def download(self, target_path, prefix=None, delimiter=None, include=None, exclude=None, start_callback=None):
"""
Download to a local path the selected objects in the container.
:param prefix: download only the objects which name starts with this prefix. Useful for pseudo-hierarchical navigation.
:param delimiter: download distinct objects with names up to either the first delimiter or the end. Useful for pseudo-hierarchical navigation.
"""
for obj_name in self.list_objects(prefix, delimiter):
if include is not None and not self.__filename_match(obj_name, include)\
or exclude is not None and self.__filename_match(obj_name, exclude):
continue
path = os.path.normpath(obj_name)
path = os.path.normpath("{}/{}".format(target_path, obj_name))
if os.path.relpath(path, target_path).startswith(".."):
get_logger(__name__).warn("Object output path goes outside the target path: {}".format(path))
continue
#print "***", obj_name, "--->", path
base_path = os.path.dirname(path)
if not os.path.exists(base_path):
os.makedirs(base_path)
prefixed_obj_name = (prefix or "") + obj_name
obj = self.get_object(prefixed_obj_name)
if obj.exists():
start_callback(obj_name)
obj.get_data(path)
def new_function(*args, **kw):
from wok.logger import get_logger
get_logger(None, "wok").debug("acquire %s" % f.__name__)
lock.acquire()
try:
return f(*args, ** kw)
finally:
lock.release()
get_logger(None, "wok").debug("release %s" % f.__name__)
def load(self):
try:
with open(self.path, "r") as f:
v = json.load(f)
return Data.create(v)
except Exception as e:
from wok.logger import get_logger
msg = ["Error loading configuration from ",
self.path, ":\n\n", str(e), "\n"]
get_logger(__name__).error("".join(msg))
raise
def sync_function(obj, *args, **kw):
#log.debug("<ACQUIRE %s>" % f.__name__)
obj._acquire()
try:
return f(obj, *args, ** kw)
finally:
try:
#log.debug("<RELEASE %s>" % f.__name__)
obj._release()
except ThreadError:
from wok.logger import get_logger
get_logger(__name__).warn("<RELEASE ERROR {}.{}>".format(obj.__class__.__name__, f.__name__))
def sync_function(obj, *args, **kw):
#log.debug("<ACQUIRE %s>" % f.__name__)
obj._acquire()
try:
return f(obj, *args, **kw)
finally:
try:
#log.debug("<RELEASE %s>" % f.__name__)
obj._release()
except ThreadError:
from wok.logger import get_logger
get_logger(__name__).warn("<RELEASE ERROR {}.{}>".format(
obj.__class__.__name__, f.__name__))
def merge_into(self, conf):
try:
f = open(self.path, "r")
v = json.load(f)
cf = DataFactory.from_native(v)
conf.merge(cf)
f.close()
except Exception as e:
from wok import logger
msg = ["Error loading configuration from ", self.path, ":\n\n", str(e), "\n"]
logger.get_logger("config").error("".join(msg))
raise
def sync_function(obj, *args, **kw):
#log.debug("<ACQUIRE %s>" % f.__name__)
obj._acquire()
try:
return f(obj, *args, ** kw)
finally:
try:
#log.debug("<RELEASE %s>" % f.__name__)
obj._release()
except:
from wok.logger import get_logger
get_logger(name = "synchronized").error(
"<RELEASE ERROR %s.%s>" % (
obj.__class__.__name__, f.__name__))
def get_annotations_to_save(self, keys, annotations, names=None, values=None):
assert (names is None and values is None) or (names is not None and values is not None and len(names) == len(values))
if names is None:
names = []
values = []
else:
names = [name for name in names]
values = [value for value in values]
ann_keys = keys
if ann_keys is None:
ann_keys = []
elif Data.is_list(ann_keys):
ann_keys = ann_keys.to_native()
else:
log = logger.get_logger(__name__)
log.warn("Wrong type for 'project.annotations', expecting a list but found:\n{0}".format(repr(ann_keys)))
