From 31fd13fe37ae48a2e48aca35334ff886f7bdd639 Mon Sep 17 00:00:00 2001
From: Arber Shabhasa <ashabhasa@gmail.com>
Date: Wed, 3 Apr 2024 17:16:15 +0200
Subject: [PATCH] Naive saolution to xml with closing tags containing
 whitespaces

---
 lib/saxy/parser/builder.ex |  5 ++++-
 test/saxy_test.exs         | 18 ++++++++++++++++--
 2 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/lib/saxy/parser/builder.ex b/lib/saxy/parser/builder.ex
index d5c205a..2ebb5fd 100644
--- a/lib/saxy/parser/builder.ex
+++ b/lib/saxy/parser/builder.ex
@@ -1184,7 +1184,7 @@ defmodule Saxy.Parser.Builder do
         lookahead buffer, @streaming do
           ">" <> rest ->
             [open_tag | stack] = state.stack
-            ending_tag = binary_part(original, pos, len)
+            ending_tag = binary_part(original, pos, len) |> String.trim()
             pos = pos + len + 1
 
             if open_tag == ending_tag do
@@ -1207,6 +1207,9 @@ defmodule Saxy.Parser.Builder do
           char <> rest when is_ascii_name_char(char) ->
             close_tag_name(rest, more?, original, pos, state, len + 1)
 
+          char <> rest when is_whitespace(char) ->
+            close_tag_name(rest, more?, original, pos, state, len + 1)
+
           token in unquote(utf8_binaries()) when more? ->
             halt!(close_tag_name(token, more?, original, pos, state, len))
 
diff --git a/test/saxy_test.exs b/test/saxy_test.exs
index e24d1d6..b579494 100644
--- a/test/saxy_test.exs
+++ b/test/saxy_test.exs
@@ -31,11 +31,11 @@ defmodule SaxyTest do
     for fixture <- @fixtures do
       stream = stream_fixture(fixture)
       element_stream = Saxy.stream_events(stream)
-      assert [_ | _] = Enum.to_list element_stream
+      assert [_ | _] = Enum.to_list(element_stream)
     end
 
     assert_raise Saxy.ParseError, fn ->
-      Enum.to_list Saxy.stream_events stream_fixture "incorrect.xml"
+      Enum.to_list(Saxy.stream_events(stream_fixture("incorrect.xml")))
     end
   end
 
@@ -73,6 +73,20 @@ defmodule SaxyTest do
            ]
   end
 
+  test "parse_string/4 parses XML binary with closing tags containing whitespaces" do
+    data = "<foo>Some Data</foo   >"
+
+    assert {:ok, state} = parse(data, StackHandler, [], expand_entity: :keep)
+
+    assert state == [
+             end_document: {},
+             end_element: "foo",
+             characters: "Some Data",
+             start_element: {"foo", []},
+             start_document: []
+           ]
+  end
+
   test "handles trailing Unicode codepoints during streaming" do
     data = "<foo>𠜎𠜱𠝹𠱓</foo>"
     stream = for byte <- :binary.bin_to_list(data), do: <<byte>>