Move to a PEP 517 pyproject.toml build, drop setup.py

.github/workflows/ci.yml

@@ -8,6 +8,10 @@ on:
     branches: [master]
 
 env:
+  # cibuildwheel 3.0 moved PyPy into an opt-in "enable group", which silently
+  # dropped the PyPy wheels this project had shipped up to 16.0.0 (17.0.0 and
+  # 17.0.1 published none). Ask for them back explicitly.
+  CIBW_ENABLE: pypy
   CIBW_MANYLINUX_X86_64_IMAGE: manylinux2014
   CIBW_MANYLINUX_I686_IMAGE: manylinux2014
   CIBW_MANYLINUX_PYPY_X86_64_IMAGE: manylinux2014
@@ -98,7 +102,7 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install --upgrade setuptools wheel twine
+        pip install --upgrade build twine
     - name: Download artifacts from build jobs
       uses: actions/download-artifact@v4
       with:
@@ -134,7 +138,7 @@ jobs:
         draft: false
         prerelease: false
     - name: Build sdist
-      run: python setup.py sdist
+      run: python -m build --sdist
     - name: Publish package distributions to PyPI
       uses: pypa/gh-action-pypi-publish@release/v1
 

CHANGELOG.md

@@ -1,3 +1,5 @@
+# Changelog
+
 ## 17.0.0
  - Upgrade to Unicode 17.0.0
 

README.md

@@ -17,7 +17,8 @@ Pre-compiled wheel packages are available on [PyPI] and can be installed via pip
 Testing
 =======
 
-We run the tests using `tox`. This can be installed as usual with `pip install tox`.
+We run the tests using `tox`. This can be installed as usual with `pip install tox`,
+or with `pip install --group dev` to pick it up from `pyproject.toml`.
 
 Without any options, `tox` will run the tests against all of the library's
 target Python versions. Any missing versions will be skipped.

pyproject.toml

@@ -0,0 +1,56 @@
+[build-system]
+# setuptools 77.0.1 is the first release that combines declarative ``ext-modules``
+# (added in 74.1) with PEP 639 license metadata, which is what lets this
+# project drop setup.py entirely.
+requires = ["setuptools>=77.0.1"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "unicodedata2"
+version = "17.0.1"
+description = "Unicodedata backport updated to the latest Unicode version."
+authors = [
+    { name = "Mike Kaplinskiy", email = "mike.kaplinskiy@gmail.com" },
+]
+license = "Apache-2.0"
+license-files = ["LICENSE"]
+# README.md and CHANGELOG.md are concatenated so both are shown on PyPI.
+dynamic = ["readme"]
+
+[project.optional-dependencies]
+# Installed into every cibuildwheel test environment via CIBW_TEST_EXTRAS, so
+# keep it to what "pytest tests/" actually loads. pytest-randomly is a plugin
+# and shuffles test order just by being installed; coverage and pytest-xdist
+# needed flags nothing passes.
+testing = [
+    "pytest",
+    "pytest-randomly",
+]
+
+# Local development only. PEP 735 dependency groups are never written into
+# the published wheel/sdist metadata, so this adds nothing for installers of
+# unicodedata2 -- it just gives contributors "pip install --group dev".
+[dependency-groups]
+dev = ["tox"]
+
+[project.urls]
+Homepage = "http://github.com/fonttools/unicodedata2"
+Download = "http://github.com/fonttools/unicodedata2"
+
+[tool.setuptools]
+platforms = ["any"]
+# The distribution is a single top-level extension module: there are no Python
+# packages or modules to ship. Saying so explicitly keeps auto-discovery from
+# looking at the unicodedata2/ source directory or at tests/.
+packages = []
+
+[tool.setuptools.dynamic]
+readme = { file = ["README.md", "CHANGELOG.md"], content-type = "text/markdown" }
+
+[[tool.setuptools.ext-modules]]
+name = "unicodedata2"
+sources = [
+    "unicodedata2/unicodedata.c",
+    "unicodedata2/unicodectype.c",
+]
+include-dirs = ["unicodedata2"]

tox.ini

@@ -1,5 +1,5 @@
 [tox]
-envlist = py{38,39,310,311,312,313,314}, pypy{37,38,39}
+envlist = py{38,39,310,311,312,313,314}, pypy{39,310,311}
 skip_missing_interpreters = true
 