ann_keys = []
for ann_key in ann_keys:
default = None
if isinstance(ann_key, basestring):
key = name = ann_key
elif isinstance(ann_key, list) and len(ann_key) == 2:
key = ann_key[0]
name = ann_key[1]
value = annotations[key] if key in annotations else default
names += [name]
values += [value]
return names, values
def __init__(self, desc, name=None, logger=None, base_path=None):
self._log = logger or woklogger.get_logger(_LOGGER_NAME)
self.name = name or desc.get("name")
self._flow_loader = None
platform = desc.get("platform")
self.path = desc.get("path")
self.conf = desc.get("conf", default=Data.element)
if isinstance(self.conf, basestring):
self.conf = ConfigLoader(os.path.join(base_path or "",
self.conf)).load()
self.conf_rules = []
if self.path is not None:
self.conf_rules += [
ConfRule(dict(set=[[rtconf.PROJECT_PATH, self.path]]),
base_path, platform)
]
self.conf_rules += [
ConfRule(rule, base_path, platform)
for rule in desc.get("conf_rules", default=Data.list)
]
self.flows = desc.get("flows", default=Data.list)
def main():
conf = OptionsConfig()
if "wok" not in conf:
print("Missing wok configuration")
exit(-1)
conf.expand_vars()
logger.initialize(conf.get("wok.logging"))
log = logger.get_logger("wok.server.start")
# run the server
retcode = 0
try:
server = WokServer(conf, app)
server.init()
server.run()
except Exception as e:
log.exception(e)
retcode = -1
exit(retcode)
def main():
conf = OptionsConfig()
if "wok" not in conf:
print("Missing wok configuration")
exit(-1)
conf.expand_vars()
logger.initialize(conf.get("wok.logging"))
log = logger.get_logger("wok.server.start")
# run the server
retcode = 0
try:
server = WokServer(conf, app)
server.init()
server.run()
except Exception as e:
log.exception(e)
retcode = -1
exit(retcode)
def __init__(self, conf):
if "sources" in conf:
self.sources = conf["sources"]
else:
self.sources = DataElement()
self.log = logger.get_logger(conf.get("log"), name = "em_file")
def __log(msg, level=None):
from wok.logger import get_logger
log = get_logger("wok.data")
if level is None:
import logging
level = logging.WARNING
log.log(level, msg)
def __log(msg, level=None):
from wok.logger import get_logger
log = get_logger("wok.data")
if level is None:
import logging
level = logging.WARNING
log.log(level, msg)
def load_conf(self, conf_files=None, path_env="WEB_CONF_PATH", files_env="WEB_CONF_FILES", default_ext=".cfg"):
logger = get_logger("wok.server", level="info")
logger.info("Loading Flask configuration ...")
conf_path = os.environ.get(path_env, os.getcwd())
conf_files = _get_conf_files(conf_path, conf_files or [], os.environ.get(files_env), default_ext)
for cfg_path in conf_files:
logger.info("+++ {}".format(cfg_path))
self.config.from_pyfile(cfg_path)
def __init__(self, conf, base_path=None):
self.conf = conf
self.base_path = base_path or os.getcwd()
self._log = woklogger.get_logger(_LOGGER_NAME)
self._projects = {}
def __init__(self, conf, base_path=None):
self.conf = conf
self.base_path = base_path or os.getcwd()
self._log = woklogger.get_logger(_LOGGER_NAME)
self._projects = {}
def __init__(self, args_usage="", epilog="", logger_name=None):
self.args_usage = args_usage
self.epilog = epilog
if logger_name is None:
if hasattr(self, "cmd_name"):
logger_name = self.cmd_name
else:
logger_name = os.path.splitext(os.path.basename(sys.argv[0]))[0]
# Determine binary path
script_path = os.path.dirname(sys.argv[0])
if script_path:
self.launchers_path = os.path.abspath(script_path)
else:
self.launchers_path = os.getcwd()
# Determine default paths
self.root_path = os.path.normpath(os.path.join(self.launchers_path, ".."))
