workflow.py 17.1 KB
Newer Older
Jerome Mariette's avatar
Jerome Mariette committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
#
# Copyright (C) 2012 INRA
# 
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# 
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#

import inspect
import os
import sys
import uuid
import pkgutil
import tempfile
import pickle
import time
import threading
Jerome Mariette's avatar
Jerome Mariette committed
27
28
from ConfigParser import ConfigParser, NoOptionError
import types
Jerome Mariette's avatar
Jerome Mariette committed
29
30
31
32

import jflow.components
import jflow.utils as utils
from jflow.workflows_manager import WorkflowsManager, JFlowConfigReader
Jerome Mariette's avatar
Jerome Mariette committed
33
from jflow.parameter import Parameter, date
Jerome Mariette's avatar
Jerome Mariette committed
34
35
36
37
38
39
40
41
42
43
44
45

from weaver.script import ABSTRACTIONS
from weaver.script import DATASETS
from weaver.script import FUNCTIONS
from weaver.script import NESTS
from weaver.script import OPTIONS
from weaver.script import STACKS
from weaver.nest import Nest
from weaver.options import Options
from cctools.makeflow import MakeflowLog
from cctools.makeflow.log import Node

Jerome Mariette's avatar
Jerome Mariette committed
46

Jerome Mariette's avatar
Jerome Mariette committed
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
class Workflow(threading.Thread):
    """
    """

    PROPERTIES_FILE_NAME = "workflow.properties"
    MAKEFLOW_LOG_FILE_NAME = "Makeflow.makeflowlog"
    DUMP_FILE_NAME = "workflow.dump"
    WORKING = ".working"
    STATUS_STARTED = "started"
    STATUS_COMPLETED = "completed"
    STATUS_FAILED = "failed"
    STATUS_ABORTED = "aborted"
    
    def __init__(self, args={}, id=None):
        """ 
        """
        # define as a thread
        threading.Thread.__init__(self)
        self.jflow_config_reader = JFlowConfigReader()
        self.manager = WorkflowsManager()
        self.components = []
        self.makes = {}
        self.globals = {}
        self.options = Options()
        self.status = self.STATUS_STARTED
        self.start_time = None
        self.end_time = None
        self.comp_pckg = self._import_components()
        self.engine_arguments = ''
        self.component_nameids_is_init = False
        self.component_nameids = {}
        # try to parse engine arguments
        try:
            type, options = self.jflow_config_reader.get_batch()
            if type: self.engine_arguments += ' -T ' + type
            if options : self.engine_arguments += ' -B "' + options + '"'
        except: self.engine_arguments = None
Jerome Mariette's avatar
Jerome Mariette committed
84
85
86
87
88
89
        try:
            # Add config information
            self.name, self.description, self.parameters = self._get_from_config()
        except:
            raise IOError(self.__class__.__name__ + " workflow property file not found or invalid.")
        
Jerome Mariette's avatar
Jerome Mariette committed
90
        self.id = id
Jerome Mariette's avatar
Jerome Mariette committed
91
        self.args = self._extend_and_format_args(args)
Jerome Mariette's avatar
Jerome Mariette committed
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
        if self.id is not None:
            self.directory = self.manager.get_workflow_directory(self.name, self.id)
            if not os.path.isdir(self.directory):
                os.makedirs(self.directory, 0751)
            self._serialize()

    def __setstate__(self, state):
        self.__dict__ = state.copy()
        threading.Thread.__init__(self, name=self.name)
        
    def __getstate__(self):
        """
        Threading uses Lock Object, do not consider these objects when serializing a workflow
        """
        odict = self.__dict__.copy()
        del odict['_Thread__started']
        del odict['_Thread__block']
        del odict['_Thread__stderr']
        return odict
    
    def add_component(self, component_name, args=[], kwargs={}, component_prefix="default"):
        # first build and check if this component is OK
        my_pckge = __import__(self.comp_pckg[component_name], globals(), locals(), [component_name], -1)
        # build the object and define required field
        cmpt_object = getattr(my_pckge, component_name)()
        cmpt_object.output_directory = self.get_component_output_directory(component_name, component_prefix)
        cmpt_object.prefix = component_prefix
        if kwargs: cmpt_object.define_parameters(**kwargs)
        else: cmpt_object.define_parameters(*args)
        
