From 1f07377cae026cd3c069bdd23001dda5daaf7ece Mon Sep 17 00:00:00 2001
From: Marcel Otto <marcelotto@gmx.de>
Date: Sat, 6 Jul 2019 01:59:03 +0200
Subject: [PATCH] Fix unicode escaping issue in RDF.Literal.matches

---
 lib/rdf/literal.ex         | 23 ++++++++++++++++++++---
 test/unit/literal_test.exs |  5 +++++
 2 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/lib/rdf/literal.ex b/lib/rdf/literal.ex
index 6e03a93..2b1a1af 100644
--- a/lib/rdf/literal.ex
+++ b/lib/rdf/literal.ex
@@ -372,14 +372,31 @@ defmodule RDF.Literal do
   defp xpath_regex_pattern(pattern, flags) do
     with {:ok, regex} <-
            pattern
-           |> convert_utf16_escaping()
+           |> convert_utf_escaping()
            |> Regex.compile(xpath_regex_flags(flags)) do
       {:regex, regex}
     end
   end
 
-  defp convert_utf16_escaping(pattern) do
-    String.replace(pattern, ~r/\\U(([0-9]|[A-F]|[a-f]){2})(([0-9]|[A-F]|[a-f]){6})/, "\\u{\\3}")
+  @doc false
+  def convert_utf_escaping(string) do
+    require Integer
+
+    xpath_unicode_regex = ~r/(\\*)\\U([0-9]|[A-F]|[a-f]){2}(([0-9]|[A-F]|[a-f]){6})/
+    [first | possible_matches] =
+      Regex.split(xpath_unicode_regex, string, include_captures: true)
+
+    [first |
+      Enum.map_every(possible_matches, 2, fn possible_xpath_unicode ->
+        [_, escapes, _, codepoint, _] = Regex.run(xpath_unicode_regex, possible_xpath_unicode)
+        if escapes |> String.length() |> Integer.is_odd() do
+          "#{escapes}\\u{#{codepoint}}"
+        else
+          "\\" <> possible_xpath_unicode
+        end
+      end)
+    ]
+    |> Enum.join()
   end
 
   defp xpath_regex_flags(flags) do
diff --git a/test/unit/literal_test.exs b/test/unit/literal_test.exs
index a5b69e5..e9eae1d 100644
--- a/test/unit/literal_test.exs
+++ b/test/unit/literal_test.exs
@@ -252,7 +252,12 @@ defmodule RDF.LiteralTest do
         {~L"abracadabra", ~L"^bra",   false},
         {@poem, ~L"Kaum.*krähen",     false},
         {@poem, ~L"^Kaum.*gesehen,$", false},
+        {~L"foobar", ~L"foo$",        false},
+
+        {~L"noe\u0308l", ~L"noe\\u0308l", true},
+        {~L"noe\\u0308l", ~L"noe\\\\u0308l", true},
         {~L"\u{01D4B8}", ~L"\\U0001D4B8", true},
+        {~L"\\U0001D4B8", ~L"\\\U0001D4B8", true},
 
         {~L"abracadabra"en, ~L"bra", true},
         {"abracadabra", "bra",       true},