#!/usr/bin/env python
+# collection-merge
+#
+# Merge two or more collections together. Can also be used to extract specific
+# files from a collection to produce a new collection.
+#
+# input:
+# An array of collections or collection/file paths in script_parameter["input"]
+#
+# output:
+# A manifest with the collections merged. Duplicate file names will
+# have their contents concatenated in the order that they appear in the input
+# array.
+
import arvados
import md5
import subst
if fn in s.files():
merged += s.files()[fn].as_manifest()
-crm = arvados.CollectionReader(merged)
-
-combined = crm.manifest_text(strip=True)
-
-m = hashlib.new('md5')
-m.update(combined)
-
-uuid = "{}+{}".format(m.hexdigest(), len(combined))
-
-collection = arvados.api().collections().create(
- body={
- 'uuid': uuid,
- 'manifest_text': crm.manifest_text(),
- }).execute()
-
-for s in src:
- l = arvados.api().links().create(body={
- "link": {
- "tail_uuid": s,
- "head_uuid": uuid,
- "link_class": "provenance",
- "name": "provided"
- }}).execute()
-
-arvados.current_task().set_output(uuid)
+arvados.current_task().set_output(merged)
--- /dev/null
+#!/usr/bin/env python
+
+#
+# decompress-all.py
+#
+# Decompress all compressed files in the collection using the "dtrx" tool and
+# produce a new collection with the contents. Uncompressed files
+# are passed through.
+#
+# input:
+# A collection at script_parameters["input"]
+#
+# output:
+# A manifest of the uncompressed contents of the input collection.
+
+import arvados
+import re
+import subprocess
+import os
+
+arvados.job_setup.one_task_per_input_file(if_sequence=0, and_end_task=True,
+ input_as_path=True)
+
+task = arvados.current_task()
+
+input_file = task['parameters']['input']
+
+result = re.match(r"(^[a-f0-9]{32}\+\d+)(\+\S+)*(/.*)(/[^/]+)$", input_file)
+
+outdir = os.path.join(task.tmpdir, "output")
+os.makedirs(outdir)
+os.chdir(outdir)
+
+if result != None:
+ cr = arvados.CollectionReader(result.group(1))
+ streamname = result.group(3)[1:]
+ filename = result.group(4)[1:]
+
+ subprocess.call(["mkdir", "-p", streamname])
+ os.chdir(streamname)
+ streamreader = filter(lambda s: s.name() == streamname, cr.all_streams())[0]
+ filereader = streamreader.files()[filename]
+ rc = subprocess.call(["dtrx", "-r", "-n", "-q", arvados.get_task_param_mount('input')])
+ if rc == 0:
+ out = arvados.CollectionWriter()
+ out.write_directory_tree(outdir, max_manifest_depth=0)
+ task.set_output(out.finish())
+ else:
+ task.set_output(streamname + filereader.as_manifest()[1:])
+else:
+ sys.exit(1)
import time
import arvados.commands.put as put
import signal
+import stat
+import copy
+import traceback
+import pprint
+import multiprocessing
os.umask(0077)
os.chdir("output")
+outdir = os.getcwd()
+
+taskp = None
+jobp = arvados.current_job()['script_parameters']
if len(arvados.current_task()['parameters']) > 0:
p = arvados.current_task()['parameters']
-else:
- p = arvados.current_job()['script_parameters']
links = []
def sub_link(v):
- r = os.path.basename(v)
- os.symlink(os.path.join(os.environ['TASK_KEEPMOUNT'], v) , r)
+ r = os.path.join(outdir, os.path.basename(v))
+ os.symlink(v, r)
links.append(r)
return r
def sub_tmpdir(v):
return os.path.join(arvados.current_task().tmpdir, 'tmpdir')
+def sub_outdir(v):
+ return outdir
+
def sub_cores(v):
- return os.environ['CRUNCH_NODE_SLOTS']
+ return str(multiprocessing.cpu_count())
+
+def sub_jobid(v):
+ return os.environ['JOB_UUID']
+
+def sub_taskid(v):
+ return os.environ['TASK_UUID']
+
+def sub_jobsrc(v):
+ return os.environ['CRUNCH_SRC']
subst.default_subs["link "] = sub_link
-subst.default_subs["tmpdir"] = sub_tmpdir
+subst.default_subs["task.tmpdir"] = sub_tmpdir
+subst.default_subs["task.outdir"] = sub_outdir
+subst.default_subs["job.srcdir"] = sub_jobsrc
subst.default_subs["node.cores"] = sub_cores
+subst.default_subs["job.uuid"] = sub_jobid
+subst.default_subs["task.uuid"] = sub_taskid
rcode = 1
sp.send_signal(signum)
self.sig = signum
+def expand_item(p, c):
+ if isinstance(c, dict):
+ if "foreach" in c and "command" in c:
+ var = c["foreach"]
