Python threads.Thread类代码示例

    def worker(please_stop):
        pending = Queue("pending ids", max=BATCH_SIZE*3, silent=False)

        pending_thread = Thread.run(
            "get pending",
            get_pending,
            source=source,
            since=last_updated,
            pending_bugs=pending,
            please_stop=please_stop
        )
        diff_thread = Thread.run(
            "diff",
            diff,
            source,
            destination,
            pending,
            please_stop=please_stop
        )
        replication_thread = Thread.run(
            "replication",
            replicate,
            source,
            destination,
            pending,
            config.fix,
            please_stop=please_stop
        )
        pending_thread.join()
        diff_thread.join()
        pending.add(Thread.STOP)
        replication_thread.join()
        done.go()
        please_stop.go()

class TextLog_usingThread(TextLog):

    def __init__(self, logger):
        if not _Log:
            _delayed_imports()

        self.queue = _Queue("logs", max=10000, silent=True)
        self.logger = logger

        def worker(please_stop):
            while not please_stop:
                Thread.sleep(1)
                logs = self.queue.pop_all()
                for log in logs:
                    if log is Thread.STOP:
                        if DEBUG_LOGGING:
                            sys.stdout.write("TextLog_usingThread.worker() sees stop, filling rest of queue\n")
                        please_stop.go()
                    else:
                        self.logger.write(**log)

        self.thread = Thread("log thread", worker)
        self.thread.parent.remove_child(self.thread)  # LOGGING WILL BE RESPONSIBLE FOR THREAD stop()
        self.thread.start()

    def write(self, template, params):
        try:
            self.queue.add({"template": template, "params": params})
            return self
        except Exception, e:
            e = _Except.wrap(e)
            sys.stdout.write("IF YOU SEE THIS, IT IS LIKELY YOU FORGOT TO RUN Log.start() FIRST\n")
            raise e  # OH NO!

    def _worker(self, please_stop):
        curr = "0.0"
        acc = []
        last_count_written = -1
        next_write = Date.now()

        while not please_stop:
            d = self.temp_queue.pop(timeout=MINUTE)
            if d == None:
                if not acc:
                    continue
                # WRITE THE INCOMPLETE DATA TO S3, BUT NOT TOO OFTEN
                next_write = Date.now() + MINUTE
                try:
                    if last_count_written != len(acc):
                        if DEBUG:
                            Log.note("write incomplete data ({{num}} lines) to {{uid}} in S3 next (time = {{next_write}})", uid=curr, next_write=next_write, num=len(acc))
                        self.bucket.write_lines(curr, (convert.value2json(a) for a in acc))
                        last_count_written = len(acc)
                except Exception, e:
                    Log.note("Problem with write to S3", cause=e)
            elif d[UID_PATH] != curr:
                # WRITE acc TO S3 IF WE ARE MOVING TO A NEW KEY
                try:
                    if acc:
                        if DEBUG:
                            Log.note("write complete data ({{num}} lines) to {{curr}} in S3", num=len(acc), curr=curr)
                        self.bucket.write_lines(curr, (convert.value2json(a) for a in acc))
                        last_count_written = 0
                    curr = d[UID_PATH]
                    acc = [d]
                except Exception, e:
                    Log.warning("Can not store data", cause=e)
                    Thread.sleep(30*MINUTE)

def main():

    try:
        settings = startup.read_settings(defs=[{
            "name": ["--id"],
            "help": "id(s) to process.  Use \"..\" for a range.",
            "type": str,
            "dest": "id",
            "required": False
        }])
        constants.set(settings.constants)
        Log.start(settings.debug)

        if settings.args.id:
            etl_one(settings)
            return

        hg = HgMozillaOrg(settings=settings.hg)
        resources = Dict(hg=dictwrap(hg))
        stopper = Signal()
        for i in range(coalesce(settings.param.threads, 1)):
            ETL(
                name="ETL Loop " + unicode(i),
                work_queue=settings.work_queue,
                resources=resources,
                workers=settings.workers,
                settings=settings.param,
                please_stop=stopper
            )

        Thread.wait_for_shutdown_signal(stopper, allow_exit=True)
    except Exception, e:
        Log.error("Problem with etl", e)

class Log_usingThread(BaseLog):

    def __init__(self, logger):
        # DELAYED LOAD FOR THREADS MODULE
        from pyLibrary.thread.threads import Queue

        self.queue = Queue("logs", max=10000, silent=True)
        self.logger = logger

        def worker(please_stop):
            while not please_stop:
                Thread.sleep(1)
                logs = self.queue.pop_all()
                for log in logs:
                    if log is Thread.STOP:
                        if DEBUG_LOGGING:
                            sys.stdout.write("Log_usingThread.worker() sees stop, filling rest of queue\n")
                        please_stop.go()
                    else:
                        self.logger.write(**log)

        self.thread = Thread("log thread", worker)
        self.thread.start()

    def write(self, template, params):
        try:
            self.queue.add({"template": template, "params": params})
            return self
        except Exception, e:
            sys.stdout.write("IF YOU SEE THIS, IT IS LIKELY YOU FORGOT TO RUN Log.start() FIRST\n")
            raise e  # OH NO!

