def MapOverRange(futurekey, startbyte, endbyte, weight, **kwargs):
logdebug("Enter MapOverRange: %s, %s, %s" % (startbyte, endbyte, weight))
linitialresult = initialresult if not initialresult is None else 0
loncombineresultsf = oncombineresultsf if oncombineresultsf else lambda a, b: a + b
try:
# open file at gcspath for read
with gcs.open(gcspath) as gcsfile:
page, ranges = hwalk(gcsfile, pagesize, 2, startbyte, endbyte)
if pagemapf:
lonallchildsuccessf = GenerateOnAllChildSuccess(futurekey, linitialresult, loncombineresultsf)
taskkwargs["futurename"] = "pagemap %s of %s,%s" % (len(page), startbyte, endbyte)
future(pagemapf, parentkey=futurekey, onallchildsuccessf=lonallchildsuccessf, weight = len(page), **taskkwargs)(page)
else:
setlocalprogress(futurekey, len(page))
if ranges:
newweight = (weight - len(page)) / len(ranges) if not weight is None else None
for arange in ranges:
taskkwargs["futurename"] = "shard %s" % (arange)
lonallchildsuccessf = GenerateOnAllChildSuccess(futurekey, linitialresult if pagemapf else len(page), loncombineresultsf)
future(MapOverRange, parentkey=futurekey, onallchildsuccessf=lonallchildsuccessf, weight = newweight, **taskkwargs)(arange[0], arange[1], weight = newweight)
if ranges or pagemapf:
raise FutureReadyForResult("still going")
else:
return len(page)
finally:
logdebug("Leave MapOverRange: %s, %s, %s" % (startbyte, endbyte, weight))
def ndbshardedpagemap(pagemapf=None, ndbquery=None, initialshards = 10, pagesize = 100, **taskkwargs):
@task(**taskkwargs)
def MapOverRange(keyrange, **kwargs):
logdebug("Enter MapOverRange: %s" % keyrange)
_fixkeyend(keyrange, kind)
filteredquery = keyrange.filter_ndb_query(ndbquery)
logdebug (filteredquery)
keys, _, more = filteredquery.fetch_page(pagesize, keys_only=True)
if pagemapf:
pagemapf(keys)
if more and keys:
newkeyrange = KeyRange(keys[-1], keyrange.key_end, keyrange.direction, False, keyrange.include_end)
krlist = newkeyrange.split_range()
logdebug("krlist: %s" % krlist)
for kr in krlist:
MapOverRange(kr)
logdebug("Leave MapOverRange: %s" % keyrange)
kind = ndbquery.kind
krlist = KeyRange.compute_split_points(kind, initialshards)
logdebug("first krlist: %s" % krlist)
for kr in krlist:
MapOverRange(kr)
def OnProgressF(futurekey):
futureobj = futurekey.get() if futurekey else None
if futureobj.parentkey:
taskkwargs = futureobj.get_taskkwargs()
logdebug("Enter OnProgressF: %s" % futureobj)
@task(**taskkwargs)
def UpdateParent(parentkey):
logdebug("***************************************************")
logdebug("Enter UpdateParent: %s" % parentkey)
logdebug("***************************************************")
parent = parentkey.get()
logdebug("1: %s" % parent)
if parent:
logdebug("2")
# if not parent.has_result():
progress = 0
for childfuture in get_children(parentkey):
logdebug("3: %s" % childfuture)
progress += childfuture.get_progress()
logdebug("4: %s" % progress)
parent.set_progress(progress)
UpdateParent(futureobj.parentkey)
def runtask(*args, **kwargs):
pickled = cloudpickle.dumps((f, args, kwargs, passthroughargs))
logdebug("task pickle length: %s" % len(pickled))
if get_dump():
logdebug("f:")
dumper(f)
logdebug("args:")
dumper(args)
logdebug("kwargs:")
dumper(kwargs)
logdebug("passthroughargs:")
dumper(passthroughargs)
try:
task = taskqueue.Task(payload=pickled, **taskkwargscopy)
