wix: add a hook for a prebuild script to inject extra libraries

I need this to build packages for Google so we can bundle some
extensions in the installed image. My assumption is that this is most
interesting for the .msi images so I only wired it up there. I'm not
thrilled with the interface this provides, but it was an easy way to
retain debug messages on Windows while also having enough structure to
know what lines are actually module names for py2exe.

Still pending on my end: I need to bundle a couple of config files,
and at least one data file. I'm open to advice on how to do those
things, and how to do this better.

Differential Revision: https://phab.mercurial-scm.org/D6164

contrib/packaging/hgpackaging/py2exe.py

@@ -1,135 +1,146 b''
 # py2exe.py - Functionality for performing py2exe builds.
 #
 # Copyright 2019 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 # no-check-code because Python 3 native.
 
 import os
 import pathlib
 import subprocess
 
 from .downloads import (
     download_entry,
 )
 from .util import (
     extract_tar_to_directory,
     extract_zip_to_directory,
     python_exe_info,
 )
 
 
 def build_py2exe(source_dir: pathlib.Path, build_dir: pathlib.Path,
                  python_exe: pathlib.Path, build_name: str,
                  venv_requirements_txt: pathlib.Path,
                  extra_packages=None, extra_excludes=None,
-                 extra_dll_excludes=None):
+                 extra_dll_excludes=None,
+                 extra_packages_script=None):
     """Build Mercurial with py2exe.
 
     Build files will be placed in ``build_dir``.
 
     py2exe's setup.py doesn't use setuptools. It doesn't have modern logic
     for finding the Python 2.7 toolchain. So, we require the environment
     to already be configured with an active toolchain.
     """
     if 'VCINSTALLDIR' not in os.environ:
         raise Exception('not running from a Visual C++ build environment; '
                         'execute the "Visual C++ <version> Command Prompt" '
                         'application shortcut or a vcsvarsall.bat file')
 
     # Identity x86/x64 and validate the environment matches the Python
     # architecture.
     vc_x64 = r'\x64' in os.environ['LIB']
 
     py_info = python_exe_info(python_exe)
 
     if vc_x64:
         if py_info['arch'] != '64bit':
             raise Exception('architecture mismatch: Visual C++ environment '
                             'is configured for 64-bit but Python is 32-bit')
     else:
         if py_info['arch'] != '32bit':
             raise Exception('architecture mismatch: Visual C++ environment '
                             'is configured for 32-bit but Python is 64-bit')
 
     if py_info['py3']:
         raise Exception('Only Python 2 is currently supported')
 
     build_dir.mkdir(exist_ok=True)
 
     gettext_pkg, gettext_entry = download_entry('gettext', build_dir)
     gettext_dep_pkg = download_entry('gettext-dep', build_dir)[0]
     virtualenv_pkg, virtualenv_entry = download_entry('virtualenv', build_dir)
     py2exe_pkg, py2exe_entry = download_entry('py2exe', build_dir)
 
     venv_path = build_dir / ('venv-%s-%s' % (build_name,
                                              'x64' if vc_x64 else 'x86'))
 
     gettext_root = build_dir / (
         'gettext-win-%s' % gettext_entry['version'])
 
     if not gettext_root.exists():
         extract_zip_to_directory(gettext_pkg, gettext_root)
         extract_zip_to_directory(gettext_dep_pkg, gettext_root)
 
     # This assumes Python 2. We don't need virtualenv on Python 3.
     virtualenv_src_path = build_dir / (
         'virtualenv-%s' % virtualenv_entry['version'])
     virtualenv_py = virtualenv_src_path / 'virtualenv.py'
 
     if not virtualenv_src_path.exists():
         extract_tar_to_directory(virtualenv_pkg, build_dir)
 
     py2exe_source_path = build_dir / ('py2exe-%s' % py2exe_entry['version'])
 
     if not py2exe_source_path.exists():
         extract_zip_to_directory(py2exe_pkg, build_dir)
 
     if not venv_path.exists():
         print('creating virtualenv with dependencies')
         subprocess.run(
             [str(python_exe), str(virtualenv_py), str(venv_path)],
             check=True)
 
     venv_python = venv_path / 'Scripts' / 'python.exe'
     venv_pip = venv_path / 'Scripts' / 'pip.exe'
 
     subprocess.run([str(venv_pip), 'install', '-r', str(venv_requirements_txt)],
                    check=True)
 
     # Force distutils to use VC++ settings from environment, which was
     # validated above.
     env = dict(os.environ)
     env['DISTUTILS_USE_SDK'] = '1'
     env['MSSdk'] = '1'
 