    def __init__(
        self,
        exchange,  # name of the Pulse exchange
        topic,  # message name pattern to subscribe to  ('#' is wildcard)
        target=None,  # WILL BE CALLED WITH PULSE PAYLOADS AND ack() IF COMPLETE$ED WITHOUT EXCEPTION
        target_queue=None,  # (aka self.queue) WILL BE FILLED WITH PULSE PAYLOADS
        host='pulse.mozilla.org',  # url to connect,
        port=5671,  # tcp port
        user=None,
        password=None,
        vhost="/",
        start=0,  # USED AS STARTING POINT FOR ASSIGNING THE _meta.count ATTRIBUTE
        ssl=True,
        applabel=None,
        heartbeat=False,  # True to also get the Pulse heartbeat message
        durable=False,  # True to keep queue after shutdown
        serializer='json',
        broker_timezone='GMT',
        settings=None
    ):
        self.target_queue = target_queue
        self.pulse_target = target
        if (target_queue == None and target == None) or (target_queue != None and target != None):
            Log.error("Expecting a queue (for fast digesters) or a target (for slow digesters)")

        Thread.__init__(self, name="Pulse consumer for " + settings.exchange, target=self._worker)
        self.settings = settings
        settings.callback = self._got_result
        settings.user = coalesce(settings.user, settings.username)
        settings.applabel = coalesce(settings.applable, settings.queue, settings.queue_name)
        settings.topic = topic

        self.pulse = ModifiedGenericConsumer(settings, connect=True, **settings)
        self.count = coalesce(start, 0)
        self.start()

    def __init__(self, host, index, alias=None, name=None, port=9200, settings=None):
        global _elasticsearch
        if hasattr(self, "settings"):
            return

        from pyLibrary.queries.containers.lists import ListContainer
        from pyLibrary.env import elasticsearch as _elasticsearch

        self.settings = settings
        self.default_name = coalesce(name, alias, index)
        self.default_es = _elasticsearch.Cluster(settings=settings)
        self.todo = Queue("refresh metadata", max=100000, unique=True)

        self.meta=Dict()
        table_columns = metadata_tables()
        column_columns = metadata_columns()
        self.meta.tables = ListContainer("meta.tables", [], wrap({c.name: c for c in table_columns}))
        self.meta.columns = ListContainer("meta.columns", [], wrap({c.name: c for c in column_columns}))
        self.meta.columns.insert(column_columns)
        self.meta.columns.insert(table_columns)
        # TODO: fix monitor so it does not bring down ES
        if ENABLE_META_SCAN:
            self.worker = Thread.run("refresh metadata", self.monitor)
        else:
            self.worker = Thread.run("refresh metadata", self.not_monitor)
        return

def main():
    try:
        settings = startup.read_settings()
        Log.start(settings.debug)
        constants.set(settings.constants)

        with startup.SingleInstance(flavor_id=settings.args.filename):
            with aws.s3.Bucket(settings.destination) as bucket:

                if settings.param.debug:
                    if settings.source.durable:
                        Log.error("Can not run in debug mode with a durable queue")
                    synch = SynchState(bucket.get_key(SYNCHRONIZATION_KEY, must_exist=False))
                else:
                    synch = SynchState(bucket.get_key(SYNCHRONIZATION_KEY, must_exist=False))
                    if settings.source.durable:
                        synch.startup()

                queue = PersistentQueue(settings.param.queue_file)
                if queue:
                    last_item = queue[len(queue) - 1]
                    synch.source_key = last_item._meta.count + 1

                with pulse.Consumer(settings=settings.source, target=None, target_queue=queue, start=synch.source_key):
                    Thread.run("pulse log loop", log_loop, settings, synch, queue, bucket)
                    Thread.wait_for_shutdown_signal(allow_exit=True)
                    Log.warning("starting shutdown")

                queue.close()
                Log.note("write shutdown state to S3")
                synch.shutdown()

    except Exception, e:
        Log.error("Problem with etl", e)

    def _insert_loop(self, please_stop=None):
        bad_count = 0
        while not please_stop:
            try:
                Thread.sleep(seconds=1)
                messages = wrap(self.queue.pop_all())
                if not messages:
                    continue

                for g, mm in jx.groupby(messages, size=self.batch_size):
                    scrubbed = []
                    try:
                        for i, message in enumerate(mm):
                            if message is Thread.STOP:
                                please_stop.go()
                                return
                            scrubbed.append(_deep_json_to_string(message, depth=3))
                    finally:
                        self.es.extend(scrubbed)
                    bad_count = 0
            except Exception, e:
                Log.warning("Problem inserting logs into ES", cause=e)
                bad_count += 1
                if bad_count > MAX_BAD_COUNT:
                    Log.warning("Given up trying to write debug logs to ES index {{index}}", index=self.es.settings.index)
                Thread.sleep(seconds=30)

class Log_usingLogger(BaseLog):
    def __init__(self, settings):
        self.logger = logging.Logger("unique name", level=logging.INFO)
        self.logger.addHandler(make_log_from_settings(settings))

