def test_das_sinfo(self):
"Test das_sinfo function"
das = {'system':['dbs', 'dbs', 'phedex'], 'api':['one', 'two', 'thr']}
row = {'das':das, 'foo':[{'dbs':1}, {'a':1}, {'phedex':1}]}
result = das_sinfo(row)
expect = {'dbs':set(['one', 'two']), 'phedex':set(['three'])}
self.assertEqual(expect, expect)
def test_das_sinfo(self):
"Test das_sinfo function"
das = {
'system': ['dbs', 'dbs', 'phedex'],
'api': ['one', 'two', 'thr']
}
row = {'das': das, 'foo': [{'dbs': 1}, {'a': 1}, {'phedex': 1}]}
result = das_sinfo(row)
expect = {'dbs': set(['one', 'two']), 'phedex': set(['three'])}
self.assertEqual(expect, expect)
def get_from_cache(self, dasquery, idx=0, limit=0, collection='merge'):
"""
Look-up results from the merge cache and yield them for
further processing.
"""
das_timer('DASCore::get_from_cache', self.verbose)
msg = 'col=%s, query=%s, idx=%s, limit=%s'\
% (collection, dasquery, idx, limit)
self.logger.info(msg)
fields = dasquery.mongo_query.get('fields', None)
if dasquery.mapreduce:
res = self.rawcache.map_reduce(dasquery.mapreduce, dasquery)
elif dasquery.aggregators:
# extract das information from rawcache
rows = self.rawcache.get_from_cache(\
dasquery, collection=collection)
first = rows.next()
sinfo = das_sinfo(first)
# to perform aggregation we need:
# - loop over all aggregator functions
# - loop over all data-services
# - loop over all APIs within a data-services
# the code below does that, it applies aggregator
# to selected (based on key/srv/api) records
res = []
_id = 0
time0 = time.time()
expire = 300 # min expire
for func, key in dasquery.aggregators:
afunc = getattr(das_aggregator, 'das_%s' % func)
found = False
for srv, apis, in sinfo.items():
for api in apis:
rows = self.rawcache.get_from_cache(\
dasquery, collection=collection)
gen = api_rows(rows, api)
data = afunc(key, gen)
ctime = time.time() - time0
das = dasheader(srv, dasquery, expire, api=api,
ctime=ctime)
if isinstance(data, dict) and data['value'] != 'N/A':
aggr = {'_id':_id, 'function': func,
'key': key, 'result': data}
aggr.update(das)
res.append(aggr)
_id += 1
found = True
if not found: # when we got nothing add empty result record
empty = {'value':'N/A'}
ctime = time.time() - time0
das = dasheader('das', dasquery, expire, api='das_core',
ctime=ctime)
rec = {'_id':0, 'function':func, 'key':key, 'result':empty}
rec.update(das)
res.append(rec)
elif isinstance(fields, list) and 'queries' in fields:
res = itertools.islice(self.get_queries(dasquery), idx, idx+limit)
else:
res = self.rawcache.get_from_cache(dasquery, idx, limit, \
collection=collection)
for row in res:
fix_times(row)
yield row
das_timer('DASCore::get_from_cache', self.verbose)
def get_from_cache(self, dasquery, idx=0, limit=0, collection='merge'):
"""
Look-up results from the merge cache and yield them for
further processing.
