pyodide/pyodide_build/pywasmcross.py

250 lines
7.4 KiB
Python
Raw Normal View History

2018-06-20 18:54:47 +00:00
#!/usr/bin/env python3
"""Helper for cross-compiling distutils-based Python extensions.
distutils has never had a proper cross-compilation story. This is a hack, which
miraculously works, to get around that.
The gist is:
- Compile the package natively, replacing calls to the compiler and linker with
wrappers that store the arguments in a log, and then delegate along to the
real native compiler and linker.
- Remove all of the native build products.
- Play back the log, replacing the native compiler with emscripten and
adjusting include paths and flags as necessary for cross-compiling to
emscripten. This overwrites the results from the original native compilation.
While this results in more work than strictly necessary (it builds a native
version of the package, even though we then throw it away), it seems to be the
only reliable way to automatically build a package that interleaves
configuration with build.
"""
2018-06-20 18:54:47 +00:00
import argparse
import importlib.machinery
2018-06-20 18:54:47 +00:00
import json
import os
2018-08-03 16:48:22 +00:00
from pathlib import Path
2018-06-20 18:54:47 +00:00
import re
import subprocess
import sys
# absolute import is necessary as this file will be symlinked
# under tools
from pyodide_build import common
ROOTDIR = common.ROOTDIR
2018-06-20 18:54:47 +00:00
symlinks = set(['cc', 'c++', 'ld', 'ar', 'gcc'])
def collect_args(basename):
"""
This is called when this script is called through a symlink that looks like
a compiler or linker.
It writes the arguments to the build.log, and then delegates to the real
native compiler or linker.
"""
# Remove the symlink compiler from the PATH, so we can delegate to the
# native compiler
2018-06-20 18:54:47 +00:00
env = dict(os.environ)
path = env['PATH']
while str(ROOTDIR) + ':' in path:
path = path.replace(str(ROOTDIR) + ':', '')
2018-06-20 18:54:47 +00:00
env['PATH'] = path
2018-06-20 18:54:47 +00:00
with open('build.log', 'a') as fd:
json.dump([basename] + sys.argv[1:], fd)
fd.write('\n')
sys.exit(
subprocess.run(
[basename] + sys.argv[1:],
env=env).returncode)
2018-06-20 18:54:47 +00:00
def make_symlinks(env):
"""
Makes sure all of the symlinks that make this script look like a compiler
exist.
"""
2018-08-03 16:48:22 +00:00
exec_path = Path(__file__).resolve()
2018-06-20 18:54:47 +00:00
for symlink in symlinks:
symlink_path = ROOTDIR / symlink
2018-08-03 16:48:22 +00:00
if not symlink_path.exists():
symlink_path.symlink_to(exec_path)
2018-06-20 18:54:47 +00:00
if symlink == 'c++':
var = 'CXX'
else:
var = symlink.upper()
env[var] = symlink
def capture_compile(args):
env = dict(os.environ)
make_symlinks(env)
env['PATH'] = str(ROOTDIR) + ':' + os.environ['PATH']
2018-06-20 18:54:47 +00:00
result = subprocess.run(
2018-08-03 16:48:22 +00:00
[Path(args.host) / 'bin' / 'python3',
'setup.py',
'install'], env=env)
2018-06-20 18:54:47 +00:00
if result.returncode != 0:
2018-08-03 16:48:22 +00:00
build_log_path = Path('build.log')
if build_log_path.exists():
build_log_path.unlink()
2018-06-20 18:54:47 +00:00
sys.exit(result.returncode)
def handle_command(line, args):
# This is a special case to skip the compilation tests in numpy that aren't
# actually part of the build
2018-06-20 18:54:47 +00:00
for arg in line:
if r'/file.c' in arg or '_configtest' in arg:
return
2018-10-02 13:17:14 +00:00
if re.match(r'/tmp/.*/source\.[bco]+', arg):
return
2018-06-20 18:54:47 +00:00
if arg == '-print-multiarch':
return
if line[0] == 'ar':
new_args = ['emar']
2018-06-20 18:54:47 +00:00
elif line[0] == 'c++':
new_args = ['em++']
2018-06-20 18:54:47 +00:00
else:
new_args = ['emcc']
2018-06-20 18:54:47 +00:00
# distutils doesn't use the c++ compiler when compiling c++ <sigh>
if any(arg.endswith('.cpp') for arg in line):
