11 #import arvados.command.ws as ws
15 logger = logging.getLogger('arvados.arv-run')
17 arvrun_parser = argparse.ArgumentParser()
18 arvrun_parser.add_argument('--dry-run', action="store_true", help="Print out the pipeline that would be submitted and exit")
19 arvrun_parser.add_argument('--local', action="store_true", help="Run locally using arv-crunch-job")
20 arvrun_parser.add_argument('--docker-image', type=str, default="arvados/jobs", help="Docker image to use, default arvados/jobs")
21 arvrun_parser.add_argument('--git-dir', type=str, default="", help="Git directory to use to find run-command when using --local")
22 arvrun_parser.add_argument('args', nargs=argparse.REMAINDER)
24 class ArvFile(object):
25 def __init__(self, prefix, fn):
29 class UploadFile(ArvFile):
32 def is_in_collection(root, branch):
35 fn = os.path.join(root, ".arvados#collection")
36 if os.path.exists(fn):
37 with file(fn, 'r') as f:
39 return (c["portable_data_hash"], branch)
41 sp = os.path.split(root)
42 return is_in_collection(sp[0], os.path.join(sp[1], branch))
44 def statfile(prefix, fn):
45 absfn = os.path.abspath(fn)
46 if os.path.exists(absfn):
48 if stat.S_ISREG(st.st_mode):
49 sp = os.path.split(absfn)
50 (pdh, branch) = is_in_collection(sp[0], sp[1])
52 return ArvFile(prefix, "$(file %s/%s)" % (pdh, branch))
54 # trim leading '/' for path prefix test later
55 return UploadFile(prefix, absfn[1:])
58 def main(arguments=None):
59 args = arvrun_parser.parse_args(arguments)
70 reading_into = len(slots)
73 slots[reading_into].append(c)
75 if slots[0] and len(slots[0]) > 1:
76 logger.error("Can only specify a single stdout file (run-command substitutions are permitted)")
79 patterns = [re.compile("(--[^=]+=)(.*)"),
80 re.compile("(-[^=]+=)(.*)"),
81 re.compile("(-.)(.+)")]
83 for command in slots[1:]:
84 for i in xrange(0, len(command)):
87 # parameter starts with '-' so it might be a command line
88 # parameter with a file name, do some pattern matching
93 command[i] = statfile(m.group(1), m.group(2))
96 # parameter might be a file, so test it
97 command[i] = statfile('', a)
101 files = [c for command in slots[1:] for c in command if isinstance(c, UploadFile)]
103 # Find the smallest path prefix that includes all the files that need to be uploaded.
104 # This starts at the root and iteratively removes common parent directory prefixes
105 # until all file pathes no longer have a common parent.
112 # no parent directories left
115 # path step takes next directory
116 pathstep = sp[0] + "/"
118 # check if pathstep is common prefix for all files
119 if not c.fn.startswith(pathstep):
123 # pathstep is common parent directory for all files, so remove the prefix
125 pathprefix += pathstep
127 c.fn = c.fn[len(pathstep):]
132 print("cd %s" % pathprefix)
133 print("arv-put \"%s\"" % '" "'.join([c.fn for c in files]))
136 pdh = put.main(["--portable-data-hash"]+[c.fn for c in files])
139 c.fn = "$(file %s/%s)" % (pdh, c.fn)
141 for i in xrange(1, len(slots)):
142 slots[i] = [("%s%s" % (c.prefix, c.fn)) if isinstance(c, ArvFile) else c for c in slots[i]]
145 "script": "run-command",
146 "script_version": "3609-arv-run",
147 "repository": "arvados",
148 "script_parameters": {
150 "runtime_constraints": {
151 "docker_image": args.docker_image
156 group_parser = argparse.ArgumentParser()
157 group_parser.add_argument('--group', type=str)
158 group_parser.add_argument('args', nargs=argparse.REMAINDER)
160 for s in xrange(2, len(slots)):
161 for i in xrange(0, len(slots[s])):
162 if slots[s][i] == '--':
164 groupargs = group_parser.parse_args(slots[2][i+1:])
165 component["script_parameters"][inp] = groupargs.args
167 inpgroups = inp+"_groups"
168 component["script_parameters"][inpgroups] = {"group":inp, "regex":groupargs.group}
169 slots[s] = slots[s][0:i] + [{"foreach": inpgroups, "command": "$(%s)" % inpgroups}]
170 task_foreach.append(inpgroups)
172 slots[s] = slots[s][0:i] + ["$(%s)" % inp]
173 task_foreach.append(inp)
175 if slots[s][i] == '\--':
179 component["script_parameters"]["task.stdout"] = slots[0][0]
181 task_foreach.append("stdin")
182 component["script_parameters"]["stdin"] = slots[1]
183 component["script_parameters"]["task.stdin"] = "$(stdin)"\
186 component["script_parameters"]["task.foreach"] = task_foreach
188 component["script_parameters"]["command"] = slots[2:]
195 "state":"RunningOnServer"
199 print(json.dumps(pipeline, indent=4))
201 subprocess.call(["arv-crunch-job", "--job", json.dumps(component), "--git-dir", args.git_dir])
203 api = arvados.api('v1')
204 pi = api.pipeline_instances().create(body=pipeline).execute()
205 print "Running pipeline %s" % pi["uuid"]
206 #ws.main(["--pipeline", pi["uuid"]])
208 if __name__ == '__main__':