bibcode/ads: normalize unicode

* Closes #85.
inveniosoftware · Oct 29, 2024 · 26318e4 · 26318e4
1 parent 812f640
commit 26318e4
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 0 deletions.
diff --git 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
@@ -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
@@ -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"],