コード例 #1
0
    def __init__(
        self,
        solrHost=None,
        config=None,
        addparams="",
    ):

        self.config = AppConfig(config)

        if solrHost is None:
            self.solrURL = self.config.getConfig()["SOLR"]["host"]
        else:
            self.solrURL = solrHost

        self.additionalparams = addparams

        self.startTime = time.strftime('%H:%M:%S')
        self.totalQueryTime = 0
        self.totalNumberRequests = 0

        self.timeCollection = {}
        self.initializeCollecCurrentQueryTime()

        cTimeUTC = datetime.utcnow()
        nTList = [
            str(cTimeUTC.date()), "T",
            str(cTimeUTC.hour),
            str(cTimeUTC.minute),
            str(cTimeUTC.second), "Z"
        ]
        self.currentTime = "".join(nTList)
コード例 #2
0
    def __init__(self, config: AppConfig):
        self.config = config

        self.parentPath = config.getConfig()["ParentPath"]
        #self.solrURL = config.getConfig()["SolrURL"]
        self.regex = re.compile(config.getConfig()["idregex"],
                                re.UNICODE | re.DOTALL | re.IGNORECASE)
        self.solrWrapper = SolrWrapper(config)
        self.outFile = open(config.getConfig()["outFile"], "a")
        self.outline = Template("idnotavailable: $id / file: $file $linebreak")
コード例 #3
0
    def __init__(self,
                 appConfig: AppConfig = None,
                 userArg=None,
                 passwordArg=None):
        self.config = appConfig.getConfig()

        if not self.config["MONGO"]["HOST"]["user"] is None and not self.config[
                "MONGO"]["HOST"]["password"] is None:
            uri = 'mongodb://{USER}:{PASSWORD}@{SERVER}:{PORT}/{DB}'.format(
                USER=self.config['HOST']['user'],
                PASSWORD=self.config['HOST']['password'],
                SERVER=self.config['HOST']['server'],
                PORT=self.config['HOST']['port'],
                DB=self.config['HOST']['authDB'])
        elif not userArg is None and not passwordArg is None:
            uri = 'mongodb://{USER}:{PASSWORD}@{SERVER}:{PORT}/{DB}'.format(
                USER=userArg,
                PASSWORD=passwordArg,
                SERVER=self.config["MONGO"]['HOST']['server'],
                PORT=self.config["MONGO"]['HOST']['port'],
                DB=self.config["MONGO"]['HOST']['authDB'])

        else:

            uri = 'mongodb://{SERVER}:{PORT}'.format(
                SERVER=self.config["MONGO"]['HOST']['server'],
                PORT=self.config["MONGO"]['HOST']['port'],
            )

        self.client = MongoClient(uri)
        self.database = self.client[self.config["MONGO"]['DB']['dbname']]
        self.queryCollection = self.database[self.config["MONGO"]['DB']
                                             ['querycollection']]
        self.responseCollection = self.database[self.config["MONGO"]['DB']
                                                ['responsecollection']]
コード例 #4
0
ファイル: mongo.py プロジェクト: guenterh/dataIngestion
    def __init__(self, appConfig : AppConfig = None):
        self.config = appConfig.getConfig()

        if not self.config['HOST']['user'] is None:
            uri = 'mongodb://{USER}:{PASSWORD}@{SERVER}:{PORT}/{DB}'.format(
                USER=self.config['HOST']['user'],
                PASSWORD=self.config['HOST']['password'],
                SERVER=self.config['HOST']['server'],
                PORT=self.config['HOST']['port'],
                DB=self.config['HOST']['authDB']
            )
        else:

            uri = 'mongodb://{SERVER}:{PORT}'.format(
                SERVER=self.config['HOST']['server'],
                PORT=self.config['HOST']['port'],
            )

        self.client = MongoClient( uri)
        self.database = self.client[self.config['DB']['dbname']]
        self.collection = self.database[self.config['DB']['collection']['name']]
コード例 #5
0
ファイル: mongo.py プロジェクト: swissbib/kafka-event-hub
    def __init__(self, appConfig : AppConfig = None):
        self.config = appConfig.getConfig()

