def __init__(self, toc, j, renderlist, redirects):
self.typedoc = StringIO.StringIO()
self.toc = toc
self.subs = {}
self.docParent = {}
self.docAfter = {}
self.rendered = set()
self.redirects = redirects
self.title = None
for t in j:
if "extends" in t:
for e in aslist(t["extends"]):
add_dictlist(self.subs, e, t["name"])
#if "docParent" not in t and "docAfter" not in t:
# add_dictlist(self.docParent, e, t["name"])
if t.get("docParent"):
add_dictlist(self.docParent, t["docParent"], t["name"])
if t.get("docChild"):
for c in aslist(t["docChild"]):
add_dictlist(self.docParent, t["name"], c)
if t.get("docAfter"):
add_dictlist(self.docAfter, t["docAfter"], t["name"])
_, _, metaschema_loader = schema.get_metaschema()
alltypes = schema.extend_and_specialize(j, metaschema_loader)
self.typemap = {}
self.uses = {}
self.record_refs = {}
for t in alltypes:
self.typemap[t["name"]] = t
try:
if t["type"] == "record":
self.record_refs[t["name"]] = []
for f in t.get("fields", []):
p = has_types(f)
for tp in p:
if tp not in self.uses:
self.uses[tp] = []
if (t["name"], f["name"]) not in self.uses[tp]:
_, frg1 = urlparse.urldefrag(t["name"])
_, frg2 = urlparse.urldefrag(f["name"])
self.uses[tp].append((frg1, frg2))
if tp not in basicTypes and tp not in self.record_refs[t["name"]]:
self.record_refs[t["name"]].append(tp)
except KeyError as e:
_logger.error("Did not find 'type' in %s", t)
raise
for f in alltypes:
if (f["name"] in renderlist or
((not renderlist) and
("extends" not in f) and
("docParent" not in f) and
("docAfter" not in f))):
self.render_type(f, 1)
def __init__(self, toolpath_object, validateAs, docpath):
self.names = get_schema()
self.docpath = docpath
self.tool = toolpath_object
# Validate tool documument
validate.validate_ex(self.names.get_name(validateAs, ""), self.tool)
self.validate_requirements(self.tool, "requirements")
self.validate_requirements(self.tool, "hints")
for t in self.tool.get("requirements", []):
t["_docpath"] = docpath
for t in self.tool.get("hints", []):
t["_docpath"] = docpath
# Import schema defs
self.schemaDefs = {
"Any": [
"null",
"boolean",
"int",
"long",
"float",
"double",
"bytes",
"string",
"File",
{"type": "array", "items": "Any"},
{"type": "map", "values": "Any"}
]}
sd, _ = get_feature("SchemaDefRequirement", requirements=self.tool.get("requirements"), hints=self.tool.get("hints"))
if sd:
for i in sd["types"]:
avro.schema.make_avsc_object(i, self.names)
self.schemaDefs[i["name"]] = i
# Build record schema from inputs
self.inputs_record_schema = {"name": "input_record_schema", "type": "record", "fields": []}
for i in self.tool["inputs"]:
c = copy.copy(i)
c["name"] = c["id"][1:]
del c["id"]
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
self.inputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.inputs_record_schema, self.names)
self.outputs_record_schema = {"name": "outputs_record_schema", "type": "record", "fields": []}
for i in self.tool["outputs"]:
c = copy.copy(i)
c["name"] = c["id"][1:]
del c["id"]
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
self.outputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.outputs_record_schema, self.names)
def scandeps(base, doc, reffields, urlfields, loadref):
r = []
if isinstance(doc, dict):
if "id" in doc:
if doc["id"].startswith("file://"):
df, _ = urlparse.urldefrag(doc["id"])
if base != df:
base = df
for k, v in doc.iteritems():
if k in reffields:
for u in aslist(v):
sub = loadref(base, u)
if isinstance(sub, dict):
subid = sub["id"]
else:
subid = urlparse.urljoin(base, u)
deps = {"class": "File", "path": subid}
sf = scandeps(subid, sub, reffields, urlfields, loadref)
if sf:
deps["secondaryFiles"] = sf
r.append(deps)
elif k in urlfields:
for u in aslist(v):
r.append({
"class": "File",
"path": urlparse.urljoin(base, u)
})
else:
r.extend(scandeps(base, v, reffields, urlfields, loadref))
elif isinstance(doc, list):
for d in doc:
r.extend(scandeps(base, d, reffields, urlfields, loadref))
return r
def exeval(ex, jobinput, requirements, outdir, tmpdir, context, pull_image):
if ex["engine"] == "https://w3id.org/cwl/cwl#JsonPointer":
try:
obj = {"job": jobinput, "context": context, "outdir": outdir, "tmpdir": tmpdir}
return schema_salad.ref_resolver.resolve_json_pointer(obj, ex["script"])
except ValueError as v:
raise WorkflowException("%s in %s" % (v, obj))
for r in reversed(requirements):
if r["class"] == "ExpressionEngineRequirement" and r["id"] == ex["engine"]:
runtime = []
class DR(object):
pass
dr = DR()
dr.requirements = r.get("requirements", [])
dr.hints = r.get("hints", [])
(docker_req, docker_is_req) = process.get_feature(dr, "DockerRequirement")
img_id = None
if docker_req:
img_id = docker.get_from_requirements(docker_req, docker_is_req, pull_image)
if img_id:
runtime = ["docker", "run", "-i", "--rm", img_id]
exdefs = []
for exdef in r.get("engineConfig", []):
if isinstance(exdef, dict) and "ref" in exdef:
with open(exdef["ref"][7:]) as f:
exdefs.append(f.read())
elif isinstance(exdef, basestring):
exdefs.append(exdef)
inp = {
"script": ex["script"],
"engineConfig": exdefs,
"job": jobinput,
"context": context,
"outdir": outdir,
"tmpdir": tmpdir,
}
_logger.debug("Invoking expression engine %s with %s",
runtime + aslist(r["engineCommand"]),
json.dumps(inp, indent=4))
sp = subprocess.Popen(runtime + aslist(r["engineCommand"]),
shell=False,
close_fds=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE)
(stdoutdata, stderrdata) = sp.communicate(json.dumps(inp) + "\n\n")
if sp.returncode != 0:
raise WorkflowException("Expression engine returned non-zero exit code on evaluation of\n%s" % json.dumps(inp, indent=4))
return json.loads(stdoutdata)
raise WorkflowException("Unknown expression engine '%s'" % ex["engine"])
def collect_output(self, schema, builder, outdir):
r = None
if "outputBinding" in schema:
binding = schema["outputBinding"]
if "glob" in binding:
r = []
bg = builder.do_eval(binding["glob"])
for gb in aslist(bg):
r.extend([{"path": g, "class": "File"} for g in builder.fs_access.glob(os.path.join(outdir, gb))])
for files in r:
checksum = hashlib.sha1()
with builder.fs_access.open(files["path"], "rb") as f:
contents = f.read(CONTENT_LIMIT)
if binding.get("loadContents"):
files["contents"] = contents
filesize = 0
while contents != "":
checksum.update(contents)
filesize += len(contents)
contents = f.read(1024 * 1024)
files["checksum"] = "sha1$%s" % checksum.hexdigest()
files["size"] = filesize
if "outputEval" in binding:
r = builder.do_eval(binding["outputEval"], context=r)
if schema["type"] == "File" and (not isinstance(r, dict) or "path" not in r):
raise WorkflowException("Expression must return a file object.")
if schema["type"] == "File":
if not r:
raise WorkflowException("No matches for output file with glob: '{}'".format(bg))
if len(r) > 1:
raise WorkflowException("Multiple matches for output item that is a single file.")
r = r[0]
if schema["type"] == "File" and "secondaryFiles" in binding:
r["secondaryFiles"] = []
for sf in aslist(binding["secondaryFiles"]):
if isinstance(sf, dict):
sfpath = builder.do_eval(sf, context=r["path"])
else:
sfpath = {"path": substitute(r["path"], sf), "class": "File"}
if isinstance(sfpath, list):
r["secondaryFiles"].extend(sfpath)
else:
r["secondaryFiles"].append(sfpath)
for sf in r["secondaryFiles"]:
if not builder.fs_access.exists(sf["path"]):
raise WorkflowException(
"Missing secondary file of '%s' of primary file '%s'" % (sf["path"], r["path"])
)
if not r and schema["type"] == "record":
r = {}
for f in schema["fields"]:
r[f["name"]] = self.collect_output(f, builder, outdir)
return r
def __init__(self, toc, j, renderlist, redirects):
self.typedoc = StringIO.StringIO()
self.toc = toc
self.subs = {}
self.docParent = {}
self.docAfter = {}
self.rendered = set()
self.redirects = redirects
self.title = None
for t in j:
if "extends" in t:
for e in aslist(t["extends"]):
add_dictlist(self.subs, e, t["name"])
#if "docParent" not in t and "docAfter" not in t:
# add_dictlist(self.docParent, e, t["name"])
if t.get("docParent"):
add_dictlist(self.docParent, t["docParent"], t["name"])
if t.get("docChild"):
for c in aslist(t["docChild"]):
add_dictlist(self.docParent, t["name"], c)
if t.get("docAfter"):
add_dictlist(self.docAfter, t["docAfter"], t["name"])
_, _, metaschema_loader = schema.get_metaschema()
alltypes = schema.extend_and_specialize(j, metaschema_loader)
self.typemap = {}
self.uses = {}
self.record_refs = {}
for t in alltypes:
self.typemap[t["name"]] = t
try:
if t["type"] == "record":
self.record_refs[t["name"]] = []
for f in t.get("fields", []):
p = has_types(f)
for tp in p:
if tp not in self.uses:
self.uses[tp] = []
if (t["name"], f["name"]) not in self.uses[tp]:
_, frg1 = urlparse.urldefrag(t["name"])
_, frg2 = urlparse.urldefrag(f["name"])
self.uses[tp].append((frg1, frg2))
if tp not in basicTypes and tp not in self.record_refs[
t["name"]]:
self.record_refs[t["name"]].append(tp)
except KeyError as e:
_logger.error("Did not find 'type' in %s", t)
raise
for f in alltypes:
if (f["name"] in renderlist
or ((not renderlist) and ("extends" not in f) and
("docParent" not in f) and ("docAfter" not in f))):
self.render_type(f, 1)
def exeval(ex, jobinput, requirements, outdir, tmpdir, context, pull_image):
if ex["engine"] == "https://w3id.org/cwl/cwl#JsonPointer":
try:
obj = {"job": jobinput, "context": context, "outdir": outdir, "tmpdir": tmpdir}
return schema_salad.ref_resolver.resolve_json_pointer(obj, ex["script"])
except ValueError as v:
raise WorkflowException("%s in %s" % (v, obj))
if ex["engine"] == "https://w3id.org/cwl/cwl#JavascriptEngine":
engineConfig = []
for r in reversed(requirements):
if r["class"] == "ExpressionEngineRequirement" and r["id"] == "https://w3id.org/cwl/cwl#JavascriptEngine":
engineConfig = r.get("engineConfig", [])
break
return sandboxjs.execjs(ex["script"], jshead(engineConfig, jobinput, context, tmpdir, outdir))
for r in reversed(requirements):
if r["class"] == "ExpressionEngineRequirement" and r["id"] == ex["engine"]:
runtime = []
class DR(object):
pass
dr = DR()
dr.requirements = r.get("requirements", [])
dr.hints = r.get("hints", [])
(docker_req, docker_is_req) = process.get_feature(dr, "DockerRequirement")
img_id = None
if docker_req:
img_id = docker.get_from_requirements(docker_req, docker_is_req, pull_image)
if img_id:
runtime = ["docker", "run", "-i", "--rm", img_id]
inp = {
"script": ex["script"],
"engineConfig": r.get("engineConfig", []),
"job": jobinput,
"context": context,
"outdir": outdir,
"tmpdir": tmpdir,
}
_logger.debug("Invoking expression engine %s with %s",
runtime + aslist(r["engineCommand"]),
json.dumps(inp, indent=4))
sp = subprocess.Popen(runtime + aslist(r["engineCommand"]),
shell=False,
close_fds=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE)
(stdoutdata, stderrdata) = sp.communicate(json.dumps(inp) + "\n\n")
if sp.returncode != 0:
raise WorkflowException("Expression engine returned non-zero exit code on evaluation of\n%s" % json.dumps(inp, indent=4))
return json.loads(stdoutdata)
raise WorkflowException("Unknown expression engine '%s'" % ex["engine"])
def checkFormat(actualFile, inputFormats, requirements, ontology):
for af in aslist(actualFile):
if "format" not in af:
raise validate.ValidationException("Missing required 'format' for File %s" % af)
for inpf in aslist(inputFormats):
if af["format"] == inpf or formatSubclassOf(af["format"], inpf, ontology, set()):
return
raise validate.ValidationException("Incompatible file format %s required format(s) %s" % (af["format"], inputFormats))
def __init__(self, toolpath_object, **kwargs):
(_, self.names, _) = get_schema()
self.tool = toolpath_object
self.requirements = kwargs.get("requirements", []) + self.tool.get("requirements", [])
self.hints = kwargs.get("hints", []) + self.tool.get("hints", [])
if "loader" in kwargs:
self.formatgraph = kwargs["loader"].graph
else:
self.formatgraph = None
self.validate_hints(self.tool.get("hints", []), strict=kwargs.get("strict"))
self.schemaDefs = {}
sd, _ = self.get_requirement("SchemaDefRequirement")
if sd:
sdtypes = sd["types"]
av = schema_salad.schema.make_valid_avro(sdtypes, {t["name"]: t for t in sdtypes}, set())
for i in av:
self.schemaDefs[i["name"]] = i
avro.schema.make_avsc_object(av, self.names)
# Build record schema from inputs
self.inputs_record_schema = {"name": "input_record_schema", "type": "record", "fields": []}
self.outputs_record_schema = {"name": "outputs_record_schema", "type": "record", "fields": []}
for key in ("inputs", "outputs"):
for i in self.tool[key]:
c = copy.copy(i)
doc_url, _ = urlparse.urldefrag(c['id'])
c["name"] = shortname(c["id"])
del c["id"]
if "type" not in c:
raise validate.ValidationException("Missing `type` in parameter `%s`" % c["name"])
if "default" in c and "null" not in aslist(c["type"]):
c["type"] = ["null"] + aslist(c["type"])
else:
c["type"] = c["type"]
if key == "inputs":
self.inputs_record_schema["fields"].append(c)
elif key == "outputs":
self.outputs_record_schema["fields"].append(c)
try:
self.inputs_record_schema = schema_salad.schema.make_valid_avro(self.inputs_record_schema, {}, set())
avro.schema.make_avsc_object(self.inputs_record_schema, self.names)
except avro.schema.SchemaParseException as e:
raise validate.ValidationException("Got error `%s` while prcoessing inputs of %s:\n%s" % (str(e), self.tool["id"], json.dumps(self.inputs_record_schema, indent=4)))
try:
self.outputs_record_schema = schema_salad.schema.make_valid_avro(self.outputs_record_schema, {}, set())
avro.schema.make_avsc_object(self.outputs_record_schema, self.names)
except avro.schema.SchemaParseException as e:
raise validate.ValidationException("Got error `%s` while prcoessing outputs of %s:\n%s" % (str(e), self.tool["id"], json.dumps(self.outputs_record_schema, indent=4)))
def checkFormat(actualFile, inputFormats, requirements, ontology):
for af in aslist(actualFile):
if "format" not in af:
raise validate.ValidationException(
"Missing required 'format' for File %s" % af)
for inpf in aslist(inputFormats):
if af["format"] == inpf or formatSubclassOf(
af["format"], inpf, ontology, set()):
return
raise validate.ValidationException(
"Incompatible file format %s required format(s) %s" %
(af["format"], inputFormats))
def merge_properties(a, b):
c = {}
for i in a:
if i not in b:
c[i] = a[i]
for i in b:
if i not in a:
c[i] = b[i]
for i in a:
if i in b:
c[i] = aslist(a[i]) + aslist(b[i])
return c
def merge_properties(a, b):
c = {}
for i in a:
if i not in b:
c[i] = a[i]
for i in b:
if i not in a:
c[i] = b[i]
for i in a:
if i in b:
c[i] = aslist(a[i]) + aslist(b[i])
return c
def object_from_state(state, parms, frag_only, supportsMultipleInput):
inputobj = {}
for inp in parms:
iid = inp["id"]
if frag_only:
iid = shortname(iid)
if "source" in inp:
if isinstance(inp["source"], list) and not supportsMultipleInput:
raise WorkflowException("Workflow contains multiple inbound links to a single parameter but MultipleInputFeatureRequirement is not declared.")