        # if already init, add the component to the list and check if weaver should be executed
        if self.component_nameids_is_init:
            # add the component
            self.components.append(cmpt_object)
            # if this one require a dynamic pipeline, execute the first part
            if cmpt_object.is_dynamic:
                self._execute_weaver()
        else: 
            if self._component_is_duplicated(cmpt_object):
                raise ValueError("Component " + cmpt_object.__class__.__name__ + " with prefix " + 
                                 cmpt_object.prefix + " already exist in this pipeline!")
            self.component_nameids[cmpt_object.get_nameid()] = None
        
        return cmpt_object
    
    def pre_process(self):
        pass
    
    def process(self):
        """ 
        Run the workflow, has to be implemented by subclasses
        """
        raise NotImplementedError
    
    def post_process(self):
        pass
    
    def get_temporary_file(self, suffix=".txt"):
        # first check if tmp directory exists
        if not os.path.isdir(self.jflow_config_reader.get_tmp_directory()):
            os.makedirs(self.jflow_config_reader.get_tmp_directory(), 0751)
        tempfile_name = os.path.basename(tempfile.NamedTemporaryFile(suffix=suffix).name)
        return os.path.join(self.jflow_config_reader.get_tmp_directory(), tempfile_name)

    def get_component_output_directory(self, component_name, component_prefix):
        return os.path.join(self.directory, component_name + "_" + component_prefix)
    
    def get_components_nameid(self):
        return self.component_nameids.keys()
    
    def run(self):
        """
        Only require for Threading
        """
        self.start_time = time.time()
        self.status = self.STATUS_STARTED
        self.end_time = None  
        # serialize the workflow before pre processing
        self._serialize()
        self.pre_process()
        # serialize the workflow before running
        self._serialize()
        # if not done yet, init the component nameids list before to run
        if not self.component_nameids_is_init:
            self.process()
            self.component_nameids_is_init = True
            self._serialize()
        self.process()
        if len(self.components) > 0:
            self._execute_weaver()
        self._serialize()
        self.post_process()
        if self.status == self.STATUS_STARTED: self.status = self.STATUS_COMPLETED
        self.end_time = time.time()
        self._serialize()
    
Jerome Mariette's avatar
Jerome Mariette committed
188
189
190
191
192
193
194
195
196
197
    def _extend_and_format_args(self, args):
        
        extended_args = {}
        for param in self.parameters:
            try: args[param.name] = args[param.name].encode('ascii','ignore')
            except: pass
            # if this parameter has been modified by the user
            if args.has_key(param.name):
                # if it's an append parameter, the result should be a list
                if param.action == "append":
Jerome Mariette's avatar
Jerome Mariette committed
198
199
200
201
                    if args[param.name]:
                        try: value = args[param.name].split(",") 
                        except: value = args[param.name]
                    else: value = []
Jerome Mariette's avatar
Jerome Mariette committed
202
203
204
205
206
207
208
                    extended_args[param.name] = value
                # if it's a date
                elif param.type == date:
                    if isinstance(args[param.name], str):
                        extended_args[param.name] = date(args[param.name])
                    else:
                        extended_args[param.name] = args[param.name]
209
210
211
                # if it's a boolean
                elif param.type == types.BooleanType:
                    extended_args[param.name] = args[param.name] in ['true', '1', 't', 'y', 'yes']
Jerome Mariette's avatar
Jerome Mariette committed
212
213
214
215
216
217
218
                else:
                    extended_args[param.name] = args[param.name]
            # otherwise use the default
            else:
                extended_args[param.name] = param.default
        return extended_args
    
