[mypyc] Add librt.strings.isdigit codepoint primitive

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

@@ -44,3 +44,4 @@ def read_f64_be(b: bytes, index: i64, /) -> float: ...
 # Codepoint classification helpers operating on i32 codepoints (typically
 # obtained via ord(s[i])). Negative inputs return False.
 def isspace(c: i32, /) -> bool: ...
+def isdigit(c: i32, /) -> bool: ...

mypyc/lib-rt/codepoint_extra_ops.h

@@ -13,4 +13,8 @@ static inline bool LibRTStrings_IsSpace(int32_t c) {
     return c >= 0 && Py_UNICODE_ISSPACE((Py_UCS4)c);
 }
 
+static inline bool LibRTStrings_IsDigit(int32_t c) {
+    return c >= 0 && Py_UNICODE_ISDIGIT((Py_UCS4)c);
+}
+
 #endif  // MYPYC_CODEPOINT_EXTRA_OPS_H

mypyc/lib-rt/strings/librt_strings.c

@@ -1191,6 +1191,7 @@ cp_parse_i32(PyObject *arg, int32_t *out) {
     }
 
 DEFINE_CP_BOOL_WRAPPER(isspace, LibRTStrings_IsSpace)
+DEFINE_CP_BOOL_WRAPPER(isdigit, LibRTStrings_IsDigit)
 
 static PyMethodDef librt_strings_module_methods[] = {
     {"write_i16_le", (PyCFunction) write_i16_le, METH_FASTCALL,
@@ -1256,6 +1257,9 @@ static PyMethodDef librt_strings_module_methods[] = {
     {"isspace", cp_isspace, METH_O,
      PyDoc_STR("Test whether a codepoint (i32) is Unicode whitespace.")
     },
+    {"isdigit", cp_isdigit, METH_O,
+     PyDoc_STR("Test whether a codepoint (i32) is a Unicode digit.")
+    },
     {NULL, NULL, 0, NULL}
 };
 

mypyc/primitives/librt_strings_ops.py

@@ -404,3 +404,12 @@
     error_kind=ERR_NEVER,
     dependencies=[LIBRT_STRINGS, CODEPOINT_EXTRA_OPS],
 )
+
+function_op(
+    name="librt.strings.isdigit",
+    arg_types=[int32_rprimitive],
+    return_type=bool_rprimitive,
+    c_function_name="LibRTStrings_IsDigit",
+    error_kind=ERR_NEVER,
+    dependencies=[LIBRT_STRINGS, CODEPOINT_EXTRA_OPS],
+)

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

@@ -345,3 +345,17 @@ L8:
 L9:
     r16 = LibRTStrings_IsSpace(r15)
     return r16
+
+[case testLibrtStringsIsDigitIR]
+from librt.strings import isdigit
+from mypy_extensions import i32
+
+def is_d(c: i32) -> bool:
+    return isdigit(c)
+[out]
+def is_d(c):
+    c :: i32
+    r0 :: bool
+L0:
+    r0 = LibRTStrings_IsDigit(c)
+    return r0

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

@@ -1440,19 +1440,23 @@ def test_new_without_init_is_usable() -> None:
     sw.write("hello")
     assert sw.getvalue() == "hello"
 
-[case testLibrtStringsIsSpace_librt]
+[case testLibrtStringsCodepointClassifiers_librt]
 from typing import Any
 from mypy_extensions import i32
-from librt.strings import isspace
+from librt.strings import isspace, isdigit
 
 
-def test_isspace() -> None:
-    assert not isspace(i32(-1))
-    assert not isspace(i32(-113))
-    # Verify our codepoint primitive agrees with str.isspace() across all
-    # Unicode codepoints, including the ord(chr(i)) round-trip. Any
-    # forces generic dispatch on the str side.
+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)
+    # 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.
     for i in range(0x110000):
         c = chr(i)
         a: Any = c
-        assert isspace(ord(c)) == isspace(i) == a.isspace()
+        o = ord(c)
+        assert isspace(o) == isspace(i) == a.isspace()
+        assert isdigit(o) == isdigit(i) == a.isdigit()