->
diff --git a/tests/docroot/Import/OPML/Empty.3.opml b/tests/docroot/Import/OPML/Empty.3.opml
new file mode 100644
index 00000000..59fd9b46
--- /dev/null
+++ b/tests/docroot/Import/OPML/Empty.3.opml
@@ -0,0 +1,11 @@
+<opml>
+    <body>
+        <someOtherElement>
+            <outline text="I should be skipped over!"/>
+        </someOtherElement>
+        <body>
+            <outline text="I should be skipped over, too!"/>
+        </body>
+    </body>
+</opml>
+<!-- The descendents of elements other than <outline> are ignored -->