Jerome Mariette's avatar
Jerome Mariette committed
219
220
221
222
    def _get_from_config(self):
        reader = ConfigParser()
        reader.read(self._get_property_path())
        parameters = []
Jerome Mariette's avatar
Jerome Mariette committed
223
224
225
226
227
228
229
230
231
        try:
            params = {}
            # First grab all parameters name
            for param in reader.items("parameters"):
                cvalue = reader.get("parameters", param[0])
                cparam = param[0].split(".")[0]
                ckey = param[0].split(".")[1]
                if not params.has_key(cparam):
                    params[cparam] = {ckey: cvalue}
Jerome Mariette's avatar
Jerome Mariette committed
232
                else:
Jerome Mariette's avatar
Jerome Mariette committed
233
234
                    params[cparam][ckey] = cvalue
            for param in params:
Jerome Mariette's avatar
Jerome Mariette committed
235
                oparam = Parameter(**params[param])
Jerome Mariette's avatar
Jerome Mariette committed
236
237
238
                parameters.append(oparam)
        except:
            pass
Jerome Mariette's avatar
Jerome Mariette committed
239
240
241
242
        return [reader.get("global", "name"),
                reader.get("global", "description"),
                parameters]
    
Jerome Mariette's avatar
Jerome Mariette committed
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
    def _update_status_from_log(self):
        # first update the status from weaver folders
        # TODO update self.end_time
        try:
            working_directory = os.path.join(self.directory, self.WORKING)
            make_states = []
            for wdir in os.listdir(working_directory):
                log_path = os.path.join(working_directory, wdir, self.MAKEFLOW_LOG_FILE_NAME)
                log = MakeflowLog(log_path)
                log.parse()
                make_states.append(log.state)
            if self.STATUS_ABORTED in make_states:
                self.status = self.STATUS_ABORTED
            elif self.STATUS_FAILED in make_states:
                self.status = self.STATUS_FAILED
            elif self.status != self.STATUS_COMPLETED:
                self.status = self.STATUS_STARTED
        except: pass
        self._serialize()

    def get_status(self):
        self._update_status_from_log()
        return self.status
    
    def get_component_status(self, component_nameid):
        status = {"time": 0.0,
                  "tasks": 0,
                  "waiting": 0,
                  "running": 0,
                  "failed":0,
                  "aborted": 0,
                  "completed": 0}
        try:
            log = MakeflowLog(self.component_nameids[component_nameid])
            log.parse()
            symbols = set(n.symbol for n in log.nodes if n.symbol)
            if not symbols: return None
            for n in log.nodes:
                if not n.symbol: continue
                if n.symbol == component_nameid:
                    status["tasks"] += 1
                    status["time"]  += n.elapsed_time
                    if n.state == Node.WAITING:
                        status["waiting"] += 1
                    elif n.state == Node.RUNNING:
                        status["running"] += 1
                    elif n.state == Node.FAILED:
                        status["failed"] += 1
                    elif n.state == Node.ABORTED:
                        status["aborted"] += 1
                    elif n.state == Node.COMPLETED:
                        status["completed"] += 1
        except: pass
        return status        
    
    def makeflow_pretty_print_node(self, dag, node):
        sys.stdout.write('{0:>10} {1} {2}\n'.format('NODE', node.id, node.symbol))
        
        for output_file in sorted(node.output_files):
            sys.stdout.write('{0:>10} {1:>10} {2}\n'.format('', 'OUTPUT', output_file))
    
        for input_file in sorted(node.input_files):
            sys.stdout.write('{0:>10} {1:>10} {2}\n'.format('', 'INPUT', input_file))
            
        sys.stdout.write('{0:>10} {1:>10} {2}\n'.format('', 'COMMAND', node.command))

    def _execute_weaver(self, engine_wrapper=None):
        # Add nest path and path to script to Python module path to allow
        # for importing modules outside of $PYTHONPATH
        make_directory, new_make = self._get_current_make()
        current_working_directory = os.path.join(os.path.join(self.directory, self.WORKING), make_directory)
        sys.path.insert(0, os.path.abspath(os.path.dirname(current_working_directory)))

        # Load built-ins
        self._import('abstraction', ABSTRACTIONS)
        self._import('dataset', DATASETS)
        self._import('function', FUNCTIONS)
        self._import('nest', NESTS)
        self._import('options', OPTIONS)
        self._import('stack', STACKS)
                
