From f7b71ece2ae146524c927061a6c08d00a329e3b4 Mon Sep 17 00:00:00 2001
From: Alex Ioannidis <a.ioannidis@cern.ch>
Date: Sat, 16 Sep 2023 21:57:20 +0200
Subject: [PATCH] bibcode/ads: normalize unicode

* Closes #85.
---
 idutils/normalizers.py | 3 +++
 idutils/validators.py  | 2 ++
 tests/test_idutils.py  | 6 ++++++
 3 files changed, 11 insertions(+)

diff --git a/idutils/normalizers.py b/idutils/normalizers.py
index baeb3eb..a4c8b07 100644
--- a/idutils/normalizers.py
+++ b/idutils/normalizers.py
@@ -13,6 +13,8 @@
 
 """ID normalizer helper functions."""
 
+import unicodedata
+
 import isbnlib
 
 from .proxies import custom_schemes_registry
@@ -34,6 +36,7 @@ def normalize_handle(val):
 
 def normalize_ads(val):
     """Normalize an ADS bibliographic code."""
+    val = unicodedata.normalize("NFKD", val)
     m = ads_regexp.match(val)
     return m.group(2)
 
diff --git a/idutils/validators.py b/idutils/validators.py
index 364279f..df6e96f 100644
--- a/idutils/validators.py
+++ b/idutils/validators.py
@@ -14,6 +14,7 @@
 """Utility file containing ID validators."""
 
 
+import unicodedata
 from urllib.parse import urlparse
 
 from .utils import *
@@ -187,6 +188,7 @@ def is_urn(val):
 
 def is_ads(val):
     """Test if argument is an ADS bibliographic code."""
+    val = unicodedata.normalize("NFKD", val)
     return ads_regexp.match(val)
 
 
diff --git a/tests/test_idutils.py b/tests/test_idutils.py
index c5eeaa0..1da757c 100644
--- a/tests/test_idutils.py
+++ b/tests/test_idutils.py
@@ -268,6 +268,12 @@
         "2017zndo....495787v",
         "http://ui.adsabs.harvard.edu/#abs/2017zndo....495787v",
     ),
+    (
+        "1992ApJ…400L…1W",
+        ["ads"],
+        "1992ApJ...400L...1W",
+        "http://ui.adsabs.harvard.edu/#abs/1992ApJ...400L...1W",
+    ),
     (
         "0000000218250097",
         ["orcid", "isni"],