        if not self.config['HOST']['user'] is None:
            uri = 'mongodb://{USER}:{PASSWORD}@{SERVER}:{PORT}/{DB}'.format(
                USER=self.config['HOST']['user'],
                PASSWORD=self.config['HOST']['password'],
                SERVER=self.config['HOST']['server'],
                PORT=self.config['HOST']['port'],
                DB=self.config['HOST']['authDB']
            )
        else:

            uri = 'mongodb://{SERVER}:{PORT}'.format(
                SERVER=self.config['HOST']['server'],
                PORT=self.config['HOST']['port'],
            )

        self.client = MongoClient( uri)
        self.database = self.client[self.config['DB']['dbname']]
        self.collection = self.database[self.config['DB']['collection']['name']]
コード例 #6
0
                                id = pId.group(1)
                                #self.printId(id,subobject)
                                if not self.solrWrapper.checkId(id):
                                    self.printId(id, subobject)
                            #print(strLine)
            except Exception as ex:
                print(ex)

    def printId(self, currentId, filename):

        self.outFile.write(
            self.outline.substitute({
                "id": currentId,
                "file": filename,
                "linebreak": linesep
            }))
        self.outFile.flush()

    def closeResources(self):
        if not self.outFile is None:
            self.outFile.close()


if __name__ == '__main__':

    config = AppConfig("config/files/idchecker/checker.yaml")

    checker = IDChecker(config)
    checker.checkIDs()
    checker.closeResources()
    print("work done")
コード例 #7
0
 def __init__(self, config: AppConfig):
     self.sorlURL = config.getConfig()["SolrURL"]
     self.template = Template(self.sorlURL)
コード例 #8
0
from mq.topic import TopicConsumer


class TestExchange(object):
    def __init__(self, exchangeName, config):
        print exchangeName
        print appConfig['rabbit']
        self.topicConsumer = TopicConsumer(
            config=config,
            exchangeName=exchangeName,
            bindingKey='admin.developmentParser.parser.acquilite',
            consumeFunc=self.consumeMessages,
            noArgs=True)
        self.topicConsumer.startConsuming()

    def consumeMessages(self, ch, method, properties, body):
        print method.routing_key
        print body


if __name__ == '__main__':
    exchangeName = 'sis-admin-exchange'
    from config.appConfig import AppConfig
    config = AppConfig('developmentParser')
    config.loadAll()
    appControl = TestExchange(exchangeName, config)
コード例 #9
0
class RunQueries():
    def __init__(
        self,
        solrHost=None,
        config=None,
        addparams="",
    ):

        self.config = AppConfig(config)

        if solrHost is None:
            self.solrURL = self.config.getConfig()["SOLR"]["host"]
        else:
            self.solrURL = solrHost

        self.additionalparams = addparams

        self.startTime = time.strftime('%H:%M:%S')
        self.totalQueryTime = 0
        self.totalNumberRequests = 0

        self.timeCollection = {}
        self.initializeCollecCurrentQueryTime()

        cTimeUTC = datetime.utcnow()
        nTList = [
            str(cTimeUTC.date()), "T",
            str(cTimeUTC.hour),
            str(cTimeUTC.minute),
            str(cTimeUTC.second), "Z"
        ]
        self.currentTime = "".join(nTList)

    def getURL(self):
        return self.solrURL

    def getAverTime(self):
        return self.totalQueryTime / self.totalNumberRequests

    def getStartTime(self):
        return self.startTime

    def collecCurrentQueryTime(self, queryTime):