        # Execute nest
        with Nest(current_working_directory, wrapper=engine_wrapper, path=self.jflow_config_reader.get_makeflow_path()) as nest:
            with self.options:
                if new_make:
                    try:
                        for component in self.components:
                            nest.symbol = component.get_nameid()
                            self.component_nameids[component.get_nameid()] = os.path.join(current_working_directory, self.MAKEFLOW_LOG_FILE_NAME)
                            component.execute()
                        # create the DAG
                        nest.compile()
                    except Exception as e:
                        sys.exit('Error compiling script: {0}'.format(e))
                        self.status = self.STATUS_FAILED
                        self._serialize()
                self.components = []
                # Once a weaver script is compiled, serialize the workflow
                self._serialize()
                try:
                    nest.execute(self.engine_arguments, exit_on_failure=True)
                except:
                    self.status = self.STATUS_FAILED
                    self.end_time = time.time()
                    self._serialize()
                    sys.exit()

    def _get_current_make(self):
        current_component, make_directory, new_make = [], None, False
        for component in self.components:
            current_component.append(component.get_nameid())
        for make in self.makes:
            if set(current_component) == set(self.makes[make]):
                make_directory = make
        # If the components in the queue have already been compiled
        if make_directory is None:
            make_directory = uuid.uuid4().hex[:10]
            self.makes[make_directory] = current_component
            new_make = True
        return [make_directory, new_make]
    
    def _serialize(self):
        self.dump_path = os.path.join(self.directory, self.DUMP_FILE_NAME)
        workflow_dump = open(self.dump_path, "wb")
        pickle.dump(self, workflow_dump)
        workflow_dump.close()

    def _component_is_duplicated(self, component):
        if component.get_nameid() in self.component_nameids.keys():
            return True
        return False

    def _get_property_path(self):
        """ 
        Return(type:string):  the path to the workflow properties file, None if does not exist
        """
        property_file = os.path.join(os.path.dirname(inspect.getfile(self.__class__)), self.PROPERTIES_FILE_NAME)
        return property_file if os.path.isfile(property_file) else None

    def _import_components(self):
        pckge = {}
        # first import global packages
        for importer, modname, ispkg in pkgutil.iter_modules(jflow.components.__path__, jflow.components.__name__ + "."):
            m = __import__(modname)
            for class_name, obj in inspect.getmembers(sys.modules[modname], inspect.isclass):
                if issubclass(obj, jflow.component.Component) and obj.__name__ != jflow.component.Component.__name__:
                    pckge[class_name] = modname
        # then import pipeline packages
        pipeline_dir = os.path.dirname(inspect.getfile(self.__class__))
        for importer, modname, ispkg in pkgutil.iter_modules([os.path.join(pipeline_dir, "components")], "workflows." + 
                                                             os.path.basename(pipeline_dir) + ".components."):
            m = __import__(modname)
            for class_name, obj in inspect.getmembers(sys.modules[modname], inspect.isclass):
                if issubclass(obj, jflow.component.Component) and obj.__name__ != jflow.component.Component.__name__:
                    pckge[class_name] = modname
        # finally import workflows shared packages
        workflows_dir = os.path.dirname(os.path.dirname(inspect.getfile(self.__class__)))
        for importer, modname, ispkg in pkgutil.iter_modules([os.path.join(workflows_dir, "components")], "workflows.components."):
            m = __import__(modname)
            for class_name, obj in inspect.getmembers(sys.modules[modname], inspect.isclass):
                if issubclass(obj, jflow.component.Component) and obj.__name__ != jflow.component.Component.__name__:
                    pckge[class_name] = modname
        return pckge

    def _import(self, module, symbols):
        """ Import ``symbols`` from ``module`` into global namespace. """
        # Import module
        m = 'weaver.{0}'.format(module)
        m = __import__(m, self.globals, self.globals, symbols, -1)

        # Import symbols from module into global namespace, which we store as
        # an attribute for later use (i.e. during compile)
        for symbol in symbols:
            self.globals[symbol] = getattr(m, symbol)