new_args = ['em++']
2018-06-20 18:54:47 +00:00
shared = '-shared' in line
if shared:
2018-06-22 18:49:52 +00:00
new_args.extend(args.ldflags.split())
elif new_args[0] in ('emcc', 'em++'):
2018-06-22 18:49:52 +00:00
new_args.extend(args.cflags.split())
2018-06-20 18:54:47 +00:00
# Go through and adjust arguments
2018-06-20 18:54:47 +00:00
for arg in line[1:]:
if arg.startswith('-I'):
# Don't include any system directories
if arg[2:].startswith('/usr'):
continue
2018-08-03 16:48:22 +00:00
if (str(Path(arg[2:]).resolve()).startswith(args.host) and
2018-06-20 19:05:13 +00:00
'site-packages' not in arg):
2018-06-22 18:49:52 +00:00
arg = arg.replace('-I' + args.host, '-I' + args.target)
# Don't include any system directories
2018-06-20 18:54:47 +00:00
if arg.startswith('-L/usr'):
continue
# The native build is possibly multithreaded, but the emscripten one
# definitely isn't
2018-06-20 18:54:47 +00:00
arg = re.sub(r'/python([0-9]\.[0-9]+)m', r'/python\1', arg)
if arg.endswith('.o'):
arg = arg[:-2] + '.bc'
if shared and arg.endswith('.so'):
arg = arg[:-3] + '.wasm'
output = arg
new_args.append(arg)
print(' '.join(new_args))
result = subprocess.run(new_args)
if result.returncode != 0:
sys.exit(result.returncode)
# Emscripten .so files shouldn't have the native platform slug
2018-06-20 18:54:47 +00:00
if shared:
renamed = output[:-5] + '.so'
for ext in importlib.machinery.EXTENSION_SUFFIXES:
if ext == '.so':
continue
if renamed.endswith(ext):
renamed = renamed[:-len(ext)] + '.so'
break
os.rename(output, renamed)
def replay_compile(args):
# If pure Python, there will be no build.log file, which is fine -- just do
# nothing
2018-08-03 16:48:22 +00:00
build_log_path = Path('build.log')
if build_log_path.is_file():
with open(build_log_path, 'r') as fd:
2018-06-20 18:54:47 +00:00
for line in fd:
line = json.loads(line)
handle_command(line, args)
def clean_out_native_artifacts():
for root, dirs, files in os.walk('.'):
for file in files:
2018-08-03 16:48:22 +00:00
path = Path(root) / file
if path.suffix in ('.o', '.so', '.a'):
path.unlink()
2018-06-20 18:54:47 +00:00
def install_for_distribution(args):
subprocess.check_call(
[Path(args.host) / 'bin' / 'python3',
'setup.py',
'install',
'--skip-build',
'--prefix=install',
'--old-and-unmanageable'])
2018-06-20 18:54:47 +00:00
def build_wrap(args):
2018-08-03 16:48:22 +00:00
build_log_path = Path('build.log')
if not build_log_path.is_file():
2018-06-20 18:54:47 +00:00
capture_compile(args)
clean_out_native_artifacts()
replay_compile(args)
install_for_distribution(args)
2018-06-20 18:54:47 +00:00
def make_parser(parser):
2018-09-21 14:58:09 +00:00
basename = Path(sys.argv[0]).name
if basename in symlinks:
# skip parsing of all arguments
parser._actions = []
else:
parser.description = (
'Cross compile a Python distutils package. '
'Run from the root directory of the package\'s source')
parser.add_argument(
'--cflags', type=str, nargs='?', default=common.DEFAULTCFLAGS,
help='Extra compiling flags')
parser.add_argument(
'--ldflags', type=str, nargs='?', default=common.DEFAULTLDFLAGS,
help='Extra linking flags')
parser.add_argument(
'--host', type=str, nargs='?', default=common.HOSTPYTHON,
help='The path to the host Python installation')
parser.add_argument(
'--target', type=str, nargs='?', default=common.TARGETPYTHON,
help='The path to the target Python installation')
return parser
2018-06-20 18:54:47 +00:00
2018-09-21 14:58:09 +00:00
def main(args):
2018-08-03 16:48:22 +00:00
basename = Path(sys.argv[0]).name
2018-06-20 18:54:47 +00:00
if basename in symlinks:
collect_args(basename)
else:
build_wrap(args)
if __name__ == '__main__':
2018-09-21 14:58:09 +00:00
basename = Path(sys.argv[0]).name
if basename in symlinks:
main(None)
else:
parser = make_parser(argparse.ArgumentParser())
args = parser.parse_args()
main(args)