def get_queries(self, dasquery): """ Look-up (popular) queries in DAS analytics/logging db """ das_timer('DASCore::get_queries', self.verbose) fields = dasquery.mongo_query.get('fields') spec = dasquery.mongo_query.get('spec') if 'popular' in fields: res = self.analytics.get_popular_queries(spec) else: datestamp = spec.get('date') if isinstance(datestamp, dict): value = datestamp.get('$in') res = \ self.analytics.list_queries(after=value[0], before=value[1]) elif isinstance(datestamp, int): res = self.analytics.list_queries(after=datestamp) elif not datestamp: res = self.analytics.list_queries() else: msg = 'Unsupported date value: %s' % datestamp raise Exception(msg) for row in res: rid = row.pop('_id') yield dict(das_query=row, _id=rid) das_timer('DASCore::get_queries', 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: res = [] _id = 0 for func, key in dasquery.aggregators: rows = self.rawcache.get_from_cache(\ dasquery, collection=collection) data = getattr(das_aggregator, 'das_%s' % func)(key, rows) res += \ [{'_id':_id, 'function': func, 'key': key, 'result': data}] _id += 1 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 insert_query_records(self, dasquery): """ Insert DAS query records into DAS cache and return list of services which will answer this query """ services = dasquery.services self.logger.info("Potential services = %s" % services) if not services: msg = "No data-services for query %s" % dasquery msg += "mongo_query: %s" % dasquery.mongo_query msg += "params: %s" % dasquery.params() print(dastimestamp("DAS WARNING "), msg) # get list of URI which can answer this query ack_services = [] for srv in services: gen = [t for t in getattr(getattr(self, srv), "apimap")(dasquery)] for url, api, args, iformat, expire in gen: header = dasheader(srv, dasquery, expire, api, url, ctime=0) self.rawcache.insert_query_record(dasquery, header) if srv not in ack_services: ack_services.append(srv) if not ack_services: ack_services = services if dasquery.query.find("records ") != -1: srv_status = True # skip DAS queries w/ records request # create das record with initial expire tstamp 2 min in a future # it should be sufficient for processing data-srv records expire = time.time() + 2 * 60 header = dasheader("das", dasquery, expire, api="das_core", services=dict(das=ack_services)) header["lookup_keys"] = [] self.rawcache.insert_query_record(dasquery, header) das_timer("das_record", self.verbose) return ack_services
def insert_query_records(self, dasquery): """ Insert DAS query records into DAS cache and return list of services which will answer this query """ services = dasquery.services self.logger.info('Potential services = %s' % services) if not services: msg = 'No data-services for query %s' % dasquery msg += 'mongo_query: %s' % dasquery.mongo_query msg += 'params: %s' % dasquery.params() print dastimestamp('DAS WARNING '), msg # get list of URI which can answer this query ack_services = [] for srv in services: gen = [t for t in getattr(getattr(self, srv), 'apimap')(dasquery)] for url, api, args, iformat, expire in gen: header = dasheader(srv, dasquery, expire, api, url, ctime=0) self.rawcache.insert_query_record(dasquery, header) if srv not in ack_services: ack_services.append(srv) if not ack_services: ack_services = services if dasquery.query.find('records ') != -1: srv_status = True # skip DAS queries w/ records request expire = 2*60 # 2 minutes, it should be overwriten by data-srv header = dasheader("das", dasquery, expire, api='das_core', services=dict(das=ack_services)) header['lookup_keys'] = [] self.rawcache.insert_query_record(dasquery, header) das_timer('das_record', self.verbose) return ack_services
def getdata_pycurl(url, params, headers=None, expire=3600, post=None, error_expire=300, verbose=0, ckey=None, cert=None, doseq=True, system=None): "Fetch data via pycurl library" contact = 'data-service.' if system: contact = system + ' ' + contact if isinstance(params, dict): timer_key = '%s?%s' % (url, urllib.urlencode(params, doseq=True)) else: timer_key = '%s?%s' % (url, params) das_timer(timer_key, verbose) handler = REQUEST_HANDLER try: data, expire = handler.getdata(url, params, headers, expire, post, \ error_expire, verbose, ckey, cert, doseq) except urllib2.HTTPError as httperror: msg = 'urllib2.HTTPError, system=%s, url=%s, args=%s, headers=%s' \ % (system, url, json.dumps(params), json.dumps(headers)) data = { 'error': 'Received HTTP error from %s data-service' % contact, 'reason': msg, 'ts': time.time() } try: reason = extract_http_error(httperror.read()) data.update({'reason': reason, 'request': msg}) # TODO: err variable did not exit in this function! msg += '\n' + reason except Exception as exp: data.update({'httperror': None}) msg += '\n' + str(exp) print(dastimestamp('getdata_pycurl'), msg) data = json.dumps(data) expire = expire_timestamp(error_expire) except Exception as exp: msg = 'HTTPError, system=%s, url=%s, args=%s, headers=%s' \ % (system, url, json.dumps(params), json.dumps(headers)) print(dastimestamp('getdata_pycurl'), msg + '\n' + str(exp)) data = { 'error': 'Received generic error from %s data-service' % contact, 'reason': msg, 'ts': time.time() } data = json.dumps(data) expire = expire_timestamp(error_expire) das_timer(timer_key, verbose) return data, expire
def getdata_pycurl( url, params, headers=None, expire=3600, post=None, error_expire=300, verbose=0, ckey=None, cert=None, doseq=True, system=None, ): "Fetch data via pycurl library" contact = "data-service." if system: contact = system + " " + contact timer_key = "%s?%s" % (url, urllib.urlencode(params, doseq=True)) das_timer(timer_key, verbose) handler = REQUEST_HANDLER try: data, expire = handler.getdata(url, params, headers, expire, post, error_expire, verbose, ckey, cert, doseq) except urllib2.HTTPError as httperror: msg = "urllib2.HTTPError, system=%s, url=%s, args=%s, headers=%s" % ( system, url, json.dumps(params), json.dumps(headers), ) data = {"error": "Unable to contact %s" % contact, "reason": msg, "ts": time.time()} try: reason = extract_http_error(httperror.read()) data.update({"reason": reason, "request": msg}) msg += "\n" + err except Exception as exp: data.update({"httperror": None}) msg += "\n" + str(exp) print dastimestamp("getdata_pycurl"), msg data = json.dumps(data) expire = expire_timestamp(error_expire) except Exception as exp: msg = "HTTPError, system=%s, url=%s, args=%s, headers=%s" % ( system, url, json.dumps(params), json.dumps(headers), ) print dastimestamp("getdata_pycurl"), msg + "\n" + str(exp) data = {"error": "Unable to contact %s" % contact, "reason": msg, "ts": time.time()} data = json.dumps(data) expire = expire_timestamp(error_expire) das_timer(timer_key, verbose) return data, expire
def getdata_pycurl(url, params, headers=None, expire=3600, post=None, error_expire=300, verbose=0, ckey=None, cert=None, doseq=True, system=None): "Fetch data via pycurl library" contact = 'data-service.' if system: contact = system + ' ' + contact if isinstance(params, dict): timer_key = '%s?%s' % (url, urllib.urlencode(params, doseq=True)) else: timer_key = '%s?%s' % (url, params) das_timer(timer_key, verbose) handler = REQUEST_HANDLER try: data, expire = handler.getdata(url, params, headers, expire, post, \ error_expire, verbose, ckey, cert, doseq) except urllib2.HTTPError as httperror: msg = 'urllib2.HTTPError, system=%s, url=%s, args=%s, headers=%s' \ % (system, url, json.dumps(params), json.dumps(headers)) data = {'error': 'Received HTTP error from %s data-service' % contact, 'reason': msg, 'ts':time.time()} try: reason = extract_http_error(httperror.read()) data.update({'reason': reason, 'request': msg}) # TODO: err variable did not exit in this function! msg += '\n' + reason except Exception as exp: data.update({'httperror': None}) msg += '\n' + str(exp) print(dastimestamp('getdata_pycurl'), msg) data = json.dumps(data) expire = expire_timestamp(error_expire) except Exception as exp: msg = 'HTTPError, system=%s, url=%s, args=%s, headers=%s' \ % (system, url, json.dumps(params), json.dumps(headers)) print(dastimestamp('getdata_pycurl'), msg + '\n' + str(exp)) data = {'error': 'Received generic error from %s data-service' % contact, 'reason': msg, 'ts':time.time()} data = json.dumps(data) expire = expire_timestamp(error_expire) das_timer(timer_key, verbose) return data, expire
def insert_query_records(self, dasquery): """ Insert DAS query records into DAS cache and return list of services which will answer this query """ services = dasquery.services self.logger.info('Potential services = %s' % services) if not services: msg = 'No data-services for query %s' % dasquery msg += 'mongo_query: %s' % dasquery.mongo_query msg += 'params: %s' % dasquery.params() print(dastimestamp('DAS WARNING '), msg) # get list of URI which can answer this query ack_services = [] for srv in services: gen = [t for t in getattr(getattr(self, srv), 'apimap')(dasquery)] for url, api, args, iformat, expire in gen: header = dasheader(srv, dasquery, expire, api, url, ctime=0) self.rawcache.insert_query_record(dasquery, header) if srv not in ack_services: ack_services.append(srv) if not ack_services: ack_services = services if dasquery.query.find('records ') != -1: srv_status = True # skip DAS queries w/ records request # create das record with initial expire tstamp expire = time.time() + self.init_expire header = dasheader("das", dasquery, expire, api='das_core', services=dict(das=ack_services)) header['lookup_keys'] = [] self.rawcache.insert_query_record(dasquery, header) das_timer('das_record', self.verbose) return ack_services
