# -*- coding: utf-8 -*-
#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
#
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from __future__ import unicode_literals
import errno
import imp
import sys
import warnings
import psutil
from builtins import input
from past.builtins import basestring
from datetime import datetime
from functools import reduce
import os
import re
import signal
from jinja2 import Template
from airflow import configuration
from airflow.exceptions import AirflowException
# When killing processes, time to wait after issuing a SIGTERM before issuing a
# SIGKILL.
DEFAULT_TIME_TO_WAIT_AFTER_SIGTERM = configuration.conf.getint(
'core', 'KILLED_TASK_CLEANUP_TIME'
)
[docs]def validate_key(k, max_length=250):
if not isinstance(k, basestring):
raise TypeError("The key has to be a string")
elif len(k) > max_length:
raise AirflowException(
"The key has to be less than {0} characters".format(max_length))
elif not re.match(r'^[A-Za-z0-9_\-\.]+$', k):
raise AirflowException(
"The key ({k}) has to be made of alphanumeric characters, dashes, "
"dots and underscores exclusively".format(**locals()))
else:
return True
def alchemy_to_dict(obj):
"""
Transforms a SQLAlchemy model instance into a dictionary
"""
if not obj:
return None
d = {}
for c in obj.__table__.columns:
value = getattr(obj, c.name)
if type(value) == datetime:
value = value.isoformat()
d[c.name] = value
return d
def ask_yesno(question):
yes = set(['yes', 'y'])
no = set(['no', 'n'])
done = False
print(question)
while not done:
choice = input().lower()
if choice in yes:
return True
elif choice in no:
return False
else:
print("Please respond by yes or no.")
def is_in(obj, l):
"""
Checks whether an object is one of the item in the list.
This is different from ``in`` because ``in`` uses __cmp__ when
present. Here we change based on the object itself
"""
for item in l:
if item is obj:
return True
return False
[docs]def is_container(obj):
"""
Test if an object is a container (iterable) but not a string
"""
return hasattr(obj, '__iter__') and not isinstance(obj, basestring)
[docs]def as_tuple(obj):
"""
If obj is a container, returns obj as a tuple.
Otherwise, returns a tuple containing obj.
"""
if is_container(obj):
return tuple(obj)
else:
return tuple([obj])
def chunks(items, chunk_size):
"""
Yield successive chunks of a given size from a list of items
"""
if chunk_size <= 0:
raise ValueError('Chunk size must be a positive integer')
for i in range(0, len(items), chunk_size):
yield items[i:i + chunk_size]
def reduce_in_chunks(fn, iterable, initializer, chunk_size=0):
"""
Reduce the given list of items by splitting it into chunks
of the given size and passing each chunk through the reducer
"""
if len(iterable) == 0:
return initializer
if chunk_size == 0:
chunk_size = len(iterable)
return reduce(fn, chunks(iterable, chunk_size), initializer)
def as_flattened_list(iterable):
"""
Return an iterable with one level flattened
>>> as_flattened_list((('blue', 'red'), ('green', 'yellow', 'pink')))
['blue', 'red', 'green', 'yellow', 'pink']
"""
return [e for i in iterable for e in i]
def chain(*tasks):
"""
Given a number of tasks, builds a dependency chain.
chain(task_1, task_2, task_3, task_4)
is equivalent to
task_1.set_downstream(task_2)
task_2.set_downstream(task_3)
task_3.set_downstream(task_4)
"""
for up_task, down_task in zip(tasks[:-1], tasks[1:]):
up_task.set_downstream(down_task)
def cross_downstream(from_tasks, to_tasks):
r"""
Set downstream dependencies for all tasks in from_tasks to all tasks in to_tasks.
E.g.: cross_downstream(from_tasks=[t1, t2, t3], to_tasks=[t4, t5, t6])
Is equivalent to:
t1 --> t4
\ /
t2 -X> t5
/ \
t3 --> t6
t1.set_downstream(t4)
t1.set_downstream(t5)
t1.set_downstream(t6)
t2.set_downstream(t4)
t2.set_downstream(t5)
t2.set_downstream(t6)
t3.set_downstream(t4)
t3.set_downstream(t5)
t3.set_downstream(t6)
:param from_tasks: List of tasks to start from.
:type from_tasks: List[airflow.models.BaseOperator]
:param to_tasks: List of tasks to set as downstream dependencies.
:type to_tasks: List[airflow.models.BaseOperator]
"""
for task in from_tasks:
task.set_downstream(to_tasks)
[docs]def pprinttable(rows):
"""Returns a pretty ascii table from tuples
If namedtuple are used, the table will have headers
"""
if not rows:
return
if hasattr(rows[0], '_fields'): # if namedtuple
headers = rows[0]._fields
else:
headers = ["col{}".format(i) for i in range(len(rows[0]))]
lens = [len(s) for s in headers]
for row in rows:
for i in range(len(rows[0])):
slenght = len("{}".format(row[i]))
if slenght > lens[i]:
lens[i] = slenght
formats = []
hformats = []
for i in range(len(rows[0])):
if isinstance(rows[0][i], int):
formats.append("%%%dd" % lens[i])
else:
formats.append("%%-%ds" % lens[i])
hformats.append("%%-%ds" % lens[i])
pattern = " | ".join(formats)
hpattern = " | ".join(hformats)
separator = "-+-".join(['-' * n for n in lens])
s = ""
s += separator + '\n'
s += (hpattern % tuple(headers)) + '\n'
s += separator + '\n'
def f(t):
return "{}".format(t) if isinstance(t, basestring) else t
for line in rows:
s += pattern % tuple(f(t) for t in line) + '\n'
s += separator + '\n'
return s
def reap_process_group(pid, log, sig=signal.SIGTERM,
timeout=DEFAULT_TIME_TO_WAIT_AFTER_SIGTERM):
"""
Tries really hard to terminate all children (including grandchildren). Will send
sig (SIGTERM) to the process group of pid. If any process is alive after timeout
a SIGKILL will be send.