return task.add(queue, transactional=transactional)
except taskqueue.TaskTooLargeError:
pickledf = cloudpickle.dumps(f)
pickleda = cloudpickle.dumps(args)
pickledk = cloudpickle.dumps(kwargs)
pickledp = cloudpickle.dumps(passthroughargs)
logexception(
"task too large, need to use datastore (%s, %s, %s, %s)" %
(len(pickledf), len(pickleda), len(pickledk), len(pickledp)))
if parent:
key = _TaskToRun(data=pickled, parent=parent).put()
else:
key = _TaskToRun(data=pickled).put()
rfspickled = cloudpickle.dumps((None, [key], {}, {
"_run_from_datastore": True
}))
task = taskqueue.Task(payload=rfspickled, **taskkwargscopy)
return task.add(queue, transactional=transactional)
def run_task(*args, **kwargs):
pickled = cloudpickle.dumps((f, args, kwargs, extra))
logdebug("task pickle length: %s" % len(pickled))
if get_dump():
logdebug("f:")
dumper(f)
logdebug("args:")
dumper(args)
logdebug("kwargs:")
dumper(kwargs)
logdebug("extra:")
dumper(extra)
try:
t = taskqueue.Task(payload=pickled, **task_kwargs)
return t.add(queue, transactional=transactional)
except taskqueue.TaskTooLargeError:
if parent:
key = _TaskToRun(data=pickled, parent=parent).put()
else:
key = _TaskToRun(data=pickled).put()
ds_pickled = cloudpickle.dumps((None, [key], {}, {
"_run_from_datastore": True
}))
t = taskqueue.Task(payload=ds_pickled, **task_kwargs)
return t.add(queue, transactional=transactional)
def InvokeMap(futurekey, key, **kwargs):
logdebug("Enter InvokeMap: %s" % key)
try:
obj = key.get()
if not obj:
raise RetryTaskException("couldn't get object for key %s" % key)
return mapf(futurekey, obj, **kwargs)
finally:
logdebug("Leave InvokeMap: %s" % key)
def toplevel(futurekey, *args, **kwargs):
logdebug("Enter futureparallel.toplevel: %s" % futurekey)
def OnAllChildSuccess():
logdebug("Enter OnAllChildSuccess: %s" % futurekey)
parentfuture = futurekey.get() if futurekey else None
if parentfuture and not parentfuture.has_result():
if not parentfuture.initialised or not parentfuture.readyforresult:
raise Exception("Parent not initialised, retry")
@ndb.transactional()
def get_children_trans():
return get_children(parentfuture.key)
children = get_children_trans()
logdebug("children: %s" % [child.key for child in children])
if children:
result = []
error = None
finished = True
for childfuture in children:
logdebug("childfuture: %s" % childfuture.key)
if childfuture.has_result():
try:
childresult = childfuture.get_result()
logdebug("childresult(%s): %s" %
(childfuture.status, childresult))
result += [childfuture.get_result()]
logdebug("intermediate result:%s" % result)
except Exception, ex:
logdebug("haserror:%s" % repr(ex))
error = ex
break
else:
logdebug("noresult")
finished = False
if error:
logwarning(
"Internal error, child has error in OnAllChildSuccess: %s"
% error)
parentfuture.set_failure(error)
elif finished:
logdebug("result: %s" % result)
parentfuture.set_success(result)
else:
logdebug(
"child not finished in OnAllChildSuccess, skipping"
)
else:
logwarning(
"Internal error, parent has no children in OnAllChildSuccess"
)
parentfuture.set_failure(Exception("no children found"))
def dofuturendbshardedmap(futurekey):
logdebug(taskkwargs)
linitialresult = initialresult if not initialresult is None else 0
loncombineresultsf = oncombineresultsf if oncombineresultsf else lambda a, b: a + b
