def import_orders(
ctx: Context,
admin_url: str,
) -> Iterator[Records[ShopifyOrder]]:
_, _, shop_name = split_admin_url(admin_url)
url, auth = url_and_auth_from_admin_url(admin_url)
endpoint_url = url + "/orders.json"
latest_updated_at = ctx.get_state_value(
"latest_updated_at") or DEFAULT_MIN_DATE
params = {
"order": "updated_at asc",
"updated_at_min": latest_updated_at,
"status": "any",
"limit": 250,
}
conn = JsonHttpApiConnection()
while ctx.should_continue():
resp = conn.get(endpoint_url, params, auth=auth)
json_resp = resp.json()
assert isinstance(json_resp, dict)
records = json_resp["orders"]
if len(records) == 0:
# All done
break
new_latest_updated_at = max([o["updated_at"] for o in records])
ctx.emit_state({"latest_updated_at": new_latest_updated_at})
yield records
# Shopify has cursor-based pagination now, so we can safely paginate results
next_page = get_next_page_link(resp.headers)
if not next_page:
# No more pages
break
endpoint_url = next_page
params = {}
def marketstack_import_eod_prices(
ctx: Context,
tickers_input: Optional[Reference[Ticker]],
access_key: str,
from_date: date = MIN_DATE,
tickers: Optional[List] = None,
) -> Iterator[Records[EodPrice]]:
# access_key = ctx.get_param("access_key")
use_https = False # TODO: when do we want this True?
default_from_date = from_date
assert access_key is not None
if tickers_input is not None:
tickers = list(tickers_input.as_dataframe()["symbol"])
if not tickers:
return
ticker_latest_dates_imported = (
ctx.get_state_value("ticker_latest_dates_imported") or {})
conn = JsonHttpApiConnection(date_format="%Y-%m-%d")
if use_https:
endpoint_url = HTTPS_MARKETSTACK_API_BASE_URL + "eod"
else:
endpoint_url = MARKETSTACK_API_BASE_URL + "eod"
for ticker in tickers:
assert isinstance(ticker, str)
latest_date_imported = ensure_date(
ticker_latest_dates_imported.get(ticker, default_from_date))
max_date = latest_date_imported
params = {
"limit": 1000,
"offset": 0,
"access_key": access_key,
"symbols": ticker,
"date_from": latest_date_imported,
}
while ctx.should_continue():
resp = conn.get(endpoint_url, params)
json_resp = resp.json()
assert isinstance(json_resp, dict)
records = json_resp["data"]
if len(records) == 0:
# All done
break
yield records
# Update state
max_date = max(max_date,
max(ensure_date(r["date"]) for r in records))
ticker_latest_dates_imported[ticker] = max_date + timedelta(days=1)
ctx.emit_state_value("ticker_latest_dates_imported",
ticker_latest_dates_imported)
# Setup for next page
params["offset"] = params["offset"] + len(records)
def alphavantage_import_eod_prices(
ctx: Context,
tickers_input: Optional[Reference[Ticker]],
api_key: str,
tickers: Optional[List] = None,
) -> Iterator[Records[AlphavantageEodPrice]]:
assert api_key is not None
tickers = prepare_tickers(tickers, tickers_input)
if not tickers:
return None
ticker_latest_dates_imported = (
ctx.get_state_value("ticker_latest_dates_imported") or {})
conn = JsonHttpApiConnection()
def fetch_prices(params: Dict, tries: int = 0) -> Optional[Records]:
if tries > 2:
return None
resp = conn.get(ALPHAVANTAGE_API_BASE_URL, params, stream=True)
try:
record = resp.json()
# Json response means error
if is_alphavantage_error(record):