self.bin_path = os.path.join(self.root_path, "bin")
self.data_path = os.path.join(self.root_path, "data")
self.conf_path = os.path.join(self.root_path, "conf")
self.workflows_path = os.path.join(self.root_path, "workflows")
self.runtime_path = os.path.join(self.root_path, "runtime")
# Parse arguments
parser = ArgumentParser(
prog = "run " + self.cmd_name,
epilog = self.epilog,
formatter_class=RawDescriptionHelpFormatter)
parser.add_argument('--version', action='version', version='%(prog)s ' + VERSION)
self.add_arguments(parser)
parser.add_argument("-L", "--log-level", dest="log_level",
default=None, choices=["debug", "info", "warn", "error", "critical", "notset"],
help="Define the logging level")
self.args = parser.parse_args()
# Create preliminary logger
level = self.args.log_level or "info"
logger.initialize(level=level)
self.log = logger.get_logger(logger_name, level=level)
def main():
conf = Config()
initialize({"log":{"level":"debug"}})
log = get_logger("mrna_preproc_counts")
log.setLevel(logging.DEBUG)
log.info("Querying log2r tumour units ...")
es = EntityServer(conf["entities"])
em = es.manager()
def __init__(self, conf, logger_name=None):
self._conf = conf
if hasattr(self, "plugin_required_conf"):
missing_conf = []
for arg in self.plugin_required_conf:
if arg not in conf:
missing_conf += [arg]
if len(missing_conf) > 0:
raise MissingConfigParamsError(missing_conf)
if logger_name is not None:
self._log = logger.get_logger(logger_name)
def __init__(self, config, paths, logger=None):
"""
:param config: intogensm.config.OncodriveClustConfig
:param paths: PathsConfig
:param logger: Optional logger
"""
self.config = config
self.paths = paths
self.logger = logger or get_logger("oncodriveclust")
self._load_config()
def __init__(self, conf, logger_name=None): self._conf = conf if hasattr(self, "plugin_required_conf"): missing_conf = [] for arg in self.plugin_required_conf: if arg not in conf: missing_conf += [arg] if len(missing_conf) > 0: raise MissingConfigParamsError(missing_conf) if logger_name is not None: self._log = logger.get_logger(logger_name)
def run(self):
"Run the command and execute the command"
self.build_conf()
# Expand configuration variables
self.conf = self.conf_builder.get_conf()
self.conf.expand_vars()
#self.log.debug(repr(self.conf))
# Validate and process configuration
self.process_conf()
# Regenerate configuration
self.conf = self.conf_builder.get_conf()
# Final logging configuration
log = logger.get_logger("")
log.removeHandler(log.handlers[0])
logging_conf = self.conf.get("wok.logging")
logger.initialize(logging_conf)
# Show some debugging information
self.log.debug("Root path = {}".format(self.root_path))
self.log.debug("Conf path = {}".format(self.conf_path))
self.log.debug("Data path = {}".format(self.data_path))
self.log.debug("Workflows path = {}".format(self.workflows_path))
self.log.debug("Runtime path = {}".format(self.runtime_path))
self.log.debug("Results path = {}".format(self.results_path))
self.log.debug("Temp path = {}".format(self.temp_path))
if len(self.user_conf_files) > 0:
self.log.debug("User defined configuration files:\n{}".format(
"\n".join(" {}".format(cf) for cf in self.user_conf_files)))
self.log.debug("Effective configuration: " + str(self.conf))
# Execute
try:
self.execute()
except Exception as ex:
self.log.exception(ex)
return -1
return 0
def load_conf(self,
conf_files=None,
path_env="WEB_CONF_PATH",
files_env="WEB_CONF_FILES",
default_ext=".cfg"):
logger = get_logger("wok.server", level="info")
logger.info("Loading Flask configuration ...")
conf_path = os.environ.get(path_env, os.getcwd())
conf_files = _get_conf_files(conf_path, conf_files or [],
os.environ.get(files_env), default_ext)
for cfg_path in conf_files:
logger.info("+++ {}".format(cfg_path))
self.config.from_pyfile(cfg_path)
def __init__(self, app=None, start_engine=True, **kwargs):
self.app = app
self._start_engine = start_engine
self.engine = None
self.lock = Lock()
self._initialized = False
self.logger = logger.get_logger("wok.server", level="info")
if app is not None:
self.init_app(app, **kwargs)
def __init__(self, app=None, start_engine=True, **kwargs):
self.app = app
self._start_engine = start_engine
self.engine = None
self.lock = Lock()
self._initialized = False
self.logger = logger.get_logger("wok.server", level="info")
if app is not None:
self.init_app(app, **kwargs)
def download(self,
target_path,
prefix=None,
delimiter=None,
include=None,
exclude=None,
start_callback=None):
"""
Download to a local path the selected objects in the container.
