Merge branch 'master' into singularity-support

Author: Anton Khodak

.travis.yml

@@ -1,5 +1,6 @@
 sudo: false
 language: python
+cache: pip
 python:
   - 2.7
   - 3.3
@@ -10,9 +11,13 @@ os:
   - linux
 install:
   - pip install tox-travis
+jobs:
+  include:
+    - stage: release-test
+      script: RELEASE_SKIP=head ./release-test.sh
 script: tox
 branches:
   only:
     - master
 notifications:
-  email: false
+  email: false

Makefile

@@ -57,7 +57,7 @@ install: FORCE
 dist: dist/${MODULE}-$(VERSION).tar.gz
 
 dist/${MODULE}-$(VERSION).tar.gz: $(SOURCES)
-	./setup.py sdist
+	./setup.py sdist bdist_wheel
 
 ## clean       : clean up all temporary / machine-generated files
 clean: FORCE

appveyor.yml

@@ -31,6 +31,7 @@ install:
   - "python -c \"import struct; print(struct.calcsize('P') * 8)\""
 
 build_script:
+  - "%CMD_IN_ENV% pip install -U setuptools pip"
   - "%CMD_IN_ENV% pip install ."
 
 

cwltool/argparser.py

@@ -9,6 +9,7 @@
 import tempfile
 from io import open
 
+import datetime
 import requests
 from typing import (Dict, List, Text, Any, MutableMapping)
 
@@ -180,8 +181,9 @@ def add_volumes(self, pathmapper, runtime):
                         docker_windows_path_adjust(createtmp),
                         docker_windows_path_adjust(vol.target)))
 
-    def create_runtime(self, env, rm_container=True, **kwargs):
-        # type: (MutableMapping[Text, Text], bool, **Any) -> List
+    def create_runtime(self, env, rm_container=True, record_container_id=False, cidfile_dir="",
+                       cidfile_prefix="", **kwargs):
+        # type: (MutableMapping[Text, Text], bool, bool, Text, Text, **Any) -> List
         user_space_docker_cmd = kwargs.get("user_space_docker_cmd")
         if user_space_docker_cmd:
             runtime = [user_space_docker_cmd, u"run"]
@@ -236,6 +238,26 @@ def create_runtime(self, env, rm_container=True, **kwargs):
         # runtime.append("--env=HOME=/tmp")
         runtime.append(u"--env=HOME=%s" % self.builder.outdir)
 
+        # add parameters to docker to write a container ID file
+        if record_container_id:
+            if cidfile_dir != "":
+                if not os.path.isdir(cidfile_dir):
+                    _logger.error("--cidfile-dir %s error:\n%s", cidfile_dir,
+                                  cidfile_dir + " is not a directory or "
+                                                "directory doesn't exist, please check it first")
+                    exit(2)
+                if not os.path.exists(cidfile_dir):
+                    _logger.error("--cidfile-dir %s error:\n%s", cidfile_dir,
+                                  "directory doesn't exist, please create it first")
+                    exit(2)
+            else:
+                cidfile_dir = os.getcwd()
+            cidfile_name = datetime.datetime.now().strftime("%Y%m%d%H%M%S-%f") + ".cid"
+            if cidfile_prefix != "":
+                cidfile_name = str(cidfile_prefix + "-" + cidfile_name)
+            cidfile_path = os.path.join(cidfile_dir, cidfile_name)
+            runtime.append(u"--cidfile=%s" % cidfile_path)
+
         for t, v in self.environment.items():
             runtime.append(u"--env=%s=%s" % (t, v))
 

cwltool/docker.py

@@ -153,7 +153,8 @@ def revmap_file(builder, outdir, f):
         return f
 
 
-    raise WorkflowException(u"Output File object is missing both `location` and `path` fields: %s" % f)
+    raise WorkflowException(u"Output File object is missing both 'location' "
+            "and 'path' fields: %s" % f)
 
 
 class CallbackJob(object):

