better parsing of year as integer in refs pipeline

2 files changed, 8 insertions, 4 deletions

diff --git a/fatcat_scholar/schema.py b/fatcat_scholar/schema.py
index bc6b016..0fcf56e 100644
--- a/fatcat_scholar/schema.py
+++ b/fatcat_scholar/schema.py
@@ -301,9 +301,12 @@ class RefTarget(BaseModel):
 
 
 def clean_small_int(raw: Optional[str]) -> Optional[int]:
-    if not raw or not raw.isdigit():
+    if not raw or not raw.strip().isdigit():
+        return None
+    try:
+        val = int(raw.strip())
+    except ValueError:
         return None
-    val = int(raw)
     if abs(val) > 30000:
         return None
     return val
@@ -318,6 +321,7 @@ def test_clean_small_int() -> None:
     assert clean_small_int("1200003") == None
     assert clean_small_int("-123") == None
     assert clean_small_int("48844") == None
+    assert clean_small_int("1990²") == None
 
 
 def doi_split_prefix(doi: str) -> str:
diff --git a/fatcat_scholar/transform.py b/fatcat_scholar/transform.py
index 13bedb9..3a7102a 100644
--- a/fatcat_scholar/transform.py
+++ b/fatcat_scholar/transform.py
@@ -798,8 +798,8 @@ def refs_from_crossref(
             ref_container_name = series_title
 
         year = ref.get("year")
-        if year and year.isdigit():
-            year = int(year)
+        if year:
+            year = clean_small_int(year)
         else:
             year = None
         date = ref.get("date")