:param prefix: download only the objects which name starts with this prefix. Useful for pseudo-hierarchical navigation.
:param delimiter: download distinct objects with names up to either the first delimiter or the end. Useful for pseudo-hierarchical navigation.
"""
for obj_name in self.list_objects(prefix, delimiter):
if include is not None and not self.__filename_match(obj_name, include)\
or exclude is not None and self.__filename_match(obj_name, exclude):
continue
path = os.path.normpath(obj_name)
path = os.path.normpath("{}/{}".format(target_path, obj_name))
if os.path.relpath(path, target_path).startswith(".."):
get_logger(__name__).warn(
"Object output path goes outside the target path: {}".
format(path))
continue
#print "***", obj_name, "--->", path
base_path = os.path.dirname(path)
if not os.path.exists(base_path):
os.makedirs(base_path)
prefixed_obj_name = (prefix or "") + obj_name
obj = self.get_object(prefixed_obj_name)
if obj.exists():
start_callback(obj_name)
obj.get_data(path)
def load(self):
log = get_logger(__name__)
log.info("Loading project data definitions ...")
log.debug("Paths:\n{0}".format("\n ".join(self.paths)))
log.debug("Includes:\n{0}".format("\n ".join(self.includes)))
log.debug("Excludes:\n{0}".format("\n ".join(self.excludes)))
# compile regular expressions
includes = [re.compile(inc) for inc in self.includes]
excludes = [re.compile(exc) for exc in self.excludes]
# scan paths
project_ids = set()
projects = []
for scan_path in self.paths:
for path, project in list_projects(log, scan_path):
if "id" not in project:
log.warn("Discarding project that doesn't have 'id': {0}".format(path))
continue
if "files" not in project:
log.warn("Discarding project that doesn't have 'files': {0}".format(path))
continue
project_id = project["id"]
project_name = ": " + project["name"] if "name" in project else ""
if match_id(project_id, includes) and not match_id(project_id, excludes):
if project_id in project_ids:
msg = "Duplicated project id at {0}".format(path)
if self.ignore_duplicates:
log.warn(msg)
continue
else:
raise Exception(msg)
log.info("Included {0}{1}".format(project_id, project_name))
self.__format_project(log, project, base_path=os.path.dirname(path))
projects += [project]
else:
log.info("Excluded {0}{1}".format(project_id, project_name))
return projects
def __init__(self, conf):
Synchronizable.__init__(self)
self.conf = conf
wok_conf = conf["wok"]
self._log = logger.get_logger(wok_conf.get("log"), "wok-engine")
self._work_path = wok_conf.get("work_path", os.path.join(os.getcwd(), "wok"))
self._flow_path = wok_conf.get("flow_path")
if self._flow_path is None:
self._flow_path = [os.curdir]
elif not isinstance(self._flow_path, (list, DataList)):
raise Exception('wok.flow_path: A list of paths expected. Example ["path1", "path2"]')
self._flow_loader = FlowLoader(self._flow_path)
sb = ["wok.flow_path:\n"]
for uri, ff in self._flow_loader.flow_files.items():
sb += ["\t", uri, "\t", ff[0], "\n"]
self._log.debug("".join(sb))
self._instances = []
self._instances_map = {}
#self._lock = Lock()
self._cvar = Condition(self._lock)
self._run_thread = None
self._running = False
self._job_task_map = {}
self._logs_threads = []
self._logs_queue = Queue()
self._join_thread = None
self._join_queue = Queue()
self._notified = False
self._job_mgr = self._create_job_manager(wok_conf)
self._storage = self._create_storage(wok_conf)
#TODO self._db = SqliteEngineDB(self)
self._restore_state()
def __init__(self, name, conf_builder, project, flow, container_name,
engine):
self.name = name
self.conf_builder = conf_builder
self.project = project
self.root_flow = flow
self.container_name = container_name
self.engine = engine
self.title = flow.title
self.created = datetime.now()
self.flow_uri = "{}:{}".format(project.name,
flow.name) # FIXME flow.uri
self._log = logger.get_logger("wok.case")
self._initialize_conf()
# reset state
self._state = runstates.PAUSED
# components by name
self._component_by_cname = None
# create nodes tree
self.root_node = self._create_tree(self.root_flow)