 [testenv]

unicodedata2/_unicodedata2_compat.h

@@ -3,6 +3,9 @@
 
 /*
  * Compatibility shims
+ *
+ * This header must be included *after* "Python.h": it detects what the host
+ * Python's headers already provide and only fills in the gaps.
  */
 
 
@@ -18,8 +21,77 @@ static inline void _Py_SET_TYPE(PyObject *ob, PyTypeObject *type)
 
 
 #ifdef PYPY_VERSION
-#include "pypy_ctype.h"
 typedef Py_ssize_t Py_ssize_clean_t;
 #endif
 
+
+/* ----------------------------------------------------------------------- *
+ * ASCII upper-casing for the character name lookup code.
+ *
+ * _gethash() and _cmpname() fold the ASCII case of character names so that
+ * lookup() is case insensitive.  CPython does that with a 256 byte table,
+ * reachable as Py_TOUPPER() because "Python.h" pulls in <cpython/pyctype.h>
+ * on every non-limited-API build.  PyPy's cpyext headers define neither the
+ * macro nor the table, so we provide an equivalent one ourselves.
+ *
+ * The detection tests Py_TOUPPER, the macro, rather than _Py_ctype_toupper,
+ * which is a linker symbol the preprocessor cannot see.  Because this file
+ * is included after "Python.h" and nothing else defines Py_TOUPPER first,
+ * "is the macro defined here?" is exactly "does the host Python provide
+ * upper-casing?".
+ *
+ * Either way the argument is masked with Py_CHARMASK() inside the macro, so
+ * callers pass a plain char and must not mask it again.
+ * ----------------------------------------------------------------------- */
+#ifdef Py_TOUPPER
+
+#define UNICODEDATA2_TOUPPER(c) Py_TOUPPER(c)
+
+#else
+
+/* Same mapping as CPython's _Py_ctype_toupper: ASCII 'a'-'z' fold to 'A'-'Z',
+   every other byte value is passed through unchanged.  Kept private to
+   unicodedata2 on purpose -- it deliberately does not define a symbol in
+   CPython's reserved _Py_ namespace, and nothing outside this extension can
+   link against it. */
+static const unsigned char unicodedata2_toupper_table[256] = {
+    0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
+    0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
+    0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
+    0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
+    0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27,
+    0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f,
+    0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
+    0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f,
+    0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
+    0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f,
+    0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
+    0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f,
+    0x60, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
+    0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f,
+    0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
+    0x58, 0x59, 0x5a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f,
+    0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
+    0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f,
+    0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
+    0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
+    0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
+    0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
+    0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
+    0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,
+    0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,
+    0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,
+    0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,
+    0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,
+    0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,
+    0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef,
+    0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,
+    0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,
+};
+
+#define UNICODEDATA2_TOUPPER(c) \
+    (unicodedata2_toupper_table[Py_CHARMASK(c)])
+
+#endif  /* Py_TOUPPER */
+
 #endif

unicodedata2/unicodedata.c

@@ -890,7 +890,7 @@ _gethash(const char *s, int len, int scale)
     unsigned long h = 0;
     unsigned long ix;
     for (i = 0; i < len; i++) {
-        h = (h * scale) + (unsigned char) Py_TOUPPER(Py_CHARMASK(s[i]));
+        h = (h * scale) + (unsigned char) UNICODEDATA2_TOUPPER(s[i]);
         ix = h & 0xff000000;
         if (ix)
             h = (h ^ ((ix>>24) & 0xff)) & 0x00ffffff;
@@ -1049,7 +1049,7 @@ _cmpname(PyObject *self, int code, const char* name, int namelen)
     if (!_getucname(self, code, buffer, NAME_MAXLEN, 1))
         return 0;
     for (i = 0; i < namelen; i++) {
-        if (Py_TOUPPER(Py_CHARMASK(name[i])) != buffer[i])
+        if (UNICODEDATA2_TOUPPER(name[i]) != buffer[i])
             return 0;
     }
     return buffer[namelen] == '\0';