X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/6599088b45103087b4be743fd51a8330e694e57f..2feec8cbac7ef28a47f3c3a9d071b070ef38cb6e:/sdk/cwl/arvados_cwl/runner.py diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py index 144fd53ffb..087fed3e16 100644 --- a/sdk/cwl/arvados_cwl/runner.py +++ b/sdk/cwl/arvados_cwl/runner.py @@ -1,10 +1,15 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + import os import urlparse from functools import partial import logging import json -import re -from cStringIO import StringIO +import subprocess + +from StringIO import StringIO from schema_salad.sourceline import SourceLine @@ -13,39 +18,47 @@ from cwltool.draft2tool import CommandLineTool import cwltool.workflow from cwltool.process import get_feature, scandeps, UnsupportedRequirement, normalizeFilesDirs, shortname from cwltool.load_tool import fetch_document -from cwltool.pathmapper import adjustFileObjs, adjustDirObjs +from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class from cwltool.utils import aslist from cwltool.builder import substitute +from cwltool.pack import pack import arvados.collection import ruamel.yaml as yaml from .arvdocker import arv_docker_get_image -from .pathmapper import ArvPathMapper +from .pathmapper import ArvPathMapper, trim_listing from ._version import __version__ from . import done logger = logging.getLogger('arvados.cwl-runner') -cwltool.draft2tool.ACCEPTLIST_RE = re.compile(r".*") +def trim_anonymous_location(obj): + """Remove 'location' field from File and Directory literals. -def trim_listing(obj): - """Remove 'listing' field from Directory objects that are keep references. + To make internal handling easier, literals are assigned a random id for + 'location'. However, when writing the record back out, this can break + reproducibility. Since it is valid for literals not have a 'location' + field, remove it. - When Directory objects represent Keep references, it redundant and - potentially very expensive to pass fully enumerated Directory objects - between instances of cwl-runner (e.g. a submitting a job, or using the - RunInSingleContainer feature), so delete the 'listing' field when it is - safe to do so. """ - if obj.get("location", "").startswith("keep:") and "listing" in obj: - del obj["listing"] if obj.get("location", "").startswith("_:"): del obj["location"] +def find_defaults(d, op): + if isinstance(d, list): + for i in d: + find_defaults(i, op) + elif isinstance(d, dict): + if "default" in d: + op(d) + else: + for i in d.itervalues(): + find_defaults(i, op) + def upload_dependencies(arvrunner, name, document_loader, - workflowobj, uri, loadref_run): + workflowobj, uri, loadref_run, include_primary=True): """Upload the dependencies of the workflowobj document to Keep. Returns a pathmapper object mapping local paths to keep references. Also @@ -92,13 +105,35 @@ def upload_dependencies(arvrunner, name, document_loader, normalizeFilesDirs(sc) - if "id" in workflowobj: + if include_primary and "id" in workflowobj: sc.append({"class": "File", "location": workflowobj["id"]}) + if "$schemas" in workflowobj: + for s in workflowobj["$schemas"]: + sc.append({"class": "File", "location": s}) + + def capture_default(obj): + remove = [False] + def add_default(f): + if "location" not in f and "path" in f: + f["location"] = f["path"] + del f["path"] + if "location" in f and not arvrunner.fs_access.exists(f["location"]): + # Remove from sc + sc[:] = [x for x in sc if x["location"] != f["location"]] + # Delete "default" from workflowobj + remove[0] = True + visit_class(obj["default"], ("File", "Directory"), add_default) + if remove[0]: + del obj["default"] + + find_defaults(workflowobj, capture_default) + mapper = ArvPathMapper(arvrunner, sc, "", "keep:%s", "keep:%s/%s", - name=name) + name=name, + single_collection=True) def setloc(p): if "location" in p and (not p["location"].startswith("_:")) and (not p["location"].startswith("keep:")): @@ -106,10 +141,18 @@ def upload_dependencies(arvrunner, name, document_loader, adjustFileObjs(workflowobj, setloc) adjustDirObjs(workflowobj, setloc) + if "$schemas" in workflowobj: + sch = [] + for s in workflowobj["$schemas"]: + sch.append(mapper.mapper(s).resolved) + workflowobj["$schemas"] = sch + return mapper def upload_docker(arvrunner, tool): + """Uploads Docker images used in CommandLineTool objects.""" + if isinstance(tool, CommandLineTool): (docker_req, docker_is_req) = get_feature(tool, "DockerRequirement") if docker_req: @@ -122,52 +165,123 @@ def upload_docker(arvrunner, tool): for s in tool.steps: upload_docker(arvrunner, s.embedded_tool) -def upload_instance(arvrunner, name, tool, job_order): - upload_docker(arvrunner, tool) - - for t in tool.tool["inputs"]: - def setSecondary(fileobj): - if isinstance(fileobj, dict) and fileobj.get("class") == "File": - if "secondaryFiles" not in fileobj: - fileobj["secondaryFiles"] = [{"location": substitute(fileobj["location"], sf), "class": "File"} for sf in t["secondaryFiles"]] - - if isinstance(fileobj, list): - for e in fileobj: - setSecondary(e) - - if shortname(t["id"]) in job_order and t.get("secondaryFiles"): - setSecondary(job_order[shortname(t["id"])]) - - workflowmapper = upload_dependencies(arvrunner, - name, - tool.doc_loader, - tool.tool, - tool.tool["id"], - True) - jobmapper = upload_dependencies(arvrunner, - os.path.basename(job_order.get("id", "#")), - tool.doc_loader, - job_order, - job_order.get("id", "#"), - False) - - if "id" in job_order: - del job_order["id"] - - return workflowmapper - -def arvados_jobs_image(arvrunner): - img = "arvados/jobs:"+__version__ +def packed_workflow(arvrunner, tool): + """Create a packed workflow. + + A "packed" workflow is one where all the components have been combined into a single document.""" + + return pack(tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), + tool.tool["id"], tool.metadata) + +def tag_git_version(packed): + if tool.tool["id"].startswith("file://"): + path = os.path.dirname(tool.tool["id"][7:]) + try: + githash = subprocess.check_output(['git', 'log', '--first-parent', '--max-count=1', '--format=%H'], stderr=subprocess.STDOUT, cwd=path).strip() + except (OSError, subprocess.CalledProcessError): + pass + else: + packed["http://schema.org/version"] = githash + + +def upload_job_order(arvrunner, name, tool, job_order): + """Upload local files referenced in the input object and return updated input + object with 'location' updated to the proper keep references. + """ + + for t in tool.tool["inputs"]: + def setSecondary(fileobj): + if isinstance(fileobj, dict) and fileobj.get("class") == "File": + if "secondaryFiles" not in fileobj: + fileobj["secondaryFiles"] = [{"location": substitute(fileobj["location"], sf), "class": "File"} for sf in t["secondaryFiles"]] + + if isinstance(fileobj, list): + for e in fileobj: + setSecondary(e) + + if shortname(t["id"]) in job_order and t.get("secondaryFiles"): + setSecondary(job_order[shortname(t["id"])]) + + jobmapper = upload_dependencies(arvrunner, + name, + tool.doc_loader, + job_order, + job_order.get("id", "#"), + False) + + if "id" in job_order: + del job_order["id"] + + # Need to filter this out, gets added by cwltool when providing + # parameters on the command line. + if "job_order" in job_order: + del job_order["job_order"] + + return job_order + +def upload_workflow_deps(arvrunner, tool, override_tools): + # Ensure that Docker images needed by this workflow are available + + upload_docker(arvrunner, tool) + + document_loader = tool.doc_loader + + def upload_tool_deps(deptool): + if "id" in deptool: + upload_dependencies(arvrunner, + "%s dependencies" % (shortname(deptool["id"])), + document_loader, + deptool, + deptool["id"], + False, + include_primary=False) + document_loader.idx[deptool["id"]] = deptool + override_tools[deptool["id"]] = json.dumps(deptool) + + tool.visit(upload_tool_deps) + +def arvados_jobs_image(arvrunner, img): + """Determine if the right arvados/jobs image version is available. If not, try to pull and upload it.""" + try: arv_docker_get_image(arvrunner.api, {"dockerPull": img}, True, arvrunner.project_uuid) except Exception as e: raise Exception("Docker image %s is not available\n%s" % (img, e) ) return img +def upload_workflow_collection(arvrunner, name, packed): + collection = arvados.collection.Collection(api_client=arvrunner.api, + keep_client=arvrunner.keep_client, + num_retries=arvrunner.num_retries) + with collection.open("workflow.cwl", "w") as f: + f.write(json.dumps(packed, indent=2, sort_keys=True, separators=(',',': '))) + + filters = [["portable_data_hash", "=", collection.portable_data_hash()], + ["name", "like", name+"%"]] + if arvrunner.project_uuid: + filters.append(["owner_uuid", "=", arvrunner.project_uuid]) + exists = arvrunner.api.collections().list(filters=filters).execute(num_retries=arvrunner.num_retries) + + if exists["items"]: + logger.info("Using collection %s", exists["items"][0]["uuid"]) + else: + collection.save_new(name=name, + owner_uuid=arvrunner.project_uuid, + ensure_unique_name=True, + num_retries=arvrunner.num_retries) + logger.info("Uploaded to %s", collection.manifest_locator()) + + return collection.portable_data_hash() + + class Runner(object): + """Base class for runner processes, which submit an instance of + arvados-cwl-runner and wait for the final result.""" + def __init__(self, runner, tool, job_order, enable_reuse, output_name, output_tags, submit_runner_ram=0, - name=None): + name=None, on_error=None, submit_runner_image=None, + intermediate_output_ttl=0): self.arvrunner = runner self.tool = tool self.job_order = job_order @@ -178,11 +292,14 @@ class Runner(object): self.output_name = output_name self.output_tags = output_tags self.name = name + self.on_error = on_error + self.jobs_image = submit_runner_image or "arvados/jobs:"+__version__ + self.intermediate_output_ttl = intermediate_output_ttl if submit_runner_ram: self.submit_runner_ram = submit_runner_ram else: - self.submit_runner_ram = 1024 + self.submit_runner_ram = 3000 if self.submit_runner_ram <= 0: raise Exception("Value of --submit-runner-ram must be greater than zero") @@ -190,20 +307,9 @@ class Runner(object): def update_pipeline_component(self, record): pass - def arvados_job_spec(self, *args, **kwargs): - if self.name is None: - self.name = self.tool.tool.get("label") or os.path.basename(self.tool.tool["id"]) - - # Need to filter this out, gets added by cwltool when providing - # parameters on the command line. - if "job_order" in self.job_order: - del self.job_order["job_order"] - - workflowmapper = upload_instance(self.arvrunner, self.name, self.tool, self.job_order) - adjustDirObjs(self.job_order, trim_listing) - return workflowmapper - def done(self, record): + """Base method for handling a completed runner.""" + try: if record["state"] == "Complete": if record.get("exit_code") is not None: