From 2aaab566d7aa9e27625811e5a356254a9f29c036 Mon Sep 17 00:00:00 2001
From: Kasia Kozlowska <36536946+KasiaKoz@users.noreply.github.com>
Date: Fri, 2 Feb 2024 13:33:55 +0000
Subject: [PATCH] Fix reading java array type (#216)

* fix reading java array type

* update CHANGELOG.md
---
 CHANGELOG.md                      |  1 +
 genet/input/matsim_reader.py      |  2 +-
 tests/test_input_matsim_reader.py | 22 ++++++++++++++++++++++
 3 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 49546920..243dbb5f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,7 @@
 * Fixed summary report:
   * Intermodal Access/Egress reporting is more general (not expecting just car and bike mode access to PT) [#204](https://github.com/arup-group/genet/pull/204)
   * Node/Links numbers were reported incorrectly (switched) [#207](https://github.com/arup-group/genet/pull/207)
+* Fixed reading `java.lang.Array` types in MATSim xml files [#216](https://github.com/arup-group/genet/pull/216)
 
 ### Changed
 * GeNet's pre-baked python scripts have been retired in favour of CLI [#194](https://github.com/arup-group/genet/pull/194)
diff --git a/genet/input/matsim_reader.py b/genet/input/matsim_reader.py
index 667880c8..11948d18 100644
--- a/genet/input/matsim_reader.py
+++ b/genet/input/matsim_reader.py
@@ -138,7 +138,7 @@ def _read_additional_attrib_text(elem):
             f"Elem {elem.attrib['name']} is being read as None. Defaulting value to empty string."
         )
     elif ("," in elem.text) and elem.attrib["name"] != "geometry":
-        t = set(elem.text.split(","))
+        t = set(elem.text.strip("{}").replace(("'"), "").replace(" ", "").split(","))
     else:
         t = elem.text
     return t
diff --git a/tests/test_input_matsim_reader.py b/tests/test_input_matsim_reader.py
index 865d186a..7dae80d2 100644
--- a/tests/test_input_matsim_reader.py
+++ b/tests/test_input_matsim_reader.py
@@ -1038,3 +1038,25 @@ def make_network_with_elevations_xml_string(nodes_with_elevations_dict, link_dic
         network_xml += " </attributes> </link>"
     network_xml += " </links> </network>"
     return network_xml
+
+
+class ElemWithText:
+    def __init__(self, text):
+        self.text = text
+        self.attrib = {"name": "doesn't matter"}
+
+
+@pytest.mark.parametrize(
+    "case,input_elem_text,expected_output",
+    [
+        ("is_just_a_string", "hello", "hello"),
+        ("is_a_java_array", "{'hello', 'yes'}", {"hello", "yes"}),
+        ("is_none", None, ""),
+    ],
+)
+def test_read_additional_attrib_text(case, input_elem_text, expected_output):
+    elem = ElemWithText(input_elem_text)
+    output = matsim_reader._read_additional_attrib_text(elem=elem)
+    assert output == expected_output, AssertionError(
+        f"Elem for case {case} did not produce the expected result"
+    )