目录
第二部分:代理商和团队
我们在这里做什么?
所以,第二部分。如前所述,我们将执行以下操作:
让我们在aiohttp上为alphavantage编写一个小型客户端,其中包含对所需端点的请求。
我们将建立一个代理商,负责收集证券数据和元数据。
, , , faust , , kafka, ( click), - , .
AlphaVantage
, aiohttp alphavantage.
Spoiler
import urllib.parse as urlparse
from io import StringIO
from typing import Any, Dict, List, Union
import aiohttp
import pandas as pd
import stringcase
from loguru import logger
from horton.config import API_ENDPOINT
class AlphaVantageClient:
def __init__(
self,
session: aiohttp.ClientSession,
api_key: str,
api_endpoint: str = API_ENDPOINT,
):
self._query_params = {"datatype": "json", "apikey": api_key}
self._api_endpoint = api_endpoint
self._session = session
@logger.catch
def _format_fields(self, data: Dict[str, Any]) -> Dict[str, Any]:
formatted_data = {}
for field, item in data.items():
formatted_data[stringcase.snakecase(field)] = item
return formatted_data
@logger.catch
async def _construct_query(
self, function: str, to_json: bool = True, **kwargs
) -> Union[Dict[str, Any], str]:
path = "query/"
async with self._session.get(
urlparse.urljoin(self._api_endpoint, path),
params={"function": function, **kwargs, **self._query_params},
) as response:
data = (await response.json()) if to_json else (await response.text())
if to_json:
data = self._format_fields(data)
return data
@logger.catch
async def get_securities(self, state: str = "active") -> List[Dict[str, str]]:
data = await self._construct_query("LISTING_STATUS", state=state, to_json=False)
data = pd.read_csv(StringIO(data))
securities = data.to_dict("records")
for index, security in enumerate(securities):
security = self._format_fields(security)
security["_type"] = "physical"
securities[index] = security
return securities
@logger.catch
async def get_security_overview(self, symbol: str) -> Dict[str, str]:
return await self._construct_query("OVERVIEW", symbol=symbol)
@logger.catch
async def get_historical_data(self, symbol: str) -> Dict[str, Any]:
return await self._construct_query(
"TIME_SERIES_DAILY_ADJUSTED", symbol=symbol, outputsize="full"
)
@logger.catch
async def get_last_price_data(self, symbol: str) -> Dict[str, Any]:
return await self._construct_query("GLOBAL_QUOTE", symbol=symbol)
@logger.catch
async def get_indicator_data(
self, symbol: str, indicator: str, **indicator_options
) -> Dict[str, Any]:
return await self._construct_query(
indicator, symbol=symbol, **indicator_options
)
:
API AlphaVantage ,
construct_query
http .snake_case
.logger.catch .
P.S. alphavantage config.yml, HORTON_SERVICE_APIKEY
. .
CRUD-
securities .
- , .
get_app()
Spoiler
import faust
from horton.config import KAFKA_BROKERS
def get_app():
return faust.App("horton", broker=KAFKA_BROKERS)
app = get_app()
collect_securities_topic = app.topic("collect_securities", internal=True)
@app.agent(collect_securities_topic)
async def collect_securities(stream: StreamT[None]) -> AsyncIterable[bool]:
pass
, faust- - . , ... , , internal -.
kafka, , . , , :)
internal, faust, , , , faust, : retention, retention policy (- delete, compact), - (partitions, , , faust).
, , , . , (, - retention policy) .
:
app = get_app()
@app.agent()
async def collect_securities(stream: StreamT[None]) -> AsyncIterable[bool]:
pass
, , :)
app = get_app()
collect_securities_topic = app.topic("collect_securities", internal=True)
@app.agent(collect_securities_topic)
async def collect_securities(stream: StreamT[None]) -> AsyncIterable[bool]:
async with aiohttp.ClientSession() as session:
async for _ in stream:
logger.info("Start collect securities")
client = AlphaVantageClient(session, API_KEY)
securities = await client.get_securities()
for security in securities:
await SecurityCRUD.update_one(
{"symbol": security["symbol"], "exchange": security["exchange"]}, security, upsert=True
)
yield True
, aiohttp . , , , - , ( , concurrency ).