cwltool/draft2tool.py

@@ -0,0 +1,170 @@
+import logging
+import tempfile
+import threading
+
+import os
+from abc import ABCMeta, abstractmethod
+
+from typing import Dict, Text, Any, Tuple, Set, List
+
+from .builder import Builder
+from .errors import WorkflowException
+from .mutation import MutationManager
+from .job import JobBase
+from .process import relocateOutputs, cleanIntermediate, Process
+from . import loghandler
+
+_logger = logging.getLogger("cwltool")
+
+class JobExecutor(object):
+    __metaclass__ = ABCMeta
+
+    def __init__(self):
+        # type: (...) -> None
+        self.final_output = []  # type: List
+        self.final_status = []  # type: List
+        self.output_dirs = set()  # type: Set
+
+    def __call__(self, *args, **kwargs):
+        return self.execute(*args, **kwargs)
+
+    def output_callback(self, out, processStatus):
+        self.final_status.append(processStatus)
+        self.final_output.append(out)
+
+    @abstractmethod
+    def run_jobs(self,
+                 t,  # type: Process
+                 job_order_object,  # type: Dict[Text, Any]
+                 logger,
+                 **kwargs  # type: Any
+                 ):
+        pass
+
+    def execute(self, t,  # type: Process
+                job_order_object,  # type: Dict[Text, Any]
+                logger=_logger,
+                **kwargs  # type: Any
+                ):
+        # type: (...) -> Tuple[Dict[Text, Any], Text]
+
+        if "basedir" not in kwargs:
+            raise WorkflowException("Must provide 'basedir' in kwargs")
+
+        finaloutdir = os.path.abspath(kwargs.get("outdir")) if kwargs.get("outdir") else None
+        kwargs["outdir"] = tempfile.mkdtemp(prefix=kwargs["tmp_outdir_prefix"]) if kwargs.get(
+            "tmp_outdir_prefix") else tempfile.mkdtemp()
+        self.output_dirs.add(kwargs["outdir"])
+        kwargs["mutation_manager"] = MutationManager()
+
+        jobReqs = None
+        if "cwl:requirements" in job_order_object:
+            jobReqs = job_order_object["cwl:requirements"]
+        elif ("cwl:defaults" in t.metadata and "cwl:requirements" in t.metadata["cwl:defaults"]):
+            jobReqs = t.metadata["cwl:defaults"]["cwl:requirements"]
+        if jobReqs:
+            for req in jobReqs:
+                t.requirements.append(req)
+
+        self.run_jobs(t, job_order_object, logger, **kwargs)
+
+        if self.final_output and self.final_output[0] and finaloutdir:
+            self.final_output[0] = relocateOutputs(self.final_output[0], finaloutdir,
+                                                   self.output_dirs, kwargs.get("move_outputs"),
+                                                   kwargs["make_fs_access"](""))
+
+        if kwargs.get("rm_tmpdir"):
+            cleanIntermediate(self.output_dirs)
+
+        if self.final_output and self.final_status:
+            return (self.final_output[0], self.final_status[0])
+        else:
+            return (None, "permanentFail")
+
+
+class SingleJobExecutor(JobExecutor):
+    def run_jobs(self,
+                 t,  # type: Process
+                 job_order_object,  # type: Dict[Text, Any]
+                 logger,
+                 **kwargs  # type: Any
+                 ):
+        jobiter = t.job(job_order_object,
+                        self.output_callback,
+                        **kwargs)
+
+        try:
+            for r in jobiter:
+                if r:
+                    builder = kwargs.get("builder", None)  # type: Builder
+                    if builder is not None:
+                        r.builder = builder
+                    if r.outdir:
+                        self.output_dirs.add(r.outdir)
+                    r.run(**kwargs)
+                else:
+                    logger.error("Workflow cannot make any more progress.")
+                    break
+        except WorkflowException:
+            raise
+        except Exception as e:
+            logger.exception("Got workflow error")
+            raise WorkflowException(Text(e))
+
+
+class MultithreadedJobExecutor(JobExecutor):
+    def __init__(self):
+        super(MultithreadedJobExecutor, self).__init__()
+        self.threads = set()
+        self.exceptions = []
+
+    def run_job(self,
+                job, # type:  JobBase
+                **kwargs  # type:  Any
+                ):
+        # type: (...) -> None
+        def runner():
+            try:
+                job.run(**kwargs)
+            except WorkflowException as e:
+                self.exceptions.append(e)
+            except Exception as e:
+                self.exceptions.append(WorkflowException(Text(e)))
+
+            self.threads.remove(thread)
+
+        thread = threading.Thread(target=runner)
+        thread.daemon = True
+        self.threads.add(thread)
+        thread.start()
+
+    def wait_for_next_completion(self):  # type: () -> None
+        if self.exceptions:
+            raise self.exceptions[0]
+
+    def run_jobs(self,
+                 t,  # type: Process
+                 job_order_object,  # type: Dict[Text, Any]
+                 logger,
+                 **kwargs  # type: Any
+                 ):
+
+        jobiter = t.job(job_order_object, self.output_callback, **kwargs)
+
+        for r in jobiter:
+            if r:
+                builder = kwargs.get("builder", None)  # type: Builder
+                if builder is not None:
+                    r.builder = builder
+                if r.outdir:
+                    self.output_dirs.add(r.outdir)
+                self.run_job(r, **kwargs)
+            else:
+                if len(self.threads):
+                    self.wait_for_next_completion()
+                else:
+                    logger.error("Workflow cannot make any more progress.")
+                    break
+
+        while len(self.threads) > 0:
+            self.wait_for_next_completion()