def MapOverRange(futurekey, keyrange, weight, **kwargs):
logdebug("Enter MapOverRange: %s" % keyrange)
try:
_fixkeyend(keyrange, kind)
filteredquery = keyrange.filter_ndb_query(ndbquery)
logdebug (filteredquery)
keys, _, more = filteredquery.fetch_page(pagesize, keys_only=True)
lonallchildsuccessf = GenerateOnAllChildSuccess(futurekey, 0 if pagemapf else len(keys), lambda a, b: a + b)
if pagemapf:
futurename = "pagemap %s of %s" % (len(keys), keyrange)
lonallchildsuccessf = GenerateOnAllChildSuccess(futurekey, linitialresult, loncombineresultsf)
future(pagemapf, parentkey=futurekey, futurename=futurename, onallchildsuccessf=lonallchildsuccessf, weight = len(keys), **taskkwargs)(keys)
else:
setlocalprogress(futurekey, len(keys))
if more and keys:
lonallchildsuccessf = GenerateOnAllChildSuccess(futurekey, linitialresult if pagemapf else len(keys), loncombineresultsf)
newkeyrange = KeyRange(keys[-1], keyrange.key_end, keyrange.direction, False, keyrange.include_end)
krlist = newkeyrange.split_range()
logdebug("krlist: %s" % krlist)
newweight = (weight / len(krlist)) - len(keys) if weight else None
for kr in krlist:
futurename = "shard %s" % (kr)
future(MapOverRange, parentkey=futurekey, futurename=futurename, onallchildsuccessf = lonallchildsuccessf, weight = newweight, **taskkwargs)(kr, weight = newweight)
#
if pagemapf or (more and keys):
# if (more and keys):
raise FutureReadyForResult("still going")
else:
return len(keys)#(len(keys), 0, keyrange)
# return len(keys)
finally:
logdebug("Leave MapOverRange: %s" % keyrange)
for kr in krlist:
lonallchildsuccessf = GenerateOnAllChildSuccess(futurekey, linitialresult, loncombineresultsf)
futurename = "shard %s" % (kr)
newweight = weight / len(krlist) if weight else None
future(MapOverRange, parentkey=futurekey, futurename=futurename, onallchildsuccessf=lonallchildsuccessf, weight = newweight, **taskkwargs)(kr, weight = newweight)
raise FutureReadyForResult("still going")
def higherlevelcompose(lop, rop):
try:
retval = None
if lop and rop:
blobnames = [lop.get("blobname"), rop.get("blobname")]
blobs = getblobsbyname(gcsbucket, *blobnames)
if len(blobs) == 2:
ltotalcomponent_count = sum(
[blob.component_count for blob in blobs])
logdebug("ltotalcomponent_count: %s" %
ltotalcomponent_count)
if ltotalcomponent_count > 1020:
logdebug("doing copying")
newblobnames = [
"%s-copy" % blobname
for blobname in blobnames
]
for ix, blob in enumerate(blobs):
try:
copyblob(gcsbucket, blob,
newblobnames[ix])
except Exception:
logexception("deleteblobs(copy)")
try:
deleteblobs(gcsbucket, blobs)
except Exception:
logexception("deleteblobs(copy)")
blobnames = newblobnames
blobs = getblobsbyname(gcsbucket, *blobnames)
if len(blobs) == 2:
llocalfilename = gcstargetfilename if istop else GenerateStableId(
blobnames[0] + blobnames[1])
lfilename = "%s/%s-%s" % (gcstargetprefix,
"composed",
llocalfilename)
retval = composeblobs(gcsbucket, lfilename,
blobs)
retval["count"] = lop.get(
"count", 0) + rop.get("count", 0)
try:
deleteblobs(gcsbucket, blobs)
except Exception:
logexception("deleteblobs")
else:
raise Exception("Can't load blobs")
else:
retval = lop if lop else rop
return retval
except Exception, ex:
logexception("higherlevelcompose")
raise ex
def InvokeMap(key, **kwargs):
logdebug("Enter InvokeMap: %s" % key)