        if queryTime >= 0 and queryTime <= 100:
            self.timeCollection['0_100'] += 1
        elif queryTime >= 101 and queryTime <= 200:
            self.timeCollection['100_200'] += 1
        elif queryTime >= 201 and queryTime <= 300:
            self.timeCollection['200_300'] += 1
        elif queryTime >= 301 and queryTime <= 400:
            self.timeCollection['300_400'] += 1
        elif queryTime >= 401 and queryTime <= 500:
            self.timeCollection['400_500'] += 1
        elif queryTime >= 501 and queryTime <= 600:
            self.timeCollection['500_600'] += 1
        elif queryTime >= 601 and queryTime <= 700:
            self.timeCollection['600_700'] += 1
        elif queryTime >= 701 and queryTime <= 800:
            self.timeCollection['700_800'] += 1
        elif queryTime >= 801 and queryTime <= 900:
            self.timeCollection['800_900'] += 1
        elif queryTime >= 901 and queryTime <= 1000:
            self.timeCollection['900_1000'] += 1
        elif queryTime >= 1001 and queryTime <= 1200:
            self.timeCollection['1000_1200'] += 1
        elif queryTime >= 1201 and queryTime <= 1400:
            self.timeCollection['1200_1400'] += 1
        elif queryTime >= 1401 and queryTime <= 1600:
            self.timeCollection['1400_1600'] += 1
        elif queryTime >= 1601 and queryTime <= 1800:
            self.timeCollection['1600_1800'] += 1
        elif queryTime >= 1801 and queryTime <= 2000:
            self.timeCollection['1800_2000'] += 1
        elif queryTime >= 2001 and queryTime <= 2400:
            self.timeCollection['2000_2400'] += 1
        elif queryTime >= 2401 and queryTime <= 2800:
            self.timeCollection['2400_2800'] += 1
        elif queryTime >= 2801 and queryTime <= 3200:
            self.timeCollection['2800_3200'] += 1
        elif queryTime >= 3201 and queryTime <= 3800:
            self.timeCollection['3200_3800'] += 1
        elif queryTime >= 3801 and queryTime <= 4400:
            self.timeCollection['3800_4400'] += 1
        elif queryTime >= 4401 and queryTime <= 5000:
            self.timeCollection['4400_5000'] += 1
        elif queryTime >= 5001 and queryTime <= 6000:
            self.timeCollection['5000_6000'] += 1
        elif queryTime >= 6001 and queryTime <= 7000:
            self.timeCollection['6000_7000'] += 1
        elif queryTime >= 7001 and queryTime <= 8000:
            self.timeCollection['7000_8000'] += 1
        elif queryTime >= 8001 and queryTime <= 9000:
            self.timeCollection['8000_9000'] += 1
        elif queryTime >= 9001 and queryTime <= 10000:
            self.timeCollection['9000_10000'] += 1
        elif queryTime >= 10001 and queryTime <= 15000:
            self.timeCollection['10000_15000'] += 1
        elif queryTime >= 15001 and queryTime <= 20000:
            self.timeCollection['15000_20000'] += 1
        elif queryTime >= 20001:
            self.timeCollection['gt20000'] += 1

    def initializeCollecCurrentQueryTime(self):

        self.timeCollection['0_100'] = 0
        self.timeCollection['100_200'] = 0
        self.timeCollection['200_300'] = 0
        self.timeCollection['300_400'] = 0
        self.timeCollection['400_500'] = 0
        self.timeCollection['500_600'] = 0
        self.timeCollection['600_700'] = 0
        self.timeCollection['700_800'] = 0
        self.timeCollection['800_900'] = 0
        self.timeCollection['900_1000'] = 0
        self.timeCollection['1000_1200'] = 0
        self.timeCollection['1200_1400'] = 0
        self.timeCollection['1400_1600'] = 0
        self.timeCollection['1600_1800'] = 0
        self.timeCollection['1800_2000'] = 0
        self.timeCollection['2000_2400'] = 0
        self.timeCollection['2400_2800'] = 0
        self.timeCollection['2800_3200'] = 0
        self.timeCollection['3200_3800'] = 0
        self.timeCollection['3800_4400'] = 0
        self.timeCollection['4400_5000'] = 0
        self.timeCollection['5000_6000'] = 0
        self.timeCollection['6000_7000'] = 0
        self.timeCollection['7000_8000'] = 0
        self.timeCollection['8000_9000'] = 0
        self.timeCollection['9000_10000'] = 0
        self.timeCollection['10000_15000'] = 0
        self.timeCollection['15000_20000'] = 0
        self.timeCollection['gt20000'] = 0

    def getDistribution(self):
        distribution = []
        for k, v in self.timeCollection.items():
            distribution.append(" (" + k + "=>" + str(v) + ") ")

        return ''.join(distribution)

    def executeRequest(self, query):

        if not self.additionalparams == "":
            query += "&" + self.additionalparams

        result = requests.get(self.solrURL, params=query.encode("utf-8"))
        io = StringIO(result.text)
        result = json.load(io)
        queryTime = result["responseHeader"]["QTime"]

        self.totalQueryTime += queryTime
        self.totalNumberRequests += 1
        self.collecCurrentQueryTime(queryTime)

        return result