def worker(self, srv, dasquery): """Main worker function which calls data-srv call function""" self.logger.info('##### %s ######\n' % srv) das_timer(srv, self.verbose) getattr(getattr(self, srv), 'call')(dasquery) das_timer(srv, self.verbose)
def getdata_urllib(url, params, headers=None, expire=3600, post=None, error_expire=300, verbose=0, ckey=None, cert=None, doseq=True, system=None, tstamp=None): """ Invoke URL call and retrieve data from data-service based on provided URL and set of parameters. Use post=True to invoke POST request. """ contact = 'data-service.' if system: contact = system + ' ' + contact timer_key = '%s?%s' % (url, urllib.urlencode(params, doseq=True)) das_timer(timer_key, verbose) encoded_data = urllib.urlencode(params, doseq=doseq) if not post: url = url + '?' + encoded_data if not headers: headers = {} if tstamp and 'If-Modified-Since' not in headers.keys(): headers['If-Modified-Since'] = http_timestamp(tstamp) if verbose: print('+++ getdata, url=%s, headers=%s' % (url, headers)) req = urllib2.Request(url) for key, val in headers.items(): req.add_header(key, val) if verbose > 1: handler = urllib2.HTTPHandler(debuglevel=1) opener = urllib2.build_opener(handler) urllib2.install_opener(opener) if ckey and cert: handler = HTTPSClientAuthHandler(ckey, cert, verbose) opener = urllib2.build_opener(handler) urllib2.install_opener(opener) try: time0 = time.time() if post: data = urllib2.urlopen(req, encoded_data) else: data = urllib2.urlopen(req) data_srv_time = time.time() - time0 info = data.info() code = data.getcode() if verbose > 1: print("+++ response code:", code) print("+++ response info\n", info) try: # get HTTP header and look for Expires e_time = expire_timestamp(\ info.__dict__['dict']['expires']) if e_time < expire_timestamp(data_srv_time): expire = max(e_time, expire_timestamp(expire)) elif e_time > time.time(): expire = e_time except Exception as _exp: pass except urllib2.HTTPError as httperror: msg = 'HTTPError, url=%s, args=%s, headers=%s' \ % (url, params, headers) data = { 'error': 'Received HTTP error from %s data-service' % contact, 'reason': msg } try: err = '%s %s' % (contact, extract_http_error(httperror.read())) data.update({'error': err}) msg += '\n' + err except Exception as exp: data.update({'httperror': None}) msg += '\n' + str(exp) print(msg) data = json.dumps(data) expire = expire_timestamp(error_expire) except Exception as exp: msg = 'HTTPError, url=%s, args=%s, headers=%s' \ % (url, params, headers) print(msg + '\n' + str(exp)) data = { 'error': 'Received generic error from %s data-service' % contact, 'reason': msg } data = json.dumps(data) expire = expire_timestamp(error_expire) das_timer(timer_key, verbose) return data, expire
def call(self, query, **kwds): """ Top level DAS api which execute a given query using underlying data-services. It follows the following steps: - parse input query - identify data-sercices based on selection keys and where clause conditions - construct DAS workflow and execute data-service API calls. At this step individual data-services store results into DAS cache. Return status 0/1 depending on success of the calls, can be used by workers on cache server. kwds is provided for compatibility with web layer, e.g. it may invoke this method with additional pid parameter. """ def update_das_query(dasquery, status, reason=None): "Update DAS query record with given status and reason" self.rawcache.update_query_record(dasquery, status, reason=reason) self.rawcache.add_to_record(dasquery, {"das.timer": get_das_timer()}, system="das") # make sure that das record is updated, we use 7 iteration which # sum up into 1 minute to cover default syncdelay value of mongo # server (in a future it would be better to find programatically # this syncdelay value, but it seems pymongo driver does not # provide any API for it. for idx in range(0, 7): spec = {"qhash": dasquery.qhash, "das.system": ["das"]} res = self.rawcache.col.find_one(spec) if res: dbstatus = res.get("das", {}).get("status", None) if dbstatus == status: break msg = "qhash %s, das.status=%s, status=%s, wait for update" % (dasquery.qhash, dbstatus, status) print(dastimestamp("DAS WARNING"), msg) self.rawcache.update_query_record(dasquery, status, reason=reason) time.sleep(idx * idx) self.logger.info("input query=%s" % query) das_timer("DASCore::call", self.verbose) if isinstance(query, object) and hasattr(query, "__class__") and query.__class__.