コード例 #1
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_transformation_catalog(self):
        tc = TransformationCatalog()
        wf = Workflow("wf")

        try:
            wf.add_transformation_catalog(tc)
        except:
            pytest.fail("should not have raised exception")
コード例 #2
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_replica_catalog(self):
        rc = ReplicaCatalog()
        wf = Workflow("wf")

        try:
            wf.add_replica_catalog(rc)
        except:
            pytest.fail("should not have raised exception")
コード例 #3
0
def wf3():
    wf = Workflow("test")
    wf.add_jobs(Job("ls"))
    wf.add_site_catalog(SiteCatalog())
    wf.add_transformation_catalog(TransformationCatalog())
    wf.add_replica_catalog(ReplicaCatalog())

    return wf
コード例 #4
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_dependency_invalid_job(self):
        wf = Workflow("wf")
        job = Job("t")

        with pytest.raises(ValueError) as e:
            wf.add_dependency(job)

        assert "The given job does not have an id" in str(e)
コード例 #5
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_site_catalog(self):
        sc = SiteCatalog()
        wf = Workflow("wf")

        try:
            wf.add_site_catalog(sc)
        except:
            pytest.fail("should not have raised exception")
コード例 #6
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_dependency_invalid_parent(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        parent = Job("t")

        with pytest.raises(ValueError) as e:
            wf.add_dependency(job, parents=[parent])

        assert "One of the given parents does not have an id" in str(e)
コード例 #7
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_dependency_invalid_child(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        child = Job("t")

        with pytest.raises(ValueError) as e:
            wf.add_dependency(job, children=[child])

        assert "One of the given children does not have an id" in str(e)
コード例 #8
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_job_id_assignment_by_workflow(self):
        wf = Workflow("wf")
        j1 = Job("t1", _id="a")
        j2 = Job("t2")
        j3 = Job("t3", _id="b")
        j4 = Job("t4")
        j5 = Job("t5")
        wf.add_jobs(j1, j2, j3, j4, j5)

        assert j2._id == "ID0000001"
        assert j4._id == "ID0000002"
        assert j5._id == "ID0000003"
コード例 #9
0
def dump(obj: Workflow, fp: TextIO, _format="yml", *args, **kwargs) -> None:
    """
    Serialize ``obj`` as a :py:class:`~Pegasus.api.worklfow.Workflow` formatted stream to ``fp`` (a ``.write()``-supporting file-like object).

    :param obj: Workflow to serialize
    :type obj: Workflow
    :param fp: file like object to serialize to
    :type fp: TextIO
    :param _format: format to write to if fp does not have an extension; can be one of ["yml" | "yaml" | "json"], defaults to "yml"
    :type _format: str
    :rtype: NoReturn
    """
    obj.write(fp, _format=_format)
コード例 #10
0
def dumps(obj: Workflow, _format="yml", *args, **kwargs) -> str:
    """
    Serialize ``obj`` to a :py:class:`~Pegasus.api.workflow.Workflow` formatted ``str``.

    :param obj: Workflow to serialize
    :type obj: Workflow
    :param _format: format to write to if fp does not have an extension; can be one of ["yml" | "yaml" | "json"], defaults to "yml"
    :type _format: str
    :return: Workflow serialized as a string
    :rtype: str
    """
    with StringIO() as s:
        obj.write(s, _format=_format)
        s.seek(0)
        return s.read()
コード例 #11
0
def wf2():
    _in = File("in")
    stdin = File("stdin")
    stdout = File("stdout")
    stderr = File("stderr")
    out = File("out")
    out2 = File("out2")

    pegasus_workflow = File("pegasus_workflow")
    condor_workflow = File("condor_workflow")

    j1 = (Job("tr", _id="1", node_label="test").add_args(
        "-i", _in, "-o", out).add_inputs(_in).add_outputs(out).set_stdin(
            stdin).set_stdout(stdout).set_stderr(stderr))

    j2 = (Job("tr2", _id="2", node_label="test").add_args(
        "-i", out, "-o", out2).add_inputs(out).add_outputs(out2))

    sbwf_pegasus = SubWorkflow(pegasus_workflow,
                               False,
                               _id="unplanned",
                               node_label="test").add_args("-flag", "-flag2")