cwltool/executors.py

@@ -185,7 +185,7 @@ def interpolate(scan, rootvars,
             e = evaluator(scan[w[0] + 1:w[1]], jslib, rootvars, fullJS=fullJS,
                           timeout=timeout, force_docker_pull=force_docker_pull,
                           debug=debug, js_console=js_console)
-            if w[0] == 0 and w[1] == len(scan):
+            if w[0] == 0 and w[1] == len(scan) and len(parts) <= 1:
                 return e
             leaf = json.dumps(e, sort_keys=True)
             if leaf[0] == '"':

cwltool/expression.py

@@ -3,7 +3,9 @@
 from typing import Callable as tCallable
 from typing import Any, Dict, Text, Tuple, Union
 
-from . import load_tool, main, workflow
+from . import load_tool, workflow
+from .argparser import get_default_args
+from .executors import SingleJobExecutor
 from .process import Process
 
 
@@ -35,13 +37,21 @@ class Factory(object):
     def __init__(self,
                  makeTool=workflow.defaultMakeTool,  # type: tCallable[[Any], Process]
                  # should be tCallable[[Dict[Text, Any], Any], Process] ?
-                 executor=main.single_job_executor,  # type: tCallable[...,Tuple[Dict[Text,Any], Text]]
+                 executor=None,  # type: tCallable[...,Tuple[Dict[Text,Any], Text]]
                  **execkwargs  # type: Any
                  ):
         # type: (...) -> None
         self.makeTool = makeTool
+        if executor is None:
+            executor = SingleJobExecutor()
         self.executor = executor
-        self.execkwargs = execkwargs
+
+        kwargs = get_default_args()
+        kwargs.pop("job_order")
+        kwargs.pop("workflow")
+        kwargs.pop("outdir")
+        kwargs.update(execkwargs)
+        self.execkwargs = kwargs
 
     def make(self, cwl):
         """Instantiate a CWL object from a CWl document."""

cwltool/factory.py

@@ -14,6 +14,7 @@
 import tempfile
 from abc import ABCMeta, abstractmethod
 from io import open
+from threading import Lock
 
 import shellescape
 from typing import (IO, Any, Callable, Dict, Iterable, List, MutableMapping, Text,
@@ -31,6 +32,8 @@
 
 needs_shell_quoting_re = re.compile(r"""(^$|[\s|&;()<>\'"$@])""")
 
+job_output_lock = Lock()
+
 FORCE_SHELLED_POPEN = os.getenv("CWLTOOL_FORCE_SHELL_POPEN", "0") == "1"
 
 SHELL_COMMAND_TEMPLATE = """#!/bin/bash
@@ -266,7 +269,8 @@ def _execute(self, runtime, env, rm_tmpdir=True, move_outputs="move"):
         if _logger.isEnabledFor(logging.DEBUG):
             _logger.debug(u"[job %s] %s", self.name, json.dumps(outputs, indent=4))
 
-        self.output_callback(outputs, processStatus)
+        with job_output_lock:
+            self.output_callback(outputs, processStatus)
 
         if self.stagedir and os.path.exists(self.stagedir):
             _logger.debug(u"[job %s] Removing input staging directory %s", self.name, self.stagedir)
@@ -320,13 +324,16 @@ def get_from_requirements(self, r, req, pull_image, dry_run=False):
         pass
 
     @abstractmethod
-    def create_runtime(self, env, rm_container=True, **kwargs):
-        # type: (MutableMapping[Text, Text], bool, **Any) -> List
+    def create_runtime(self, env, rm_container, record_container_id, cidfile_dir,
+                       cidfile_prefix, **kwargs):
+        # type: (MutableMapping[Text, Text], bool, bool, Text, Text, **Any) -> List
         pass
 
     def run(self, pull_image=True, rm_container=True,
+            record_container_id=False, cidfile_dir="",
+            cidfile_prefix="",
             rm_tmpdir=True, move_outputs="move", **kwargs):
-        # type: (bool, bool, bool, Text, **Any) -> None
+        # type: (bool, bool, bool, Text, Text, bool, Text, **Any) -> None
 
         (docker_req, docker_is_req) = get_feature(self, "DockerRequirement")
 
@@ -372,7 +379,7 @@ def run(self, pull_image=True, rm_container=True,
                         "--user-space-docker-cmd.: {1}".format(container, e))
 
         self._setup(kwargs)
-        runtime = self.create_runtime(env, rm_container, **kwargs)
+        runtime = self.create_runtime(env, rm_container, record_container_id, cidfile_dir, cidfile_prefix, **kwargs)
         runtime.append(img_id)
 
         self._execute(runtime, env, rm_tmpdir=rm_tmpdir, move_outputs=move_outputs)