# calculate components configuration
self._apply_component_conf()
# connect ports and create dependencies
self._connect_ports_and_create_deps()
# calculate the list of tasks in dependency order
self.tasks = self._tasks_in_dependency_order()
# list of used platforms
self.platforms = self._used_platforms()
# calculate priorities
for node in self.root_node.children:
self._calculate_priorities(node)
# Track the number of active workitems
self.num_active_workitems = 0
# Whether this case is marked for removal when all the jobs finish
self.removed = False
def __init__(self, name, engine, conf):
self.name = name
self.engine = engine
self.conf = conf
self.__next_id = 1
self._log = logger.get_logger(conf.get("log"), name)
if "work_path" not in self.conf:
raise Exception("Missing 'work_path' in configuration")
self._work_path = conf["work_path"]
self._output_path = self.conf.get("output_path", os.path.join(self._work_path, "output"))
def run(self):
"Run the command and execute the command"
self.build_conf()
# Expand configuration variables
self.engine_conf = self.engine_conf_args.validated_conf()
self.case_conf = self.case_conf_args.validated_conf(expand_vars=False)
# Validate and process configuration
self.process_conf()
# Regenerate configuration
self.engine_conf = self.engine_conf_builder.get_conf()
self.case_conf = self.case_conf_builder.get_conf()
# Final logging configuration
log = logger.get_logger("")
log.removeHandler(log.handlers[0])
logging_conf = self.engine_conf.get("wok.logging")
logger.initialize(logging_conf)
# Show some debugging information
self.log.debug("Root path = {}".format(self.root_path))
self.log.debug("Conf path = {}".format(self.conf_path))
#self.log.debug("Runtime path = {}".format(self.runtime_path))
#self.log.debug("Results path = {}".format(self.results_path))
#self.log.debug("Temp path = {}".format(self.temp_path))
self.engine_conf_args.log_debug()
self.case_conf_args.log_debug()
# Execute
try:
self.execute()
except BaseException as ex:
self.log.error(str(ex))
from traceback import format_exc
self.log.debug(format_exc())
return -1
return 0
def __init__(self, name, conf_builder, project, flow, container_name, engine):
self.name = name
self.conf_builder = conf_builder
self.project = project
self.root_flow = flow
self.container_name = container_name
self.engine = engine
self.title = flow.title
self.created = datetime.now()
self.flow_uri = "{}:{}".format(project.name, flow.name) # FIXME flow.uri
self._log = logger.get_logger("wok.case")
self._initialize_conf()
# reset state
self._state = runstates.PAUSED
# components by name
self._component_by_cname = None
# create nodes tree
self.root_node = self._create_tree(self.root_flow)
# calculate components configuration
self._apply_component_conf()
# connect ports and create dependencies
self._connect_ports_and_create_deps()
# calculate the list of tasks in dependency order
self.tasks = self._tasks_in_dependency_order()
# list of used platforms
self.platforms = self._used_platforms()
# calculate priorities
for node in self.root_node.children:
self._calculate_priorities(node)
# Track the number of active workitems
self.num_active_workitems = 0
# Whether this case is marked for removal when all the jobs finish
self.removed = False
def _init_conf(self, app, conf_files, path_var, files_var, args_var):
self.logger.info("Checking Wok configuration files ...")
args = []
conf_path = app.config.get(path_var,
os.environ.get(path_var, os.getcwd()))
wok_conf_files = _get_conf_files(
conf_path, conf_files or [],
app.config.get(files_var, os.environ.get(files_var)), ".conf")
for path in wok_conf_files:
if not os.path.exists(path):
self.logger.error("--- {} (not found)".format(path))
continue
self.logger.info("+++ {}".format(path))
args += ["-c", path]
env_args = os.environ.get(args_var)
if env_args is not None:
env_args = env_args.strip().split(" ")
wok_conf_args = app.config.get(args_var, env_args)
if wok_conf_args is not None:
args += wok_conf_args
self.logger.debug("Arguments: {}".format(" ".join(wok_conf_args)))
self.logger.info("Loading Wok configuration ...")