    sbwf_condor = SubWorkflow(condor_workflow,
                              True,
                              _id="planned",
                              node_label="test")

    return (Workflow("test", infer_dependencies=False).add_jobs(
        j1, j2, sbwf_pegasus, sbwf_condor).add_dependency(j1, children=[j2]))
コード例 #12
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_write_hierarchical_workflow_when_catalogs_are_inlined(
            self, sc, tc):
        wf = Workflow("test")
        wf.add_jobs(SubWorkflow("file", False))

        if sc:
            wf.add_site_catalog(sc)

        if tc:
            wf.add_transformation_catalog(tc)

        with pytest.raises(PegasusError) as e:
            wf.write()

        assert (
            "Site Catalog and Transformation Catalog must be written as a separate"
            in str(e))
コード例 #13
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_duplicate_transformation_catalog(self):
        tc = TransformationCatalog()
        wf = Workflow("wf")
        wf.add_transformation_catalog(tc)

        with pytest.raises(DuplicateError) as e:
            wf.add_transformation_catalog(tc)

        assert "a TransformationCatalog has already" in str(e)
コード例 #14
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_duplicate_replica_catalog(self):
        rc = ReplicaCatalog()
        wf = Workflow("wf")
        wf.add_replica_catalog(rc)

        with pytest.raises(DuplicateError) as e:
            wf.add_replica_catalog(rc)

        assert "a ReplicaCatalog has already" in str(e)
コード例 #15
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_duplicate_site_catalog(self):
        sc = SiteCatalog()
        wf = Workflow("wf")
        wf.add_site_catalog(sc)

        with pytest.raises(DuplicateError) as e:
            wf.add_site_catalog(sc)

        assert "a SiteCatalog has already" in str(e)
コード例 #16
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_dependency_parents(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        parents = [
            Job("t", _id="parent1"),
            Job("t", _id="parent2"),
            Job("t", _id="parent3"),
        ]

        wf.add_jobs(job, *parents)

        wf.add_dependency(job, parents=[parents[0]])
        wf.add_dependency(job, parents=parents[1:])

        for parent in parents:
            assert wf.dependencies[parent._id] == _JobDependency(
                parent._id, {job._id})
コード例 #17
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_write_valid_hierarchical_workflow(self, mocker):
        mocker.patch("Pegasus.api.workflow.Workflow.write")

        try:
            wf = Workflow("test")
            wf.add_jobs(SubWorkflow("file", False))
            wf.write(file="workflow.yml", _format="yml")
        except PegasusError:
            pytest.fail("shouldn't have thrown PegasusError")

        Pegasus.api.workflow.Workflow.write.assert_called_once_with(
            file="workflow.yml", _format="yml")
コード例 #18
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_dependency_children(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        children = [
            Job("t", _id="child1"),
            Job("t", _id="child2"),
            Job("t", _id="child3"),
        ]

        wf.add_jobs(job, *children)

        wf.add_dependency(job, children=[children[0]])
        assert wf.dependencies[job._id] == _JobDependency(
            job._id, {children[0]._id})

        wf.add_dependency(job, children=children[1:])
        assert wf.dependencies[job._id] == _JobDependency(
            job._id, {child._id
                      for child in children})
コード例 #19
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_duplicate_parent_dependency(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        parent = Job("t", _id="parent")

        wf.add_jobs(job, parent)

        with pytest.raises(DuplicateError) as e:
            wf.add_dependency(job, parents=[parent, parent])

        assert (
            "A dependency already exists between parent id: parent and job id: job"
            in str(e))
コード例 #20
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_duplicate_child_dependency(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        child = Job("t", _id="child")

        wf.add_jobs(job, child)

        with pytest.raises(DuplicateError) as e:
            wf.add_dependency(job, children=[child, child])

        assert (
            "A dependency already exists between job id: job and child id: child"
            in str(e))
コード例 #21
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_infer_dependencies_when_job_uses_stdin_stdout_and_stderr(self):
        wf = Workflow("wf")
        j1 = Job("t1", _id="j1").add_outputs(File("f1"))
        j2 = Job("t1",
                 _id="j2").set_stdin(*j1.get_outputs()).set_stdout(File("f2"))
        j3 = Job("t1", _id="j3").add_inputs(*j2.get_outputs())
        wf.add_jobs(j1, j2, j3)