+    if extra_packages_script:
+        more_packages = set(subprocess.check_output(
+            extra_packages_script,
+            cwd=build_dir).split(b'\0')[-1].strip().decode('utf-8').splitlines())
+        if more_packages:
+            if not extra_packages:
+                extra_packages = more_packages
+            else:
+                extra_packages |= more_packages
+
     if extra_packages:
         env['HG_PY2EXE_EXTRA_PACKAGES'] = ' '.join(sorted(extra_packages))
     if extra_excludes:
         env['HG_PY2EXE_EXTRA_EXCLUDES'] = ' '.join(sorted(extra_excludes))
     if extra_dll_excludes:
         env['HG_PY2EXE_EXTRA_DLL_EXCLUDES'] = ' '.join(
             sorted(extra_dll_excludes))
 
     py2exe_py_path = venv_path / 'Lib' / 'site-packages' / 'py2exe'
     if not py2exe_py_path.exists():
         print('building py2exe')
         subprocess.run([str(venv_python), 'setup.py', 'install'],
                        cwd=py2exe_source_path,
                        env=env,
                        check=True)
 
     # Register location of msgfmt and other binaries.
     env['PATH'] = '%s%s%s' % (
         env['PATH'], os.pathsep, str(gettext_root / 'bin'))
 
     print('building Mercurial')
     subprocess.run(
         [str(venv_python), 'setup.py',
          'py2exe',
          'build_doc', '--html'],
         cwd=str(source_dir),
         env=env,
         check=True)

contrib/packaging/hgpackaging/wix.py

@@ -1,301 +1,308 b''
 # wix.py - WiX installer functionality
 #
 # Copyright 2019 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 # no-check-code because Python 3 native.
 
 import os
 import pathlib
 import re
 import subprocess
 import tempfile
 import xml.dom.minidom
 
 from .downloads import (
     download_entry,
 )
 from .py2exe import (
     build_py2exe,
 )
 from .util import (
     extract_zip_to_directory,
     sign_with_signtool,
 )
 
 
 SUPPORT_WXS = [
     ('contrib.wxs', r'contrib'),
     ('dist.wxs', r'dist'),
     ('doc.wxs', r'doc'),
     ('help.wxs', r'mercurial\help'),
     ('i18n.wxs', r'i18n'),
     ('locale.wxs', r'mercurial\locale'),
     ('templates.wxs', r'mercurial\templates'),
 ]
 
 
 EXTRA_PACKAGES = {
     'distutils',
     'pygments',
 }
 
 
 def find_version(source_dir: pathlib.Path):
     version_py = source_dir / 'mercurial' / '__version__.py'
 
     with version_py.open('r', encoding='utf-8') as fh:
         source = fh.read().strip()
 
     m = re.search('version = b"(.*)"', source)
     return m.group(1)
 
 
 def normalize_version(version):
     """Normalize Mercurial version string so WiX accepts it.
 
     Version strings have to be numeric X.Y.Z.
     """
 
     if '+' in version:
         version, extra = version.split('+', 1)
     else:
         extra = None
 
     # 4.9rc0
     if version[:-1].endswith('rc'):
         version = version[:-3]
 
     versions = [int(v) for v in version.split('.')]
     while len(versions) < 3:
         versions.append(0)
 
     major, minor, build = versions[:3]
 
     if extra:
         # <commit count>-<hash>+<date>
         build = int(extra.split('-')[0])
 
     return '.'.join('%d' % x for x in (major, minor, build))
 
 
 def ensure_vc90_merge_modules(build_dir):
     x86 = (
         download_entry('vc9-crt-x86-msm', build_dir,
                        local_name='microsoft.vcxx.crt.x86_msm.msm')[0],
         download_entry('vc9-crt-x86-msm-policy', build_dir,
                        local_name='policy.x.xx.microsoft.vcxx.crt.x86_msm.msm')[0]
     )
 
     x64 = (
         download_entry('vc9-crt-x64-msm', build_dir,
                        local_name='microsoft.vcxx.crt.x64_msm.msm')[0],
         download_entry('vc9-crt-x64-msm-policy', build_dir,
                        local_name='policy.x.xx.microsoft.vcxx.crt.x64_msm.msm')[0]
     )
     return {
         'x86': x86,
         'x64': x64,
     }
 
 
 def run_candle(wix, cwd, wxs, source_dir, defines=None):
     args = [
         str(wix / 'candle.exe'),
         '-nologo',
         str(wxs),
         '-dSourceDir=%s' % source_dir,
     ]
 
     if defines:
         args.extend('-d%s=%s' % define for define in sorted(defines.items()))
 
     subprocess.run(args, cwd=str(cwd), check=True)
 
 
 def make_post_build_signing_fn(name, subject_name=None, cert_path=None,
                                cert_password=None, timestamp_url=None):
     """Create a callable that will use signtool to sign hg.exe."""
 
