Merge pull request #264 from harfbuzz/py-limited-api

enable Py_LIMITED_API to strip down build matrix to one abi3 wheel per platform

Author: anthrotype

pyproject.toml

@@ -1,22 +1,20 @@
 [build-system]
 requires = [
-    # pin setuptools on pypy to workaround this bug: https://github.com/pypa/distutils/issues/283
-    "setuptools >= 36.4, < 72.2; platform_python_implementation == 'PyPy'",
-    "setuptools >= 36.4; platform_python_implementation != 'PyPy'",
+    "setuptools >= 36.4",
     "wheel",
     "setuptools_scm >= 2.1",
-    "cython >= 0.28.1",
+    "cython >= 3.1",
     "pkgconfig"
 ]
 build-backend = "setuptools.build_meta"
 
 [tool.setuptools_scm]
 
 [tool.cibuildwheel]
-# Skip building PyPy 3.8 wheels, the build currently fails
 # Also skip wheels for free-threaded CPython 3.14 (e.g. 'cp314t') until we
 # actually support them. We currently aren't thread-safe as we define HB_NO_MT=1.
-skip = ["pp38-*", "cp3??t-*"]
+# CPython abi3 wheels only need universal2 on macOS, skip the single-arch builds.
+skip = ["cp38-*", "cp39-*", "cp3??t-*", "cp*-macosx_x86_64", "cp*-macosx_arm64"]
 enable = ["pypy"]
 test-requires = "pytest"
 test-command = "pytest {project}/tests"
@@ -27,6 +25,21 @@ archs = ["x86_64", "universal2", "arm64"]
 [tool.cibuildwheel.linux]
 archs = ["native"]
 
+# Run abi3audit after the default repair commands to scan for abi3 violations
+# https://github.com/pypa/abi3audit
+# Only on Unix platforms (Linux/macOS) since Windows has no default repair command
+[[tool.cibuildwheel.overrides]]
+select = "cp*-{linux,macosx}*"
+inherit.repair-wheel-command = "append"
+repair-wheel-command = "pipx run abi3audit --strict --report {wheel}"
+
+# Disable Limited API for PyPy as the build currently fails, see:
+# https://github.com/harfbuzz/uharfbuzz/issues/262#issuecomment-3415144557
+# https://cibuildwheel.pypa.io/en/stable/configuration/#overrides
+[[tool.cibuildwheel.overrides]]
+select = "pp*"
+environment = { USE_PY_LIMITED_API = "0" }
+
 
 [tool.black]
 extend-exclude = "harfbuzz"

requirements-dev.txt

@@ -1,4 +1,4 @@
-cython>=0.29.1
+cython>=3.1
 
 # we need wheel >= 0.31.0 to support Markdown long_description
 wheel>=0.31

setup.cfg

@@ -1,5 +1,2 @@
-[sdist]
-formats = zip
-
 [metadata]
 license_file = LICENSE

setup.py

@@ -10,10 +10,10 @@
 from setuptools import Extension, setup
 
 
-def bool_from_environ(key: str):
+def bool_from_environ(key: str, default: bool = False):
     value = os.environ.get(key)
     if not value:
-        return False
+        return default
     if value == "1":
         return True
     if value == "0":
@@ -32,6 +32,12 @@ def bool_from_environ(key: str):
 use_system_libraries = bool_from_environ("USE_SYSTEM_LIBS")
 use_cython_linetrace = bool_from_environ("CYTHON_LINETRACE")
 use_cython_annotate = bool_from_environ("CYTHON_ANNOTATE")
+# Python Limited API for stable ABI support is enabled by default.
+# Set USE_PY_LIMITED_API=0 to turn it off.
+# https://docs.python.org/3.14/c-api/stable.html#limited-c-api
+use_py_limited_api = bool_from_environ("USE_PY_LIMITED_API", default=True)
+# NOTE: this must be kept in sync with python_requires='>=3.10' below
+limited_api_min_version = "0x030A0000"
 
 
 def _configure_extensions_with_system_libs() -> List[Extension]:
@@ -51,6 +57,9 @@ def _configure_extensions_with_system_libs() -> List[Extension]:
     if use_cython_linetrace:
         define_macros.append(("CYTHON_TRACE_NOGIL", "1"))
 
+    if use_py_limited_api:
+        define_macros.append(("Py_LIMITED_API", limited_api_min_version))
+
     extension = Extension(
         "uharfbuzz._harfbuzz",
         define_macros=define_macros,
@@ -61,6 +70,7 @@ def _configure_extensions_with_system_libs() -> List[Extension]:
         language="c++",
         libraries=libraries,
         library_dirs=library_dirs,
+        py_limited_api=use_py_limited_api,
     )
 
     extension_test = Extension(
@@ -74,6 +84,7 @@ def _configure_extensions_with_system_libs() -> List[Extension]:
         language="c++",
         libraries=libraries,
         library_dirs=library_dirs,
+        py_limited_api=use_py_limited_api,
     )
 
     return [extension, extension_test]
@@ -87,6 +98,9 @@ def _configure_extensions_with_vendored_libs() -> List[Extension]:
     if use_cython_linetrace:
         define_macros.append(("CYTHON_TRACE_NOGIL", "1"))
 