# TODO: Log this failure?
print(f"Error for {params} {record}")
return None
if is_alphavantage_rate_limit(record):
time.sleep(60)
return fetch_prices(params, tries=tries + 1)
except:
pass
# print(resp.raw.read().decode("utf8"))
# resp.raw.seek(0)
records = list(read_csv(resp.iter_lines()))
return records
for ticker in tickers:
assert isinstance(ticker, str)
latest_date_imported = ensure_datetime(
ticker_latest_dates_imported.get(ticker, MIN_DATETIME))
assert latest_date_imported is not None
if utcnow() - ensure_utc(latest_date_imported) < timedelta(days=1):
# Only check once a day
continue
params = prepare_params_for_ticker(ticker,
ticker_latest_dates_imported)
params["apikey"] = api_key
records = fetch_prices(params)
if records:
# Symbol not included
for r in records:
r["symbol"] = ticker
yield records
# Update state
ticker_latest_dates_imported[ticker] = utcnow()
ctx.emit_state_value("ticker_latest_dates_imported",
ticker_latest_dates_imported)
if not ctx.should_continue():
break
def import_observations(
ctx: Context,
api_key: str,
series_id: str,
) -> Iterator[Records[FredObservation]]:
latest_fetched_at = ctx.get_state_value("latest_fetched_at")
if latest_fetched_at:
# Two year curing window (to be safe)
obs_start = ensure_datetime(latest_fetched_at) - timedelta(days=365 *
2)
else:
obs_start = MIN_DATE
params = {
"file_type": "json",
"api_key": api_key,
"series_id": series_id,
"observation_start": obs_start,
"offset": 0,
"limit": 100000,
}
conn = JsonHttpApiConnection(date_format="%Y-%m-%d")
endpoint_url = FRED_API_BASE_URL + "series/observations"
while True:
resp = conn.get(endpoint_url, params)
json_resp = resp.json()
assert isinstance(json_resp, dict)
records = json_resp["observations"]
if len(records) == 0:
# All done
break
for r in records:
r["series_id"] = params[
"series_id"] # Add series ID to data so we know what the data is
r["value"] = (
None if r["value"] == "." else r["value"]
) # FRED quirk, returns empty decimal number "." instead of null
yield records
num_returned = len(records)
if num_returned < json_resp["limit"]:
# we got back less than limit, so must be done (no other way to tell?)
break
params["offset"] += num_returned
# We only update date if we have fetched EVERYTHING available as of now
ctx.emit_state_value("latest_fetched_at", utcnow())
def marketstack_import_tickers(
ctx: Context,
access_key: str,
exchanges: List = ["XNYS", "XNAS"],
) -> Iterator[Records[MarketstackTicker]]:
use_https = False # TODO: when do we want this True?
# default_from_date = ctx.get_param("from_date", MIN_DATE)
assert access_key is not None
assert isinstance(exchanges, list)
last_imported_at = ensure_datetime(
ctx.get_state_value("last_imported_at") or "2020-01-01 00:00:00")
assert last_imported_at is not None
last_imported_at = ensure_utc(last_imported_at)
if utcnow() - last_imported_at < timedelta(days=1): # TODO: from config
return
conn = JsonHttpApiConnection()
if use_https:
endpoint_url = HTTPS_MARKETSTACK_API_BASE_URL + "tickers"
else:
endpoint_url = MARKETSTACK_API_BASE_URL + "tickers"
for exchange in exchanges:
params = {
"limit": 1000,
"offset": 0,
"access_key": access_key,
"exchange": exchange,
}
while ctx.should_continue():
resp = conn.get(endpoint_url, params)
json_resp = resp.json()
assert isinstance(json_resp, dict)
records = json_resp["data"]
if len(records) == 0:
# All done
break
# Add a flattened exchange indicator
for r in records:
r["exchange_acronym"] = r.get("stock_exchange",
{}).get("acronym")
yield records
# Setup for next page
params["offset"] = params["offset"] + len(records)
ctx.emit_state_value("last_imported_at", utcnow())
def alphavantage_import_company_overview(
ctx: Context,
tickers_input: Optional[Reference[Ticker]],
api_key: str,
tickers: Optional[List] = None,
) -> Iterator[Records[AlphavantageCompanyOverview]]:
assert api_key is not None
tickers = prepare_tickers(tickers, tickers_input)
if tickers is None:
# We didn't get an input block for tickers AND
# the config is empty, so we are done
return None
ticker_latest_dates_imported = (
ctx.get_state_value("ticker_latest_dates_imported") or {})
conn = JsonHttpApiConnection()
batch_size = 50
records = []
tickers_updated = []
def fetch_overview(params: Dict, tries: int = 0) -> Optional[Dict]:
if tries > 2:
return None
resp = conn.get(ALPHAVANTAGE_API_BASE_URL, params)
record = resp.json()
# Alphavantage returns 200 and json error message on failure
if is_alphavantage_error(record):
# TODO: Log this failure?
# print(f"Error for ticker {params['symbol']}: {record}")
return None
if is_alphavantage_rate_limit(record):
time.sleep(20)
return fetch_overview(params, tries=tries + 1)
return record
for i, ticker in enumerate(tickers):
assert isinstance(ticker, str)
latest_date_imported = ensure_datetime(
ticker_latest_dates_imported.get(ticker, MIN_DATETIME))
assert latest_date_imported is not None
# Refresh at most once a day
# TODO: make this configurable instead of hard-coded 1 day
if utcnow() - ensure_utc(latest_date_imported) < timedelta(days=1):
continue
params = {
"apikey": api_key,
"symbol": ticker,
"function": "OVERVIEW",
}
record = fetch_overview(params)
if not record:
continue
# Clean up json keys to be more DB friendly
record = {title_to_snake_case(k): v for k, v in record.items()}
records.append(record)
tickers_updated.append(ticker)
if len(records) >= batch_size or i == len(tickers) - 1:
yield records
# Update state
for updated_ticker in tickers_updated:
ticker_latest_dates_imported[updated_ticker] = utcnow()
ctx.emit_state_value("ticker_latest_dates_imported",
ticker_latest_dates_imported)
if not ctx.should_continue():
break
records = []
tickers_updated = []
def import_subscription_items(
ctx: Context,
api_key: str,
curing_window_days: int = 90
) -> Iterator[Records[StripeSubscriptionItemRaw]]:
"""
# TODO: repeated code
"""
latest_full_import_at = ctx.get_state_value("latest_full_import_at")
latest_full_import_at = ensure_datetime(latest_full_import_at)
current_starting_after = ctx.get_state_value("current_starting_after")
params = {
"limit": 100,
"status": "all",
}
# if earliest_created_at_imported <= latest_full_import_at - timedelta(days=int(curing_window_days)):
if latest_full_import_at and curing_window_days:
# Import only more recent than latest imported at date, offset by a curing window
# (default 90 days) to capture updates to objects (refunds, etc)
params["created[gte]"] = int(
(latest_full_import_at -
timedelta(days=int(curing_window_days))).timestamp())
if current_starting_after:
params["starting_after"] = current_starting_after
conn = JsonHttpApiConnection()
endpoint_url = STRIPE_API_BASE_URL + "subscriptions"
all_done = False
while ctx.should_continue():
resp = conn.get(endpoint_url, params, auth=HTTPBasicAuth(api_key, ""))
json_resp = resp.json()
assert isinstance(json_resp, dict)
records = json_resp["data"]
if len(records) == 0:
# All done
all_done = True
break
for record in records:
item_params = {
"limit": 100,
"subscription": record["id"],
}
while True:
items_url = STRIPE_API_BASE_URL + "subscription_items"
items_resp = conn.get(items_url,
item_params,
auth=HTTPBasicAuth(api_key, ""))
items_json_resp = items_resp.json()
assert isinstance(items_json_resp, dict)
items = items_json_resp["data"]
if len(items) == 0:
# All done
break
yield items
if not items_json_resp.get("has_more"):
break
latest_item_id = items[-1]["id"]
item_params["starting_after"] = latest_item_id
if not ctx.should_continue():
break
latest_object_id = records[-1]["id"]
if not json_resp.get("has_more"):
all_done = True
break
params["starting_after"] = latest_object_id
ctx.emit_state_value("current_starting_after", latest_object_id)
else:
# Don't update any state, we just timed out
return
# We only update state if we have fetched EVERYTHING available as of now
if all_done:
ctx.emit_state_value("latest_imported_at", utcnow())
# IMPORTANT: we reset the starting after cursor so we start from the beginning again on next run
ctx.emit_state_value("current_starting_after", None)
def stripe_importer(
endpoint: str,
ctx: Context,
api_key: str,
curing_window_days: int = None,
extra_params: Dict = None,
):
"""
Stripe only allows fetching records in one order: from newest to oldest,
so we use its cursor based pagination to iterate once through all historical.
Stripe doesn't have a way to request by "updated at" times, so we must
refresh old records according to our own logic, using a "curing window"
to re-import records up to one year (the default) old.
"""
latest_full_import_at = ctx.get_state_value("latest_full_import_at")
latest_full_import_at = ensure_datetime(latest_full_import_at)
current_starting_after = ctx.get_state_value("current_starting_after")
params = {
"limit": 100,
}
if extra_params:
params.update(extra_params)
# if earliest_created_at_imported <= latest_full_import_at - timedelta(days=int(curing_window_days)):
if latest_full_import_at and curing_window_days:
# Import only more recent than latest imported at date, offset by a curing window
# (default 90 days) to capture updates to objects (refunds, etc)
params["created[gte]"] = int(
(latest_full_import_at -
timedelta(days=int(curing_window_days))).timestamp())
if current_starting_after:
params["starting_after"] = current_starting_after
conn = JsonHttpApiConnection()
endpoint_url = STRIPE_API_BASE_URL + endpoint
all_done = False
while ctx.should_continue():
resp = conn.get(endpoint_url, params, auth=HTTPBasicAuth(api_key, ""))
json_resp = resp.json()
assert isinstance(json_resp, dict)
records = json_resp["data"]
if len(records) == 0:
# All done
all_done = True
break
# Return acutal data
yield records
latest_object_id = records[-1]["id"]
if not json_resp.get("has_more"):
all_done = True
break
params["starting_after"] = latest_object_id
ctx.emit_state_value("current_starting_after", latest_object_id)
else:
# Don't update any state, we just timed out
return
# We only update state if we have fetched EVERYTHING available as of now
if all_done:
ctx.emit_state_value("latest_imported_at", utcnow())
# IMPORTANT: we reset the starting after cursor so we start from the beginning again on next run
ctx.emit_state_value("current_starting_after", None)