        # manually call _infer_dependencies() as it is only called when
        # wf.write() is called
        wf._infer_dependencies()

        assert wf.dependencies["j1"] == _JobDependency("j1", {"j2"})
        assert wf.dependencies["j2"] == _JobDependency("j2", {"j3"})
コード例 #22
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_dependency_parents_and_children(self):
        wf = Workflow("wf")
        job = Job("t", _id="job")
        parents = [Job("t", _id="parent1"), Job("t", _id="parent2")]

        children = [Job("t", _id="child1"), Job("t", _id="child2")]

        wf.add_jobs(*parents, *children)

        # add nothing
        wf.add_dependency(job)
        assert len(wf.dependencies) == 0

        wf.add_dependency(job, parents=parents, children=children)

        for parent in parents:
            assert wf.dependencies[parent._id] == _JobDependency(
                parent._id, {job._id})

        assert wf.dependencies[job._id] == _JobDependency(
            job._id, {child._id
                      for child in children})
コード例 #23
0
def wf1():
    in1 = File("in1", size=2048).add_metadata(createdBy="ryan")
    in2 = File("in2")
    stdin = File("stdin").add_metadata(size=1024)
    stdout = File("stdout").add_metadata(size=1024)
    stderr = File("stderr").add_metadata(size=1024)
    out = File("out").add_metadata(size=1024)
    out2 = File("out2").add_metadata(size=1024)

    pegasus_workflow = File("pegasus_workflow").add_metadata(size=2048)
    condor_workflow = File("condor_workflow").add_metadata(size=2048)

    j1 = (Job("tr", _id="1", node_label="test").add_args(
        "-i", in1, "-o", out, "-n", 1, 1.1).add_inputs(in1).add_inputs(
            in2, bypass_staging=True).add_outputs(out).set_stdin(
                stdin).set_stdout(stdout).set_stderr(stderr).add_shell_hook(
                    EventType.START,
                    "/cmd2").add_env(JAVA_HOME="/usr/bin/java").add_metadata(
                        xtra_info="123"))

    j2 = (Job("tr2", _id="2", node_label="test").add_args(
        "-i", out, "-o", out2).add_inputs(out).add_outputs(out2))

    sbwf_pegasus = SubWorkflow(pegasus_workflow,
                               False,
                               _id="unplanned",
                               node_label="test").add_args("-flag", "-flag2")

    sbwf_condor = SubWorkflow(condor_workflow,
                              True,
                              _id="planned",
                              node_label="test")

    return (Workflow("test", infer_dependencies=False).add_shell_hook(
        EventType.START, "/cmd").add_dagman_profile(retry=1).add_metadata(
            author="ryan").add_jobs(j1, j2, sbwf_pegasus,
                                    sbwf_condor).add_dependency(j1,
                                                                children=[j2]))
コード例 #24
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_infer_dependencies_fork_join_wf(self):
        wf = Workflow("wf")

        f1 = File("f1")
        f2 = File("f2")
        f3 = File("f3")
        f4 = File("f4")

        fork = Job("t1", _id="fork").add_outputs(f1, f2)
        work1 = Job("t1", _id="work1").add_inputs(f1).add_outputs(f3)
        work2 = Job("t1", _id="work2").add_inputs(f2).add_outputs(f4)
        join = Job("t1", _id="join").add_inputs(f3, f4)
        wf.add_jobs(fork, work1, work2, join)

        # manually call _infer_dependencies() as it is only called when
        # wf.write() is called
        wf._infer_dependencies()

        assert wf.dependencies["fork"] == _JobDependency(
            "fork", {"work1", "work2"})
        assert wf.dependencies["work1"] == _JobDependency("work1", {"join"})
        assert wf.dependencies["work2"] == _JobDependency("work2", {"join"})
コード例 #25
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_invalid_transformation_catalog(self):
        wf = Workflow("wf")
        with pytest.raises(TypeError) as e:
            wf.add_transformation_catalog(123)

        assert "invalid catalog: 123" in str(e)
コード例 #26
0
def _to_wf(d: dict) -> Workflow:
    """Convert dict to Workflow