+ items = get_items(p, p[var])
+ r = []
+ for i in items:
+ params = copy.copy(p)
+ params[var] = i
+ r.extend(expand_list(params, c["command"]))
+ return r
+ elif isinstance(c, list):
+ return expand_list(p, c)
+ elif isinstance(c, str) or isinstance(c, unicode):
+ return [subst.do_substitution(p, c)]
+
+ return []
+
+def expand_list(p, l):
+ return [exp for arg in l for exp in expand_item(p, arg)]
+
+def get_items(p, value):
+ if isinstance(value, list):
+ return expand_list(p, value)
+
+ fn = subst.do_substitution(p, value)
+ mode = os.stat(fn).st_mode
+ prefix = fn[len(os.environ['TASK_KEEPMOUNT'])+1:]
+ if mode != None:
+ if stat.S_ISDIR(mode):
+ items = ["$(dir %s/%s/)" % (prefix, l) for l in os.listdir(fn)]
+ elif stat.S_ISREG(mode):
+ with open(fn) as f:
+ items = [line for line in f]
+ return items
+ else:
+ return None
+
+stdoutname = None
+stdoutfile = None
+
try:
- cmd = []
- for c in p["command"]:
- cmd.append(subst.do_substitution(p, c))
-
- stdoutname = None
- stdoutfile = None
- if "stdout" in p:
- stdoutname = subst.do_substitution(p, p["stdout"])
+ if "task.foreach" in jobp:
+ if arvados.current_task()['sequence'] == 0:
+ var = jobp["task.foreach"]
+ items = get_items(jobp, jobp[var])
+ print("run-command: parallelizing on %s with items %s" % (var, items))
+ if items != None:
+ for i in items:
+ params = copy.copy(jobp)
+ params[var] = i
+ arvados.api().job_tasks().create(body={
+ 'job_uuid': arvados.current_job()['uuid'],
+ 'created_by_job_task_uuid': arvados.current_task()['uuid'],
+ 'sequence': 1,
+ 'parameters': params
+ }
+ ).execute()
+ arvados.current_task().set_output(None)
+ sys.exit(0)
+ else:
+ sys.exit(1)
+ else:
+ p = jobp
+
+ cmd = expand_list(p, p["command"])
+
+ if "save.stdout" in p:
+ stdoutname = subst.do_substitution(p, p["save.stdout"])
stdoutfile = open(stdoutname, "wb")
print("run-command: {}{}".format(' '.join(cmd), (" > " + stdoutname) if stdoutname != None else ""))
+except Exception as e:
+ print("run-command: caught exception:")
+ traceback.print_exc(file=sys.stdout)
+ print("run-command: task parameters was:")
+ pprint.pprint(p)
+ sys.exit(1)
+
+try:
sp = subprocess.Popen(cmd, shell=False, stdout=stdoutfile)
sig = SigHandler()
print("run-command: completed with exit code %i (%s)" % (rcode, "success" if rcode == 0 else "failed"))
except Exception as e:
- print("run-command: caught exception: {}".format(e))
+ print("run-command: caught exception:")
+ traceback.print_exc(file=sys.stdout)
# restore default signal handlers.
signal.signal(signal.SIGINT, signal.SIG_DFL)
for l in links:
os.unlink(l)
-print("run-command: the follow output files will be saved to keep:")
+print("run-command: the following output files will be saved to keep:")
subprocess.call(["find", ".", "-type", "f", "-printf", "run-command: %12.12s %h/%f\\n"])
print("run-command: terminating on signal 2")
sys.exit(2)
except Exception as e:
- print("run-command: caught exception: {}".format(e))
+ print("run-command: caught exception:")
+ traceback.print_exc(file=sys.stdout)
time.sleep(5)
sys.exit(rcode)
--- /dev/null
+#!/usr/bin/python
+
+import arvados
+import re
+import hashlib
+import string
+
+api = arvados.api('v1')
+
+piece = 0
+manifest_text = ""
+
+# Look for paired reads
+
+inp = arvados.CollectionReader(arvados.getjobparam('reads'))
+
+prog = re.compile(r'(.*?)_1.fastq(.gz)?$')
+
+manifest_list = []
+
+chunking = False #arvados.getjobparam('chunking')
+
+def nextline(reader, start):
+ n = -1
+ while True:
+ r = reader.readfrom(start, 128)
+ if r == '':
+ break
+ n = string.find(r, "\n")
+ if n > -1:
+ break
+ else:
+ start += 128
+ return n
+
+def splitfastq(p):
+ for i in xrange(0, len(p)):
+ p[i]["start"] = 0
+ p[i]["end"] = 0
+
+ count = 0
+ recordsize = [0, 0]
+
+ global piece
+ finish = False
+ while not finish:
+ for i in xrange(0, len(p)):
+ recordsize[i] = 0
+
+ # read next 4 lines
+ for i in xrange(0, len(p)):
+ for ln in xrange(0, 4):
+ r = nextline(p[i]["reader"], p[i]["end"]+recordsize[i])
+ if r == -1:
+ finish = True
+ break
+ recordsize[i] += (r+1)
+
+ splitnow = finish
+ for i in xrange(0, len(p)):