parser = ArgumentParser()
wok_args = Arguments(parser, case_name_args=False, logger=self.logger)
wok_args.initialize(parser.parse_args(args))
self.conf_builder = wok_args.engine_conf_builder
self.conf = wok_args.engine_conf.expand_vars()
self.case_conf_builder = wok_args.case_conf_builder
self.case_conf = wok_args.case_conf.expand_vars()
# initialize logging according to the configuration
log = logger.get_logger("")
log.removeHandler(log.handlers[0])
logging_conf = self.conf.clone().expand_vars().get("wok.logging")
logger.initialize(logging_conf)
self.logger.debug(repr(self.conf))
def _init_conf(self, app, conf_files, path_var, files_var, args_var):
self.logger.info("Checking Wok configuration files ...")
args = []
conf_path = app.config.get(path_var, os.environ.get(path_var, os.getcwd()))
wok_conf_files = _get_conf_files(conf_path, conf_files or [],
app.config.get(files_var, os.environ.get(files_var)), ".conf")
for path in wok_conf_files:
if not os.path.exists(path):
self.logger.error("--- {} (not found)".format(path))
continue
self.logger.info("+++ {}".format(path))
args += ["-c", path]
env_args = os.environ.get(args_var)
if env_args is not None:
env_args = env_args.strip().split(" ")
wok_conf_args = app.config.get(args_var, env_args)
if wok_conf_args is not None:
args += wok_conf_args
self.logger.debug("Arguments: {}".format(" ".join(wok_conf_args)))
self.logger.info("Loading Wok configuration ...")
parser = ArgumentParser()
wok_args = Arguments(parser, case_name_args=False, logger=self.logger)
wok_args.initialize(parser.parse_args(args))
self.conf_builder = wok_args.engine_conf_builder
self.conf = wok_args.engine_conf.expand_vars()
self.case_conf_builder = wok_args.case_conf_builder
self.case_conf = wok_args.case_conf.expand_vars()
# initialize logging according to the configuration
log = logger.get_logger("")
log.removeHandler(log.handlers[0])
logging_conf = self.conf.clone().expand_vars().get("wok.logging")
logger.initialize(logging_conf)
self.logger.debug(repr(self.conf))
def __init__(self, conf, flow = None):
self.conf = conf
wok_conf = conf["wok"]
self._log = logger.get_logger(wok_conf.get("log"), "engine")
self._instance_name = wok_conf["__instance.name"]
self._work_path = wok_conf.get("work_path", os.path.join(os.getcwd(), "wok"))
self._output_path = os.path.join(self._work_path, "output")
self._ports_path = os.path.join(self._work_path, "ports")
self._tasks_path = os.path.join(self._work_path, "tasks")
if "port_map" in wok_conf:
self._port_data_conf = wok_conf["port_map"]
else:
self._port_data_conf = wok_conf.create_element()
self._autorm_task = wok_conf.get("auto_remove.task", False, dtype=bool)
self._clean = wok_conf.get("clean", True, dtype=bool)
self._stop_on_errors = wok_conf.get("stop_on_errors", True, dtype=bool)
self._maxpar = wok_conf.get("defaults.maxpar", 0, dtype=int)
self._wsize = wok_conf.get("defaults.wsize", 0, dtype=int)
self._start_module = wok_conf.get("start_module")
self._flow = None
self._state = WokEngine.S_UNINITIALIZED
self._run_thread = None
self._job_sched = self._create_job_scheduler(wok_conf)
self._run_lock = Lock()
if flow is not None:
self._initialize(flow)
def __init__(self, name, conf):
self._conf = conf
self._name = conf.get("name", name)
self._log = logger.get_logger("wok.platform.{}".format(name))
if "work_path" not in self._conf:
raise MissingConfigParamError("work_path")
self._work_path = conf["work_path"]
if not os.path.exists(self._work_path):
os.makedirs(self._work_path)
self._data = self._create_data_provider()
self._storage = self._create_storage()
self._job_manager = self._create_job_manager()
self._callbacks = CallbackManager(delegates=[(events.JOB_UPDATE, self._job_manager.callbacks)])
def __init__(self, desc, name=None, logger=None, base_path=None):
self._log = logger or woklogger.get_logger(_LOGGER_NAME)
self.name = name or desc.get("name")
self._flow_loader = None
platform = desc.get("platform")
self.path = desc.get("path")
self.conf = desc.get("conf", default=Data.element)
if isinstance(self.conf, basestring):
self.conf = ConfigLoader(os.path.join(base_path or "", self.conf)).load()
self.conf_rules = []
if self.path is not None:
self.conf_rules += [ConfRule(dict(set=[[rtconf.PROJECT_PATH, self.path]]), base_path, platform)]
self.conf_rules += [ConfRule(rule, base_path, platform) for rule in desc.get("conf_rules", default=Data.list)]
self.flows = desc.get("flows", default=Data.list)
def db_update(engine, session, logger=None):
if logger is None:
logger = woklogger.get_logger(__name__)
version_param = session.query(DbParam).filter(DbParam.name == "version").first()
if version_param is None:
version_param = DbParam(name="version", value=1)
session.add(version_param)
if version_param.value < DB_VERSION:
logger.info("Updating the server database ...")
