stop using before-build, but copy data in actions

Author: taku910

.github/workflows/wheel.yml

@@ -64,7 +64,10 @@ jobs:
 
       - name: Build wheels
         working-directory: ${{github.workspace}}/python
-        run: python -m cibuildwheel --output-dir wheelhouse
+        run: |
+          mkdir -p src/sentencepiece/package_data
+          cp -f ../data/*.bin src/sentencepiece/package_data
+          python -m cibuildwheel --output-dir wheelhouse
         env:
           CIBW_ARCHS_LINUX: auto aarch64
           CIBW_ARCHS_MACOS: x86_64 universal2 arm64

python/build_sdist.sh

@@ -8,7 +8,4 @@ do
   cp -f -R "../${i}" sentencepiece
 done
 
-mkdir -p sentencepiece/data
-cp -f ../data/*.bin sentencepiece/data
-
 python3 setup.py sdist

python/pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 name = "sentencepiece"
 version = "0.2.1"
 authors = [
-  { name="Taku Kudo", email="taku@gogle.com" },
+  { name="Taku Kudo", email="taku@google.com" },
 ]
 description = "Unsupervised text tokenizer and detokenizer."
 readme = "README.md"

python/setup.py

@@ -115,10 +115,15 @@ def copy_package_data():
 
   package_data = os.path.join('src', 'sentencepiece', 'package_data')
 
+  if not os.path.exists(package_data):
+    os.makedirs(package_data)
+
+  if glob.glob(os.path.join(package_data, '*.bin')):
+    return
+
   def find_targets(roots):
     for root in roots:
-      data = glob.glob(os.path.join(root, '*.bin'))
-      if len(data) != 0:
+      if glob.glob(os.path.join(root, '*.bin')):
         return data
     return []
 
@@ -128,12 +133,9 @@ def find_targets(roots):
       '../data',
   ])
 
-  if len(data) == 0 and is_sentencepiece_installed():
+  if not data and is_sentencepiece_installed():
     data = find_targets(run_pkg_config('datadir'))
 
-  if not os.path.exists(package_data):
-    os.makedirs(package_data)
-
   for filename in data:
     print('copying {} -> {}'.format(filename, package_data))
     shutil.copy(filename, package_data)