8488: Determine actual output dir from reading logs. Need to set

[arvados.git] / sdk / cwl / arvados_cwl / __init__.py

diff --git a/sdk/cwl/arvados_cwl/__init__.py b/sdk/cwl/arvados_cwl/__init__.py
index 86619d5bb4e4278f8f7c6c553882d9d2d155dda0..105d754aa424901201189741cf4d354f35c92c01 100644 (file)
--- a/sdk/cwl/arvados_cwl/__init__.py
+++ b/sdk/cwl/arvados_cwl/__init__.py
@@ -25,10 +25,15 @@ from arvados.api import OrderedJsonModel
 logger = logging.getLogger('arvados.cwl-runner')
 logger.setLevel(logging.INFO)
 
-crunchrunner_pdh = "721abe848fd8e6e6d1c99b920e6b7a2c+140"
+crunchrunner_pdh = "83db29f08544e1c319572a6bd971088a+140"
 crunchrunner_download = "https://cloud.curoverse.com/collections/download/qr1hi-4zz18-n3m1yxd0vx78jic/1i1u2qtq66k1atziv4ocfgsg5nu5tj11n4r6e0bhvjg03rix4m/crunchrunner"
 certs_download = "https://cloud.curoverse.com/collections/download/qr1hi-4zz18-n3m1yxd0vx78jic/1i1u2qtq66k1atziv4ocfgsg5nu5tj11n4r6e0bhvjg03rix4m/ca-certificates.crt"
 
+tmpdirre = re.compile(r"\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.tmpdir\)=(.*)")
+outdirre = re.compile(r"\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.outdir\)=(.*)")
+keepre = re.compile(r"\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.keep\)=(.*)")
+
+
 def arv_docker_get_image(api_client, dockerRequirement, pull_image):
     if "dockerImageId" not in dockerRequirement and "dockerPull" in dockerRequirement:
         dockerRequirement["dockerImageId"] = dockerRequirement["dockerPull"]
@@ -194,8 +199,26 @@ class ArvadosJob(object):
             try:
                 outputs = {}
                 if record["output"]:
-                    self.builder.outdir = "keep:" + record["output"]
-                    outputs = self.collect_outputs(self.builder.outdir)
+                    logc = arvados.collection.Collection(record["log"])
+                    log = logc.open(logc.keys()[0])
+                    tmpdir = None
+                    outdir = None
+                    keepdir = None
+                    for l in log.readlines():
+                        g = tmpdirre.match(l)
+                        if g:
+                            tmpdir = g.group(1)
+                        g = outdirre.match(l)
+                        if g:
+                            outdir = g.group(1)
+                        g = keepre.match(l)
+                        if g:
+                            keepdir = g.group(1)
+                        if tmpdir and outdir and keepdir:
+                            break
+
+                    self.builder.outdir = outdir
+                    outputs = self.collect_outputs("keep:" + record["output"])
             except Exception as e:
                 logger.exception("Got exception while collecting job outputs:")
                 processStatus = "permanentFail"