connections = aslist(inp["source"])
for src in connections:
if src in state and state[src] is not None:
if not match_types(inp["type"], state[src], iid, inputobj,
inp.get("linkMerge", ("merge_nested" if len(connections) > 1 else None)),
valueFrom=inp.get("valueFrom")):
raise WorkflowException("Type mismatch between source '%s' (%s) and sink '%s' (%s)" % (src, state[src].parameter["type"], inp["id"], inp["type"]))
elif src not in state:
raise WorkflowException("Connect source '%s' on parameter '%s' does not exist" % (src, inp["id"]))
else:
return None
elif "default" in inp:
inputobj[iid] = inp["default"]
elif "valueFrom" in inp:
inputobj[iid] = None
else:
raise WorkflowException("Value for %s not specified" % (inp["id"]))
return inputobj
def object_from_state(state, parms, frag_only):
inputobj = {}
for inp in parms:
iid = inp["id"]
if frag_only:
(_, iid) = urlparse.urldefrag(iid)
iid = iid.split(".")[-1]
if "source" in inp:
connections = aslist(inp["source"])
for src in connections:
if src in state and state[src] is not None:
if not match_types(
inp["type"], state[src], iid, inputobj,
inp.get("linkMerge",
("merge_nested"
if len(connections) > 1 else None))):
raise WorkflowException(
"Type mismatch between source '%s' (%s) and sink '%s' (%s)"
% (src, state[src].parameter["type"], inp["id"],
inp["type"]))
elif src not in state:
raise WorkflowException(
"Connect source '%s' on parameter '%s' does not exist"
% (src, inp["id"]))
else:
return None
elif "default" in inp:
inputobj[iid] = inp["default"]
else:
raise WorkflowException("Value for %s not specified" % (inp["id"]))
return inputobj
def adjust_for_scatter(self, steps):
(scatterSpec, _) = self.get_requirement("ScatterFeatureRequirement")
for step in steps:
if scatterSpec and "scatter" in step.tool:
inputparms = copy.deepcopy(step.tool["inputs"])
outputparms = copy.deepcopy(step.tool["outputs"])
scatter = aslist(step.tool["scatter"])
inp_map = {i["id"]: i for i in inputparms}
for s in scatter:
if s not in inp_map:
raise WorkflowException("Invalid Scatter parameter '%s'" % s)
inp_map[s]["type"] = {"type": "array", "items": inp_map[s]["type"]}
if step.tool.get("scatterMethod") == "nested_crossproduct":
nesting = len(scatter)
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
step.tool["inputs"] = inputparms
step.tool["outputs"] = outputparms
def adjust_for_scatter(self, steps):
(scatterSpec, _) = self.get_requirement("ScatterFeatureRequirement")
for step in steps:
if scatterSpec and "scatter" in step.tool:
inputparms = copy.deepcopy(step.tool["inputs"])
outputparms = copy.deepcopy(step.tool["outputs"])
scatter = aslist(step.tool["scatter"])
inp_map = {i["id"]: i for i in inputparms}
for s in scatter:
if s not in inp_map:
raise WorkflowException(
"Invalid Scatter parameter '%s'" % s)
inp_map[s]["type"] = {
"type": "array",
"items": inp_map[s]["type"]
}
if step.tool.get("scatterMethod") == "nested_crossproduct":
nesting = len(scatter)
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
step.tool["inputs"] = inputparms
step.tool["outputs"] = outputparms
def scandeps(base, doc, reffields, urlfields, loadref):
r = []
if isinstance(doc, dict):
if "id" in doc:
if doc["id"].startswith("file://"):
df, _ = urlparse.urldefrag(doc["id"])
if base != df:
r.append({
"class": "File",
"path": df
})
base = df
for k, v in doc.iteritems():
if k in reffields:
for u in aslist(v):
if isinstance(u, dict):
r.extend(scandeps(base, u, reffields, urlfields, loadref))
else:
sub = loadref(base, u)
subid = urlparse.urljoin(base, u)
deps = {
"class": "File",
"path": subid
}
sf = scandeps(subid, sub, reffields, urlfields, loadref)
if sf:
deps["secondaryFiles"] = sf
r.append(deps)
elif k in urlfields:
for u in aslist(v):
r.append({
"class": "File",
"path": urlparse.urljoin(base, u)
})
else:
r.extend(scandeps(base, v, reffields, urlfields, loadref))
elif isinstance(doc, list):
for d in doc:
r.extend(scandeps(base, d, reffields, urlfields, loadref))
return r
def try_make_job(self, step, basedir, **kwargs):
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
try:
inputobj = object_from_state(self.state, inputparms, False)
if inputobj is None:
_logger.debug("[workflow %s] job step %s not ready", id(self),
step.id)
return
_logger.debug("[step %s] starting job step %s of workflow %s",
id(step), step.id, id(self))
if step.submitted:
return
callback = functools.partial(self.receive_output, step,
outputparms)
if "scatter" in step.tool:
scatter = aslist(step.tool["scatter"])
method = step.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException(
"Must specify scatterMethod when scattering over multiple inputs"
)
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter,
callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj, basedir,
scatter, callback,
**kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir,
scatter, callback, 0,
**kwargs)
else:
jobs = step.job(inputobj, basedir, callback, **kwargs)
step.submitted = True
for j in jobs:
yield j
except Exception as e:
_logger.exception("Unhandled exception")
self.processStatus = "permanentFail"
step.completed = True
def exeval(ex, jobinput, requirements, docpath, context, pull_image):
for r in reversed(requirements):
if r["class"] == "ExpressionEngineRequirement" and r["id"] == ex[
"engine"]:
if r["id"][0] != "#":
with open(os.path.join(docpath, r["id"])) as f:
ex_obj = yaml.load(f)
sch = process.get_schema()
validate.validate_ex(
sch.get_name("ExpressionEngineRequirement", ""), ex_obj)
r = ex_obj
runtime = []
img_id = docker.get_from_requirements(r.get("requirements"),
r.get("hints"), pull_image)
if img_id:
runtime = ["docker", "run", "-i", "--rm", img_id]
exdefs = []
for exdef in r.get("expressionDefs", []):
if isinstance(exdef, dict) and "ref" in exdef:
with open(os.path.join(r["_docpath"], exdef["ref"])) as f:
exdefs.append(f.read())
elif isinstance(exdef, basestring):
exdefs.append(exdef)
inp = {
"script": ex["script"],
"expressionDefs": exdefs,
"job": jobinput,
"context": context
}
_logger.debug(json.dumps(inp))
sp = subprocess.Popen(runtime + aslist(r["engineCommand"]),
shell=False,
close_fds=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE)
(stdoutdata, stderrdata) = sp.communicate(json.dumps(inp) + "\n\n")
if sp.returncode != 0:
raise WorkflowException(
"Expression engine returned non-zero exit code.")
return json.loads(stdoutdata)
raise WorkflowException("Unknown expression engine '%s'" % ex["engine"])
def try_make_job(self, step, basedir, **kwargs):
_logger.debug("Try to make job %s", step.id)
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
try:
inputobj = self.object_from_state(inputparms, False)
if inputobj is None:
return
if step.submitted:
return
callback = functools.partial(self.receive_output, step,
outputparms)
(scatterSpec,
_) = self.get_requirement("ScatterFeatureRequirement")
if scatterSpec and "scatter" in step.tool:
scatter = aslist(step.tool["scatter"])
method = step.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException(
"Must specify scatterMethod when scattering over multiple inputs"
)
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter,
callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj, basedir,
scatter, callback,
**kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir,
scatter, callback, 0,
**kwargs)
else:
jobs = step.job(inputobj, basedir, callback, **kwargs)
step.submitted = True
for j in jobs:
yield j
except Exception as e:
_logger.error(e)
self.processStatus = "permanentFail"
step.completed = True
def exeval(ex, jobinput, requirements, docpath, context, pull_image):
if ex["engine"] == "JsonPointer":
return ref_resolver.resolve_pointer({"job": jobinput, "context": context}, ex["script"])
for r in reversed(requirements):
if r["class"] == "ExpressionEngineRequirement" and r["id"] == ex["engine"]:
if r["id"][0] != "#":
with open(os.path.join(docpath, r["id"])) as f:
ex_obj = yaml.load(f)
sch = process.get_schema()
validate.validate_ex(sch.get_name("ExpressionEngineRequirement", ""), ex_obj)
r = ex_obj
runtime = []
img_id = docker.get_from_requirements(r.get("requirements"), r.get("hints"), pull_image)
if img_id:
runtime = ["docker", "run", "-i", "--rm", img_id]
exdefs = []
for exdef in r.get("expressionDefs", []):
if isinstance(exdef, dict) and "ref" in exdef:
with open(os.path.join(r["_docpath"], exdef["ref"])) as f:
exdefs.append(f.read())
elif isinstance(exdef, basestring):
exdefs.append(exdef)
inp = {
"script": ex["script"],
"expressionDefs": exdefs,
"job": jobinput,
"context": context
}
_logger.debug(json.dumps(inp))
sp = subprocess.Popen(runtime + aslist(r["engineCommand"]),
shell=False,
close_fds=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE)
(stdoutdata, stderrdata) = sp.communicate(json.dumps(inp) + "\n\n")
if sp.returncode != 0:
raise WorkflowException("Expression engine returned non-zero exit code.")
