def create_artifact_cache(log, artifact_root, spec, task_name):
"""Returns an artifact cache for the specified spec.
spec can be:
- a path to a file-based cache root.
- a URL of a RESTful cache root.
- a bar-separated list of URLs, where we'll pick the one with the best ping times.
- A list of the above, for a combined cache.
"""
if not spec:
raise ValueError('Empty artifact cache spec')
if isinstance(spec, basestring):
if spec.startswith('/'):
log.info('%s using local artifact cache at %s' % (task_name, spec))
return FileBasedArtifactCache(log, artifact_root, spec)
elif spec.startswith('http://') or spec.startswith('https://'):
# Caches are supposed to be close, and we don't want to waste time pinging on no-op builds.
# So we ping twice with a short timeout.
pinger = Pinger(timeout=0.5, tries=2)
best_url = select_best_url(spec, pinger, log)
if best_url:
log.info('%s using remote artifact cache at %s' % (task_name, best_url))
return RESTfulArtifactCache(log, artifact_root, best_url)
else:
log.warn('No reachable artifact cache in %s.' % spec)
return None
else:
raise ValueError('Invalid artifact cache spec: %s' % spec)
elif isinstance(spec, (list, tuple)):
caches = [ create_artifact_cache(log, artifact_root, x, task_name) for x in spec ]
return CombinedArtifactCache(caches)
def __init__(self, target, root_dir, extra_targets=None, builder=None, conn_timeout=None):
self._config = Config.load()
self._target = target
self._root = root_dir
self._key_generator = CacheKeyGenerator()
self._extra_targets = list(extra_targets) if extra_targets is not None else []
self._resolver = MultiResolver.from_target(self._config, target, conn_timeout=conn_timeout)
self._builder = builder or PEXBuilder(tempfile.mkdtemp())
artifact_cache_root = os.path.join(self._config.get('python-setup', 'artifact_cache'),
'%s' % PythonIdentity.get())
self._artifact_cache = FileBasedArtifactCache(None, self._root, artifact_cache_root,
self._builder.add_dependency_file)
def __init__(self,
target,
root_dir,
extra_targets=None,
builder=None,
conn_timeout=None):
self._config = Config.load()
self._target = target
self._root = root_dir
self._key_generator = CacheKeyGenerator()
self._extra_targets = list(
extra_targets) if extra_targets is not None else []
self._resolver = MultiResolver.from_target(self._config,
target,
conn_timeout=conn_timeout)
self._builder = builder or PEXBuilder(tempfile.mkdtemp())
artifact_cache_root = os.path.join(
self._config.get('python-setup', 'artifact_cache'),
'%s' % PythonIdentity.get())
self._artifact_cache = FileBasedArtifactCache(
None, self._root, artifact_cache_root,
self._builder.add_dependency_file)
def create_artifact_cache(log, artifact_root, spec):
"""Returns an artifact cache for the specified spec.
spec can be:
- a path to a file-based cache root.
- a URL of a RESTful cache root.
- a bar-separated list of URLs, where we'll pick the one with the best ping times.
- A list of the above, for a combined cache.