+    if use_py_limited_api:
+        define_macros.append(("Py_LIMITED_API", limited_api_min_version))
+
     extra_compile_args = []
     extra_link_args = []
     libraries = []
@@ -124,6 +138,7 @@ def _configure_extensions_with_vendored_libs() -> List[Extension]:
         libraries=libraries,
         extra_compile_args=extra_compile_args,
         extra_link_args=extra_link_args,
+        py_limited_api=use_py_limited_api,
     )
 
     extension_test = Extension(
@@ -138,6 +153,7 @@ def _configure_extensions_with_vendored_libs() -> List[Extension]:
         libraries=libraries,
         extra_compile_args=extra_compile_args,
         extra_link_args=extra_link_args,
+        py_limited_api=use_py_limited_api,
     )
 
     return [extension, extension_test]
@@ -164,10 +180,11 @@ def configure_extensions() -> List[Extension]:
     packages=["uharfbuzz"],
     zip_safe=False,
     setup_requires=["setuptools_scm"],
-    python_requires=">=3.8",
+    python_requires=">=3.10",
     ext_modules=cythonize(
         configure_extensions(),
         annotate=use_cython_annotate,
         compiler_directives={"linetrace": use_cython_linetrace},
     ),
+    options={"bdist_wheel": {"py_limited_api": "cp310"}} if use_py_limited_api else {},
 )

src/uharfbuzz/_harfbuzz.pyx

@@ -7,15 +7,16 @@ from .charfbuzz cimport *
 from libc.stdlib cimport free, malloc, calloc
 from libc.string cimport const_char
 from cpython.pycapsule cimport PyCapsule_GetPointer, PyCapsule_IsValid
-from cpython.unicode cimport (
-    PyUnicode_1BYTE_DATA, PyUnicode_2BYTE_DATA, PyUnicode_4BYTE_DATA,
-    PyUnicode_1BYTE_KIND, PyUnicode_2BYTE_KIND, PyUnicode_4BYTE_KIND,
-    PyUnicode_KIND, PyUnicode_GET_LENGTH, PyUnicode_FromKindAndData
-)
+from cpython.unicode cimport PyUnicode_GetLength, PyUnicode_AsUCS4Copy
+from cpython.mem cimport PyMem_Free
 from typing import Callable, Dict, List, Sequence, Tuple, Union, NamedTuple
 from pathlib import Path
 from functools import wraps
 
+# Declare Limited API types and functions (Python 3.3+)
+cdef extern from "Python.h":
+    ctypedef uint32_t Py_UCS4
+
 
 DEF STATIC_ARRAY_SIZE = 128
 
@@ -341,38 +342,27 @@ cdef class Buffer:
 
     def add_str(self, text: str,
                 item_offset: int = 0, item_length: int = -1) -> None:
+        cdef Py_UCS4* ucs4_buffer
+        cdef Py_ssize_t text_length
 
-        cdef Py_ssize_t length = PyUnicode_GET_LENGTH(text)
-        cdef int kind = PyUnicode_KIND(text)
-
-        if kind == PyUnicode_1BYTE_KIND:
-            hb_buffer_add_latin1(
-                self._hb_buffer,
-                <uint8_t*>PyUnicode_1BYTE_DATA(text),
-                length,
-                item_offset,
-                item_length,
-            )
-        elif kind == PyUnicode_2BYTE_KIND:
-            hb_buffer_add_utf16(
-                self._hb_buffer,
-                <uint16_t*>PyUnicode_2BYTE_DATA(text),
-                length,
-                item_offset,
-                item_length,
-            )
-        elif kind == PyUnicode_4BYTE_KIND:
+        ucs4_buffer = PyUnicode_AsUCS4Copy(text)
+        if ucs4_buffer == NULL:
+            raise MemoryError()
+        try:
+            text_length = PyUnicode_GetLength(text)
+            if text_length == -1:
+                raise ValueError("Invalid Unicode string")
             hb_buffer_add_utf32(
                 self._hb_buffer,
-                <uint32_t*>PyUnicode_4BYTE_DATA(text),
-                length,
+                <uint32_t*>ucs4_buffer,
+                text_length,
                 item_offset,
-                item_length,
+                item_length
             )
-        else:
-            raise AssertionError(kind)
-        if not hb_buffer_allocation_successful(self._hb_buffer):
-            raise MemoryError()
+            if not hb_buffer_allocation_successful(self._hb_buffer):
+                raise MemoryError()
+        finally:
+            PyMem_Free(ucs4_buffer)
 
     def guess_segment_properties(self) -> None:
         hb_buffer_guess_segment_properties(self._hb_buffer)
@@ -952,7 +942,7 @@ cdef class Face:
     def get_name(self, name_id: OTNameIdPredefined | int, language: str | None = None) -> str | None:
         cdef bytes packed
         cdef hb_language_t lang
-        cdef uint32_t *text
+        cdef char *text
         cdef unsigned int length
 
         if language is None:
@@ -961,12 +951,18 @@ cdef class Face:
             packed = language.encode()
             lang = hb_language_from_string(<char*>packed, -1)
 
-        length = hb_ot_name_get_utf32(self._hb_face, name_id, lang, NULL, NULL)
+        length = hb_ot_name_get_utf8(self._hb_face, name_id, lang, NULL, NULL)
         if length:
             length += 1  # for the null terminator
-            text = <uint32_t*>malloc(length * sizeof(uint32_t))
-            hb_ot_name_get_utf32(self._hb_face, name_id, lang, &length, text)
-            return PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, text, length)
+            text = <char*>malloc(length * sizeof(char))
+            if text == NULL:
+                raise MemoryError()
+            try:
+                hb_ot_name_get_utf8(self._hb_face, name_id, lang, &length, text)
+                result = text[:length].decode("utf-8")
+                return result
+            finally:
+                free(text)
         return None