return json.loads(stdoutdata)
raise WorkflowException("Unknown expression engine '%s'" % ex["engine"])
def add_schemas(self, ns, base_url):
for sch in aslist(ns):
self.graph.parse(urlparse.urljoin(base_url, sch))
for s, _, _ in self.graph.triples( (None, RDF.type, RDF.Property) ):
self._add_properties(s)
for s, _, o in self.graph.triples( (None, RDFS.subPropertyOf, None) ):
self._add_properties(s)
self._add_properties(o)
for s, _, _ in self.graph.triples( (None, RDFS.range, None) ):
self._add_properties(s)
for s, _, _ in self.graph.triples( (None, RDF.type, OWL.ObjectProperty) ):
self._add_properties(s)
for s, _, _ in self.graph.triples( (None, None, None) ):
self.idx[str(s)] = True
def add_schemas(self, ns, base_url):
for sch in aslist(ns):
self.graph.parse(urlparse.urljoin(base_url, sch))
for s, _, _ in self.graph.triples((None, RDF.type, RDF.Property)):
self._add_properties(s)
for s, _, o in self.graph.triples((None, RDFS.subPropertyOf, None)):
self._add_properties(s)
self._add_properties(o)
for s, _, _ in self.graph.triples((None, RDFS.range, None)):
self._add_properties(s)
for s, _, _ in self.graph.triples(
(None, RDF.type, OWL.ObjectProperty)):
self._add_properties(s)
for s, _, _ in self.graph.triples((None, None, None)):
self.idx[str(s)] = True
def _draftDraft3dev1toDev2(doc, loader, baseuri):
doc = _updateDev2Script(doc)
if isinstance(doc, basestring):
return doc
# Convert expressions
if isinstance(doc, dict):
if "@import" in doc:
r, _ = loader.resolve_ref(doc["@import"], base_url=baseuri)
return _draftDraft3dev1toDev2(r, loader, r["id"])
for a in doc:
doc[a] = _draftDraft3dev1toDev2(doc[a], loader, baseuri)
if "class" in doc and (doc["class"] in ("CommandLineTool", "Workflow",
"ExpressionTool")):
added = False
if "requirements" in doc:
for r in doc["requirements"]:
if r["class"] == "ExpressionEngineRequirement":
if "engineConfig" in r:
doc["requirements"].append({
"class":
"InlineJavascriptRequirement",
"expressionLib": [
updateScript(sc)
for sc in aslist(r["engineConfig"])
]
})
added = True
doc["requirements"] = [
rq for rq in doc["requirements"]
if rq["class"] != "ExpressionEngineRequirement"
]
break
else:
doc["requirements"] = []
if not added:
doc["requirements"].append(
{"class": "InlineJavascriptRequirement"})
elif isinstance(doc, list):
return [_draftDraft3dev1toDev2(a, loader, baseuri) for a in doc]
return doc
def try_make_job(self, step, basedir, **kwargs):
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
supportsMultipleInput = bool(self.workflow.get_requirement("MultipleInputFeatureRequirement")[0])
try:
inputobj = object_from_state(self.state, inputparms, False, supportsMultipleInput)
if inputobj is None:
_logger.debug("[workflow %s] job step %s not ready", id(self), step.id)
return
_logger.debug("[step %s] starting job step %s of workflow %s", id(step), step.id, id(self))
if step.submitted:
return
callback = functools.partial(self.receive_output, step, outputparms)
if "scatter" in step.tool:
scatter = aslist(step.tool["scatter"])
method = step.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException("Must specify scatterMethod when scattering over multiple inputs")
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter, callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj, basedir, scatter, callback, **kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir, scatter, callback, 0, **kwargs)
else:
jobs = step.job(inputobj, basedir, callback, **kwargs)
step.submitted = True
for j in jobs:
yield j
except WorkflowException:
raise
except Exception as e:
_logger.exception("Unhandled exception")
self.processStatus = "permanentFail"
step.completed = True
def add_schemas(self, ns, base_url):
for sch in aslist(ns):
try:
self.graph.parse(urlparse.urljoin(base_url, sch), format="xml")
except xml.sax.SAXParseException:
self.graph.parse(urlparse.urljoin(base_url, sch), format="turtle")
for s, _, _ in self.graph.triples( (None, RDF.type, RDF.Property) ):
self._add_properties(s)
for s, _, o in self.graph.triples( (None, RDFS.subPropertyOf, None) ):
self._add_properties(s)
self._add_properties(o)
for s, _, _ in self.graph.triples( (None, RDFS.range, None) ):
self._add_properties(s)
for s, _, _ in self.graph.triples( (None, RDF.type, OWL.ObjectProperty) ):
self._add_properties(s)
for s, _, _ in self.graph.triples( (None, None, None) ):
self.idx[str(s)] = True
def add_schemas(self, ns, base_url):
for sch in aslist(ns):
try:
self.graph.parse(urlparse.urljoin(base_url, sch), format="xml")
except xml.sax.SAXParseException:
self.graph.parse(urlparse.urljoin(base_url, sch),
format="turtle")
for s, _, _ in self.graph.triples((None, RDF.type, RDF.Property)):
self._add_properties(s)
for s, _, o in self.graph.triples((None, RDFS.subPropertyOf, None)):
self._add_properties(s)
self._add_properties(o)
for s, _, _ in self.graph.triples((None, RDFS.range, None)):
self._add_properties(s)
for s, _, _ in self.graph.triples(
(None, RDF.type, OWL.ObjectProperty)):
self._add_properties(s)
for s, _, _ in self.graph.triples((None, None, None)):
self.idx[str(s)] = True
def try_make_job(self, step, basedir, **kwargs):
_logger.debug("Try to make job %s", step.id)
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
try:
inputobj = self.object_from_state(inputparms, False)
if inputobj is None:
return
if step.submitted:
return
callback = functools.partial(self.receive_output, step, outputparms)
(scatterSpec, _) = self.get_requirement("ScatterFeatureRequirement")
if scatterSpec and "scatter" in step.tool:
scatter = aslist(step.tool["scatter"])
method = step.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException("Must specify scatterMethod when scattering over multiple inputs")
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter, callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj, basedir, scatter, callback, **kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir, scatter, callback, 0, **kwargs)
else:
jobs = step.job(inputobj, basedir, callback, **kwargs)
step.submitted = True
for j in jobs:
yield j
except Exception as e:
_logger.error(e)
self.processStatus = "permanentFail"
step.completed = True
def __init__(self, toc, j):
self.typedoc = StringIO.StringIO()
self.toc = toc
self.subs = {}
self.docParent = {}
self.docAfter = {}
for t in j:
if "extends" in t:
for e in aslist(t["extends"]):
add_dictlist(self.subs, e, t["name"])
if "docParent" not in t and "docAfter" not in t:
add_dictlist(self.docParent, e, t["name"])
if t.get("docParent"):
add_dictlist(self.docParent, t["docParent"], t["name"])
if t.get("docAfter"):
add_dictlist(self.docAfter, t["docAfter"], t["name"])
_, _, metaschema_loader = schema.get_metaschema()
alltypes = schema.extend_and_specialize(j, metaschema_loader)
self.typemap = {}
self.uses = {}
for t in alltypes:
self.typemap[t["name"]] = t
if t["type"] == "https://w3id.org/cwl/salad#record":
for f in t["fields"]:
p = has_types(f)
for tp in p:
if tp not in self.uses:
self.uses[tp] = []
if (t["name"], f["name"]) not in self.uses[tp]:
_, frg1 = urlparse.urldefrag(t["name"])
_, frg2 = urlparse.urldefrag(f["name"])
self.uses[tp].append((frg1, frg2))
for f in alltypes:
if ("extends" not in f) and ("docParent" not in f) and ("docAfter" not in f):
self.render_type(f, 1)
def object_from_state(self, parms, frag_only):
inputobj = {}
for inp in parms:
iid = inp["id"]
if frag_only:
(_, iid) = urlparse.urldefrag(iid)
iid = iid.split(".")[-1]
if "source" in inp:
connections = aslist(inp["source"])
for src in connections:
if src in self.state and self.state[src] is not None:
if not self.match_types(inp["type"], self.state[src], iid, inputobj,
inp.get("linkMerge", ("merge_nested" if len(connections) > 1 else None))):
raise WorkflowException("Type mismatch between source '%s' (%s) and sink '%s' (%s)" % (src, self.state[src].parameter["type"], inp["id"], inp["type"]))
elif src not in self.state:
raise WorkflowException("Connect source '%s' on parameter '%s' does not exist" % (src, inp["id"]))
else:
return None
elif "default" in inp:
inputobj[iid] = inp["default"]
else:
raise WorkflowException("Value for %s not specified" % (inp["id"]))
return inputobj
def _draftDraft3dev1toDev2(doc, loader, baseuri):
doc = _updateDev2Script(doc)
if isinstance(doc, basestring):
return doc
# Convert expressions
if isinstance(doc, dict):
if "@import" in doc:
r, _ = loader.resolve_ref(doc["@import"], base_url=baseuri)
return _draftDraft3dev1toDev2(r, loader, r["id"])
for a in doc:
doc[a] = _draftDraft3dev1toDev2(doc[a], loader, baseuri)
if "class" in doc and (doc["class"] in ("CommandLineTool", "Workflow", "ExpressionTool")):
added = False
if "requirements" in doc:
for r in doc["requirements"]:
if r["class"] == "ExpressionEngineRequirement":
if "engineConfig" in r:
doc["requirements"].append({
"class":"InlineJavascriptRequirement",
"expressionLib": [updateScript(sc) for sc in aslist(r["engineConfig"])]
})
added = True
doc["requirements"] = [rq for rq in doc["requirements"] if rq["class"] != "ExpressionEngineRequirement"]
break
else:
doc["requirements"] = []
if not added:
doc["requirements"].append({"class":"InlineJavascriptRequirement"})
elif isinstance(doc, list):
return [_draftDraft3dev1toDev2(a, loader, baseuri) for a in doc]
return doc
def resolve_all(self, document, base_url, file_base=None):
loader = self
metadata = {}
if file_base is None:
file_base = base_url
if isinstance(document, dict):
# Handle $import and $include
if ('$import' in document or '$include' in document):
return self.resolve_ref(document, file_base)
elif isinstance(document, list):
pass
else:
return document, metadata
newctx = None
if isinstance(document, dict):
# Handle $base, $profile, $namespaces, $schemas and $graph
if "$base" in document:
base_url = document["$base"]
if "$profile" in document:
if not newctx:
newctx = SubLoader(self)
prof = self.fetch(document["$profile"])
newctx.add_namespaces(document.get("$namespaces", {}),
document["$profile"])
newctx.add_schemas(document.get("$schemas", []),
document["$profile"])
if "$namespaces" in document:
if not newctx:
newctx = SubLoader(self)
newctx.add_namespaces(document["$namespaces"])
if "$schemas" in document:
if not newctx:
newctx = SubLoader(self)
newctx.add_schemas(document["$schemas"], file_base)
if newctx:
loader = newctx
if "$graph" in document:
metadata = {k: v for k, v in document.items() if k != "$graph"}
document = document["$graph"]
metadata, _ = loader.resolve_all(metadata, base_url, file_base)
if isinstance(document, dict):
for identifer in loader.identity_links:
if identifer in document:
if isinstance(document[identifer], basestring):
document[identifer] = loader.expand_url(
document[identifer], base_url, scoped=True)
if document[identifer] not in loader.idx or isinstance(
loader.idx[document[identifer]], basestring):
loader.idx[document[identifer]] = document
base_url = document[identifer]
elif isinstance(document[identifer], list):
for n, v in enumerate(document[identifer]):
document[identifer][n] = loader.expand_url(
document[identifer][n], base_url, scoped=True)
if document[identifer][n] not in loader.idx:
loader.idx[document[identifer]
[n]] = document[identifer][n]
for d in document:
d2 = loader.expand_url(d, "", scoped=False, vocab_term=True)
if d != d2:
document[d2] = document[d]
del document[d]
for d in loader.url_fields:
if d in document:
if isinstance(document[d], basestring):
document[d] = loader.expand_url(
document[d],
base_url,
scoped=False,
vocab_term=(d in loader.vocab_fields))
elif isinstance(document[d], list):
document[d] = [
loader.expand_url(
url,
base_url,
scoped=False,
vocab_term=(d in loader.vocab_fields))
if isinstance(url, basestring) else url
for url in document[d]
]
try:
for key, val in document.items():
document[key], _ = loader.resolve_all(
val, base_url, file_base)
except validate.ValidationException as v:
_logger.debug("loader is %s", id(loader))
raise validate.ValidationException(
"(%s) (%s) Validation error in field %s:\n%s" %
(id(loader), file_base, key, validate.indent(str(v))))
elif isinstance(document, list):
i = 0
try:
while i < len(document):
val = document[i]
if isinstance(val, dict) and "$import" in val:
l, _ = loader.resolve_ref(val, file_base)
if isinstance(l, list):
del document[i]
for item in aslist(l):
document.insert(i, item)
i += 1
else:
document[i] = l
i += 1
else:
document[i], _ = loader.resolve_all(
val, base_url, file_base)
i += 1
except validate.ValidationException as v:
raise validate.ValidationException(
"(%s) (%s) Validation error in position %i:\n%s" %
(id(loader), file_base, i, validate.indent(str(v))))