__name__ == "DASQuery": dasquery = query else: dasquery = DASQuery(query) for col in ["merge", "cache"]: self.rawcache.remove_expired(dasquery, col) query = dasquery.mongo_query spec = query.get("spec") fields = query.get("fields") if fields == ["records"]: msg = "look-up all records in cache" self.logger.info(msg) return "in cache" if spec == dict(records="*"): self.logger.info("look-up everything in cache") return "in cache" for record in self.rawcache.find_specs(dasquery): status = record["das"]["status"] msg = "found query %s in cache, status=%s\n" % (record["query"], status) self.logger.info(msg) print(dastimestamp("DAS INFO"), msg) return status self.logger.info(dasquery) das_timer("das_record", self.verbose) services = self.insert_query_records(dasquery) if not services: msg = "unable to locate data-services to fulfill this request" msg += ", will iterate over all registered services" print(dastimestamp("DAS WARNING "), dasquery, msg) services = dasquery.services if dasquery.services else self.systems try: if self.multitask: jobs = [] for srv in sorted(services): jobs.append(self.taskmgr.spawn(self.worker, srv, dasquery)) self.taskmgr.joinall(jobs) else: for srv in services: self.worker(srv, dasquery) except Exception as exc: print_exc(exc) return "fail" self.logger.info("\n##### merging ######\n") update_das_query(dasquery, "merging") das_timer("merge", self.verbose) for attempt in range(0, 4): # try couple of times to avoid DB problems time.sleep(attempt) status = self.rawcache.merge_records(dasquery, attempt) if status == "ok": break das_timer("merge", self.verbose) # check if we have service records and properly setup status self.logger.info("\n##### check services ######\n") das_services = self.rawcache.check_services(dasquery) reason = "" status = "ok" if not das_services: if "records" in dasquery.query: status = "ok" # keep status ok for 'records' queries else: reason = "no data records found in DAS cache" status = "fail" print(dastimestamp("DAS ERROR "), dasquery, reason) update_das_query(dasquery, status, reason) das_timer("DASCore::call", self.verbose) return status
def getdata_urllib(url, params, headers=None, expire=3600, post=None, error_expire=300, verbose=0, ckey=None, cert=None, doseq=True, system=None, tstamp=None): """ Invoke URL call and retrieve data from data-service based on provided URL and set of parameters. Use post=True to invoke POST request. """ contact = 'data-service.' if system: contact = system + ' ' + contact timer_key = '%s?%s' % (url, urllib.urlencode(params, doseq=True)) das_timer(timer_key, verbose) encoded_data = urllib.urlencode(params, doseq=doseq) if not post: url = url + '?' + encoded_data if not headers: headers = {} if tstamp and 'If-Modified-Since' not in headers.keys(): headers['If-Modified-Since'] = http_timestamp(tstamp) if verbose: print('+++ getdata, url=%s, headers=%s' % (url, headers)) req = urllib2.Request(url) for key, val in headers.items(): req.add_header(key, val) if verbose > 1: handler = urllib2.HTTPHandler(debuglevel=1) opener = urllib2.build_opener(handler) urllib2.install_opener(opener) if ckey and cert: handler = HTTPSClientAuthHandler(ckey, cert, verbose) opener = urllib2.build_opener(handler) urllib2.install_opener(opener) try: time0 = time.time() if post: data = urllib2.urlopen(req, encoded_data) else: data = urllib2.urlopen(req) data_srv_time = time.time()-time0 info = data.info() code = data.getcode() if verbose > 1: print("+++ response code:", code) print("+++ response info\n", info) try: # get HTTP header and look for Expires e_time = expire_timestamp(\ info.__dict__['dict']['expires']) if e_time < expire_timestamp(data_srv_time): expire = max(e_time, expire_timestamp(expire)) elif e_time > time.time(): expire = e_time except Exception as _exp: pass except urllib2.HTTPError as httperror: msg = 'HTTPError, url=%s, args=%s, headers=%s' \ % (url, params, headers) data = {'error': 'Received HTTP error from %s data-service' % contact, 'reason': msg} try: err = '%s %s' % (contact, extract_http_error(httperror.read())) data.update({'error':err}) msg += '\n' + err except Exception as exp: data.update({'httperror': None}) msg += '\n' + str(exp) print(msg) data = json.dumps(data) expire = expire_timestamp(error_expire) except Exception as exp: msg = 'HTTPError, url=%s, args=%s, headers=%s' \ % (url, params, headers) print(msg + '\n' + str(exp)) data = {'error': 'Received generic error from %s data-service' % contact, 'reason': msg} data = json.dumps(data) expire = expire_timestamp(error_expire) das_timer(timer_key, verbose) return data, expire