     def post_build_sign(source_dir, build_dir, dist_dir, version):
         description = '%s %s' % (name, version)
 
         sign_with_signtool(dist_dir / 'hg.exe', description,
                            subject_name=subject_name, cert_path=cert_path,
                            cert_password=cert_password,
                            timestamp_url=timestamp_url)
 
     return post_build_sign
 
 
 LIBRARIES_XML = '''
 <?xml version="1.0" encoding="utf-8"?>
 <Wix xmlns="http://schemas.microsoft.com/wix/2006/wi">
 
   <?include {wix_dir}/guids.wxi ?>
   <?include {wix_dir}/defines.wxi ?>
 
   <Fragment>
     <DirectoryRef Id="INSTALLDIR" FileSource="$(var.SourceDir)">
       <Directory Id="libdir" Name="lib" FileSource="$(var.SourceDir)/lib">
         <Component Id="libOutput" Guid="$(var.lib.guid)" Win64='$(var.IsX64)'>
         </Component>
       </Directory>
     </DirectoryRef>
   </Fragment>
 </Wix>
 '''.lstrip()
 
 
 def make_libraries_xml(wix_dir: pathlib.Path, dist_dir: pathlib.Path):
     """Make XML data for library components WXS."""
     # We can't use ElementTree because it doesn't handle the
     # <?include ?> directives.
     doc = xml.dom.minidom.parseString(
         LIBRARIES_XML.format(wix_dir=str(wix_dir)))
 
     component = doc.getElementsByTagName('Component')[0]
 
     f = doc.createElement('File')
     f.setAttribute('Name', 'library.zip')
     f.setAttribute('KeyPath', 'yes')
     component.appendChild(f)
 
     lib_dir = dist_dir / 'lib'
 
     for p in sorted(lib_dir.iterdir()):
         if not p.name.endswith(('.dll', '.pyd')):
             continue
 
         f = doc.createElement('File')
         f.setAttribute('Name', p.name)
         component.appendChild(f)
 
     return doc.toprettyxml()
 
 
 def build_installer(source_dir: pathlib.Path, python_exe: pathlib.Path,
-                    msi_name='mercurial', version=None, post_build_fn=None):
+                    msi_name='mercurial', version=None, post_build_fn=None,
+                    extra_packages_script=None):
     """Build a WiX MSI installer.
 
     ``source_dir`` is the path to the Mercurial source tree to use.
     ``arch`` is the target architecture. either ``x86`` or ``x64``.
     ``python_exe`` is the path to the Python executable to use/bundle.
     ``version`` is the Mercurial version string. If not defined,
     ``mercurial/__version__.py`` will be consulted.
     ``post_build_fn`` is a callable that will be called after building
     Mercurial but before invoking WiX. It can be used to e.g. facilitate
     signing. It is passed the paths to the Mercurial source, build, and
     dist directories and the resolved Mercurial version.
+    ``extra_packages_script`` is a command to be run to inject extra packages
+    into the py2exe binary. It should stage packages into the virtualenv and
+    print a null byte followed by a newline-separated list of packages that
+    should be included in the exe.
     """
     arch = 'x64' if r'\x64' in os.environ.get('LIB', '') else 'x86'
 
     hg_build_dir = source_dir / 'build'
     dist_dir = source_dir / 'dist'
     wix_dir = source_dir / 'contrib' / 'packaging' / 'wix'
 
     requirements_txt = wix_dir / 'requirements.txt'
 
     build_py2exe(source_dir, hg_build_dir,
                  python_exe, 'wix', requirements_txt,
-                 extra_packages=EXTRA_PACKAGES)
+                 extra_packages=EXTRA_PACKAGES,
+                 extra_packages_script=extra_packages_script)
 
     version = version or normalize_version(find_version(source_dir))
     print('using version string: %s' % version)
 
     if post_build_fn:
         post_build_fn(source_dir, hg_build_dir, dist_dir, version)
 
     build_dir = hg_build_dir / ('wix-%s' % arch)
 
     build_dir.mkdir(exist_ok=True)
 
     wix_pkg, wix_entry = download_entry('wix', hg_build_dir)
     wix_path = hg_build_dir / ('wix-%s' % wix_entry['version'])
 
     if not wix_path.exists():
         extract_zip_to_directory(wix_pkg, wix_path)
 
     ensure_vc90_merge_modules(hg_build_dir)
 
     source_build_rel = pathlib.Path(os.path.relpath(source_dir, build_dir))
 
     defines = {'Platform': arch}
 
     for wxs, rel_path in SUPPORT_WXS:
         wxs = wix_dir / wxs
         wxs_source_dir = source_dir / rel_path
         run_candle(wix_path, build_dir, wxs, wxs_source_dir, defines=defines)
 