for identifer in loader.identity_links:
if identifer in metadata:
if isinstance(metadata[identifer], basestring):
metadata[identifer] = loader.expand_url(
metadata[identifer], base_url, scoped=True)
loader.idx[metadata[identifer]] = document
return document, metadata
def __init__(self, toolpath_object, **kwargs):
(_, self.names, _) = get_schema()
self.tool = toolpath_object
self.requirements = kwargs.get("requirements", []) + self.tool.get(
"requirements", [])
self.hints = kwargs.get("hints", []) + self.tool.get("hints", [])
if "loader" in kwargs:
self.formatgraph = kwargs["loader"].graph
self.validate_hints(self.tool.get("hints", []),
strict=kwargs.get("strict"))
self.schemaDefs = {}
sd, _ = self.get_requirement("SchemaDefRequirement")
if sd:
sdtypes = sd["types"]
av = schema_salad.schema.make_valid_avro(
sdtypes, {t["name"]: t
for t in sdtypes}, set())
for i in av:
self.schemaDefs[i["name"]] = i
avro.schema.make_avsc_object(av, self.names)
# Build record schema from inputs
self.inputs_record_schema = {
"name": "input_record_schema",
"type": "record",
"fields": []
}
self.outputs_record_schema = {
"name": "outputs_record_schema",
"type": "record",
"fields": []
}
for key in ("inputs", "outputs"):
for i in self.tool[key]:
c = copy.copy(i)
doc_url, _ = urlparse.urldefrag(c['id'])
c["name"] = shortname(c["id"])
del c["id"]
if "type" not in c:
raise validate.ValidationException(
"Missing `type` in parameter `%s`" % c["name"])
if "default" in c and "null" not in aslist(c["type"]):
c["type"] = ["null"] + aslist(c["type"])
else:
c["type"] = c["type"]
if key == "inputs":
self.inputs_record_schema["fields"].append(c)
elif key == "outputs":
self.outputs_record_schema["fields"].append(c)
try:
self.inputs_record_schema = schema_salad.schema.make_valid_avro(
self.inputs_record_schema, {}, set())
avro.schema.make_avsc_object(self.inputs_record_schema, self.names)
except avro.schema.SchemaParseException as e:
raise validate.ValidationException(
"Got error `%s` while prcoessing inputs of %s:\n%s" %
(str(e), self.tool["id"],
json.dumps(self.inputs_record_schema, indent=4)))
try:
self.outputs_record_schema = schema_salad.schema.make_valid_avro(
self.outputs_record_schema, {}, set())
avro.schema.make_avsc_object(self.outputs_record_schema,
self.names)
except avro.schema.SchemaParseException as e:
raise validate.ValidationException(
"Got error `%s` while prcoessing outputs of %s:\n%s" %
(str(e), self.tool["id"],
json.dumps(self.outputs_record_schema, indent=4)))
def try_make_job(self, step, basedir, **kwargs):
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
supportsMultipleInput = bool(self.workflow.get_requirement("MultipleInputFeatureRequirement")[0])
try:
inputobj = object_from_state(self.state, inputparms, False, supportsMultipleInput)
if inputobj is None:
_logger.debug("[workflow %s] job step %s not ready", self.name, step.id)
return
_logger.debug("[step %s] starting job step %s of workflow %s", id(step), step.id, id(self))
if step.submitted:
return
callback = functools.partial(self.receive_output, step, outputparms)
valueFrom = {i["id"]: i["valueFrom"] for i in step.tool["inputs"] if "valueFrom" in i}
if len(valueFrom) > 0 and not bool(self.workflow.get_requirement("StepInputExpressionRequirement")[0]):
raise WorkflowException("Workflow step contains valueFrom but StepInputExpressionRequirement not in requirements")
vfinputs = {shortname(k): v for k,v in inputobj.iteritems()}
def valueFromFunc(k, v):
if k in valueFrom:
return expression.do_eval(valueFrom[k], vfinputs, self.workflow.requirements,
None, None, {}, context=v)
else:
return v
if "scatter" in step.tool:
scatter = aslist(step.tool["scatter"])
method = step.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException("Must specify scatterMethod when scattering over multiple inputs")
if "valueFrom" not in kwargs:
kwargs["valueFrom"] = valueFromFunc
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter,
callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj,
basedir, scatter, callback, **kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir,
scatter, callback, 0, **kwargs)
else:
_logger.debug("[workflow %s] Job is input %s", self.name, json.dumps(inputobj, indent=4))
inputobj = {k: valueFromFunc(k, v) for k,v in inputobj.items()}
_logger.debug("[workflow %s] Evaluated job input to %s", self.name, json.dumps(inputobj, indent=4))
jobs = step.job(inputobj, basedir, callback, **kwargs)
step.submitted = True
for j in jobs:
yield j
except WorkflowException:
raise
except Exception as e:
_logger.exception("Unhandled exception")
self.processStatus = "permanentFail"
step.completed = True
def resolve_all(self, document, base_url, file_base=None):
loader = self
metadata = {}
if file_base is None:
file_base = base_url
if isinstance(document, dict):
# Handle $import and $include
if ('$import' in document or '$include' in document):
return self.resolve_ref(document, file_base)
elif isinstance(document, list):
pass
else:
return document, metadata
newctx = None
if isinstance(document, dict):
# Handle $base, $profile, $namespaces, $schemas and $graph
if "$base" in document:
base_url = document["$base"]
if "$profile" in document:
if not newctx:
newctx = SubLoader(self)
prof = self.fetch(document["$profile"])
newctx.add_namespaces(document.get("$namespaces", {}), document["$profile"])
newctx.add_schemas(document.get("$schemas", []), document["$profile"])
if "$namespaces" in document:
if not newctx:
newctx = SubLoader(self)
newctx.add_namespaces(document["$namespaces"])
if "$schemas" in document:
if not newctx:
newctx = SubLoader(self)
newctx.add_schemas(document["$schemas"], file_base)
if newctx:
loader = newctx
if "$graph" in document:
metadata = {k: v for k,v in document.items() if k != "$graph"}
document = document["$graph"]
metadata, _ = loader.resolve_all(metadata, base_url, file_base)
if isinstance(document, dict):
for identifer in loader.identity_links:
if identifer in document:
if isinstance(document[identifer], basestring):
document[identifer] = loader.expand_url(document[identifer], base_url, scoped=True)
if document[identifer] not in loader.idx or isinstance(loader.idx[document[identifer]], basestring):
loader.idx[document[identifer]] = document
base_url = document[identifer]
elif isinstance(document[identifer], list):
for n, v in enumerate(document[identifer]):
document[identifer][n] = loader.expand_url(document[identifer][n], base_url, scoped=True)
if document[identifer][n] not in loader.idx:
loader.idx[document[identifer][n]] = document[identifer][n]
for d in document:
d2 = loader.expand_url(d, "", scoped=False, vocab_term=True)
if d != d2:
document[d2] = document[d]
del document[d]
for d in loader.url_fields:
if d in document:
if isinstance(document[d], basestring):
document[d] = loader.expand_url(document[d], base_url, scoped=False, vocab_term=(d in loader.vocab_fields))
elif isinstance(document[d], list):
document[d] = [loader.expand_url(url, base_url, scoped=False, vocab_term=(d in loader.vocab_fields)) if isinstance(url, basestring) else url for url in document[d] ]
try:
for key, val in document.items():
document[key], _ = loader.resolve_all(val, base_url, file_base)
except validate.ValidationException as v:
_logger.debug("loader is %s", id(loader))
raise validate.ValidationException("(%s) (%s) Validation error in field %s:\n%s" % (id(loader), file_base, key, validate.indent(str(v))))
elif isinstance(document, list):
i = 0
try:
while i < len(document):
val = document[i]
if isinstance(val, dict) and "$import" in val:
l, _ = loader.resolve_ref(val, file_base)
if isinstance(l, list):
del document[i]
for item in aslist(l):
document.insert(i, item)
i += 1
else:
document[i] = l
i += 1
else:
document[i], _ = loader.resolve_all(val, base_url, file_base)
i += 1
except validate.ValidationException as v:
raise validate.ValidationException("(%s) (%s) Validation error in position %i:\n%s" % (id(loader), file_base, i, validate.indent(str(v))))
return document, metadata
def collect_output(self, schema, builder, outdir):
r = None
if "outputBinding" in schema:
binding = schema["outputBinding"]
if "glob" in binding:
r = []
bg = builder.do_eval(binding["glob"])
for gb in aslist(bg):
r.extend([{
"path": g,
"class": "File"
} for g in builder.fs_access.glob(os.path.join(outdir, gb))
])
for files in r:
checksum = hashlib.sha1()
with builder.fs_access.open(files["path"], "rb") as f:
contents = f.read(CONTENT_LIMIT)
if binding.get("loadContents"):
files["contents"] = contents
filesize = 0
while contents != "":
checksum.update(contents)
filesize += len(contents)
contents = f.read(1024 * 1024)
files["checksum"] = "sha1$%s" % checksum.hexdigest()
files["size"] = filesize
if "outputEval" in binding:
r = builder.do_eval(binding["outputEval"], context=r)
if schema["type"] == "File" and (not isinstance(r, dict)
or "path" not in r):
raise WorkflowException(
"Expression must return a file object.")
if schema["type"] == "File":
if not r:
raise WorkflowException(
"No matches for output file with glob: {}.".format(
binding["glob"]))
if len(r) > 1:
raise WorkflowException(
"Multiple matches for output item that is a single file."
)
r = r[0]
if schema["type"] == "File" and "secondaryFiles" in binding:
r["secondaryFiles"] = []
for sf in aslist(binding["secondaryFiles"]):
if isinstance(sf, dict):
sfpath = builder.do_eval(sf, context=r["path"])
else:
sfpath = {
"path": substitute(r["path"], sf),
"class": "File"
}
if isinstance(sfpath, list):
r["secondaryFiles"].extend(sfpath)
else:
r["secondaryFiles"].append(sfpath)
for sf in r["secondaryFiles"]:
if not builder.fs_access.exists(sf["path"]):
raise WorkflowException(
"Missing secondary file of '%s' of primary file '%s'"
% (sf["path"], r["path"]))
if not r and schema["type"] == "record":
r = {}
for f in schema["fields"]:
r[f["name"]] = self.collect_output(f, builder, outdir)
return r
def __init__(self, toolpath_object, pos, **kwargs):
try:
makeTool = kwargs.get("makeTool")
self.embedded_tool = makeTool(toolpath_object["run"], **kwargs)
except validate.ValidationException as v:
raise WorkflowException(
"Tool definition %s failed validation:\n%s" %
(toolpath_object["run"]["id"], validate.indent(str(v))))
if "id" in toolpath_object:
self.id = toolpath_object["id"]
else:
self.id = "#step" + str(pos)
for field in ("inputs", "outputs"):
for i in toolpath_object[field]:
inputid = i["id"]
(_, d) = urlparse.urldefrag(inputid)
frag = d.split(".")[-1]
p = urlparse.urljoin(toolpath_object["run"].get("id", self.id),
"#" + frag)
found = False
for a in self.embedded_tool.tool[field]:
if a["id"] == p:
i.update(a)
found = True
if not found:
raise WorkflowException(
"Did not find %s parameter '%s' in workflow step" %
(field, p))
i["id"] = inputid
super(WorkflowStep, self).__init__(toolpath_object,
"Process",
do_validate=False,
**kwargs)
if self.embedded_tool.tool["class"] == "Workflow":
(feature,
_) = self.get_requirement("SubworkflowFeatureRequirement")
if not feature:
raise WorkflowException(
"Workflow contains embedded workflow but SubworkflowFeatureRequirement not declared"
)
if "scatter" in self.tool:
(feature, _) = self.get_requirement("ScatterFeatureRequirement")
if not feature:
raise WorkflowException(
"Workflow contains scatter but ScatterFeatureRequirement not declared"
)
inputparms = copy.deepcopy(self.tool["inputs"])
outputparms = copy.deepcopy(self.tool["outputs"])
scatter = aslist(self.tool["scatter"])
method = self.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException(
"Must specify scatterMethod when scattering over multiple inputs"
)
inp_map = {i["id"]: i for i in inputparms}
for s in scatter:
if s not in inp_map:
raise WorkflowException("Invalid Scatter parameter '%s'" %
s)
inp_map[s]["type"] = {
"type": "array",
"items": inp_map[s]["type"]
}
if self.tool.get("scatterMethod") == "nested_crossproduct":
nesting = len(scatter)
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
self.tool["inputs"] = inputparms
self.tool["outputs"] = outputparms
def __init__(self, toolpath_object, validateAs, do_validate=True, **kwargs):
(_, self.names) = get_schema()
self.tool = toolpath_object
if do_validate:
try:
# Validate tool documument
validate.validate_ex(self.names.get_name(validateAs, ""), self.tool, strict=kwargs.get("strict"))
except validate.ValidationException as v:
raise validate.ValidationException("Could not validate %s as %s:\n%s" % (self.tool.get("id"), validateAs, validate.indent(str(v))))
self.requirements = kwargs.get("requirements", []) + self.tool.get("requirements", [])