def call(self, query, **kwds): """ Top level DAS api which execute a given query using underlying data-services. It follows the following steps: - parse input query - identify data-sercices based on selection keys and where clause conditions - construct DAS workflow and execute data-service API calls. At this step individual data-services store results into DAS cache. Return status 0/1 depending on success of the calls, can be used by workers on cache server. kwds is provided for compatibility with web layer, e.g. it may invoke this method with additional pid parameter. """ def update_das_query(dasquery, status, reason=None): "Update DAS query record with given status and reason" self.rawcache.update_query_record(dasquery, status, reason=reason) self.rawcache.add_to_record(\ dasquery, {'das.timer': get_das_timer()}, system='das') self.logger.info('input query=%s' % query) das_timer('DASCore::call', self.verbose) if isinstance(query, object) and hasattr(query, '__class__')\ and query.__class__.__name__ == 'DASQuery': dasquery = query else: dasquery = DASQuery(query) for col in ['merge', 'cache']: self.rawcache.remove_expired(dasquery, col) query = dasquery.mongo_query spec = query.get('spec') fields = query.get('fields') if fields == ['records']: msg = 'look-up all records in cache' self.logger.info(msg) return 'in cache' if spec == dict(records='*'): self.logger.info("look-up everything in cache") return 'in cache' for record in self.rawcache.find_specs(dasquery): status = record['das']['status'] msg = 'found query %s in cache, status=%s\n' \ % (record['query'], status) self.logger.info(msg) print(dastimestamp('DAS INFO'), msg) return status self.logger.info(dasquery) das_timer('das_record', self.verbose) services = self.insert_query_records(dasquery) if not services: msg = 'unable to locate data-services to fulfill this request' msg += ', will iterate over all registered services' print(dastimestamp('DAS WARNING '), dasquery, msg) services = dasquery.services if dasquery.services else self.systems try: if self.multitask: jobs = [] for srv in sorted(services): jobs.append(self.taskmgr.spawn(self.worker, srv, dasquery)) self.taskmgr.joinall(jobs) else: for srv in services: self.worker(srv, dasquery) except Exception as exc: print_exc(exc) return 'fail' self.logger.info('\n##### merging ######\n') update_das_query(dasquery, 'merging') das_timer('merge', self.verbose) # check that all query record statuses are ok, i.e. we did insert records # this status is set by self.rawcache.update_cache for idx in range(self.collect_wait_time): records = self.rawcache.find_query_record(dasquery) statuses = [] for row in records: system = row['das']['system'] status = row['das']['status'] self.logger.info("### query record status %s %s %s" % (dasquery.qhash, system, status)) statuses.append(status) all_statuses = sorted(list(set(statuses))) # at this point we're looking that all services will have 'ok' and das status will be 'merging' if len(all_statuses) == 2 and all_statuses == ['merging', 'ok']: break time.sleep(1) # now we can merge records status = self.rawcache.merge_records(dasquery) das_timer('merge', self.verbose) # check if we have service records and properly setup status self.logger.info('\n##### check services ######\n') das_services = self.rawcache.check_services(dasquery) reason = '' status = 'ok' if not das_services: if 'records' in dasquery.query: status = 'ok' # keep status ok for 'records' queries else: reason = 'no data records found in DAS cache' status = 'fail' print(dastimestamp('DAS ERROR '), dasquery, reason) update_das_query(dasquery, status, reason) das_timer('DASCore::call', self.verbose) return status
def __init__(self, config=None, debug=0, nores=False, logger=None, engine=None, multitask=True): if config: dasconfig = config else: dasconfig = das_readconfig() verbose = dasconfig['verbose'] self.stdout = debug if isinstance(debug, int) and debug: self.verbose = debug dasconfig['verbose'] = debug else: self.verbose = verbose das_timer('DASCore::init', self.verbose) self.operators = das_operators() self.collect_wait_time = dasconfig['das'].get('collect_wait_time', 120) # set noresults option self.noresults = False if nores: dasconfig['write_cache'] = True self.noresults = nores self.init_expire = dasconfig['das'].get('init_expire', 5 * 60) self.multitask = dasconfig['das'].get('multitask', True) if debug or self.verbose: self.multitask = False # in verbose mode do not use multitask dasconfig['das']['multitask'] = False if not multitask: # explicitly call DASCore ctor self.multitask = False dasconfig['das']['multitask'] = False dasconfig['engine'] = engine if self.multitask: nworkers = dasconfig['das'].get('core_workers', 5) # if engine: # thr_name = 'DASCore:PluginTaskManager' # self.taskmgr = PluginTaskManager(\ # engine, nworkers=nworkers, name=thr_name) # self.taskmgr.subscribe() # else: # thr_name = 'DASCore:TaskManager' # self.taskmgr = TaskManager(nworkers=nworkers, name=thr_name) thr_name = 'DASCore:TaskManager' self.taskmgr = TaskManager(nworkers=nworkers, name=thr_name) else: self.taskmgr = None if logger: self.logger = logger else: self.logger = PrintManager('DASCore', self.verbose) # define Mapping/Analytics/Parser in this order since Parser depends # on first two dasmapping = DASMapping(dasconfig) dasconfig['dasmapping'] = dasmapping self.mapping = dasmapping self.keylearning = DASKeyLearning(dasconfig) dasconfig['keylearning'] = self.keylearning # init DAS cache self.rawcache = DASMongocache(dasconfig) dasconfig['rawcache'] = self.rawcache # plug-in architecture: loop over registered data-services in # dasconfig; load appropriate module/class; register data # service with DASCore. self.systems = dasmapping.list_systems() # pointer to the DAS top level directory dasroot = '/'.join(__file__.split('/')[:-3]) for name in self.systems: try: klass = 'DAS/services/%s/%s_service.py' \ % (name, name) srvfile = os.path.join(dasroot, klass) with open(srvfile) as srvclass: for line in srvclass: if line.find('(DASAbstractService)') != -1: klass = line.split('(DASAbstractService)')[0] klass = klass.split('class ')[-1] break mname = 'DAS.services.