try:
obj = key.get()
if not obj:
if not skipmissing:
raise RetryTaskException("couldn't get object for key %s" % key)
# else just skip
else:
mapf(obj, **kwargs)
finally:
logdebug("Leave InvokeMap: %s" % key)
def toplevel(futurekey, *args, **kwargs):
logdebug("Enter futuresequence.toplevel: %s" % futurekey)
def childonsuccessforindex(index, results):
logdebug("Enter childonsuccessforindex: %s, %s, %s" %
(futurekey, index, json.dumps(results, indent=2)))
def childonsuccess(childfuturekey):
logdebug("Enter childonsuccess: %s, %s, %s" %
(futurekey, index, childfuturekey))
logdebug("results: %s" % json.dumps(results, indent=2))
try:
childfuture = GetFutureAndCheckReady(childfuturekey)
try:
result = childfuture.get_result()
except Exception, ex:
toplevelfuture = futurekey.get()
if toplevelfuture:
toplevelfuture.set_failure(ex)
else:
raise Exception(
"Can't load toplevel future for failure")
else:
logdebug("result: %s" % json.dumps(result, indent=2))
newresults = results + [result]
islast = (index == (len(flist) - 1))
if islast:
logdebug("islast")
toplevelfuture = futurekey.get()
if toplevelfuture:
logdebug("setting top level success")
toplevelfuture.set_success_and_readyforesult(
newresults)
else:
raise Exception(
"Can't load toplevel future for success")
else:
logdebug("not last")
taskkwargs["futurename"] = "%s [%s]" % (
futurenameprefix if futurenameprefix else "-",
index + 1)
future(flist[index + 1],
parentkey=futurekey,
onsuccessf=childonsuccessforindex(
index + 1, newresults),
weight=weight /
len(flist) if weight else None,
timeoutsec=timeoutsec,
maxretries=maxretries,
**taskkwargs)(newresults)
def task(f=None, **kw):
if not f:
return functools.partial(task, **kw)
task_kwargs = deepcopy(kw)
queue = task_kwargs.pop("queue", "default")
transactional = task_kwargs.pop("transactional", False)
parent = task_kwargs.pop("parent", None)
include_headers = task_kwargs.pop("includeheaders", False)
log_name = task_kwargs.pop(
"logname", "%s/%s" %
(getattr(f, '__module__', 'none'), getattr(f, '__name__', 'none')))
task_kwargs["headers"] = dict(_TASKQUEUE_HEADERS)
url = get_enqueue_url(log_name) # _DEFAULT_ENQUEUE_URL % logname
task_kwargs["url"] = url.lower()
logdebug(task_kwargs)
extra = {"includeheaders": include_headers}
@functools.wraps(f)
def run_task(*args, **kwargs):
pickled = cloudpickle.dumps((f, args, kwargs, extra))
logdebug("task pickle length: %s" % len(pickled))
if get_dump():
logdebug("f:")
dumper(f)
logdebug("args:")
dumper(args)
logdebug("kwargs:")
dumper(kwargs)
logdebug("extra:")
dumper(extra)
try:
t = taskqueue.Task(payload=pickled, **task_kwargs)
return t.add(queue, transactional=transactional)
except taskqueue.TaskTooLargeError:
if parent:
key = _TaskToRun(data=pickled, parent=parent).put()
else:
key = _TaskToRun(data=pickled).put()
ds_pickled = cloudpickle.dumps((None, [key], {}, {
"_run_from_datastore": True
}))
t = taskqueue.Task(payload=ds_pickled, **task_kwargs)
return t.add(queue, transactional=transactional)
return run_task
def getvalue(*args, **kwargs):
lcachekey = cachekey if cachekey else make_flash(f, *args, **kwargs)
retval = memcache.get(lcachekey) # @UndefinedVariable
if retval is None:
logdebug("MISS: %s" % lcachekey)
retval = f(*args, **kwargs)
memcache.add(key=lcachekey, value=retval,
time=expiresec) # @UndefinedVariable
else:
logdebug("HIT: %s" % lcachekey)
return retval
def MapOverRange(startpos, endpos, **kwargs):
logdebug("Enter MapOverRange: %s, %s" % (startpos, endpos))
# open file at gcspath for read
with gcs.open(gcspath) as gcsfile:
page, ranges = hwalk(gcsfile, pagesize, initialshards, startpos, endpos)
if ranges:
for arange in ranges:
MapOverRange(arange[0], arange[1])
if pagemapf:
pagemapf(page)
logdebug("Leave MapOverRange: %s, %s" % (startpos, endpos))
def childonsuccess(childfuturekey):
logdebug("Enter childonsuccess: %s, %s, %s" %
(futurekey, index, childfuturekey))
logdebug("results: %s" % json.dumps(results, indent=2))
try:
childfuture = GetFutureAndCheckReady(childfuturekey)
try:
result = childfuture.get_result()
except Exception, ex:
toplevelfuture = futurekey.get()
if toplevelfuture:
toplevelfuture.set_failure(ex)
else:
raise Exception(
"Can't load toplevel future for failure")
else:
def _set_local_progress_for_success(self):
progressObj = self._get_progressobject()
logdebug("progressObj = %s" % progressObj)
weight = self.get_weight(progressObj)
weight = weight or 1
logdebug("weight = %s" % weight)
localprogress = self.get_localprogress(progressObj)
logdebug("localprogress = %s" % localprogress)
if localprogress < weight and not self.GetChildren():
logdebug("No children, we can auto set localprogress from weight")
self.set_localprogress(weight)
def _futurewrapper(headers):
if maxretries:
lretryCount = 0
try:
lretryCount = int(
headers.get("X-Appengine-Taskretrycount",
0)) if headers else 0
except:
logexception(
"Failed trying to get retry count, using 0")
if lretryCount > maxretries:
raise PermanentTaskFailure(
"Too many retries of Future")
logdebug("inner, futurekey=%s" % futurekey)
futureobj2 = futurekey.get()
if futureobj2:
futureobj2.set_weight(weight) # if weight >= 1 else 1)
else:
raise Exception("Future not ready yet")
try:
logdebug("args, kwargs=%s, %s" % (args, kwargs))
result = f(futurekey, *args, **kwargs)
except FutureReadyForResult:
futureobj3 = futurekey.get()
if futureobj3:
futureobj3.set_readyforesult()
except FutureNotReadyForResult:
futureobj4 = futurekey.get()
if futureobj4:
futureobj4.set_initialised()
except PermanentTaskFailure, ptf:
try:
futureobj5 = futurekey.get()
if futureobj5:
futureobj5.set_failure(ptf)
finally:
raise ptf
def _launch_task(pickled, name, headers):
try:
# Add some task debug information.
# dheaders = []
# for key, value in headers.items():
# k = key.lower()
# if k.startswith("x-appengine-") and k not in _SKIP_HEADERS:
# dheaders.append("%s:%s" % (key, value))
# logdebug(", ".join(dheaders))
logdebug(", ".join(
["%s:%s" % (key, value) for key, value in headers.items()]))
if not isFromTaskQueue(headers):
raise PermanentTaskFailure(
'Detected an attempted XSRF attack: we are not executing from a task queue.'
)
logdebug('before run "%s"' % name)
_run(pickled, headers)
logdebug('after run "%s"' % name)
except PermanentTaskFailure:
logexception("Aborting task")
except:
logexception("failure")
raise
def intask(self, nameprefix, f, *args, **kwargs):
taskkwargs = self.get_taskkwargs()
name = ""
if nameprefix:
name = "%s-%s" % (nameprefix, self.key.id())
taskkwargs["name"] = name
elif taskkwargs.get("name"):
del taskkwargs["name"]
taskkwargs["transactional"] = False
@task(**taskkwargs)
def dof():
f(*args, **kwargs)
try:
# run the wrapper task, and if it fails due to a name clash just skip it (it was already kicked off by an earlier
# attempt to construct this future).
# logdebug("about to run task %s" % name)
dof()
except taskqueue.TombstonedTaskError:
logdebug("skip adding task %s (already been run)" % name)
except taskqueue.TaskAlreadyExistsError:
logdebug("skip adding task %s (already running)" % name)
def MapOverRange(keyrange, **kwargs):
logdebug("Enter MapOverRange: %s" % keyrange)
_fixkeyend(keyrange, kind)
filteredquery = keyrange.filter_ndb_query(ndbquery)
logdebug (filteredquery)
keys, _, more = filteredquery.fetch_page(pagesize, keys_only=True)
if pagemapf:
pagemapf(keys)
if more and keys:
newkeyrange = KeyRange(keys[-1], keyrange.key_end, keyrange.direction, False, keyrange.include_end)
krlist = newkeyrange.split_range()
logdebug("krlist: %s" % krlist)
for kr in krlist:
MapOverRange(kr)
logdebug("Leave MapOverRange: %s" % keyrange)
def set_localprogress(self, value):
obj = self._get_progressobject()
local = self.get_localprogress(obj)
calculated = self.get_calculatedprogress(obj)
if local != value:
# haschildren = self.GetChildren()
# logdebug("haschildren: %s" % haschildren)
obj.localprogress = value
logdebug("localprogress: %s" % value)
# if not haschildren:
changed = value > calculated
if changed:
logdebug("setting calculated progress")
obj.calculatedprogress = value
obj.put()
if changed:
logdebug("kicking off calculate parent progress")
self._calculate_parent_progress()
self._callOnProgress()
def set_localprogress(self, value):
progressobj = self._get_progressobject()
localprogress = self.get_localprogress(progressobj)
calculatedprogress = self.get_calculatedprogress(progressobj)
if localprogress != value:
# haschildren = self.GetChildren()
# logdebug("haschildren: %s" % haschildren)
progressobj.localprogress = value
logdebug("localprogress: %s" % value)
# if not haschildren:
lneedupd = value > calculatedprogress
if lneedupd:
logdebug("setting calculated progress")
progressobj.calculatedprogress = value
progressobj.put()
if lneedupd:
logdebug("kicking off calculate parent progress")
self._calculate_parent_progress()
self._callOnProgress()
def getvalue(*args, **kwargs):
key = cachekey if cachekey else make_flash(f, args, kwargs)
logdebug("Enter gcscacher.getvalue: %s" % key)
bucket = bucketname if bucketname else os.environ.get(
'BUCKET_NAME',
app_identity.get_default_gcs_bucket_name())
lpicklepath = "/%s/gcscache/%s.pickle" % (bucket, key)
logdebug("picklepath: %s" % lpicklepath)
lsaved = None
try:
#1: Get the meta info
with gcs.open(lpicklepath) as picklefile:
lsaved = pickle.load(picklefile)
except gcs.NotFoundError:
pass
lexpireat = lsaved.get("expireat") if lsaved else None
lcontent = None
lcacheIsValid = False
if lsaved and not (lexpireat and lexpireat < get_utcnow_unixtimestampusec()):
lcontent = lsaved.get("content")
lcacheIsValid = True
if not lcacheIsValid:
logdebug("GCS Cache miss")
lcontent = f(*args, **kwargs)
logdebug("write content back to gcs")
ltosave = {
"expireat": get_utcnow_unixtimestampusec() + (expiresec * 1000000) if expiresec else None,
"content": lcontent
}
with gcs.open(lpicklepath, "w") as picklefilewrite:
cloudpickle.dump(ltosave, picklefilewrite)
else:
logdebug("GCS Cache hit")
logdebug("Leave gcscacher.getvalue: %s" % key)
return lcontent
if numfiles > 32:
ranges = CalculateFileRanges(startindex, finishindex, 2)
logdebug("ranges:%s" % ranges)
for r in ranges:
futurename = "split %s" % (r, )
future(GCSCombineToTarget, futurename=futurename, onallchildsuccessf=onallchildsuccessf, parentkey=futurekey, weight = r[1]-r[0], **taskkwargs)(r[0], r[1], False)
raise FutureReadyForResult()
else:
lblobs = list(listbucket(gcsbucket, gcssourceprefix))[startindex:finishindex]
lfilename = "%s/%s" % (gcstargetprefix, gcstargetfilename if istop else "composed-%s-%s" % (startindex, finishindex))
