RustPython · Copilot · Mar 27, 2026 · Mar 27, 2026 · Mar 27, 2026 · Mar 29, 2026
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/crates/sre_engine/Cargo.toml b/crates/sre_engine/Cargo.toml
@@ -16,6 +16,7 @@ harness = false
 
 [dependencies]
 rustpython-wtf8 = { workspace = true }
+unic-ucd-category = { workspace = true }
 num_enum = { workspace = true }
 bitflags = { workspace = true }
 optional = { workspace = true }

diff --git a/crates/sre_engine/src/string.rs b/crates/sre_engine/src/string.rs
@@ -1,4 +1,5 @@
 use rustpython_wtf8::Wtf8;
+use unic_ucd_category::GeneralCategory;
 
 #[derive(Debug, Clone, Copy)]
 pub struct StringCursor {
@@ -441,9 +442,19 @@ pub(crate) const fn is_uni_linebreak(ch: u32) -> bool {
 }
 #[inline]
 pub(crate) fn is_uni_alnum(ch: u32) -> bool {
-    // TODO: check with cpython
     char::try_from(ch)
-        .map(|x| x.is_alphanumeric())
+        .map(|c| match GeneralCategory::of(c) {
+            GeneralCategory::UppercaseLetter
+            | GeneralCategory::LowercaseLetter
+            | GeneralCategory::TitlecaseLetter
+            | GeneralCategory::ModifierLetter
+            | GeneralCategory::OtherLetter
+            | GeneralCategory::DecimalNumber
+            | GeneralCategory::LetterNumber
+            | GeneralCategory::OtherNumber => true,
+            GeneralCategory::Unassigned => c.is_alphanumeric(),
+            _ => false,
+        })
         .unwrap_or(false)
 }
 #[inline]

diff --git a/crates/vm/src/builtins/str.rs b/crates/vm/src/builtins/str.rs
@@ -945,7 +945,19 @@ impl PyStr {
 
     #[pymethod]
     fn isalnum(&self) -> bool {
-        !self.data.is_empty() && self.char_all(char::is_alphanumeric)
+        !self.data.is_empty()
+            && self.char_all(|c| match GeneralCategory::of(c) {
+                GeneralCategory::UppercaseLetter
+                | GeneralCategory::LowercaseLetter
+                | GeneralCategory::TitlecaseLetter
+                | GeneralCategory::ModifierLetter
+                | GeneralCategory::OtherLetter
+                | GeneralCategory::DecimalNumber
+                | GeneralCategory::LetterNumber
+                | GeneralCategory::OtherNumber => true,
+                GeneralCategory::Unassigned => c.is_alphanumeric(),
+                _ => false,
+            })
     }
 
     #[pymethod]
@@ -1056,7 +1068,16 @@ impl PyStr {
 
     #[pymethod]
     fn isalpha(&self) -> bool {
-        !self.data.is_empty() && self.char_all(char::is_alphabetic)
+        !self.data.is_empty()
+            && self.char_all(|c| match GeneralCategory::of(c) {
+                GeneralCategory::UppercaseLetter
+                | GeneralCategory::LowercaseLetter
+                | GeneralCategory::TitlecaseLetter
+                | GeneralCategory::ModifierLetter
+                | GeneralCategory::OtherLetter => true,
+                GeneralCategory::Unassigned => c.is_alphabetic(),
+                _ => false,
+            })
     }
 
     #[pymethod]

diff --git a/extra_tests/snippets/builtin_str.py b/extra_tests/snippets/builtin_str.py
@@ -839,3 +839,13 @@ class MyString(str):
 assert id(b) != id(b * 1)
 assert id(b) != id(1 * b)
 assert id(b) != id(b * 2)
+
+
+# Regression tests for isalpha/isalnum Unicode General Category correctness.
+# These characters are in letter categories (Ll/Lo) and should return True,
+# but were missed in older Unicode tables used by unic-ucd-category.
+# See: https://github.com/RustPython/RustPython/pull/7520#issuecomment-4148322294
+for _cp in [1376, 1416, 1519, 2160, 2161, 2162, 2163, 2164, 2165, 2166]:
+    _c = chr(_cp)
+    assert _c.isalpha(), f"U+{_cp:04X} should be isalpha"
+    assert _c.isalnum(), f"U+{_cp:04X} should be isalnum"
diff --git a/extra_tests/snippets/builtin_str_unicode.py b/extra_tests/snippets/builtin_str_unicode.py
@@ -11,6 +11,7 @@
 
 assert c == "👋👋👋"
 
+import re
 import unicodedata
 
 assert unicodedata.category("a") == "Ll"
@@ -38,3 +39,10 @@
 # TODO: add east_asian_width and mirrored
 # assert unicodedata.ucd_3_2_0.east_asian_width('\u231a') == 'N'
 # assert not unicodedata.ucd_3_2_0.mirrored("\u0f3a")
+
+# U+0345 COMBINING GREEK YPOGEGRAMMENI (category Mn) should not be alphanumeric.
+# CPython's isalpha/isalnum use Unicode letter categories (Lu/Ll/Lt/Lm/Lo),
+# not the broader Unicode Alphabetic derived property.
+assert not "\u0345".isalpha(), "isalpha should not match Mn category characters"
+assert not "\u0345".isalnum(), "isalnum should not match Mn category characters"
+assert not re.match(r"\w", "\u0345"), r"\w should not match U+0345 (category Mn)"