%s.%s_service' % (name, name) module = __import__(mname, fromlist=[klass]) obj = getattr(module, klass)(dasconfig) setattr(self, name, obj) except IOError as err: if debug > 1: # we have virtual services, so IOError can be correct print_exc(err) try: mname = 'DAS.services.generic_service' module = __import__(mname, fromlist=['GenericService']) obj = module.GenericService(name, dasconfig) setattr(self, name, obj) except Exception as exc: print_exc(exc) msg = "Unable to load %s data-service plugin" % name raise Exception(msg) except Exception as exc: print_exc(exc) msg = "Unable to load %s data-service plugin" % name raise Exception(msg) # loop over systems and get system keys, add mapping keys to final list self.service_keys = {} self.service_parameters = {} for name in self.systems: skeys = list(getattr(self, name).keys()) self.service_keys[getattr(self, name).name] = skeys sparams = getattr(self, name).parameters() self.service_parameters[getattr(self, name).name] = sparams self.service_keys['special'] = das_special_keys() self.dasconfig = dasconfig das_timer('DASCore::init', 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 call(self, query, add_to_analytics=True, **kwds): """ Top level DAS api which execute a given query using underlying data-services. It follows the following steps: - parse input query - identify data-sercices based on selection keys and where clause conditions - construct DAS workflow and execute data-service API calls. At this step individual data-services store results into DAS cache. Return status 0/1 depending on success of the calls, can be used by workers on cache server. kwds is provided for compatibility with web layer, e.g. it may invoke this method with additional pid parameter. """ def update_das_query(dasquery, status, reason=None): "Update DAS query record with given status and reason" self.rawcache.update_query_record(dasquery, status, reason=reason) self.rawcache.add_to_record(\ dasquery, {'das.timer': get_das_timer()}, system='das') # make sure that das record is updated, we use 7 iteration which # sum up into 1 minute to cover default syncdelay value of mongo # server (in a future it would be better to find programatically # this syncdelay value, but it seems pymongo driver does not # provide any API for it. for idx in xrange(1, 7): spec = {'qhash':dasquery.qhash, 'das.system':['das']} res = self.rawcache.col.find_one(spec) if res: dbstatus = res.get('das', {}).get('status', None) if dbstatus == status: break msg = 'qhash %s, das.status=%s, status=%s, wait for update' \ % (dasquery.qhash, dbstatus, status) print dastimestamp('DAS WARNING'), msg time.sleep(idx*idx) self.rawcache.update_query_record(dasquery, status, reason=reason) self.logger.info('input query=%s' % query) das_timer('DASCore::call', self.verbose) if isinstance(query, object) and hasattr(query, '__class__')\ and query.__class__.__name__ == 'DASQuery': dasquery = query else: dasquery = DASQuery(query) for col in ['merge', 'cache']: self.rawcache.remove_expired(dasquery, col) if add_to_analytics: dasquery.add_to_analytics() query = dasquery.mongo_query spec = query.get('spec') fields = query.get('fields') if fields == ['records']: msg = 'look-up all records in cache' self.logger.info(msg) return 'in cache' if spec == dict(records='*'): self.logger.info("look-up everything in cache") return 'in cache' for record in self.rawcache.find_specs(dasquery): status = record['das']['status'] msg = 'found query %s in cache, status=%s\n' \ % (record['query'], status) self.logger.info(msg) print dastimestamp('DAS INFO'), msg return status self.logger.info(dasquery) das_timer('das_record', self.verbose) services = self.insert_query_records(dasquery) if not services: msg = 'unable to locate data-services to fulfill this request' msg += ', will iterate over all registered services' print dastimestamp('DAS WARNING '), dasquery, msg services = dasquery.services if dasquery.services else self.systems try: if self.multitask: jobs = [] for srv in sorted(services): jobs.append(self.taskmgr.spawn(self.worker, srv, dasquery)) self.taskmgr.joinall(jobs) else: for srv in services: self.worker(srv, dasquery) except Exception as exc: print_exc(exc) return 'fail' self.logger.info('\n##### merging ######\n') update_das_query(dasquery, 'merging') das_timer('merge', self.verbose) self.rawcache.merge_records(dasquery) das_timer('merge', self.verbose) # check if we have service records and properly setup status self.logger.info('\n##### check services ######\n') das_services = self.rawcache.check_services(dasquery) reason = '' status = 'ok' if not das_services: if 'records' in dasquery.query: status = 'ok' # keep status ok for 'records' queries else: reason = 'no data records found in DAS cache' status = 'fail' print dastimestamp('DAS ERROR '), dasquery, reason update_das_query(dasquery, status, reason) das_timer('DASCore::call', self.verbose) return status