for i in range(version_param.value, DB_VERSION):
logger.info(" {} --> {} ...".format(i, i + 1))
sts = __CHANGES[i - 1]
if isinstance(sts, basestring):
sts = [sts]
for sql in sts:
logger.debug(" " + sql)
engine.execute(sql)
version_param.value = DB_VERSION
def __init__(self, name, conf):
self._conf = conf
self._name = conf.get("name", name)
self._log = logger.get_logger("wok.platform.{}".format(name))
if "work_path" not in self._conf:
raise MissingConfigParamError("work_path")
self._work_path = conf["work_path"]
if not os.path.exists(self._work_path):
os.makedirs(self._work_path)
self._data = self._create_data_provider()
self._storage = self._create_storage()
self._job_manager = self._create_job_manager()
self._callbacks = CallbackManager(
delegates=[(events.JOB_UPDATE, self._job_manager.callbacks)])
def __init__(self, name, conf):
self._name = name
self._conf = conf
self._log = logger.get_logger("wok.jobs.{}".format(name))
if "work_path" not in self._conf:
raise MissingConfigParamError("work_path")
self._work_path = conf["work_path"]
if not os.path.exists(self._work_path):
os.makedirs(self._work_path)
db_path = os.path.join(self._work_path, "jobs.db")
# TODO if not in recover_mode then delete jobs.db
self._db = create_engine("sqlite:///{}".format(db_path))
self._session_factory = create_session_factory(self._db)
self._callbacks = CallbackManager(valid_events=[events.JOB_UPDATE])
# Lock for database jobs state changes
self._lock = threading.Lock()
def __init__(self, name, conf):
self._name = name
self._conf = conf
self._log = logger.get_logger("wok.jobs.{}".format(name))
if "work_path" not in self._conf:
raise MissingConfigParamError("work_path")
self._work_path = conf["work_path"]
if not os.path.exists(self._work_path):
os.makedirs(self._work_path)
db_path = os.path.join(self._work_path, "jobs.db")
# TODO if not in recover_mode then delete jobs.db
self._db = create_engine("sqlite:///{}".format(db_path))
self._session_factory = create_session_factory(self._db)
self._callbacks = CallbackManager(valid_events=[events.JOB_UPDATE])
# Lock for database jobs state changes
self._lock = threading.Lock()
def db_update(engine, session, logger=None):
if logger is None:
logger = woklogger.get_logger(__name__)
version_param = session.query(DbParam).filter(
DbParam.name == "version").first()
if version_param is None:
version_param = DbParam(name="version", value=1)
session.add(version_param)
if version_param.value < DB_VERSION:
logger.info("Updating the server database ...")