self.hints = kwargs.get("hints", []) + self.tool.get("hints", [])
self.validate_hints(self.tool.get("hints", []), strict=kwargs.get("strict"))
self.schemaDefs = {}
sd, _ = self.get_requirement("SchemaDefRequirement")
if sd:
for i in sd["types"]:
avro.schema.make_avsc_object(i, self.names)
self.schemaDefs[i["name"]] = i
# Build record schema from inputs
self.inputs_record_schema = {"name": "input_record_schema", "type": "record", "fields": []}
for i in self.tool["inputs"]:
c = copy.copy(i)
doc_url, fragment = urlparse.urldefrag(c['id'])
c["name"] = fragment
del c["id"]
if "type" not in c:
raise validate.ValidationException("Missing `type` in parameter `%s`" % c["name"])
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
else:
c["type"] = c["type"]
self.inputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.inputs_record_schema, self.names)
self.outputs_record_schema = {"name": "outputs_record_schema", "type": "record", "fields": []}
for i in self.tool["outputs"]:
c = copy.copy(i)
doc_url, fragment = urlparse.urldefrag(c['id'])
c["name"] = fragment
del c["id"]
if "type" not in c:
raise validate.ValidationException("Missing `type` in parameter `%s`" % c["name"])
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
else:
c["type"] = c["type"]
self.outputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.outputs_record_schema, self.names)
def resolve_all(self, document, base_url):
loader = self
metadata = {}
if isinstance(document, dict):
inc = '@include' in document
if '@import' in document or '@include' in document:
document, _ = self.resolve_ref(document, base_url)
else:
for identifer in self.identity_links:
if identifer in document:
if isinstance(document[identifer], basestring):
document[identifer] = self.expand_url(document[identifer], base_url, scoped=True)
if document[identifer] not in self.idx or isinstance(self.idx[document[identifer]], basestring):
self.idx[document[identifer]] = document
base_url = document[identifer]
elif isinstance(document[identifer], list):
for n, v in enumerate(document[identifer]):
document[identifer][n] = self.expand_url(document[identifer][n], base_url, scoped=True)
if document[identifer][n] not in self.idx:
self.idx[document[identifer][n]] = document[identifer][n]
if inc:
return document, {}
if isinstance(document, dict) and "@context" in document:
loader = Loader(self.ctx)
loader.idx = self.idx
loader.add_context(document["@context"])
if "@base" in loader.ctx:
base_url = loader.ctx["@base"]
if "@graph" in document:
metadata = {k: v for k,v in document.items() if k != "@graph"}
document = document["@graph"]
metadata, _ = self.resolve_all(metadata, base_url)
elif isinstance(document, list):
pass
else:
return document, metadata
try:
if isinstance(document, dict):
for d in document:
d2 = self.expand_url(d, "", scoped=False, vocab_term=True)
if d != d2:
document[d2] = document[d]
del document[d]
for d in loader.url_fields:
if d in document:
if isinstance(document[d], basestring):
document[d] = loader.expand_url(document[d], base_url, scoped=False, vocab_term=(d in loader.vocab_fields))
elif isinstance(document[d], list):
document[d] = [loader.expand_url(url, base_url, scoped=False, vocab_term=(d in loader.vocab_fields)) if isinstance(url, basestring) else url for url in document[d] ]
for key, val in document.items():
document[key], _ = loader.resolve_all(val, base_url)
elif isinstance(document, list):
i = 0
while i < len(document):
val = document[i]
if isinstance(val, dict) and "@import" in val and val.get("inline"):
l, _ = loader.resolve_all(val, base_url)
del document[i]
for item in aslist(l):
document.insert(i, item)
i += 1
else:
document[i], _ = loader.resolve_all(val, base_url)
i += 1
except validate.ValidationException as v:
if isinstance(key, basestring):
raise validate.ValidationException("Validation error in field %s:\n%s" % (key, validate.indent(str(v))))
else:
raise validate.ValidationException("Validation error in position %i:\n%s" % (key, validate.indent(str(v))))
return document, metadata
def exeval(ex, jobinput, requirements, outdir, tmpdir, context, pull_image):
if ex["engine"] == "cwl:JsonPointer":
try:
obj = {
"job": jobinput,
"context": context,
"outdir": outdir,
"tmpdir": tmpdir
}
return avro_ld.ref_resolver.resolve_json_pointer(obj, ex["script"])
except ValueError as v:
raise WorkflowException("%s in %s" % (v, obj))
for r in reversed(requirements):
if r["class"] == "ExpressionEngineRequirement" and r["id"] == ex[
"engine"]:
runtime = []
class DR(object):
pass
dr = DR()
dr.requirements = r.get("requirements", [])
dr.hints = r.get("hints", [])
(docker_req,
docker_is_req) = process.get_feature(dr, "DockerRequirement")
if docker_req:
img_id = docker.get_from_requirements(docker_req,
docker_is_req,
pull_image)
if img_id:
runtime = ["docker", "run", "-i", "--rm", img_id]
exdefs = []
for exdef in r.get("engineConfig", []):
if isinstance(exdef, dict) and "ref" in exdef:
with open(exdef["ref"][7:]) as f:
exdefs.append(f.read())
elif isinstance(exdef, basestring):
exdefs.append(exdef)
inp = {
"script": ex["script"],
"engineConfig": exdefs,
"job": jobinput,
"context": context,
"outdir": outdir,
"tmpdir": tmpdir,
}
_logger.debug("Invoking expression engine %s with %s",
runtime + aslist(r["engineCommand"]),
json.dumps(inp, indent=4))
sp = subprocess.Popen(runtime + aslist(r["engineCommand"]),
shell=False,
close_fds=True,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE)
(stdoutdata, stderrdata) = sp.communicate(json.dumps(inp) + "\n\n")
if sp.returncode != 0:
raise WorkflowException(
"Expression engine returned non-zero exit code on evaluation of\n%s"
% json.dumps(inp, indent=4))
return json.loads(stdoutdata)
raise WorkflowException("Unknown expression engine '%s'" % ex["engine"])
def bind_input(self, schema, datum, lead_pos=[], tail_pos=[]):
bindings = []
binding = None
if "inputBinding" in schema and isinstance(schema["inputBinding"],
dict):
binding = copy.copy(schema["inputBinding"])
if "position" in binding:
binding["position"] = aslist(lead_pos) + aslist(
binding["position"]) + aslist(tail_pos)
else:
binding["position"] = aslist(lead_pos) + [0] + aslist(tail_pos)
if "valueFrom" in binding:
binding["do_eval"] = binding["valueFrom"]
binding["valueFrom"] = datum
# Handle union types
if isinstance(schema["type"], list):
for t in schema["type"]:
if isinstance(t, basestring) and self.names.has_name(t, ""):
avsc = self.names.get_name(t, "")
elif isinstance(t,
dict) and "name" in t and self.names.has_name(
t["name"], ""):
avsc = self.names.get_name(t["name"], "")
else:
avsc = avro.schema.make_avsc_object(t, self.names)
if validate.validate(avsc, datum):
schema = copy.deepcopy(schema)
schema["type"] = t
return self.bind_input(schema,
datum,
lead_pos=lead_pos,
tail_pos=tail_pos)
raise validate.ValidationException("'%s' is not a valid union %s" %
(datum, schema["type"]))
elif isinstance(schema["type"], dict):
st = copy.deepcopy(schema["type"])
if binding and "inputBinding" not in st and "itemSeparator" not in binding and st[
"type"] in ("array", "map"):
st["inputBinding"] = {}
bindings.extend(
self.bind_input(st,
datum,
lead_pos=lead_pos,
tail_pos=tail_pos))
else:
if schema["type"] in self.schemaDefs:
schema = self.schemaDefs[schema["type"]]
if schema["type"] == "record":
for f in schema["fields"]:
if f["name"] in datum:
bindings.extend(
self.bind_input(f,
datum[f["name"]],
lead_pos=lead_pos,
tail_pos=f["name"]))
else:
datum[f["name"]] = f.get("default")
if schema["type"] == "map":
for n, item in datum.items():
b2 = None
if binding:
b2 = copy.deepcopy(binding)
b2["valueFrom"] = [n, item]
bindings.extend(
self.bind_input(
{
"type": schema["values"],
"inputBinding": b2
},
item,
lead_pos=n,
tail_pos=tail_pos))
binding = None
if schema["type"] == "array":
for n, item in enumerate(datum):
b2 = None
if binding:
b2 = copy.deepcopy(binding)
b2["valueFrom"] = item
bindings.extend(
self.bind_input(
{
"type": schema["items"],
"inputBinding": b2
},
item,
lead_pos=n,
tail_pos=tail_pos))
binding = None
if schema["type"] == "File":
self.files.append(datum)
if binding and binding.get("loadContents"):
with self.fs_access.open(datum["path"], "rb") as f:
datum["contents"] = f.read(CONTENT_LIMIT)
if "secondaryFiles" in schema:
if "secondaryFiles" not in datum:
datum["secondaryFiles"] = []
for sf in aslist(schema["secondaryFiles"]):
if isinstance(sf, dict) or "$(" in sf or "${" in sf:
sfpath = self.do_eval(sf, context=datum)
if isinstance(sfpath, basestring):
sfpath = {"path": sfpath, "class": "File"}
else:
sfpath = {
"path": substitute(datum["path"], sf),
"class": "File"
}
if isinstance(sfpath, list):
datum["secondaryFiles"].extend(sfpath)
else:
datum["secondaryFiles"].append(sfpath)
for sf in datum.get("secondaryFiles", []):
self.files.append(sf)
# Position to front of the sort key
if binding:
for bi in bindings:
bi["position"] = binding["position"] + bi["position"]
bindings.append(binding)
return bindings
def try_make_job(self, step, basedir, **kwargs):
inputobj = {}
_logger.debug("Try to make job %s", step.id)
(scatterSpec, _) = get_feature("Scatter", requirements=step.tool.get("requirements"), hints=step.tool.get("hints"))
if scatterSpec:
inputparms = copy.deepcopy(step.tool["inputs"])
outputparms = copy.deepcopy(step.tool["outputs"])
scatter = aslist(scatterSpec["scatter"])
inp_map = {i["id"]: i for i in inputparms}
for s in scatter:
if s not in inp_map:
raise WorkflowException("Invalid Scatter parameter '%s'" % s)
inp_map[s]["type"] = {"type": "array", "items": inp_map[s]["type"]}
if scatterSpec.get("scatterMethod") == "nested_crossproduct":
nesting = len(scatter)
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
else:
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
for inp in inputparms:
_logger.debug("Trying input %s", inp)
iid = inp["id"]
if "connect" in inp:
connections = inp["connect"]
for connection in aslist(connections):
src = connection["source"]
if src in self.state and self.state[src] is not None:
if not self.match_types(inp["type"], self.state[src], inp["id"], inputobj):
raise WorkflowException("Type mismatch between source '%s' (%s) and sink '%s' (%s)" % (src, self.state[src].parameter["type"], inp["id"], inp["type"]))
elif src not in self.state:
raise WorkflowException("Connect source '%s' on parameter '%s' does not exist" % (src, inp["id"]))
else:
return
elif "default" in inp:
inputobj[iid] = inp["default"]
else:
raise WorkflowException("Value for %s not specified" % (inp["id"]))
_logger.info("Creating job with input: %s", pprint.pformat(inputobj))
callback = functools.partial(self.receive_output, step, outputparms)
if scatterSpec:
method = scatterSpec.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException("Must specify scatterMethod when scattering over multiple inputs")
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter, callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj, basedir, scatter, callback, **kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir, scatter, callback, 0, **kwargs)
else:
jobs = step.job(inputobj, basedir, callback, **kwargs)
for j in jobs:
yield j
def extendsfrom(item, ex):
if "extends" in item:
for e in aslist(item["extends"]):
ex.insert(0, self.typemap[e])
extendsfrom(self.typemap[e], ex)
def process_type(t, g, context, defaultBase, namespaces, defaultPrefix):
if t["type"] == "record":
recordname = t["name"]
_logger.debug("Processing record %s\n", t)
classnode = URIRef(recordname)
g.add((classnode, RDF.type, RDFS.Class))
split = urlparse.urlsplit(recordname)
if "jsonldPrefix" in t:
predicate = "%s:%s" % (t["jsonldPrefix"], recordname)
elif split.scheme:
(ns, ln) = rdflib.namespace.split_uri(unicode(recordname))
predicate = recordname
recordname = ln
else:
predicate = "%s:%s" % (defaultPrefix, recordname)
if context.get(recordname, predicate) != predicate:
raise Exception("Predicate collision on '%s', '%s' != '%s'" % (recordname, context[t["name"]], predicate))
if not recordname:
raise Exception()
_logger.debug("Adding to context '%s' %s (%s)", recordname, predicate, type(predicate))
context[recordname] = predicate
for i in t.get("fields", []):
fieldname = i["name"]
_logger.debug("Processing field %s", i)
v = pred(t, i, fieldname, context, defaultPrefix, namespaces)
if isinstance(v, basestring):
v = v if v[0] != "@" else None
else:
v = v["_@id"] if v.get("_@id", "@")[0] != "@" else None
if v:
(ns, ln) = rdflib.namespace.split_uri(unicode(v))
if ns[0:-1] in namespaces:
propnode = namespaces[ns[0:-1]][ln]
else:
propnode = URIRef(v)
g.add((propnode, RDF.type, RDF.Property))
g.add((propnode, RDFS.domain, classnode))