def getdata_urllib( url, params, headers=None, expire=3600, post=None, error_expire=300, verbose=0, ckey=None, cert=None, doseq=True, system=None, ): """ Invoke URL call and retrieve data from data-service based on provided URL and set of parameters. Use post=True to invoke POST request. """ contact = "data-service." if system: contact = system + " " + contact timer_key = "%s?%s" % (url, urllib.urlencode(params, doseq=True)) das_timer(timer_key, verbose) encoded_data = urllib.urlencode(params, doseq=doseq) if not post: url = url + "?" + encoded_data if not headers: headers = {} if verbose: print "+++ getdata, url=%s, headers=%s" % (url, headers) req = urllib2.Request(url) for key, val in headers.iteritems(): req.add_header(key, val) if verbose > 1: handler = urllib2.HTTPHandler(debuglevel=1) opener = urllib2.build_opener(handler) urllib2.install_opener(opener) if ckey and cert: handler = HTTPSClientAuthHandler(ckey, cert, verbose) opener = urllib2.build_opener(handler) urllib2.install_opener(opener) try: time0 = time.time() if post: data = urllib2.urlopen(req, encoded_data) else: data = urllib2.urlopen(req) data_srv_time = time.time() - time0 try: # get HTTP header and look for Expires e_time = expire_timestamp(data.info().__dict__["dict"]["expires"]) if e_time < expire_timestamp(data_srv_time): expire = max(e_time, expire_timestamp(expire)) elif e_time > time.time(): expire = e_time except Exception as _exp: pass except urllib2.HTTPError as httperror: msg = "HTTPError, url=%s, args=%s, headers=%s" % (url, params, headers) data = {"error": "Unable to contact %s" % contact, "reason": msg} try: err = "%s %s" % (contact, extract_http_error(httperror.read())) data.update({"error": err}) msg += "\n" + err except Exception as exp: data.update({"httperror": None}) msg += "\n" + str(exp) print msg data = json.dumps(data) expire = expire_timestamp(error_expire) except Exception as exp: msg = "HTTPError, url=%s, args=%s, headers=%s" % (url, params, headers) print msg + "\n" + str(exp) data = {"error": "Unable to contact %s" % contact, "reason": msg} data = json.dumps(data) expire = expire_timestamp(error_expire) das_timer(timer_key, verbose) return data, expire
def call(self, query, add_to_analytics=True, **kwds): """ Top level DAS api which execute a given query using underlying data-services. It follows the following steps: - parse input query - identify data-sercices based on selection keys and where clause conditions - construct DAS workflow and execute data-service API calls. At this step individual data-services store results into DAS cache. Return status 0/1 depending on success of the calls, can be used by workers on cache server. kwds is provided for compatibility with web layer, e.g. it may invoke this method with additional pid parameter. """ self.logger.info('input query=%s' % query) das_timer('DASCore::call', self.verbose) services = [] if isinstance(query, object) and hasattr(query, '__class__')\ and query.__class__.__name__ == 'DASQuery': dasquery = query else: dasquery = DASQuery(query, mongoparser=self.mongoparser) if add_to_analytics: dasquery.add_to_analytics() query = dasquery.mongo_query if dasquery.mongo_query.has_key('system'): system = query['system'] if isinstance(system, str) or isinstance(system, unicode): services = [system] elif isinstance(system, list): services = system else: msg = 'Unsupported system=%s type=%s in DAS query' \ % (system, type(system)) raise Exception(msg) spec = query.get('spec') fields = query.get('fields') if fields == ['records']: msg = 'look-up all records in cache' self.logger.info(msg) return 'in cache' if spec == dict(records='*'): self.logger.info("look-up everything in cache") return 'in cache' for record in self.rawcache.find_specs(dasquery): status = record['das']['status'] msg = 'found query %s in cache, status=%s\n' \ % (record['query'], status) self.logger.info(msg) return status similar_dasquery = self.rawcache.similar_queries(dasquery) if similar_dasquery: for record in self.rawcache.find_specs(similar_dasquery): if record: try: status = record['das']['status'] except: status = 'N/A' msg = 'Fail to look-up