:param log: log handler
:param pid: pid to kill
:param sig: signal type
:param timeout: how much time a process has to terminate
"""
def on_terminate(p):
log.info("Process %s (%s) terminated with exit code %s", p, p.pid, p.returncode)
if pid == os.getpid():
raise RuntimeError("I refuse to kill myself")
parent = psutil.Process(pid)
children = parent.children(recursive=True)
children.append(parent)
try:
pg = os.getpgid(pid)
except OSError as err:
# Skip if not such process - we experience a race and it just terminated
if err.errno == errno.ESRCH:
return
raise
log.info("Sending %s to GPID %s", sig, pg)
os.killpg(os.getpgid(pid), sig)
gone, alive = psutil.wait_procs(children, timeout=timeout, callback=on_terminate)
if alive:
for p in alive:
log.warn("process %s (%s) did not respond to SIGTERM. Trying SIGKILL", p, pid)
os.killpg(os.getpgid(pid), signal.SIGKILL)
gone, alive = psutil.wait_procs(alive, timeout=timeout, callback=on_terminate)
if alive:
for p in alive:
log.error("Process %s (%s) could not be killed. Giving up.", p, p.pid)
def parse_template_string(template_string):
if "{{" in template_string: # jinja mode
return None, Template(template_string)
else:
return template_string, None
class AirflowImporter(object):
"""
Importer that dynamically loads a class and module from its parent. This
allows Airflow to support ``from airflow.operators import BashOperator``
even though BashOperator is actually in
``airflow.operators.bash_operator``.
The importer also takes over for the parent_module by wrapping it. This is
required to support attribute-based usage:
.. code:: python
from airflow import operators
operators.BashOperator(...)
"""
def __init__(self, parent_module, module_attributes):
"""
:param parent_module: The string package name of the parent module. For
example, 'airflow.operators'
:type parent_module: str
:param module_attributes: The file to class mappings for all importable
classes.
:type module_attributes: str
"""
self._parent_module = parent_module
self._attribute_modules = self._build_attribute_modules(module_attributes)
self._loaded_modules = {}
# Wrap the module so we can take over __getattr__.
sys.modules[parent_module.__name__] = self
@staticmethod
def _build_attribute_modules(module_attributes):
"""
Flips and flattens the module_attributes dictionary from:
module => [Attribute, ...]
To:
Attribute => module
This is useful so that we can find the module to use, given an
attribute.
"""
attribute_modules = {}
for module, attributes in list(module_attributes.items()):
for attribute in attributes:
attribute_modules[attribute] = module
return attribute_modules
def _load_attribute(self, attribute):
"""
Load the class attribute if it hasn't been loaded yet, and return it.
"""
module = self._attribute_modules.get(attribute, False)
if not module:
# This shouldn't happen. The check happens in find_modules, too.
raise ImportError(attribute)
elif module not in self._loaded_modules:
# Note that it's very important to only load a given modules once.
# If they are loaded more than once, the memory reference to the
# class objects changes, and Python thinks that an object of type
# Foo that was declared before Foo's module was reloaded is no
# longer the same type as Foo after it's reloaded.
path = os.path.realpath(self._parent_module.__file__)
folder = os.path.dirname(path)
f, filename, description = imp.find_module(module, [folder])
self._loaded_modules[module] = imp.load_module(module, f, filename, description)
# This functionality is deprecated, and AirflowImporter should be
# removed in 2.0.
warnings.warn(
"Importing '{i}' directly from '{m}' has been "
"deprecated. Please import from "
"'{m}.[operator_module]' instead. Support for direct "
"imports will be dropped entirely in Airflow 2.0.".format(
i=attribute, m=self._parent_module.__name__),
DeprecationWarning)
loaded_module = self._loaded_modules[module]
return getattr(loaded_module, attribute)
def __getattr__(self, attribute):
"""
Get an attribute from the wrapped module. If the attribute doesn't
exist, try and import it as a class from a submodule.
This is a Python trick that allows the class to pretend it's a module,
so that attribute-based usage works:
from airflow import operators
operators.BashOperator(...)
It also allows normal from imports to work:
from airflow.operators.bash_operator import BashOperator
"""
if hasattr(self._parent_module, attribute):
# Always default to the parent module if the attribute exists.
return getattr(self._parent_module, attribute)
elif attribute in self._attribute_modules:
# Try and import the attribute if it's got a module defined.
loaded_attribute = self._load_attribute(attribute)
setattr(self, attribute, loaded_attribute)
return loaded_attribute
raise AttributeError
def render_log_filename(ti, try_number, filename_template):
"""
Given task instance, try_number, filename_template, return the rendered log filename
:param ti: task instance
:param try_number: try_number of the task
:param filename_template: filename template, which can be jinja template or python string template
"""
filename_template, filename_jinja_template = parse_template_string(filename_template)
if filename_jinja_template:
jinja_context = ti.get_template_context()
jinja_context['try_number'] = try_number
return filename_jinja_template.render(**jinja_context)
return filename_template.format(dag_id=ti.dag_id,
task_id=ti.task_id,
execution_date=ti.execution_date.isoformat(),
try_number=try_number)