From a36dcc9535d701cf2d98848e5692cbc9a53c23bf Mon Sep 17 00:00:00 2001
From: Shadowfacts <me@shadowfacts.net>
Date: Mon, 11 Apr 2022 16:25:08 -0400
Subject: [PATCH] Handle errors when parsing XML

---
 lib/parser/atom.ex | 19 +++++++++++++------
 lib/parser/rss2.ex | 19 +++++++++++++------
 lib/xml.ex         | 12 ++++++++----
 3 files changed, 34 insertions(+), 16 deletions(-)

diff --git a/lib/parser/atom.ex b/lib/parser/atom.ex
index 16d61d3..a95d5ae 100644
--- a/lib/parser/atom.ex
+++ b/lib/parser/atom.ex
@@ -12,15 +12,22 @@ defmodule FeedParser.Parser.Atom do
   def accepts(data, content_type) do
     case content_type do
       "application/atom+xml" ->
-        {true, XML.parse(data)}
+        case XML.parse(data) do
+          {:error, _} -> false
+          {:ok, doc} -> {true, doc}
+        end
 
       _ when content_type in ["text/xml", "application/xml"] ->
-        doc = XML.parse(data)
+        case XML.parse(data) do
+          {:error, _} ->
+            false
 
-        if XML.xmlElement(doc, :name) == :feed do
-          {true, doc}
-        else
-          false
+          {:ok, doc} ->
+            if XML.xmlElement(doc, :name) == :feed do
+              {true, doc}
+            else
+              false
+            end
         end
 
       _ ->
diff --git a/lib/parser/rss2.ex b/lib/parser/rss2.ex
index 8f3cbc6..5650682 100644
--- a/lib/parser/rss2.ex
+++ b/lib/parser/rss2.ex
@@ -12,15 +12,22 @@ defmodule FeedParser.Parser.RSS2 do
   def accepts(data, content_type) do
     cond do
       content_type in ["application/rss+xml", "text/rss+xml"] ->
-        {true, XML.parse(data)}
+        case XML.parse(data) do
+          {:error, _} -> false
+          {:ok, doc} -> {true, doc}
+        end
 
       content_type in ["text/xml", "application/xml"] ->
-        doc = XML.parse(data)
+        case XML.parse(data) do
+          {:error, _} ->
+            false
 
-        if XML.xmlElement(doc, :name) == :rss do
-          {true, doc}
-        else
-          false
+          {:ok, doc} ->
+            if XML.xmlElement(doc, :name) == :rss do
+              {true, doc}
+            else
+              false
+            end
         end
 
       true ->
diff --git a/lib/xml.ex b/lib/xml.ex
index de8d397..c61fc71 100644
--- a/lib/xml.ex
+++ b/lib/xml.ex
@@ -9,12 +9,16 @@ defmodule FeedParser.XML do
   defrecord :xmlAttribute, extract(:xmlAttribute, from_lib: "xmerl/include/xmerl.hrl")
   defrecord :xmlText, extract(:xmlText, from_lib: "xmerl/include/xmerl.hrl")
 
-  @spec parse(data :: String.t()) :: tuple()
+  @spec parse(data :: String.t()) :: {:ok, tuple()} | {:error, String.t()}
   def parse(data) do
     {doc, _} =
-      data
-      |> :binary.bin_to_list()
-      |> :xmerl_scan.string()
+      try do
+        data
+        |> :binary.bin_to_list()
+        |> :xmerl_scan.string()
+      catch
+        :exit, reason -> {:error, "parsing XML failed: #{inspect(reason)}"}
+      end
 
     doc
   end