        # TURNS OUT LOGGERS ARE REALLY SLOW TOO
        self.queue = threads.Queue("log to classic logger", max=10000, silent=True)
        self.thread = Thread("log to logger", time_delta_pusher, appender=self.logger.info, queue=self.queue, interval=timedelta(seconds=0.3))
        self.thread.start()

    def write(self, template, params):
        # http://docs.python.org/2/library/logging.html# logging.LogRecord
        self.queue.add({"template": template, "params": params})

    def stop(self):
        try:
            if DEBUG_LOGGING:
                sys.stdout.write("Log_usingLogger sees stop, adding stop to queue\n")
            self.queue.add(Thread.STOP)  # BE PATIENT, LET REST OF MESSAGE BE SENT
            self.thread.join()
            if DEBUG_LOGGING:
                sys.stdout.write("Log_usingLogger done\n")
        except Exception, e:
            pass

        try:
            self.queue.close()
        except Exception, f:
            pass

    def get_columns(self, table_name, column_name=None, force=False):
        """
        RETURN METADATA COLUMNS
        """
        try:
            # LAST TIME WE GOT INFO FOR THIS TABLE
            short_name = join_field(split_field(table_name)[0:1])
            table = self.get_table(short_name)[0]

            if not table:
                table = Table(
                    name=short_name,
                    url=None,
                    query_path=None,
                    timestamp=Date.now()
                )
                with self.meta.tables.locker:
                    self.meta.tables.add(table)
                self._get_columns(table=short_name)
            elif force or table.timestamp == None or table.timestamp < Date.now() - MAX_COLUMN_METADATA_AGE:
                table.timestamp = Date.now()
                self._get_columns(table=short_name)

            with self.meta.columns.locker:
                columns = self.meta.columns.find(table_name, column_name)
            if columns:
                columns = jx.sort(columns, "name")
                # AT LEAST WAIT FOR THE COLUMNS TO UPDATE
                while len(self.todo) and not all(columns.get("last_updated")):
                    Log.note("waiting for columns to update {{columns|json}}", columns=[c.table+"."+c.es_column for c in columns if not c.last_updated])
                    Thread.sleep(seconds=1)
                return columns
        except Exception, e:
            Log.error("Not expected", cause=e)

 def __init__(self, name, in_queue, out_queue, function):
     Thread.__init__(self, name, self.event_loop)
     self.in_queue = in_queue
     self.out_queue = out_queue
     self.function = function
     self.num_runs = 0
     self.start()

def wait_for_queue(work_queue):
    """
    SLEEP UNTIL WORK QUEU IS EMPTY ENOUGH FOR MORE
    """
    # return
    while True:
        if len(work_queue) < MAX_QUEUE_SIZE:
            break
        Log.note("sleep for 5min")
        Thread.sleep(seconds=5 * 60)

 def worker(please_stop):
     while not please_stop:
         try:
             response = requests.get("http://169.254.169.254/latest/meta-data/spot/termination-time")
             if response.status_code != 400:
                 please_stop.go()
                 return
         except Exception, e:
             Thread.sleep(seconds=61, please_stop=please_stop)
         Thread.sleep(seconds=11, please_stop=please_stop)

    def create_index(
        self,
        index,
        alias=None,
        create_timestamp=None,
        schema=None,
        limit_replicas=None,
        read_only=False,
        tjson=False,
        settings=None
    ):
        if not alias:
            alias = settings.alias = settings.index
            index = settings.index = proto_name(alias, create_timestamp)

        if settings.alias == index:
            Log.error("Expecting index name to conform to pattern")

        if settings.schema_file:
            Log.error('schema_file attribute not supported.  Use {"$ref":<filename>} instead')

        if schema == None:
            Log.error("Expecting a schema")
        elif isinstance(schema, basestring):
            schema = convert.json2value(schema, leaves=True)
        else:
            schema = convert.json2value(convert.value2json(schema), leaves=True)

        if limit_replicas:
            # DO NOT ASK FOR TOO MANY REPLICAS
            health = self.get("/_cluster/health")
            if schema.settings.index.number_of_replicas >= health.number_of_nodes:
                Log.warning("Reduced number of replicas: {{from}} requested, {{to}} realized",
                    {"from": schema.settings.index.number_of_replicas},
                    to= health.number_of_nodes - 1
                )
                schema.settings.index.number_of_replicas = health.number_of_nodes - 1

        self.post(
            "/" + index,
            data=schema,
            headers={"Content-Type": "application/json"}
        )

        # CONFIRM INDEX EXISTS
        while True:
            try:
                state = self.get("/_cluster/state", retry={"times": 5}, timeout=3)
                if index in state.metadata.indices:
                    break
                Log.note("Waiting for index {{index}} to appear", index=index)
            except Exception, e:
                Log.warning("Problem while waiting for index {{index}} to appear", index=index, cause=e)
            Thread.sleep(seconds=1)