     # candle.exe doesn't like when we have an open handle on the file.
     # So use TemporaryDirectory() instead of NamedTemporaryFile().
     with tempfile.TemporaryDirectory() as td:
         td = pathlib.Path(td)
 
         tf = td / 'library.wxs'
         with tf.open('w') as fh:
             fh.write(make_libraries_xml(wix_dir, dist_dir))
 
         run_candle(wix_path, build_dir, tf, dist_dir, defines=defines)
 
     source = wix_dir / 'mercurial.wxs'
     defines['Version'] = version
     defines['Comments'] = 'Installs Mercurial version %s' % version
     defines['VCRedistSrcDir'] = str(hg_build_dir)
 
     run_candle(wix_path, build_dir, source, source_build_rel, defines=defines)
 
     msi_path = source_dir / 'dist' / (
         '%s-%s-%s.msi' % (msi_name, version, arch))
 
     args = [
         str(wix_path / 'light.exe'),
         '-nologo',
         '-ext', 'WixUIExtension',
         '-sw1076',
         '-spdb',
         '-o', str(msi_path),
     ]
 
     for source, rel_path in SUPPORT_WXS:
         assert source.endswith('.wxs')
         args.append(str(build_dir / ('%s.wixobj' % source[:-4])))
 
     args.extend([
         str(build_dir / 'library.wixobj'),
         str(build_dir / 'mercurial.wixobj'),
     ])
 
     subprocess.run(args, cwd=str(source_dir), check=True)
 
     print('%s created' % msi_path)
 
     return {
         'msi_path': msi_path,
     }
 
 
 def build_signed_installer(source_dir: pathlib.Path, python_exe: pathlib.Path,
                            name: str, version=None, subject_name=None,
                            cert_path=None, cert_password=None,
-                           timestamp_url=None):
+                           timestamp_url=None, extra_packages_script=None):
     """Build an installer with signed executables."""
 
     post_build_fn = make_post_build_signing_fn(
         name,
         subject_name=subject_name,
         cert_path=cert_path,
         cert_password=cert_password,
         timestamp_url=timestamp_url)
 
     info = build_installer(source_dir, python_exe=python_exe,
                            msi_name=name.lower(), version=version,
-                           post_build_fn=post_build_fn)
+                           post_build_fn=post_build_fn,
+                           extra_packages_script=extra_packages_script)
 
     description = '%s %s' % (name, version)
 
     sign_with_signtool(info['msi_path'], description,
                        subject_name=subject_name, cert_path=cert_path,
                        cert_password=cert_password, timestamp_url=timestamp_url)

contrib/packaging/wix/build.py

@@ -1,65 +1,71 b''
 #!/usr/bin/env python3
 # Copyright 2019 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 # no-check-code because Python 3 native.
 
 """Code to build Mercurial WiX installer."""
 
 import argparse
 import os
 import pathlib
 import sys
 
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
 
     parser.add_argument('--name',
                         help='Application name',
                         default='Mercurial')
     parser.add_argument('--python',
                         help='Path to Python executable to use',
                         required=True)
     parser.add_argument('--sign-sn',
                         help='Subject name (or fragment thereof) of certificate '
                              'to use for signing')
     parser.add_argument('--sign-cert',
                         help='Path to certificate to use for signing')
     parser.add_argument('--sign-password',
                         help='Password for signing certificate')
     parser.add_argument('--sign-timestamp-url',
                         help='URL of timestamp server to use for signing')
     parser.add_argument('--version',
                         help='Version string to use')
+    parser.add_argument('--extra-packages-script',
+                        help=('Script to execute to include extra packages in '
+                              'py2exe binary.'))
 
     args = parser.parse_args()
 
     here = pathlib.Path(os.path.abspath(os.path.dirname(__file__)))
     source_dir = here.parent.parent.parent
 
     sys.path.insert(0, str(source_dir / 'contrib' / 'packaging'))
 
     from hgpackaging.wix import (
         build_installer,
         build_signed_installer,
     )
 
     fn = build_installer
     kwargs = {
         'source_dir': source_dir,
         'python_exe': pathlib.Path(args.python),
         'version': args.version,
     }
 
+    if args.extra_packages_script:
+        kwargs['extra_packages_script'] = args.extra_packages_script
+
     if args.sign_sn or args.sign_cert:
         fn = build_signed_installer
         kwargs['name'] = args.name
         kwargs['subject_name'] = args.sign_sn
         kwargs['cert_path'] = args.sign_cert
         kwargs['cert_password'] = args.sign_password
         kwargs['timestamp_url'] = args.sign_timestamp_url
 
     fn(**kwargs)