"""
if not spec:
raise ValueError('Empty artifact cache spec')
if isinstance(spec, basestring):
if spec.startswith('/'):
return FileBasedArtifactCache(log, artifact_root, spec)
elif spec.startswith('http://') or spec.startswith('https://'):
return RESTfulArtifactCache(log, artifact_root,
select_best_url(spec, Pinger(), log))
else:
raise ValueError('Invalid artifact cache spec: %s' % spec)
elif isinstance(spec, (list, tuple)):
caches = [create_artifact_cache(log, artifact_root, x) for x in spec]
return CombinedArtifactCache(caches)
class PythonChroot(object):
_VALID_DEPENDENCIES = {
PythonLibrary: 'libraries',
PythonRequirement: 'reqs',
PythonBinary: 'binaries',
PythonThriftLibrary: 'thrifts',
PythonAntlrLibrary: 'antlrs',
PythonTests: 'tests'
}
class InvalidDependencyException(Exception):
def __init__(self, target):
Exception.__init__(self, "Not a valid Python dependency! Found: %s" % target)
class BuildFailureException(Exception):
def __init__(self, msg):
Exception.__init__(self, msg)
def __init__(self, target, root_dir, extra_targets=None, builder=None, conn_timeout=None):
self._config = Config.load()
self._target = target
self._root = root_dir
self._key_generator = CacheKeyGenerator()
self._extra_targets = list(extra_targets) if extra_targets is not None else []
self._resolver = MultiResolver.from_target(self._config, target, conn_timeout=conn_timeout)
self._builder = builder or PEXBuilder(tempfile.mkdtemp())
artifact_cache_root = os.path.join(self._config.get('python-setup', 'artifact_cache'),
'%s' % PythonIdentity.get())
self._artifact_cache = FileBasedArtifactCache(None, self._root, artifact_cache_root,
self._builder.add_dependency_file)
def __del__(self):
if os.getenv('PANTS_LEAVE_CHROOT') is None:
safe_rmtree(self.path())
@property
def builder(self):
return self._builder
def debug(self, msg, indent=0):
if os.getenv('PANTS_VERBOSE') is not None:
print('%s%s' % (' ' * indent, msg))
def path(self):
return self._builder.path()
def _dump_library(self, library):
def translate_module(module):
if module is None:
module = ''
return module.replace('.', os.path.sep)
def copy_to_chroot(base, path, relative_to, add_function):
src = os.path.join(self._root, base, path)
dst = os.path.join(translate_module(relative_to), path)
add_function(src, dst)
self.debug(' Dumping library: %s [relative module: %s]' % (library, library.module))
for filename in library.sources:
copy_to_chroot(library.target_base, filename, library.module, self._builder.add_source)
for filename in library.resources:
copy_to_chroot(library.target_base, filename, library.module, self._builder.add_resource)
def _dump_requirement(self, req, dynamic, repo):
self.debug(' Dumping requirement: %s%s%s' % (str(req),
' (dynamic)' if dynamic else '', ' (repo: %s)' if repo else ''))
self._builder.add_requirement(req, dynamic, repo)
def _dump_distribution(self, dist):
self.debug(' Dumping distribution: .../%s' % os.path.basename(dist.location))
self._builder.add_distribution(dist)
def _dump_bin(self, binary_name, base):
src = os.path.join(self._root, base, binary_name)
self.debug(' Dumping binary: %s' % binary_name)
self._builder.set_executable(src, os.path.basename(src))
def _dump_thrift_library(self, library):
self._dump_built_library(library, PythonThriftBuilder(library, self._root, self._config))
def _dump_antlr_library(self, library):
self._dump_built_library(library, PythonAntlrBuilder(library, self._root))
def _dump_built_library(self, library, builder):