    :param d: Workflow represented as a dict
    :type d: dict
    :raises PegasusError: encountered error parsing
    :return: a Workflow object based on d
    :rtype: Workflow
    """

    try:
        #
        wf = Workflow(d["name"], infer_dependencies=False)

        # add rc
        if "replicaCatalog" in d:
            wf.replica_catalog = _to_rc(d["replicaCatalog"])

        # add tc
        if "transformationCatalog" in d:
            wf.transformation_catalog = _to_tc(d["transformationCatalog"])

        # add sc
        if "siteCatalog" in d:
            wf.site_catalog = _to_sc(d["siteCatalog"])

        # add jobs
        for j in d["jobs"]:
            # create appropriate job based on type
            if j["type"] == "job":
                job = Job(
                    j["name"],
                    _id=j["id"],
                    node_label=j.get("nodeLabel"),
                    namespace=j.get("namespace"),
                    version=j.get("version"),
                )
            elif j["type"] in {"pegasusWorkflow", "condorWorkflow"}:
                f = File(j["file"])

                is_planned = False if j["type"] == "pegasusWorkflow" else True

                job = SubWorkflow(
                    f, is_planned, _id=j["id"], node_label=j.get("nodeLabel")
                )

            else:
                raise ValueError

            # add args
            args = list()
            for a in j["arguments"]:
                args.append(a)

            job.args = args

            # add uses
            uses = set()
            for u in j["uses"]:
                f = File(u["lfn"], size=u.get("size"))
                try:
                    f.metadata = u["metadata"]
                except KeyError:
                    pass

                uses.add(
                    _Use(
                        f,
                        getattr(_LinkType, u["type"].upper()),
                        stage_out=u.get("stageOut"),
                        register_replica=u.get("registerReplica"),
                        bypass_staging=u.get("bypass"),
                    )
                )

            job.uses = uses

            # set stdin
            if "stdin" in j:
                for u in job.uses:
                    if u.file.lfn == j["stdin"]:
                        job.stdin = u.file
                        break

            # set stdout
            if "stdout" in j:
                for u in job.uses:
                    if u.file.lfn == j["stdout"]:
                        job.stdout = u.file
                        break

            # set stderr
            if "stderr" in j:
                for u in job.uses:
                    if u.file.lfn == j["stderr"]:
                        job.stderr = u.file
                        break

            # add profiles
            if j.get("profiles"):
                job.profiles = defaultdict(dict, j.get("profiles"))

            # add metadata
            if j.get("metadata"):
                job.metadata = j.get("metadata")

            # add hooks
            if j.get("hooks"):
                job.hooks = defaultdict(list, j.get("hooks"))

            # add job to wf
            wf.add_jobs(job)

        # add dependencies
        if d.get("jobDependencies"):
            dependencies = defaultdict(_JobDependency)
            for item in d.get("jobDependencies"):
                dependencies[item["id"]] = _JobDependency(
                    item["id"], {child for child in item["children"]}
                )

            wf.dependencies = dependencies

        # add profiles
        if d.get("profiles"):
            wf.profiles = defaultdict(dict, d.get("profiles"))

        # add metadata
        if d.get("metadata"):
            wf.metadata = d.get("metadata")

        # add hooks
        if d.get("hooks"):
            wf.hooks = defaultdict(list, d.get("hooks"))

        return wf
    except (KeyError, ValueError):
        raise PegasusError("error parsing {}".format(d))
コード例 #27
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_get_job(self):
        wf = Workflow("wf")
        j1 = Job("t1", _id="j1")
        wf.add_jobs(j1)

        assert j1 == wf.get_job("j1")
コード例 #28
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
    def test_add_job(self, job):
        wf = Workflow("wf")
        wf.add_jobs(job)

        assert job == wf.get_job("job")
コード例 #29
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
 def test_add_duplicate_job(self):
     wf = Workflow("wf")
     with pytest.raises(DuplicateError):
         wf.add_jobs(Job("t1", _id="j1"), Job("t2", _id="j1"))
コード例 #30
0
ファイル: test_workflow.py プロジェクト: ryantanaka/pegasus
 def test_get_invalid_job(self):
     wf = Workflow("wf")
     with pytest.raises(NotFoundError):
         wf.get_job("abc123")