# TODO generate range from datatype.
if isinstance(i["type"], dict) and "name" in i["type"]:
process_type(i["type"], g, context, defaultBase, namespaces, defaultPrefix)
if "extends" in t:
for e in aslist(t["extends"]):
g.add((classnode, RDFS.subClassOf, URIRef(e)))
elif t["type"] == "enum":
_logger.debug("Processing enum %s", t["name"])
for i in t["symbols"]:
pred(t, None, i, context, defaultBase, namespaces)
def process_type(t, g, context, defaultBase, namespaces, defaultPrefix):
if t["type"] == "record":
recordname = t["name"]
_logger.debug("Processing record %s\n", t)
classnode = URIRef(recordname)
g.add((classnode, RDF.type, RDFS.Class))
split = urlparse.urlsplit(recordname)
if "jsonldPrefix" in t:
predicate = "%s:%s" % (t["jsonldPrefix"], recordname)
elif split.scheme:
(ns, ln) = rdflib.namespace.split_uri(unicode(recordname))
predicate = recordname
recordname = ln
else:
predicate = "%s:%s" % (defaultPrefix, recordname)
if context.get(recordname, predicate) != predicate:
raise Exception("Predicate collision on '%s', '%s' != '%s'" %
(recordname, context[t["name"]], predicate))
if not recordname:
raise Exception()
_logger.debug("Adding to context '%s' %s (%s)", recordname, predicate,
type(predicate))
context[recordname] = predicate
for i in t.get("fields", []):
fieldname = i["name"]
_logger.debug("Processing field %s", i)
v = pred(t, i, fieldname, context, defaultPrefix, namespaces)
if isinstance(v, basestring):
v = v if v[0] != "@" else None
else:
v = v["_@id"] if v.get("_@id", "@")[0] != "@" else None
if v:
(ns, ln) = rdflib.namespace.split_uri(unicode(v))
if ns[0:-1] in namespaces:
propnode = namespaces[ns[0:-1]][ln]
else:
propnode = URIRef(v)
g.add((propnode, RDF.type, RDF.Property))
g.add((propnode, RDFS.domain, classnode))
# TODO generate range from datatype.
if isinstance(i["type"], dict) and "name" in i["type"]:
process_type(i["type"], g, context, defaultBase, namespaces,
defaultPrefix)
if "extends" in t:
for e in aslist(t["extends"]):
g.add((classnode, RDFS.subClassOf, URIRef(e)))
elif t["type"] == "enum":
_logger.debug("Processing enum %s", t["name"])
for i in t["symbols"]:
pred(t, None, i, context, defaultBase, namespaces)
def __init__(self, toolpath_object, validateAs, docpath):
self.names = get_schema()
self.docpath = docpath
self.tool = toolpath_object
#if self.tool.get("@context") != TOOL_CONTEXT_URL:
# raise Exception("Missing or invalid '@context' field in tool description document, must be %s" % TOOL_CONTEXT_URL)
# Validate tool documument
validate.validate_ex(self.names.get_name(validateAs, ""), self.tool)
self.validate_requirements(self.tool, "requirements")
self.validate_requirements(self.tool, "hints")
for t in self.tool.get("requirements", []):
t["_docpath"] = docpath
for t in self.tool.get("hints", []):
t["_docpath"] = docpath
# Import schema defs
self.schemaDefs = {
"Any": [
"null",
"boolean",
"int",
"long",
"float",
"double",
"bytes",
"string",
"File",
{"type": "array", "items": "Any"},
{"type": "map", "values": "Any"}
]}
if self.tool.get("schemaDefs"):
for i in self.tool["schemaDefs"]:
avro.schema.make_avsc_object(i, self.names)
self.schemaDefs[i["name"]] = i
# Build record schema from inputs
self.inputs_record_schema = {"name": "input_record_schema", "type": "record", "fields": []}
for i in self.tool["inputs"]:
c = copy.copy(i)
c["name"] = c["id"][1:]
del c["id"]
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
self.inputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.inputs_record_schema, self.names)
self.outputs_record_schema = {"name": "outputs_record_schema", "type": "record", "fields": []}
for i in self.tool["outputs"]:
c = copy.copy(i)
c["name"] = c["id"][1:]
del c["id"]
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
self.outputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.outputs_record_schema, self.names)
def extend_and_specialize(items, loader):
"""Apply 'extend' and 'specialize' to fully materialize derived record
types."""
types = {t["name"]: t for t in items}
n = []
for t in items:
t = copy.deepcopy(t)
if "extends" in t:
if "specialize" in t:
spec = {
sp["specializeFrom"]: sp["specializeTo"]
for sp in aslist(t["specialize"])
}
else:
spec = {}
exfields = []
exsym = []
for ex in aslist(t["extends"]):
if ex not in types:
raise Exception(
"Extends %s in %s refers to invalid base type" %
(t["extends"], t["name"]))
basetype = copy.deepcopy(types[ex])
if t["type"] == "record":
if spec:
basetype["fields"] = replace_type(
basetype.get("fields", []), spec, loader, set())
for f in basetype.get("fields", []):
if "inherited_from" not in f:
f["inherited_from"] = ex
exfields.extend(basetype.get("fields", []))
elif t["type"] == "enum":
exsym.extend(basetype.get("symbols", []))
if t["type"] == "record":
exfields.extend(t.get("fields", []))
t["fields"] = exfields
fieldnames = set()
for field in t["fields"]:
if field["name"] in fieldnames:
raise validate.ValidationException(
"Field name %s appears twice in %s" %
(field["name"], t["name"]))
else:
fieldnames.add(field["name"])
for y in [x for x in t["fields"] if x["name"] == "class"]:
y["type"] = {
"type": "enum",
"symbols": [r["name"]],
"name": r["name"] + "_class",
}
y["doc"] = "Must be `%s` to indicate this is a %s object." % (
r["name"], r["name"])
elif t["type"] == "enum":
exsym.extend(t.get("symbols", []))
t["symbol"] = exsym
types[t["name"]] = t
n.append(t)
ex_types = {t["name"]: t for t in n}
extended_by = {}
for t in n:
if "extends" in t:
for ex in aslist(t["extends"]):
if ex_types[ex].get("abstract"):
add_dictlist(extended_by, ex, ex_types[t["name"]])
add_dictlist(extended_by, avro_name(ex), ex_types[ex])
for t in n:
if "fields" in t:
t["fields"] = replace_type(t["fields"], extended_by, loader, set())
return n
def __init__(self,
toolpath_object,
validateAs,
do_validate=True,
**kwargs):
(_, self.names) = get_schema()
self.tool = toolpath_object
if do_validate:
try:
# Validate tool documument
validate.validate_ex(self.names.get_name(validateAs, ""),
self.tool,
strict=kwargs.get("strict"))
except validate.ValidationException as v:
raise validate.ValidationException(
"Could not validate %s as %s:\n%s" %
(self.tool.get("id"), validateAs, validate.indent(str(v))))
self.requirements = kwargs.get("requirements", []) + self.tool.get(
"requirements", [])
self.hints = kwargs.get("hints", []) + self.tool.get("hints", [])
self.validate_hints(self.tool.get("hints", []),
strict=kwargs.get("strict"))
self.schemaDefs = {}
sd, _ = self.get_requirement("SchemaDefRequirement")
if sd:
for i in sd["types"]:
avro.schema.make_avsc_object(i, self.names)
self.schemaDefs[i["name"]] = i
# Build record schema from inputs
self.inputs_record_schema = {
"name": "input_record_schema",
"type": "record",
"fields": []
}
for i in self.tool["inputs"]:
c = copy.copy(i)
doc_url, fragment = urlparse.urldefrag(c['id'])
c["name"] = fragment
del c["id"]
if "type" not in c:
raise validate.ValidationException(
"Missing `type` in parameter `%s`" % c["name"])
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
else:
c["type"] = c["type"]
self.inputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.inputs_record_schema, self.names)
self.outputs_record_schema = {
"name": "outputs_record_schema",
"type": "record",
"fields": []
}
for i in self.tool["outputs"]:
c = copy.copy(i)
doc_url, fragment = urlparse.urldefrag(c['id'])
c["name"] = fragment
del c["id"]
if "type" not in c:
raise validate.ValidationException(
"Missing `type` in parameter `%s`" % c["name"])
if "default" in c:
c["type"] = ["null"] + aslist(c["type"])
else:
c["type"] = c["type"]
self.outputs_record_schema["fields"].append(c)
avro.schema.make_avsc_object(self.outputs_record_schema, self.names)
def try_make_job(self, step, basedir, **kwargs):
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
supportsMultipleInput = bool(
self.workflow.get_requirement("MultipleInputFeatureRequirement")
[0])
try:
inputobj = object_from_state(self.state, inputparms, False,
supportsMultipleInput)
if inputobj is None:
_logger.debug("[workflow %s] job step %s not ready", self.name,
step.id)
return
_logger.debug("[step %s] starting job step %s of workflow %s",
id(step), step.id, id(self))
if step.submitted:
return
callback = functools.partial(self.receive_output, step,
outputparms)
valueFrom = {
i["id"]: i["valueFrom"]
for i in step.tool["inputs"] if "valueFrom" in i
}
if len(valueFrom) > 0 and not bool(
self.workflow.get_requirement(
"StepInputExpressionRequirement")[0]):
raise WorkflowException(
"Workflow step contains valueFrom but StepInputExpressionRequirement not in requirements"
)
vfinputs = {shortname(k): v for k, v in inputobj.iteritems()}
def valueFromFunc(k, v):
if k in valueFrom:
return expression.do_eval(valueFrom[k],
vfinputs,
self.workflow.requirements,
None,
None, {},
context=v)
else:
return v
if "scatter" in step.tool:
scatter = aslist(step.tool["scatter"])
method = step.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException(
"Must specify scatterMethod when scattering over multiple inputs"
)
if "valueFrom" not in kwargs:
kwargs["valueFrom"] = valueFromFunc
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir, scatter,
callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(step, inputobj, basedir,
scatter, callback,
**kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir,
scatter, callback, 0,
**kwargs)
else:
_logger.debug("[workflow %s] Job is input %s", self.name,
json.dumps(inputobj, indent=4))
inputobj = {
k: valueFromFunc(k, v)
for k, v in inputobj.items()
}
_logger.debug("[workflow %s] Evaluated job input to %s",
self.name, json.dumps(inputobj, indent=4))
jobs = step.job(inputobj, basedir, callback, **kwargs)
step.submitted = True
for j in jobs:
yield j
except WorkflowException:
raise
except Exception as e:
_logger.exception("Unhandled exception")
self.processStatus = "permanentFail"
step.completed = True
def extend_and_specialize(items, loader):
"""Apply 'extend' and 'specialize' to fully materialize derived record
types."""