# TODO(wickman): Port this over to the Installer+Distiller and stop using ArtifactCache.
absolute_sources = library.expand_files()
absolute_sources.sort()
cache_key = self._key_generator.key_for(library.id, absolute_sources)
if self._artifact_cache.has(cache_key):
self.debug(' Generating (cached) %s...' % library)
self._artifact_cache.use_cached_files(cache_key)
else:
self.debug(' Generating %s...' % library)
egg_file = builder.build_egg()
if not egg_file:
raise PythonChroot.BuildFailureException("Failed to build %s!" % library)
src_egg_file = egg_file
dst_egg_file = os.path.join(os.path.dirname(egg_file),
cache_key.hash + '_' + os.path.basename(egg_file))
self.debug(' %s => %s' % (src_egg_file, dst_egg_file))
os.rename(src_egg_file, dst_egg_file)
self._artifact_cache.insert(cache_key, [dst_egg_file])
self._builder.add_egg(dst_egg_file)
def resolve(self, targets):
children = defaultdict(OrderedSet)
def add_dep(trg):
if is_concrete(trg):
for target_type, target_key in self._VALID_DEPENDENCIES.items():
if isinstance(trg, target_type):
children[target_key].add(trg)
return
raise self.InvalidDependencyException(trg)
for target in targets:
target.walk(add_dep)
return children
def dump(self):
self.debug('Building PythonBinary %s:' % self._target)
targets = self.resolve([self._target] + self._extra_targets)
for lib in targets['libraries']:
self._dump_library(lib)
for req in targets['reqs']:
if not req.should_build():
self.debug('Skipping %s based upon version filter' % req)
continue
self._dump_requirement(req._requirement, req._dynamic, req._repository)
for dist in self._resolver.resolve(
req._requirement for req in targets['reqs'] if req.should_build()):
self._dump_distribution(dist)
if targets['thrifts']:
default_thrift_version = self._config.get('python-thrift', 'default-version', default='0.9')
thrift_versions = set()
for thr in targets['thrifts']:
self._dump_thrift_library(thr)
thrift_version = thr.thrift_version or default_thrift_version
thrift_versions.add(thrift_version)
if len(thrift_versions) > 1:
print('WARNING: Target has multiple thrift versions!')
for version in thrift_versions:
self._builder.add_requirement('thrift==%s' % version)
for dist in self._resolver.resolve('thrift==%s' % version for version in thrift_versions):
self._dump_distribution(dist)
for antlr in targets['antlrs']:
self._dump_antlr_library(antlr)
if len(targets['binaries']) > 1:
print('WARNING: Target has multiple python_binary targets!', file=sys.stderr)
for binary in targets['binaries']:
if len(binary.sources) > 0:
self._dump_bin(binary.sources[0], binary.target_base)
return self._builder
def test_local_cache(self):
with temporary_dir() as artifact_root:
with temporary_dir() as cache_root:
artifact_cache = FileBasedArtifactCache(
None, artifact_root, cache_root)
self.do_test_artifact_cache(artifact_cache)
class PythonChroot(object):
_VALID_DEPENDENCIES = {
PythonLibrary: 'libraries',
PythonRequirement: 'reqs',
PythonBinary: 'binaries',
PythonThriftLibrary: 'thrifts',
PythonAntlrLibrary: 'antlrs',
PythonTests: 'tests'
}
class InvalidDependencyException(Exception):
def __init__(self, target):
Exception.__init__(
self, "Not a valid Python dependency! Found: %s" % target)
class BuildFailureException(Exception):
def __init__(self, msg):
Exception.__init__(self, msg)
def __init__(self,
target,
root_dir,
extra_targets=None,
builder=None,
conn_timeout=None):
self._config = Config.load()
self._target = target
self._root = root_dir
self._key_generator = CacheKeyGenerator()
self._extra_targets = list(
extra_targets) if extra_targets is not None else []
self._resolver = MultiResolver.from_target(self._config,
target,
conn_timeout=conn_timeout)
self._builder = builder or PEXBuilder(tempfile.mkdtemp())
artifact_cache_root = os.path.join(
self._config.get('python-setup', 'artifact_cache'),
'%s' % PythonIdentity.get())
self._artifact_cache = FileBasedArtifactCache(
None, self._root, artifact_cache_root,
self._builder.add_dependency_file)
def __del__(self):
if os.getenv('PANTS_LEAVE_CHROOT') is None:
safe_rmtree(self.path())
@property
def builder(self):
return self._builder
def debug(self, msg, indent=0):
if os.getenv('PANTS_VERBOSE') is not None:
