From f54d8ba070f40ddddf9ef6d7777eaf2b3432f889 Mon Sep 17 00:00:00 2001
From: Taku Kudo <taku@google.com>
Date: Tue, 4 Apr 2023 03:15:11 +0000
Subject: [PATCH] includes the sentencepiece source files in python source
 package

---
 .gitignore              |  2 ++
 python/MANIFEST.in      |  1 +
 python/build_bundled.sh | 12 +++++++-----
 python/build_sdist.sh   | 11 +++++++++++
 python/setup.py         | 39 ++++++++++++++++++++++++++++++++++-----
 5 files changed, 55 insertions(+), 10 deletions(-)
 create mode 100755 python/build_sdist.sh

diff --git a/.gitignore b/.gitignore
index ecdb585..743769d 100644
--- a/.gitignore
+++ b/.gitignore
@@ -73,3 +73,5 @@ libsentencepiece_train.so*
 python/bundled
 _sentencepiece.*.so
 third_party/abseil-cpp
+
+python/sentencepiece
diff --git a/python/MANIFEST.in b/python/MANIFEST.in
index ddfa0f1..fab3380 100644
--- a/python/MANIFEST.in
+++ b/python/MANIFEST.in
@@ -1,3 +1,4 @@
 recursive-include test *.py *.model botchan.txt
 recursive-include src *.i
+recursive-include sentencepiece *
 include *.md VERSION.* build_bundled.sh
diff --git a/python/build_bundled.sh b/python/build_bundled.sh
index 996a39e..a3f6861 100755
--- a/python/build_bundled.sh
+++ b/python/build_bundled.sh
@@ -2,13 +2,15 @@
 
 VERSION="$1"
 
-mkdir -p bundled
+mkdir -p build
 
-BUILD_DIR=./bundled
-INSTALL_DIR=./bundled/root
+BUILD_DIR=./build
+INSTALL_DIR=./build/root
 
-if [ -f ../src/CMakeLists.txt ]; then
-  SRC_DIR=..
+if [ -f ./sentencepiece/src/CMakeLists.txt ]; then
+  SRC_DIR=./sentencepiece
+elif [ -f ../src/CMakeLists.txt ]; then
+  SRC_DIR=..  
 else
   # Try taged version. Othewise, use head.
   git clone https://github.com/google/sentencepiece.git -b v"${VERSION}" --depth 1 || \
diff --git a/python/build_sdist.sh b/python/build_sdist.sh
new file mode 100755
index 0000000..92da94b
--- /dev/null
+++ b/python/build_sdist.sh
@@ -0,0 +1,11 @@
+#!/bin/sh
+
+mkdir -p sentencepiece
+
+for i in CMakeLists.txt LICENSE README.md VERSION.txt cmake config.h.in  sentencepiece.pc.in src third_party
+do
+  echo "copying ../${i} sentencepiece/${i}"
+  cp -f -R "../${i}" sentencepiece
+done
+
+python3 setup.py sdist
diff --git a/python/setup.py b/python/setup.py
index e6aa585..5411231 100755
--- a/python/setup.py
+++ b/python/setup.py
@@ -78,8 +78,6 @@ class build_ext(_build_ext):
 
   def build_extension(self, ext):
     cflags, libs = get_cflags_and_libs('../build/root')
-    if len(libs) == 0:
-      cflags, libs = get_cflags_and_libs('./bundled/root')
 
     if len(libs) == 0:
       if is_sentencepiece_installed():
@@ -87,7 +85,7 @@ class build_ext(_build_ext):
         libs = run_pkg_config('libs')
       else:
         subprocess.check_call(['./build_bundled.sh', __version__])
-        cflags, libs = get_cflags_and_libs('./bundled/root')
+        cflags, libs = get_cflags_and_libs('./build/root')
 
     # Fix compile on some versions of Mac OSX
     # See: https://github.com/neulab/xnmt/issues/199
@@ -104,7 +102,7 @@ class build_ext(_build_ext):
 
 
 if os.name == 'nt':
-  # Must pre-install sentencepice into bundled directory.
+  # Must pre-install sentencepice into build directory.
   arch = 'win32'
   if sys.maxsize > 2**32:
     arch = 'amd64'
@@ -114,12 +112,43 @@ if os.name == 'nt':
         '..\\build\\root_{}\\lib\\sentencepiece.lib'.format(arch),
         '..\\build\\root_{}\\lib\\sentencepiece_train.lib'.format(arch),
     ]
-  else:
+  elif os.path.exists('..\\build\\root\\lib'):
     cflags = ['/std:c++17', '/I..\\build\\root\\include']
     libs = [
         '..\\build\\root\\lib\\sentencepiece.lib',
         '..\\build\\root\\lib\\sentencepiece_train.lib',
     ]
+  else:
+    # build library locally with cmake and vc++.
+    cmake_arch = 'Win32'
+    if arch == 'amd64':
+      cmake_arch = 'x64'
+    subprocess.check_call([
+        'cmake',
+        'sentencepiece',
+        '-A',
+        cmake_arch,
+        '-B',
+        'build',
+        '-DSPM_ENABLE_SHARED=OFF',
+        '-DCMAKE_INSTALL_PREFIX=build\\root',
+    ])
+    subprocess.check_call([
+        'cmake',
+        '--build',
+        'build',
+        '--config',
+        'Release',
+        '--target',
+        'install',
+        '--parallel',
+        '8',
+    ])
+    cflags = ['/std:c++17', '/I.\\build\\root\\include']
+    libs = [
+        '.\\build\\root\\lib\\sentencepiece.lib',
+        '.\\build\\root\\lib\\sentencepiece_train.lib',
+    ]
 
   SENTENCEPIECE_EXT = Extension(
       'sentencepiece._sentencepiece',