"""
das_timer('DASCore::get_from_cache', self.verbose)
msg = 'col=%s, query=%s, idx=%s, limit=%s'\
% (collection, dasquery, idx, limit)
self.logger.info(msg)
fields = dasquery.mongo_query.get('fields', None)
if dasquery.mapreduce:
res = self.rawcache.map_reduce(dasquery.mapreduce, dasquery)
elif dasquery.aggregators:
# extract das information from rawcache
rows = self.rawcache.get_from_cache(\
dasquery, collection=collection)
first = next(rows)
sinfo = das_sinfo(first)
# to perform aggregation we need:
# - loop over all aggregator functions
# - loop over all data-services
# - loop over all APIs within a data-services
# the code below does that, it applies aggregator
# to selected (based on key/srv/api) records
res = []
_id = 0
time0 = time.time()
expire = 300 # min expire
for func, key in dasquery.aggregators:
afunc = getattr(das_aggregator, 'das_%s' % func)
found = False
for srv, apis, in sinfo.items():
for api in apis:
rows = self.rawcache.get_from_cache(\
dasquery, collection=collection)
gen = api_rows(rows, api)
data = afunc(key, gen)
ctime = time.time() - time0
das = dasheader(srv,
dasquery,
expire,
api=api,
ctime=ctime)
if isinstance(data, dict) and data['value'] != 'N/A':
aggr = {
'_id': _id,
'function': func,
'key': key,
'result': data
}
aggr.update(das)
res.append(aggr)
_id += 1
found = True
if not found: # when we got nothing add empty result record
empty = {'value': 'N/A'}
ctime = time.time() - time0
das = dasheader('das',
dasquery,
expire,
api='das_core',
ctime=ctime)
rec = {
'_id': 0,
'function': func,
'key': key,
'result': empty
}
rec.update(das)
res.append(rec)
else:
res = self.rawcache.get_from_cache(dasquery, idx, limit, \
collection=collection)
# we assume that all records from single query will have
# identical structure, therefore it will be sufficient to update
# keylearning DB only with first record
count = 0
for row in res:
if not count:
self.keylearning.add_record(dasquery, row)
fix_times(row)
yield row
count += 1
das_timer('DASCore::get_from_cache', self.verbose)
def get_from_cache(self, dasquery, idx=0, limit=0, collection="merge"):
"""
Look-up results from the merge cache and yield them for
further processing.
"""
das_timer("DASCore::get_from_cache", self.verbose)
msg = "col=%s, query=%s, idx=%s, limit=%s" % (collection, dasquery, idx, limit)
self.logger.info(msg)
fields = dasquery.mongo_query.get("fields", None)
if dasquery.mapreduce:
res = self.rawcache.map_reduce(dasquery.mapreduce, dasquery)
elif dasquery.aggregators:
# extract das information from rawcache
rows = self.rawcache.get_from_cache(dasquery, collection=collection)
first = next(rows)
sinfo = das_sinfo(first)
# to perform aggregation we need:
# - loop over all aggregator functions
# - loop over all data-services
# - loop over all APIs within a data-services
# the code below does that, it applies aggregator
# to selected (based on key/srv/api) records
res = []
_id = 0
time0 = time.time()
expire = 300 # min expire
for func, key in dasquery.aggregators:
afunc = getattr(das_aggregator, "das_%s" % func)
found = False
for srv, apis in sinfo.items():
for api in apis:
rows = self.rawcache.get_from_cache(dasquery, collection=collection)
gen = api_rows(rows, api)
data = afunc(key, gen)
ctime = time.time() - time0
das = dasheader(srv, dasquery, expire, api=api, ctime=ctime)
if isinstance(data, dict) and data["value"] != "N/A":
aggr = {"_id": _id, "function": func, "key": key, "result": data}
aggr.update(das)
res.append(aggr)
_id += 1
found = True
if not found: # when we got nothing add empty result record
empty = {"value": "N/A"}
ctime = time.time() - time0
das = dasheader("das", dasquery, expire, api="das_core", ctime=ctime)
rec = {"_id": 0, "function": func, "key": key, "result": empty}
rec.update(das)
res.append(rec)
else:
res = self.rawcache.get_from_cache(dasquery, idx, limit, collection=collection)
# we assume that all records from single query will have
# identical structure, therefore it will be sufficient to update
# keylearning DB only with first record
count = 0
for row in res:
if not count:
self.keylearning.add_record(dasquery, row)
fix_times(row)
yield row
count += 1
das_timer("DASCore::get_from_cache", self.verbose)