From 39bacba2803260a21dc1c50ba90695baa25193b3 Mon Sep 17 00:00:00 2001
From: Roger Braun <roger@rogerbraun.net>
Date: Sat, 24 Jun 2017 14:35:32 +0200
Subject: [PATCH] More resilient xml parsing.

---
 lib/pleroma/web/ostatus/ostatus.ex       |  1 +
 lib/pleroma/web/web_finger/web_finger.ex |  2 +-
 lib/pleroma/web/xml/xml.ex               | 17 +++++++++++++----
 3 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/lib/pleroma/web/ostatus/ostatus.ex b/lib/pleroma/web/ostatus/ostatus.ex
index bb0a3b5b9..f987752a0 100644
--- a/lib/pleroma/web/ostatus/ostatus.ex
+++ b/lib/pleroma/web/ostatus/ostatus.ex
@@ -30,6 +30,7 @@ defmodule Pleroma.Web.OStatus do
     activities = Enum.map(entries, fn (entry) ->
       {:xmlObj, :string, object_type} = :xmerl_xpath.string('string(/entry/activity:object-type[1])', entry)
       {:xmlObj, :string, verb} = :xmerl_xpath.string('string(/entry/activity:verb[1])', entry)
+      Logger.debug("Handling #{verb}")
 
       case verb do
         'http://activitystrea.ms/schema/1.0/follow' ->
diff --git a/lib/pleroma/web/web_finger/web_finger.ex b/lib/pleroma/web/web_finger/web_finger.ex
index e8b738c96..7ae413c26 100644
--- a/lib/pleroma/web/web_finger/web_finger.ex
+++ b/lib/pleroma/web/web_finger/web_finger.ex
@@ -98,7 +98,7 @@ defmodule Pleroma.Web.WebFinger do
                end
 
     with {:ok, %{status_code: status_code, body: body}} when status_code in 200..299 <- response,
-         doc <- XML.parse_document(body),
+         doc when doc != :error<- XML.parse_document(body),
          {:ok, data} <- webfinger_from_xml(doc) do
       {:ok, data}
     else
diff --git a/lib/pleroma/web/xml/xml.ex b/lib/pleroma/web/xml/xml.ex
index 22faf72df..63580c1f8 100644
--- a/lib/pleroma/web/xml/xml.ex
+++ b/lib/pleroma/web/xml/xml.ex
@@ -1,4 +1,7 @@
 defmodule Pleroma.Web.XML do
+  require Logger
+
+  def string_from_xpath(xpath, :error), do: nil
   def string_from_xpath(xpath, doc) do
     {:xmlObj, :string, res} = :xmerl_xpath.string('string(#{xpath})', doc)
 
@@ -10,10 +13,16 @@ defmodule Pleroma.Web.XML do
   end
 
   def parse_document(text) do
-    {doc, _rest} = text
-    |> :binary.bin_to_list
-    |> :xmerl_scan.string
+    try do
+      {doc, _rest} = text
+      |> :binary.bin_to_list
+      |> :xmerl_scan.string
 
-    doc
+      doc
+    catch
+      :exit, error ->
+        Logger.debug("Couldn't parse xml: #{inspect(text)}")
+        :error
+    end
   end
 end