types = {t["name"]: t for t in items}
n = []
for t in items:
t = copy.deepcopy(t)
if "extends" in t:
if "specialize" in t:
spec = {sp["specializeFrom"]: sp["specializeTo"] for sp in aslist(t["specialize"])}
else:
spec = {}
exfields = []
exsym = []
for ex in aslist(t["extends"]):
if ex not in types:
raise Exception("Extends %s in %s refers to invalid base type" % (t["extends"], t["name"]))
basetype = copy.deepcopy(types[ex])
if t["type"] == "record":
if spec:
basetype["fields"] = replace_type(basetype.get("fields", []), spec, loader, set())
for f in basetype.get("fields", []):
if "inherited_from" not in f:
f["inherited_from"] = ex
exfields.extend(basetype.get("fields", []))
elif t["type"] == "enum":
exsym.extend(basetype.get("symbols", []))
if t["type"] == "record":
exfields.extend(t.get("fields", []))
t["fields"] = exfields
fieldnames = set()
for field in t["fields"]:
if field["name"] in fieldnames:
raise validate.ValidationException("Field name %s appears twice in %s" % (field["name"], t["name"]))
else:
fieldnames.add(field["name"])
for y in [x for x in t["fields"] if x["name"] == "class"]:
y["type"] = {"type": "enum",
"symbols": [r["name"]],
"name": r["name"]+"_class",
}
y["doc"] = "Must be `%s` to indicate this is a %s object." % (r["name"], r["name"])
elif t["type"] == "enum":
exsym.extend(t.get("symbols", []))
t["symbol"] = exsym
types[t["name"]] = t
n.append(t)
ex_types = {t["name"]: t for t in n}
extended_by = {}
for t in n:
if "extends" in t:
for ex in aslist(t["extends"]):
if ex_types[ex].get("abstract"):
add_dictlist(extended_by, ex, ex_types[t["name"]])
add_dictlist(extended_by, avro_name(ex), ex_types[ex])
for t in n:
if "fields" in t:
t["fields"] = replace_type(t["fields"], extended_by, loader, set())
return n
def __init__(self, toolpath_object, pos, **kwargs):
if "id" in toolpath_object:
self.id = toolpath_object["id"]
else:
self.id = "#step" + str(pos)
try:
makeTool = kwargs.get("makeTool")
runobj = None
if isinstance(toolpath_object["run"], basestring):
runobj, _ = schema_salad.schema.load_and_validate(
kwargs["loader"], kwargs["avsc_names"],
toolpath_object["run"], True)
else:
runobj = toolpath_object["run"]
self.embedded_tool = makeTool(runobj, **kwargs)
except validate.ValidationException as v:
raise WorkflowException(
"Tool definition %s failed validation:\n%s" %
(toolpath_object["run"], validate.indent(str(v))))
for field in ("inputs", "outputs"):
for i in toolpath_object[field]:
inputid = i["id"]
p = shortname(inputid)
found = False
for a in self.embedded_tool.tool[field]:
frag = shortname(a["id"])
if frag == p:
i.update(a)
found = True
if not found:
i["type"] = "Any"
#raise WorkflowException("Parameter '%s' of %s in workflow step %s does not correspond to parameter in %s" % (p, field, self.id, self.embedded_tool.tool.get("id")))
i["id"] = inputid
super(WorkflowStep, self).__init__(toolpath_object, **kwargs)
if self.embedded_tool.tool["class"] == "Workflow":
(feature,
_) = self.get_requirement("SubworkflowFeatureRequirement")
if not feature:
raise WorkflowException(
"Workflow contains embedded workflow but SubworkflowFeatureRequirement not in requirements"
)
if "scatter" in self.tool:
(feature, _) = self.get_requirement("ScatterFeatureRequirement")
if not feature:
raise WorkflowException(
"Workflow contains scatter but ScatterFeatureRequirement not in requirements"
)
inputparms = copy.deepcopy(self.tool["inputs"])
outputparms = copy.deepcopy(self.tool["outputs"])
scatter = aslist(self.tool["scatter"])
method = self.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException(
"Must specify scatterMethod when scattering over multiple inputs"
)
inp_map = {i["id"]: i for i in inputparms}
for s in scatter:
if s not in inp_map:
raise WorkflowException("Invalid Scatter parameter '%s'" %
s)
inp_map[s]["type"] = {
"type": "array",
"items": inp_map[s]["type"]
}
if self.tool.get("scatterMethod") == "nested_crossproduct":
nesting = len(scatter)
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
self.tool["inputs"] = inputparms
self.tool["outputs"] = outputparms
def extendsfrom(item, ex):
if "extends" in item:
for e in aslist(item["extends"]):
ex.insert(0, self.typemap[e])
extendsfrom(self.typemap[e], ex)
def job(self, joborder, input_basedir, output_callback, **kwargs):
builder = self._init_job(joborder, input_basedir, **kwargs)
if self.tool["baseCommand"]:
for n, b in enumerate(aslist(self.tool["baseCommand"])):
builder.bindings.append({"position": [-1000000, n], "valueFrom": b})
if self.tool.get("arguments"):
for i, a in enumerate(self.tool["arguments"]):
if isinstance(a, dict):
a = copy.copy(a)
if a.get("position"):
a["position"] = [a["position"], i]
else:
a["position"] = [0, i]
a["do_eval"] = a["valueFrom"]
a["valueFrom"] = None
builder.bindings.append(a)
else:
builder.bindings.append({"position": [0, i], "valueFrom": a})
builder.bindings.sort(key=lambda a: a["position"])
reffiles = set((f["path"] for f in builder.files))
j = self.makeJobRunner()
j.joborder = builder.job
j.stdin = None
j.stdout = None
j.successCodes = self.tool.get("successCodes")
j.temporaryFailCodes = self.tool.get("temporaryFailCodes")
j.permanentFailCodes = self.tool.get("permanentFailCodes")
j.requirements = self.requirements
j.hints = self.hints
_logger.debug(
"[job %s] initializing from %s%s",
id(j),
self.tool.get("id", ""),
" as part of %s" % kwargs["part_of"] if "part_of" in kwargs else "",
)
_logger.debug("[job %s] %s", id(j), json.dumps(joborder, indent=4))
builder.pathmapper = None
if self.tool.get("stdin"):
j.stdin = builder.do_eval(self.tool["stdin"])
if isinstance(j.stdin, dict) and "ref" in j.stdin:
j.stdin = builder.job[j.stdin["ref"][1:]]["path"]
reffiles.add(j.stdin)
if self.tool.get("stdout"):
j.stdout = builder.do_eval(self.tool["stdout"])
if os.path.isabs(j.stdout) or ".." in j.stdout:
raise validate.ValidationException("stdout must be a relative path")
builder.pathmapper = self.makePathMapper(reffiles, input_basedir, **kwargs)
builder.requirements = j.requirements
for f in builder.files:
f["path"] = builder.pathmapper.mapper(f["path"])[1]
_logger.debug("[job %s] command line bindings is %s", id(j), json.dumps(builder.bindings, indent=4))
_logger.debug(
"[job %s] path mappings is %s",
id(j),
json.dumps({p: builder.pathmapper.mapper(p) for p in builder.pathmapper.files()}, indent=4),
)
dockerReq, _ = self.get_requirement("DockerRequirement")
if dockerReq and kwargs.get("use_container"):
out_prefix = kwargs.get("tmp_outdir_prefix")
j.outdir = kwargs.get("outdir") or tempfile.mkdtemp(prefix=out_prefix)
tmpdir_prefix = kwargs.get("tmpdir_prefix")
j.tmpdir = kwargs.get("tmpdir") or tempfile.mkdtemp(prefix=tmpdir_prefix)
else:
j.outdir = builder.outdir
j.tmpdir = builder.tmpdir
createFiles, _ = self.get_requirement("CreateFileRequirement")
j.generatefiles = {}
if createFiles:
for t in createFiles["fileDef"]:
j.generatefiles[builder.do_eval(t["filename"])] = copy.deepcopy(builder.do_eval(t["fileContent"]))
j.environment = {}
evr, _ = self.get_requirement("EnvVarRequirement")
if evr:
for t in evr["envDef"]:
j.environment[t["envName"]] = builder.do_eval(t["envValue"])
j.command_line = flatten(map(builder.generate_arg, builder.bindings))
j.pathmapper = builder.pathmapper
j.collect_outputs = functools.partial(self.collect_output_ports, self.tool["outputs"], builder)
j.output_callback = output_callback
yield j
def collect_output(self, schema, builder, outdir):
r = None
if "outputBinding" in schema:
binding = schema["outputBinding"]
globpatterns = []
if "glob" in binding:
r = []
for gb in aslist(binding["glob"]):
try:
gb = builder.do_eval(gb)
globpatterns.append(gb)
if gb:
r.extend([{"path": g, "class": "File"} for g in builder.fs_access.glob(os.path.join(outdir, gb))])
except (OSError, IOError) as e:
_logger.warn(str(e))
for files in r:
checksum = hashlib.sha1()
with builder.fs_access.open(files["path"], "rb") as f:
contents = f.read(CONTENT_LIMIT)
if binding.get("loadContents"):
files["contents"] = contents
filesize = 0
while contents != "":
checksum.update(contents)
filesize += len(contents)
contents = f.read(1024*1024)
files["checksum"] = "sha1$%s" % checksum.hexdigest()
files["size"] = filesize
if "format" in schema:
files["format"] = builder.do_eval(schema["format"], context=files)
optional = False
singlefile = False
if isinstance(schema["type"], list):
if "null" in schema["type"]:
optional = True
if "File" in schema["type"]:
singlefile = True
elif schema["type"] == "File":
singlefile = True
if "outputEval" in binding:
r = builder.do_eval(binding["outputEval"], context=r)
if singlefile:
# Handle single file outputs not wrapped in a list
if r is not None and not isinstance(r, (list, tuple)):
r = [r]
if optional and r is None:
pass
elif (r is None or len(r) != 1 or not isinstance(r[0], dict) or "path" not in r[0]):
raise WorkflowException("Expression must return a file object for %s." % schema["id"])
if singlefile:
if not r and not optional:
raise WorkflowException("Did not find output file with glob pattern: '{}'".format(globpatterns))
elif not r and optional:
pass
elif isinstance(r, list):
if len(r) > 1:
raise WorkflowException("Multiple matches for output item that is a single file.")
else:
r = r[0]
if "secondaryFiles" in schema:
for primary in aslist(r):
if isinstance(primary, dict):
primary["secondaryFiles"] = []
for sf in aslist(schema["secondaryFiles"]):
if isinstance(sf, dict) or "$(" in sf or "${" in sf:
sfpath = builder.do_eval(sf, context=r)
if isinstance(sfpath, basestring):
sfpath = {"path": sfpath, "class": "File"}
else:
sfpath = {"path": substitute(primary["path"], sf), "class": "File"}
for sfitem in aslist(sfpath):
if builder.fs_access.exists(sfitem["path"]):
primary["secondaryFiles"].append(sfitem)
if not r and optional:
r = None
if not r and isinstance(schema["type"], dict) and schema["type"]["type"] == "record":
r = {}
for f in schema["type"]["fields"]:
r[shortname(f["name"])] = self.collect_output(f, builder, outdir)
return r
def bind_input(self, schema, datum, lead_pos=[], tail_pos=[]):
bindings = []
binding = None
if "inputBinding" in schema and isinstance(schema["inputBinding"], dict):
binding = copy.copy(schema["inputBinding"])
if "position" in binding:
binding["position"] = aslist(lead_pos) + aslist(binding["position"]) + aslist(tail_pos)
else:
binding["position"] = aslist(lead_pos) + [0] + aslist(tail_pos)
if "valueFrom" in binding:
binding["do_eval"] = binding["valueFrom"]
binding["valueFrom"] = datum
# Handle union types
if isinstance(schema["type"], list):
for t in schema["type"]:
if isinstance(t, basestring) and self.names.has_name(t, ""):
avsc = self.names.get_name(t, "")
elif isinstance(t, dict) and "name" in t and self.names.has_name(t["name"], ""):
avsc = self.names.get_name(t["name"], "")
else:
avsc = avro.schema.make_avsc_object(t, self.names)
if validate.validate(avsc, datum):
schema = copy.deepcopy(schema)
schema["type"] = t
return self.bind_input(schema, datum, lead_pos=lead_pos, tail_pos=tail_pos)
raise validate.ValidationException("'%s' is not a valid union %s" % (datum, schema["type"]))
elif isinstance(schema["type"], dict):
st = copy.deepcopy(schema["type"])
if (
binding
and "inputBinding" not in st
and "itemSeparator" not in binding
and st["type"] in ("array", "map")
):
st["inputBinding"] = {}
bindings.extend(self.bind_input(st, datum, lead_pos=lead_pos, tail_pos=tail_pos))
else:
if schema["type"] in self.schemaDefs:
schema = self.schemaDefs[schema["type"]]
if schema["type"] == "record":
for f in schema["fields"]:
if f["name"] in datum:
bindings.extend(self.bind_input(f, datum[f["name"]], lead_pos=lead_pos, tail_pos=f["name"]))
else:
datum[f["name"]] = f.get("default")
if schema["type"] == "map":
for n, item in datum.items():
b2 = None
if binding:
b2 = copy.deepcopy(binding)
b2["valueFrom"] = [n, item]
bindings.extend(
self.bind_input(
{"type": schema["values"], "inputBinding": b2}, item, lead_pos=n, tail_pos=tail_pos
)
)
binding = None
if schema["type"] == "array":
for n, item in enumerate(datum):
b2 = None
if binding:
b2 = copy.deepcopy(binding)
b2["valueFrom"] = item
bindings.extend(
self.bind_input(
{"type": schema["items"], "inputBinding": b2}, item, lead_pos=n, tail_pos=tail_pos
)
)
binding = None
if schema["type"] == "File":
self.files.append(datum)
if binding:
if binding.get("loadContents"):
with self.fs_access.open(datum["path"], "rb") as f:
datum["contents"] = f.read(CONTENT_LIMIT)
if "secondaryFiles" in binding:
if "secondaryFiles" not in datum:
datum["secondaryFiles"] = []
for sf in aslist(binding["secondaryFiles"]):
if isinstance(sf, dict):
sfpath = self.do_eval(sf, context=datum["path"])
else:
sfpath = {"path": substitute(datum["path"], sf), "class": "File"}
if isinstance(sfpath, list):
datum["secondaryFiles"].extend(sfpath)