print('%s%s' % (' ' * indent, msg))
def path(self):
return self._builder.path()
def _dump_library(self, library):
def translate_module(module):
if module is None:
module = ''
return module.replace('.', os.path.sep)
def copy_to_chroot(base, path, relative_to, add_function):
src = os.path.join(self._root, base, path)
dst = os.path.join(translate_module(relative_to), path)
add_function(src, dst)
self.debug(' Dumping library: %s [relative module: %s]' %
(library, library.module))
for filename in library.sources:
copy_to_chroot(library.target_base, filename, library.module,
self._builder.add_source)
for filename in library.resources:
copy_to_chroot(library.target_base, filename, library.module,
self._builder.add_resource)
def _dump_requirement(self, req, dynamic, repo):
self.debug(' Dumping requirement: %s%s%s' %
(str(req), ' (dynamic)' if dynamic else '',
' (repo: %s)' if repo else ''))
self._builder.add_requirement(req, dynamic, repo)
def _dump_distribution(self, dist):
self.debug(' Dumping distribution: .../%s' %
os.path.basename(dist.location))
self._builder.add_distribution(dist)
def _dump_bin(self, binary_name, base):
src = os.path.join(self._root, base, binary_name)
self.debug(' Dumping binary: %s' % binary_name)
self._builder.set_executable(src, os.path.basename(src))
def _dump_thrift_library(self, library):
self._dump_built_library(
library, PythonThriftBuilder(library, self._root, self._config))
def _dump_antlr_library(self, library):
self._dump_built_library(library,
PythonAntlrBuilder(library, self._root))
def _dump_built_library(self, library, builder):
# TODO(wickman): Port this over to the Installer+Distiller and stop using ArtifactCache.
absolute_sources = library.expand_files()
absolute_sources.sort()
cache_key = self._key_generator.key_for(library.id, absolute_sources)
if self._artifact_cache.has(cache_key):
self.debug(' Generating (cached) %s...' % library)
self._artifact_cache.use_cached_files(cache_key)
else:
self.debug(' Generating %s...' % library)
egg_file = builder.build_egg()
if not egg_file:
raise PythonChroot.BuildFailureException(
"Failed to build %s!" % library)
src_egg_file = egg_file
dst_egg_file = os.path.join(
os.path.dirname(egg_file),
cache_key.hash + '_' + os.path.basename(egg_file))
self.debug(' %s => %s' % (src_egg_file, dst_egg_file))
os.rename(src_egg_file, dst_egg_file)
self._artifact_cache.insert(cache_key, [dst_egg_file])
self._builder.add_egg(dst_egg_file)
def resolve(self, targets):
children = defaultdict(OrderedSet)
def add_dep(trg):
if is_concrete(trg):
for target_type, target_key in self._VALID_DEPENDENCIES.items(
):
if isinstance(trg, target_type):
children[target_key].add(trg)
return
raise self.InvalidDependencyException(trg)
for target in targets:
target.walk(add_dep)
return children
def dump(self):
self.debug('Building PythonBinary %s:' % self._target)
targets = self.resolve([self._target] + self._extra_targets)
for lib in targets['libraries']:
self._dump_library(lib)
for req in targets['reqs']:
if not req.should_build():
self.debug('Skipping %s based upon version filter' % req)
continue
self._dump_requirement(req._requirement, req._dynamic,
req._repository)
for dist in self._resolver.resolve(req._requirement
for req in targets['reqs']
if req.should_build()):
self._dump_distribution(dist)
if targets['thrifts']:
default_thrift_version = self._config.get('python-thrift',
'default-version',
default='0.9')
thrift_versions = set()
for thr in targets['thrifts']:
self._dump_thrift_library(thr)
thrift_version = thr.thrift_version or default_thrift_version
thrift_versions.add(thrift_version)
if len(thrift_versions) > 1:
print('WARNING: Target has multiple thrift versions!')
for version in thrift_versions:
self._builder.add_requirement('thrift==%s' % version)
for dist in self._resolver.resolve(
'thrift==%s' % version for version in thrift_versions):
self._dump_distribution(dist)
for antlr in targets['antlrs']:
self._dump_antlr_library(antlr)
if len(targets['binaries']) > 1:
print('WARNING: Target has multiple python_binary targets!',
file=sys.stderr)
for binary in targets['binaries']:
if len(binary.sources) > 0:
self._dump_bin(binary.sources[0], binary.target_base)
return self._builder