das.status, record=%s' % record self.logger.info(msg) msg = 'found SIMILAR query in cache,' msg += 'query=%s, status=%s\n' % (record['query'], status) self.logger.info(msg) return status self.logger.info(dasquery) params = dasquery.params() if not services: services = params['services'] self.logger.info('services = %s' % services) das_timer('das_record', self.verbose) # initial expire tstamp 1 day (long enough to be overwriten by data-srv) expire = expire_timestamp(time.time()+1*24*60*60) header = dasheader("das", dasquery, expire) header['lookup_keys'] = [] self.rawcache.insert_query_record(dasquery, header) das_timer('das_record', self.verbose) try: if self.multitask: jobs = [] for srv in services: jobs.append(self.taskmgr.spawn(self.worker, srv, dasquery)) self.taskmgr.joinall(jobs) else: for srv in services: self.worker(srv, dasquery) except Exception as exc: print_exc(exc) return 'fail' self.logger.info('\n##### merging ######\n') self.rawcache.update_query_record(dasquery, 'merging') das_timer('merge', self.verbose) self.rawcache.merge_records(dasquery) das_timer('merge', self.verbose) self.rawcache.update_query_record(dasquery, 'ok') self.rawcache.add_to_record(\ dasquery, {'das.timer': get_das_timer()}, system='das') das_timer('DASCore::call', self.verbose) return 'ok'
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 __init__(self, config=None, debug=0, nores=False, logger=None, engine=None, multitask=True): if config: dasconfig = config else: dasconfig = das_readconfig() verbose = dasconfig['verbose'] self.stdout = debug if isinstance(debug, int): self.verbose = debug dasconfig['verbose'] = debug else: self.verbose = verbose das_timer('DASCore::init', self.verbose) self.operators = das_operators() # set noresults option self.noresults = False if nores: dasconfig['write_cache'] = True self.noresults = nores self.multitask = dasconfig['das'].get('multitask', True) if debug or self.verbose: self.multitask = False # in verbose mode do not use multitask dasconfig['das']['multitask'] = False if not multitask: # explicitly call DASCore ctor, e.g. in analytics self.multitask = False dasconfig['das']['multitask'] = False dasconfig['engine'] = engine if self.multitask: nworkers = dasconfig['das'].get('core_workers', 5) if engine: thr_name = 'DASCore:PluginTaskManager' self.taskmgr = PluginTaskManager(\ engine, nworkers=nworkers, name=thr_name) self.taskmgr.subscribe() else: thr_name = 'DASCore:TaskManager' self.taskmgr = TaskManager(nworkers=nworkers, name=thr_name) else: self.taskmgr = None if logger: self.logger = logger else: self.logger = PrintManager('DASCore', self.verbose) # define Mapping/Analytics/Parser in this order since Parser depends # on first two dasmapping = DASMapping(dasconfig) dasconfig['dasmapping'] = dasmapping self.mapping = dasmapping self.analytics = DASAnalytics(dasconfig) dasconfig['dasanalytics'] = self.analytics self.keylearning = DASKeyLearning(dasconfig) dasconfig['keylearning'] = self.keylearning # init DAS cache self.rawcache = DASMongocache(dasconfig) dasconfig['rawcache'] = self.rawcache # plug-in architecture: loop over registered data-services in # dasconfig; load appropriate module/class; register data # service with DASCore. self.systems = dasmapping.list_systems() # pointer to the DAS top level directory dasroot = '/'.join(__file__.split('/')[:-3]) for name in self.systems: try: klass = 'DAS/services/%s/%s_service.py' \ % (name, name) srvfile = os.path.join(dasroot, klass) with file(srvfile) as srvclass: for line in srvclass: if line.find('(DASAbstractService)') != -1: klass = line.split('(DASAbstractService)')[0] klass = klass.split('class ')[-1] break mname = 'DAS.services.%s.%s_service' % (name, name) module = __import__(mname, fromlist=[klass]) obj = getattr(module, klass)(dasconfig) setattr(self, name, obj) SERVICES[name] = obj except IOError as err: if debug > 1: # we have virtual services, so IOError can be correct print_exc(err) try: mname = 'DAS.services.generic_service' module = __import__(mname, fromlist=['GenericService']) obj = module.GenericService(name, dasconfig) setattr(self, name, obj) except Exception as exc: print_exc(exc) msg = "Unable to load %s data-service plugin" % name raise Exception(msg) except Exception as exc: print_exc(exc) msg = "Unable to load %s data-service plugin" % name raise Exception(msg) # loop over systems and get system keys, add mapping keys to final list self.service_keys = {} self.service_parameters = {} for name in self.systems: skeys = getattr(self, name).keys() self.service_keys[getattr(self, name).name] = skeys sparams = getattr(self, name).parameters() self.service_parameters[getattr(self, name).name] = sparams self.service_keys['special'] = das_special_keys() self.dasconfig = dasconfig das_timer('DASCore::init', 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)