+ if ((p[i]["end"] - p[i]["start"]) + recordsize[i]) >= (64*1024*1024):
+ splitnow = True
+
+ if splitnow:
+ for i in xrange(0, len(p)):
+ global manifest_list
+ print "Finish piece ./_%s/%s (%s %s)" % (piece, p[i]["reader"].name(), p[i]["start"], p[i]["end"])
+ manifest = []
+ manifest.extend(["./_" + str(piece)])
+ manifest.extend([d[arvados.LOCATOR] for d in p[i]["reader"]._stream._data_locators])
+
+ print p[i]
+ print arvados.locators_and_ranges(p[i]["reader"].segments, p[i]["start"], p[i]["end"] - p[i]["start"])
+
+ manifest.extend(["{}:{}:{}".format(seg[arvados.LOCATOR]+seg[arvados.OFFSET], seg[arvados.SEGMENTSIZE], p[i]["reader"].name().replace(' ', '\\040')) for seg in arvados.locators_and_ranges(p[i]["reader"].segments, p[i]["start"], p[i]["end"] - p[i]["start"])])
+ manifest_list.append(manifest)
+ print "Finish piece %s" % (" ".join(manifest))
+ p[i]["start"] = p[i]["end"]
+ piece += 1
+ else:
+ for i in xrange(0, len(p)):
+ p[i]["end"] += recordsize[i]
+ count += 1
+ if count % 10000 == 0:
+ print "Record %s at %s" % (count, p[i]["end"])
+
+for s in inp.all_streams():
+ if s.name() == ".":
+ for f in s.all_files():
+ result = prog.match(f.name())
+ if result != None:
+ p = [{}, {}]
+ p[0]["reader"] = s.files()[result.group(0)]
+ if result.group(2) != None:
+ p[1]["reader"] = s.files()[result.group(1) + "_2.fastq" + result.group(2)]
+ else:
+ p[1]["reader"] = s.files()[result.group(1) + "_2.fastq"]
+ if chunking:
+ splitfastq(p)
+ else:
+ m0 = p[0]["reader"].as_manifest()[1:]
+ m1 = p[1]["reader"].as_manifest()[1:]
+ manifest_list.append(["./_" + str(piece), m0[:-1]])
+ manifest_list.append(["./_" + str(piece), m1[:-1]])
+ piece += 1
+
+# No pairs found so just put each fastq file into a separate directory
+if len(manifest_list) == 0:
+ for s in inp.all_streams():
+ prog = re.compile("(.*?).fastq(.gz)?$")
+ if s.name() == ".":
+ for f in s.all_files():
+ result = prog.match(f.name())
+ if result != None:
+ p = [{}]
+ p[0]["reader"] = s.files()[result.group(0)]
+ if chunking:
+ splitfastq(p)
+ else:
+ m0 = p[0]["reader"].as_manifest()[1:]
+ manifest_list.append(["./_" + str(piece), m0])
+ piece += 1
+
+manifest_text = "\n".join(" ".join(m) for m in manifest_list)
+
+arvados.current_task().set_output(manifest_text)
return os.path.splitext(os.path.basename(v))[0]
def sub_glob(v):
- return glob.glob(v)[0]
+ l = glob.glob(v)
+ if len(l) == 0:
+ raise Exception("$(glob): No match on '%s'" % v)
+ else:
+ return l[0]
default_subs = {"file ": sub_file,
"dir ": sub_dir,
# Install dependencies and set up system.
# The FUSE packages help ensure that we can install the Python SDK (arv-mount).
RUN /usr/bin/apt-get install -q -y python-dev python-llfuse python-pip \
- libio-socket-ssl-perl libjson-perl liburi-perl libwww-perl \
+ libio-socket-ssl-perl libjson-perl liburi-perl libwww-perl dtrx \
fuse libattr1-dev libfuse-dev && \
/usr/sbin/adduser --disabled-password \
--gecos 'Crunch execution user' crunch && \
# Install Arvados packages.
RUN (find /usr/src/arvados/sdk -name '*.gem' -print0 | \
- xargs -0rn 1 gem install) && \
+ xargs -0rn 1 /usr/local/rvm/bin/rvm-exec default gem install) && \
cd /usr/src/arvados/services/fuse && \
python setup.py install && \
cd /usr/src/arvados/sdk/python && \
end
def self.create(attributes)
result = $client.execute(:api_method => $arvados.pipeline_instances.create,
- :body => {
+ :body_object => {
:pipeline_instance => attributes.to_json
},
:authenticated => false,
:parameters => {
:uuid => @pi[:uuid]
},
- :body => {
+ :body_object => {
:pipeline_instance => @attributes_to_update.to_json
},
:authenticated => false,
body = {job: no_nil_values(job)}.merge(no_nil_values(create_params))
result = $client.execute(:api_method => $arvados.jobs.create,
- :body => body,
+ :body_object => body,
:authenticated => false,
:headers => {
authorization: 'OAuth2 '+ENV['ARVADOS_API_TOKEN']
msg += "Job submission was: #{body.to_json}"
$client.execute(:api_method => $arvados.logs.create,
- :body => {
+ :body_object => {
:log => {
:object_uuid => pipeline[:uuid],
:event_type => 'stderr',
projects / arvados.git / commitdiff