for i in range(version_param.value, DB_VERSION):
logger.info(" {} --> {} ...".format(i, i + 1))
sts = __CHANGES[i - 1]
if isinstance(sts, basestring):
sts = [sts]
for sql in sts:
logger.debug(" " + sql)
engine.execute(sql)
version_param.value = DB_VERSION
def initialize(self):
self.conf = self.conf_builder()
wok_conf = self.conf["wok"]
inst_conf = wok_conf["__instance"]
self._log = logger.get_logger(wok_conf.get("log"), self.name)
self._work_path = inst_conf["work_path"] # TODO deprecated
#TODO deprecated
self._port_data_conf = wok_conf.create_element()
self.default_maxpar = wok_conf.get("defaults.maxpar", 0, dtype=int)
self.default_wsize = wok_conf.get("defaults.wsize", 1, dtype=int)
self.root_flow = self.engine.flow_loader.load_from_file(self.flow_file)
wok_conf["__flow.name"] = self.root_flow.name
wok_conf["__flow.path"] = os.path.dirname(os.path.abspath(self.flow_file))
wok_conf["__flow.file"] = os.path.basename(self.flow_file)
# self._log.debug("\n" + repr(self.root_flow))
# create nodes tree
self.root_node = self._create_tree(self.root_flow, namespace = "")
# connect ports
self._connect_ports(self.root_node, namespace = self.root_node.name)
# calculcate dependencies
self._calculate_dependencies(self.root_node)
# calculate priorities
for m in self.root_node.modules:
self._calculate_priorities(m)
def __init__(self):
# Get task key and storage configuration
cmd_conf = OptionsConfig(
required=["case", "task", "index", "data.type", "storage.type"])
# Register signals
self._signal_names = {}
for signame in [x for x in dir(signal) if x.startswith("SIG")]:
try:
signum = getattr(signal, signame)
signal.signal(signum, self.__signal_handler)
self._signal_names[signum] = signame
except:
pass
# command line configuration
case_name = cmd_conf["case"]
task_cname = cmd_conf["task"]
workitem_index = cmd_conf["index"]
# initialize the data provider
provider_conf = cmd_conf["data"]
self._provider = data_provider_factory.create(provider_conf)
self._provider.start()
# initialize storage
storage_conf = cmd_conf["storage"]
self.storage = storage_factory.create(storage_conf)
self.storage = self.storage.get_container(case_name)
# load the module and task descriptors
task_desc = self._provider.load_task(case_name, task_cname)
workitem_desc = self._provider.load_workitem(case_name, task_cname,
workitem_index)
partition = workitem_desc["partition"]
# setup task configuration
self.conf = Data.create(task_desc["conf"])
self.conf["__task_index"] = workitem_index
self.conf.expand_vars()
# setup task attributes
self.case = workitem_desc["case"]
self.task = workitem_desc["task"]
self.id = workitem_desc["cname"]
self.name = workitem_desc["name"]
self.index = workitem_index
# initialize decorators
self._main = None
self._sources = []
self._foreach = None
self._begin = None
self._end = None
self._start_time = 0
self._end_time = self._start_time
# intialize task logging
log_conf = self.conf.get("logging")
logger.initialize(log_conf)
self.logger = logger.get_logger(self.name)
self.logger.debug("Task descriptor: {}".format(Data.create(task_desc)))
self.logger.debug("WorkItem descriptor: {}".format(
Data.create(workitem_desc)))
# Initialize input stream
self._stream = Stream(self._provider, task_desc["stream"])
# Initialize ports
self._ports = {}
self._in_ports = []
self._out_ports = []
self._open_ports = {}
if "ports" in task_desc and "ports" in partition:
port_descriptors = Data.create(task_desc["ports"])
for port_desc in port_descriptors.get("in", default=list):
port_desc["mode"] = PORT_MODE_IN
self._ports[port_desc["name"]] = port_desc
self._in_ports += [port_desc]
for port_desc in port_descriptors.get("out", default=list):
port_desc["mode"] = PORT_MODE_OUT
self._ports[port_desc["name"]] = port_desc
self._out_ports += [port_desc]
port_descriptors = Data.create(partition["ports"])
for port_desc in port_descriptors.get("in", default=list):
task_port_desc = self._ports[port_desc["name"]]
task_port_desc["data"] = port_desc["data"]
for port_desc in port_descriptors.get("out", default=list):
task_port_desc = self._ports[port_desc["name"]]
task_port_desc["data"] = port_desc["data"]
# Get hostname
try:
import socket
self.hostname = socket.gethostname()
except:
self.hostname = "unknown"
# The context field is free to be used by the task user to
# save variables related with the whole task life cycle.
# By default it is initialized with a dictionary but can be
# overwrote with any value by the user. Wok will never use it.
self.context = {}
def missing_fields(self, keys):
from wok.logger import get_logger
get_logger(__name__).warn("{0}.missing_fields() is deprecated, "
"use {0}.missing_keys() instead".format(
self.__class__.__name__))
return self.missing_keys(keys)