self.files.extend(sfpath)
else:
datum["secondaryFiles"].append(sfpath)
self.files.append(sfpath)
# Position to front of the sort key
if binding:
for bi in bindings:
bi["position"] = binding["position"] + bi["position"]
bindings.append(binding)
return bindings
def job(self, joborder, input_basedir, output_callback, **kwargs):
builder = self._init_job(joborder, input_basedir, **kwargs)
if self.tool["baseCommand"]:
for n, b in enumerate(aslist(self.tool["baseCommand"])):
builder.bindings.append({
"position": [-1000000, n],
"valueFrom": b
})
if self.tool.get("arguments"):
for i, a in enumerate(self.tool["arguments"]):
if isinstance(a, dict):
a = copy.copy(a)
if a.get("position"):
a["position"] = [a["position"], i]
else:
a["position"] = [0, i]
a["do_eval"] = a["valueFrom"]
a["valueFrom"] = None
builder.bindings.append(a)
else:
builder.bindings.append({
"position": [0, i],
"valueFrom": a
})
builder.bindings.sort(key=lambda a: a["position"])
reffiles = set((f["path"] for f in builder.files))
j = self.makeJobRunner()
j.builder = builder
j.joborder = builder.job
j.stdin = None
j.stdout = None
j.successCodes = self.tool.get("successCodes")
j.temporaryFailCodes = self.tool.get("temporaryFailCodes")
j.permanentFailCodes = self.tool.get("permanentFailCodes")
j.requirements = self.requirements
j.hints = self.hints
j.name = uniquename(kwargs.get("name", str(id(j))))
_logger.debug("[job %s] initializing from %s%s",
j.name,
self.tool.get("id", ""),
" as part of %s" % kwargs["part_of"] if "part_of" in kwargs else "")
_logger.debug("[job %s] %s", j.name, json.dumps(joborder, indent=4))
builder.pathmapper = None
if self.tool.get("stdin"):
j.stdin = builder.do_eval(self.tool["stdin"])
if isinstance(j.stdin, dict) and "ref" in j.stdin:
j.stdin = builder.job[j.stdin["ref"][1:]]["path"]
reffiles.add(j.stdin)
if self.tool.get("stdout"):
j.stdout = builder.do_eval(self.tool["stdout"])
if os.path.isabs(j.stdout) or ".." in j.stdout:
raise validate.ValidationException("stdout must be a relative path")
builder.pathmapper = self.makePathMapper(reffiles, input_basedir, **kwargs)
builder.requirements = j.requirements
# map files to assigned path inside a container. We need to also explicitly
# walk over input as implicit reassignment doesn't reach everything in builder.bindings
def _check_adjust(f):
if not f.get("containerfs"):
f["path"] = builder.pathmapper.mapper(f["path"])[1]
f["containerfs"] = True
return f
_logger.debug("[job %s] path mappings is %s", j.name, json.dumps({p: builder.pathmapper.mapper(p) for p in builder.pathmapper.files()}, indent=4))
adjustFileObjs(builder.files, _check_adjust)
adjustFileObjs(builder.bindings, _check_adjust)
_logger.debug("[job %s] command line bindings is %s", j.name, json.dumps(builder.bindings, indent=4))
dockerReq, _ = self.get_requirement("DockerRequirement")
if dockerReq and kwargs.get("use_container"):
out_prefix = kwargs.get("tmp_outdir_prefix")
j.outdir = kwargs.get("outdir") or tempfile.mkdtemp(prefix=out_prefix)
tmpdir_prefix = kwargs.get('tmpdir_prefix')
j.tmpdir = kwargs.get("tmpdir") or tempfile.mkdtemp(prefix=tmpdir_prefix)
else:
j.outdir = builder.outdir
j.tmpdir = builder.tmpdir
createFiles, _ = self.get_requirement("CreateFileRequirement")
j.generatefiles = {}
if createFiles:
for t in createFiles["fileDef"]:
j.generatefiles[builder.do_eval(t["filename"])] = copy.deepcopy(builder.do_eval(t["fileContent"]))
j.environment = {}
evr, _ = self.get_requirement("EnvVarRequirement")
if evr:
for t in evr["envDef"]:
j.environment[t["envName"]] = builder.do_eval(t["envValue"])
shellcmd, _ = self.get_requirement("ShellCommandRequirement")
if shellcmd:
cmd = []
for b in builder.bindings:
arg = builder.generate_arg(b)
if b.get("shellQuote", True):
arg = [shellescape.quote(a) for a in aslist(arg)]
cmd.extend(aslist(arg))
j.command_line = ["/bin/sh", "-c", " ".join(cmd)]
else:
j.command_line = flatten(map(builder.generate_arg, builder.bindings))
j.pathmapper = builder.pathmapper
j.collect_outputs = functools.partial(self.collect_output_ports, self.tool["outputs"], builder)
j.output_callback = output_callback
yield j
def __init__(self, toolpath_object, pos, **kwargs):
if "id" in toolpath_object:
self.id = toolpath_object["id"]
else:
self.id = "#step" + str(pos)
try:
makeTool = kwargs.get("makeTool")
runobj = None
if isinstance(toolpath_object["run"], basestring):
runobj, _ = schema_salad.schema.load_and_validate(kwargs["loader"],
kwargs["avsc_names"],
toolpath_object["run"],
True)
else:
runobj = toolpath_object["run"]
self.embedded_tool = makeTool(runobj, **kwargs)
except validate.ValidationException as v:
raise WorkflowException("Tool definition %s failed validation:\n%s" % (toolpath_object["run"], validate.indent(str(v))))
for field in ("inputs", "outputs"):
for i in toolpath_object[field]:
inputid = i["id"]
p = shortname(inputid)
found = False
for a in self.embedded_tool.tool[field]:
frag = shortname(a["id"])
if frag == p:
i.update(a)
found = True
if not found:
i["type"] = "Any"
#raise WorkflowException("Parameter '%s' of %s in workflow step %s does not correspond to parameter in %s" % (p, field, self.id, self.embedded_tool.tool.get("id")))
i["id"] = inputid
super(WorkflowStep, self).__init__(toolpath_object, **kwargs)
if self.embedded_tool.tool["class"] == "Workflow":
(feature, _) = self.get_requirement("SubworkflowFeatureRequirement")
if not feature:
raise WorkflowException("Workflow contains embedded workflow but SubworkflowFeatureRequirement not in requirements")
if "scatter" in self.tool:
(feature, _) = self.get_requirement("ScatterFeatureRequirement")
if not feature:
raise WorkflowException("Workflow contains scatter but ScatterFeatureRequirement not in requirements")
inputparms = copy.deepcopy(self.tool["inputs"])
outputparms = copy.deepcopy(self.tool["outputs"])
scatter = aslist(self.tool["scatter"])
method = self.tool.get("scatterMethod")
if method is None and len(scatter) != 1:
raise WorkflowException("Must specify scatterMethod when scattering over multiple inputs")
inp_map = {i["id"]: i for i in inputparms}
for s in scatter:
if s not in inp_map:
raise WorkflowException("Invalid Scatter parameter '%s'" % s)
inp_map[s]["type"] = {"type": "array", "items": inp_map[s]["type"]}
if self.tool.get("scatterMethod") == "nested_crossproduct":
nesting = len(scatter)
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
self.tool["inputs"] = inputparms
self.tool["outputs"] = outputparms
def try_make_job(self, step, basedir, **kwargs):
inputobj = {}
if "scatter" in step.tool:
if not self.check_feature("ScatterFeature", kwargs):
raise WorkflowException(
"Must include ScatterFeature in requirements.")
inputparms = copy.deepcopy(step.tool["inputs"])
outputparms = copy.deepcopy(step.tool["outputs"])
scatter = aslist(step.tool["scatter"])
inp_map = {i["id"]: i for i in inputparms}
for s in aslist(step.tool["scatter"]):
if s not in inp_map:
raise WorkflowException("Invalid Scatter parameter '%s'" %
s)
inp_map[s]["type"] = {
"type": "array",
"items": inp_map[s]["type"]
}
if step.tool.get("scatterMethod") == "nested_crossproduct":
nesting = len(aslist(step.tool["scatter"]))
else:
nesting = 1
for r in xrange(0, nesting):
for i in outputparms:
i["type"] = {"type": "array", "items": i["type"]}
else:
inputparms = step.tool["inputs"]
outputparms = step.tool["outputs"]
for inp in inputparms:
_logger.debug(inp)
iid = idk(inp["id"])
if "connect" in inp:
connections = inp["connect"]
is_array = isinstance(inp["type"],
dict) and inp["type"]["type"] == "array"
for connection in aslist(connections):
src = idk(connection["source"])
if src in self.state and self.state[src] is not None:
if self.state[src].parameter["type"] == inp["type"]:
# source and input types are the same
if is_array and iid in inputobj:
# there's already a value in the input object, so extend the existing array
inputobj[iid].extend(self.state[src].value)
else:
# simply assign the value from state to input
inputobj[iid] = copy.deepcopy(
self.state[src].value)
elif is_array and self.state[src].parameter[
"type"] == inp["type"]["items"]:
# source type is the item type on the input array
# promote single item to array entry
if iid in inputobj:
inputobj[iid].append(self.state[src].value)
else:
inputobj[iid] = [self.state[src].value]
else:
raise WorkflowException(
"Type mismatch between '%s' (%s) and '%s' (%s)"
% (src, self.state[src].parameter["type"],
idk(inp["id"]), inp["type"]))
elif src not in self.state:
raise WorkflowException(
"Connect source '%s' on parameter '%s' does not exist"
% (src, inp["id"]))
else:
return
elif "default" in inp:
inputobj[iid] = inp["default"]
else:
raise WorkflowException("Value for %s not specified" %
(inp["id"]))
_logger.info("Creating job with input: %s", inputobj)
callback = functools.partial(self.receive_output, step, outputparms)
if step.tool.get("scatter"):
method = step.tool.get("scatterMethod")
if method is None and len(aslist(step.tool["scatter"])) != 1:
raise WorkflowException(
"Must specify scatterMethod when scattering over multiple inputs"
)
if method == "dotproduct" or method is None:
jobs = dotproduct_scatter(step, inputobj, basedir,
aslist(step.tool["scatter"]),
callback, **kwargs)
elif method == "nested_crossproduct":
jobs = nested_crossproduct_scatter(
step, inputobj, basedir, aslist(step.tool["scatter"]),
callback, **kwargs)
elif method == "flat_crossproduct":
jobs = flat_crossproduct_scatter(step, inputobj, basedir,
aslist(step.tool["scatter"]),
callback, 0, **kwargs)
else:
jobs = step.job(inputobj, basedir, callback, **kwargs)
for j in jobs:
yield j
def job(self, joborder, input_basedir, output_callback, **kwargs):
builder = self._init_job(joborder, input_basedir, **kwargs)
if self.tool["baseCommand"]:
for n, b in enumerate(aslist(self.tool["baseCommand"])):
builder.bindings.append({
"position": [-1000000, n],
"valueFrom": b
})
if self.tool.get("arguments"):
for i, a in enumerate(self.tool["arguments"]):
if isinstance(a, dict):
a = copy.copy(a)
if a.get("position"):
a["position"] = [a["position"], i]
else:
a["position"] = [0, i]
a["do_eval"] = a["valueFrom"]
a["valueFrom"] = None
builder.bindings.append(a)
else:
builder.bindings.append({
"position": [0, i],
"valueFrom": a
})
builder.bindings.sort(key=lambda a: a["position"])
reffiles = set((f["path"] for f in builder.files))
j = self.makeJobRunner()
j.joborder = builder.job
j.stdin = None
j.stdout = None
j.successCodes = self.tool.get("successCodes")
j.temporaryFailCodes = self.tool.get("temporaryFailCodes")
j.permanentFailCodes = self.tool.get("permanentFailCodes")
j.requirements = self.requirements
j.hints = self.hints
_logger.debug(
"[job %s] initializing from %s%s", id(j), self.tool.get("id", ""),
" as part of %s" %
kwargs["part_of"] if "part_of" in kwargs else "")
_logger.debug("[job %s] %s", id(j), json.dumps(joborder, indent=4))
builder.pathmapper = None
if self.tool.get("stdin"):
j.stdin = builder.do_eval(self.tool["stdin"])
if isinstance(j.stdin, dict) and "ref" in j.stdin:
j.stdin = builder.job[j.stdin["ref"][1:]]["path"]
reffiles.add(j.stdin)
if self.tool.get("stdout"):
j.stdout = builder.do_eval(self.tool["stdout"])
if os.path.isabs(j.stdout) or ".." in j.stdout:
raise validate.ValidationException(
"stdout must be a relative path")
builder.pathmapper = self.makePathMapper(reffiles, input_basedir,
**kwargs)
builder.requirements = j.requirements
for f in builder.files:
f["path"] = builder.pathmapper.mapper(f["path"])[1]
_logger.debug("[job %s] command line bindings is %s", id(j),
json.dumps(builder.bindings, indent=4))
_logger.debug(
"[job %s] path mappings is %s", id(j),
json.dumps(
{
p: builder.pathmapper.mapper(p)
for p in builder.pathmapper.files()
},
indent=4))
dockerReq, _ = self.get_requirement("DockerRequirement")
if dockerReq and kwargs.get("use_container"):
out_prefix = kwargs.get("tmp_outdir_prefix")
j.outdir = kwargs.get("outdir") or tempfile.mkdtemp(
prefix=out_prefix)
tmpdir_prefix = kwargs.get('tmpdir_prefix')
j.tmpdir = kwargs.get("tmpdir") or tempfile.mkdtemp(
prefix=tmpdir_prefix)
else:
j.outdir = builder.outdir
j.tmpdir = builder.tmpdir
createFiles, _ = self.get_requirement("CreateFileRequirement")
j.generatefiles = {}
if createFiles:
for t in createFiles["fileDef"]:
j.generatefiles[builder.do_eval(
t["filename"])] = copy.deepcopy(
builder.do_eval(t["fileContent"]))
j.environment = {}
evr, _ = self.get_requirement("EnvVarRequirement")
if evr:
for t in evr["envDef"]:
j.environment[t["envName"]] = builder.do_eval(t["envValue"])
j.command_line = flatten(map(builder.generate_arg, builder.bindings))
j.pathmapper = builder.pathmapper
j.collect_outputs = functools.partial(self.collect_output_ports,
self.tool["outputs"], builder)
j.output_callback = output_callback
yield j