, ( _
, , , ) , (offset), , . , , (get_securities - active, . ) , , , , () .
!
> docker-compose up -d
... ...
> faust -A horton.agents worker --without-web -l info
P.S. - faust , .
faust', ( ) info. :
Spoiler
┌ƒaµS† v1.10.4┬───────────────────────────────────────────────────┐
│ id │ horton │
│ transport │ [URL('kafka://localhost:9092')] │
│ store │ memory: │
│ log │ -stderr- (info) │
│ pid │ 1271262 │
│ hostname │ host-name │
│ platform │ CPython 3.8.2 (Linux x86_64) │
│ drivers │ │
│ transport │ aiokafka=1.1.6 │
│ web │ aiohttp=3.6.2 │
│ datadir │ /path/to/project/horton-data │
│ appdir │ /path/to/project/horton-data/v1 │
└─────────────┴───────────────────────────────────────────────────┘
... , , ...
┌Topic Partition Set─────────┬────────────┐
│ topic │ partitions │
├────────────────────────────┼────────────┤
│ collect_securities │ {0-7} │
│ horton-__assignor-__leader │ {0} │
└────────────────────────────┴────────────┘
!!!
partition set. , , , - (8, topic_partitions - ), ( partitions) . 8 , , .
, :
> faust -A horton.agents send @collect_securities
{"topic": "collect_securities", "partition": 6, "topic_partition": ["collect_securities", 6], "offset": 0, "timestamp": ..., "timestamp_type": 0}
P.S. @
, "collect_securities".
, 6 - , kafdrop localhost:9000
, , loguru:
2020-09-23 00:26:37.304 | INFO | horton.agents:collect_securities:40 - Start collect securities
, mongo ( Robo3T Studio3T) , :
, , .
- :)
, !
, , 1/3 , , , .
, , :
collect_security_overview_topic = app.topic("collect_security_overview", internal=True)
@app.agent(collect_security_overview_topic)
async def collect_security_overview(
stream: StreamT[?],
) -> AsyncIterable[bool]:
async with aiohttp.ClientSession() as session:
async for event in stream:
...
security, (symbol) . faust Records - , .
records.py , :
import faust
class CollectSecurityOverview(faust.Record):
symbol: str
exchange: str
, faust python, , - 3.6.
, :
collect_security_overview_topic = app.topic(
"collect_security_overview", internal=True, value_type=CollectSecurityOverview
)
@app.agent(collect_security_overview_topic)
async def collect_security_overview(
stream: StreamT[CollectSecurityOverview],
) -> AsyncIterable[bool]:
async with aiohttp.ClientSession() as session:
async for event in stream:
logger.info(
"Start collect security [{symbol}] overview", symbol=event.symbol
)
client = AlphaVantageClient(session, API_KEY)
security_overview = await client.get_security_overview(event.symbol)
await SecurityCRUD.update_one({"symbol": event.symbol, "exchange": event.exchange}, security_overview)
yield True
, - value_type. , , - .
, - collect_securitites :
....
for security in securities:
await SecurityCRUD.update_one({
"symbol": security["symbol"],
"exchange": security["exchange"]
},
security,
upsert = True,
)
await collect_security_overview.cast(
CollectSecurityOverview(symbol = security["symbol"], exchange = security["exchange"])
)
....
cast - , . .
send - , . .
ask - . .
, !
, - . , faust - click. faust -A
agents.py app.command, cast collect_securitites:
@app.command()
async def start_collect_securities():
"""Collect securities and overview."""
await collect_securities.cast()
, , :
> faust -A horton.agents --help
....
Commands:
agents List agents.
clean-versions Delete old version directories.
completion Output shell completion to be evaluated by the...
livecheck Manage LiveCheck instances.
model Show model detail.
models List all available models as a tabulated list.
reset Delete local table state.
send Send message to agent/topic.
start-collect-securities Collect securities and overview.
tables List available tables.
worker Start worker instance for given app.
, , faust :
> faust -A horton.agents start-collect-securities
?
, , , sink cron- .
! :)
PS在最后一部分下,我被问到浮士德和融合卡夫卡(具有融合的特征)。似乎confluent的功能要强大得多,但要点是,浮士德时代没有对confluent的完整客户端支持-这是根据码头中对客户端限制的描述得出的。