fix: recognize valid PMID and legacy arXiv identifiers

tests/test_identifier.py

@@ -12,10 +12,30 @@
     [
         ("10.1063/5.0155600", Identifier.DOI),
         ("37526163", Identifier.PMID),
+        ("1234567", Identifier.PMID),
+        ("1", Identifier.PMID),
         ("2304.09409", Identifier.ARXIV),
+        ("2304.09409v2", Identifier.ARXIV),
+        ("hep-th/9901001", Identifier.ARXIV),
+        ("hep-th/9901001v2", Identifier.ARXIV),
+        ("math.GT/0309136", Identifier.ARXIV),
         ("Deep residual learning for image recognition", Identifier.TITLE),
     ],
 )
 def test_get_identifier_type(identifier, expected):
     """Test get_identifier_type()."""
     assert get_identifier_type(identifier) == expected
+
+
+@pytest.mark.parametrize(
+    "identifier",
+    [
+        "10.1063/5.0155600 trailing",
+        "2304.09409 trailing",
+        "hep-th/9901001 trailing",
+        "123456789",
+    ],
+)
+def test_get_identifier_type_rejects_malformed_identifiers(identifier):
+    """Test malformed identifier-like strings are not accepted by prefix."""
+    assert get_identifier_type(identifier) is None

wenxian/identifier.py

@@ -17,16 +17,19 @@ class Identifier(Enum):
 
 REGEX: dict[Identifier, re.Pattern] = {
     Identifier.DOI: re.compile(r"10\.\d{4,9}/[-._;()/:A-Z0-9]+", re.IGNORECASE),
-    Identifier.PMID: re.compile(r"\b\d{8}\b"),
-    Identifier.ARXIV: re.compile(r"\d{4}\.\d{4,5}(v\d+)?"),
+    Identifier.PMID: re.compile(r"\d{1,8}"),
+    Identifier.ARXIV: re.compile(
+        r"(?:\d{4}\.\d{4,5}|[A-Z][A-Z0-9.-]*/\d{7})(?:v\d+)?",
+        re.IGNORECASE,
+    ),
 }
 """Regex for identifiers."""
 
 
 def get_identifier_type(identifier: str) -> Identifier | None:
     """Get the type of an identifier."""
     for id_type, regex in REGEX.items():
-        if regex.match(identifier):
+        if regex.fullmatch(identifier):
             return id_type
     # Check if it could be a reasonable paper title
     # Title should have at least 3 words and be at least 10 characters