[mypyc] Add librt.strings.isalnum codepoint primitive (#21509)

3rd PR for #21418, mirroring
`librt.strings.isdigit`.

Measured on a microbenchmark this is roughly 30-40% faster for a char

Author: VaggelisD

mypy/typeshed/stubs/librt/librt/strings.pyi

@@ -45,3 +45,4 @@ def read_f64_be(b: bytes, index: i64, /) -> float: ...
 # obtained via ord(s[i])). Negative inputs return False.
 def isspace(c: i32, /) -> bool: ...
 def isdigit(c: i32, /) -> bool: ...
+def isalnum(c: i32, /) -> bool: ...

mypyc/lib-rt/codepoint_extra_ops.h

@@ -17,4 +17,8 @@ static inline bool LibRTStrings_IsDigit(int32_t c) {
     return c >= 0 && Py_UNICODE_ISDIGIT((Py_UCS4)c);
 }
 
+static inline bool LibRTStrings_IsAlnum(int32_t c) {
+    return c >= 0 && Py_UNICODE_ISALNUM((Py_UCS4)c);
+}
+
 #endif  // MYPYC_CODEPOINT_EXTRA_OPS_H

mypyc/lib-rt/strings/librt_strings.c

@@ -1192,6 +1192,7 @@ cp_parse_i32(PyObject *arg, int32_t *out) {
 
 DEFINE_CP_BOOL_WRAPPER(isspace, LibRTStrings_IsSpace)
 DEFINE_CP_BOOL_WRAPPER(isdigit, LibRTStrings_IsDigit)
+DEFINE_CP_BOOL_WRAPPER(isalnum, LibRTStrings_IsAlnum)
 
 static PyMethodDef librt_strings_module_methods[] = {
     {"write_i16_le", (PyCFunction) write_i16_le, METH_FASTCALL,
@@ -1260,6 +1261,9 @@ static PyMethodDef librt_strings_module_methods[] = {
     {"isdigit", cp_isdigit, METH_O,
      PyDoc_STR("Test whether a codepoint (i32) is a Unicode digit.")
     },
+    {"isalnum", cp_isalnum, METH_O,
+     PyDoc_STR("Test whether a codepoint (i32) is alphanumeric.")
+    },
     {NULL, NULL, 0, NULL}
 };
 

mypyc/primitives/librt_strings_ops.py

@@ -413,3 +413,12 @@
     error_kind=ERR_NEVER,
     dependencies=[LIBRT_STRINGS, CODEPOINT_EXTRA_OPS],
 )
+
+function_op(
+    name="librt.strings.isalnum",
+    arg_types=[int32_rprimitive],
+    return_type=bool_rprimitive,
+    c_function_name="LibRTStrings_IsAlnum",
+    error_kind=ERR_NEVER,
+    dependencies=[LIBRT_STRINGS, CODEPOINT_EXTRA_OPS],
+)

mypyc/test-data/irbuild-librt-strings.test

@@ -359,3 +359,17 @@ def is_d(c):
 L0:
     r0 = LibRTStrings_IsDigit(c)
     return r0
+
+[case testLibrtStringsIsAlnumIR]
+from librt.strings import isalnum
+from mypy_extensions import i32
+
+def is_an(c: i32) -> bool:
+    return isalnum(c)
+[out]
+def is_an(c):
+    c :: i32
+    r0 :: bool
+L0:
+    r0 = LibRTStrings_IsAlnum(c)
+    return r0

mypyc/test-data/run-librt-strings.test

@@ -1443,14 +1443,17 @@ def test_new_without_init_is_usable() -> None:
 [case testLibrtStringsCodepointClassifiers_librt]
 from typing import Any
 from mypy_extensions import i32
-from librt.strings import isspace, isdigit
+from librt.strings import isspace, isdigit, isalnum
+
+from testutil import assertRaises
 
 
 def test_codepoint_classifiers() -> None:
     # Negative values are not codepoints.
     for bad in (i32(-1), i32(-113)):
         assert not isspace(bad)
         assert not isdigit(bad)
+        assert not isalnum(bad)
     # Verify each codepoint primitive agrees with the matching str method
     # across all Unicode codepoints, including the ord(chr(i)) round-trip.
     # Any forces generic dispatch on the str side.
@@ -1460,3 +1463,24 @@ def test_codepoint_classifiers() -> None:
         o = ord(c)
         assert isspace(o) == isspace(i) == a.isspace()
         assert isdigit(o) == isdigit(i) == a.isdigit()
+        assert isalnum(o) == isalnum(i) == a.isalnum()
+
+
+def test_codepoint_classifiers_via_any() -> None:
+    # Iterate so the callee is opaque to mypyc and dispatch falls back to
+    # the PyMethodDef wrapper, exercising the i32 range check.
+    for fn, true_input, false_input in (
+        (isspace, " ", "a"),
+        (isdigit, "5", "a"),
+        (isalnum, "A", " "),
+    ):
+        f: Any = fn
+        assert f(ord(true_input)) is True
+        assert f(ord(false_input)) is False
+        # Negative values are valid i32, just not codepoints.
+        assert f(-1) is False
+        # Inputs outside i32 range raise OverflowError through the wrapper.
+        with assertRaises(OverflowError, "codepoint out of i32 range"):
+            f(1 << 40)
+        with assertRaises(OverflowError, "codepoint out of i32 range"):
+            f(-(1 << 40))