# lfilename = "%s/%s-%s-%s" % (gcstargetprefix, "composed", startindex, finishindex)
retval = composeblobs(gcsbucket, lfilename, lblobs)
return retval
finally:
logdebug("Leave GCSCombineToTarget: %s, %s" % (startindex, finishindex))
futurename = "gcscombinetotarget %s" % (numgcsfiles)
return future(GCSCombineToTarget, futurename=futurename, onsuccessf = onsuccessf, onfailuref = onfailuref, onprogressf = onprogressf, parentkey=parentkey, weight = numgcsfiles, **taskkwargs)(0, numgcsfiles, True)
def hwalk(afile, pagesizeinlines, numranges, startbytes, endbytes):
## print(afile, pagesizeinlines, numranges, startbytes, endbytes)
page = []
ranges = []
if startbytes <= 0:
#1: we're at the start of the file, just start here
afile.seek(0, 0)
else:
def ProcessPage(keys):
for index, key in enumerate(keys):
logdebug("Key #%s: %s" % (index, key))
InvokeMap(key)
def _fixkeyend(keyrange, kind):
if keyrange.key_start and not keyrange.key_end:
endkey = KeyRange.guess_end_key(kind, keyrange.key_start)
if endkey and endkey > keyrange.key_start:
logdebug("Fixing end: %s" % endkey)
keyrange.key_end = endkey
def GCSCombineToTarget(futurekey, startindex, finishindex, istop, **kwargs):
logdebug("Enter GCSCombineToTarget: %s, %s" % (startindex, finishindex))
try:
def higherlevelcompose(lop, rop):
try:
retval = None
if lop and rop:
blobnames = [lop.get("blobname"), rop.get("blobname")]
blobs = getblobsbyname(gcsbucket, *blobnames)
if len(blobs) == 2:
ltotalcomponent_count = sum([blob.component_count for blob in blobs])
logdebug("ltotalcomponent_count: %s" % ltotalcomponent_count)
if ltotalcomponent_count > 1020:
logdebug("doing copying")
newblobnames = ["%s-copy" % blobname for blobname in blobnames]
for ix, blob in enumerate(blobs):
try:
copyblob(gcsbucket, blob, newblobnames[ix])
except Exception:
logexception("deleteblobs(copy)")
try:
deleteblobs(gcsbucket, blobs)
except Exception:
logexception("deleteblobs(copy)")
blobnames = newblobnames
blobs = getblobsbyname(gcsbucket, *blobnames)
if len(blobs) == 2:
llocalfilename = gcstargetfilename if istop else GenerateStableId(blobnames[0] + blobnames[1])
lfilename = "%s/%s-%s" % (gcstargetprefix, "composed", llocalfilename)
retval = composeblobs(gcsbucket, lfilename, blobs)
retval["count"] = lop.get("count", 0) + rop.get("count", 0)
try:
deleteblobs(gcsbucket, blobs)
except Exception:
logexception("deleteblobs")
else:
raise Exception("Can't load blobs")
else:
retval = lop if lop else rop
return retval
except Exception, ex:
logexception("higherlevelcompose")
raise ex
onallchildsuccessf = GenerateOnAllChildSuccess(futurekey, None, higherlevelcompose, failonerror=False)
numfiles = finishindex - startindex
if numfiles > 32:
ranges = CalculateFileRanges(startindex, finishindex, 2)
logdebug("ranges:%s" % ranges)
for r in ranges:
futurename = "split %s" % (r, )
future(GCSCombineToTarget, futurename=futurename, onallchildsuccessf=onallchildsuccessf, parentkey=futurekey, weight = r[1]-r[0], **taskkwargs)(r[0], r[1], False)
raise FutureReadyForResult()
else:
lblobs = list(listbucket(gcsbucket, gcssourceprefix))[startindex:finishindex]
lfilename = "%s/%s" % (gcstargetprefix, gcstargetfilename if istop else "composed-%s-%s" % (startindex, finishindex))
# lfilename = "%s/%s-%s-%s" % (gcstargetprefix, "composed", startindex, finishindex)
retval = composeblobs(gcsbucket, lfilename, lblobs)
return retval
def InvokeMap(futurekey, line, **kwargs):
logdebug("Enter InvokeMap: %s" % line)
try:
return mapf(line, **kwargs)
finally:
logdebug("Leave InvokeMap: %s" % line)
def InvokeMap(line, **kwargs):
logdebug("Enter InvokeMap: %s" % line)
try:
mapf(line, **kwargs)
finally:
logdebug("Leave InvokeMap: %s" % line)
def ProcessPage(lines):
for index, line in enumerate(lines):
logdebug("Line #%s: %s" % (index, line))
InvokeMap(line)
