mirror of
https://github.com/freqtrade/freqtrade.git
synced 2024-11-14 04:03:55 +00:00
4d1613a432
Implementing a generator to split Lists into chunks.
216 lines
6.4 KiB
Python
216 lines
6.4 KiB
Python
"""
|
|
Various tool function for Freqtrade and scripts
|
|
"""
|
|
import gzip
|
|
import logging
|
|
import re
|
|
from datetime import datetime
|
|
from pathlib import Path
|
|
from typing import Any, Iterator, List
|
|
from typing.io import IO
|
|
|
|
import rapidjson
|
|
|
|
from freqtrade.constants import DECIMAL_PER_COIN_FALLBACK, DECIMALS_PER_COIN
|
|
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
def decimals_per_coin(coin: str):
|
|
"""
|
|
Helper method getting decimal amount for this coin
|
|
example usage: f".{decimals_per_coin('USD')}f"
|
|
:param coin: Which coin are we printing the price / value for
|
|
"""
|
|
return DECIMALS_PER_COIN.get(coin, DECIMAL_PER_COIN_FALLBACK)
|
|
|
|
|
|
def round_coin_value(value: float, coin: str, show_coin_name=True) -> str:
|
|
"""
|
|
Get price value for this coin
|
|
:param value: Value to be printed
|
|
:param coin: Which coin are we printing the price / value for
|
|
:param show_coin_name: Return string in format: "222.22 USDT" or "222.22"
|
|
:return: Formatted / rounded value (with or without coin name)
|
|
"""
|
|
if show_coin_name:
|
|
return f"{value:.{decimals_per_coin(coin)}f} {coin}"
|
|
else:
|
|
return f"{value:.{decimals_per_coin(coin)}f}"
|
|
|
|
|
|
def shorten_date(_date: str) -> str:
|
|
"""
|
|
Trim the date so it fits on small screens
|
|
"""
|
|
new_date = re.sub('seconds?', 'sec', _date)
|
|
new_date = re.sub('minutes?', 'min', new_date)
|
|
new_date = re.sub('hours?', 'h', new_date)
|
|
new_date = re.sub('days?', 'd', new_date)
|
|
new_date = re.sub('^an?', '1', new_date)
|
|
return new_date
|
|
|
|
|
|
def file_dump_json(filename: Path, data: Any, is_zip: bool = False, log: bool = True) -> None:
|
|
"""
|
|
Dump JSON data into a file
|
|
:param filename: file to create
|
|
:param data: JSON Data to save
|
|
:return:
|
|
"""
|
|
|
|
if is_zip:
|
|
if filename.suffix != '.gz':
|
|
filename = filename.with_suffix('.gz')
|
|
if log:
|
|
logger.info(f'dumping json to "{filename}"')
|
|
|
|
with gzip.open(filename, 'w') as fpz:
|
|
rapidjson.dump(data, fpz, default=str, number_mode=rapidjson.NM_NATIVE)
|
|
else:
|
|
if log:
|
|
logger.info(f'dumping json to "{filename}"')
|
|
with open(filename, 'w') as fp:
|
|
rapidjson.dump(data, fp, default=str, number_mode=rapidjson.NM_NATIVE)
|
|
|
|
logger.debug(f'done json to "{filename}"')
|
|
|
|
|
|
def json_load(datafile: IO) -> Any:
|
|
"""
|
|
load data with rapidjson
|
|
Use this to have a consistent experience,
|
|
set number_mode to "NM_NATIVE" for greatest speed
|
|
"""
|
|
return rapidjson.load(datafile, number_mode=rapidjson.NM_NATIVE)
|
|
|
|
|
|
def file_load_json(file):
|
|
|
|
if file.suffix != ".gz":
|
|
gzipfile = file.with_suffix(file.suffix + '.gz')
|
|
else:
|
|
gzipfile = file
|
|
# Try gzip file first, otherwise regular json file.
|
|
if gzipfile.is_file():
|
|
logger.debug(f"Loading historical data from file {gzipfile}")
|
|
with gzip.open(gzipfile) as datafile:
|
|
pairdata = json_load(datafile)
|
|
elif file.is_file():
|
|
logger.debug(f"Loading historical data from file {file}")
|
|
with open(file) as datafile:
|
|
pairdata = json_load(datafile)
|
|
else:
|
|
return None
|
|
return pairdata
|
|
|
|
|
|
def pair_to_filename(pair: str) -> str:
|
|
for ch in ['/', '-', ' ', '.', '@', '$', '+', ':']:
|
|
pair = pair.replace(ch, '_')
|
|
return pair
|
|
|
|
|
|
def format_ms_time(date: int) -> str:
|
|
"""
|
|
convert MS date to readable format.
|
|
: epoch-string in ms
|
|
"""
|
|
return datetime.fromtimestamp(date/1000.0).strftime('%Y-%m-%dT%H:%M:%S')
|
|
|
|
|
|
def deep_merge_dicts(source, destination):
|
|
"""
|
|
Values from Source override destination, destination is returned (and modified!!)
|
|
Sample:
|
|
>>> a = { 'first' : { 'rows' : { 'pass' : 'dog', 'number' : '1' } } }
|
|
>>> b = { 'first' : { 'rows' : { 'fail' : 'cat', 'number' : '5' } } }
|
|
>>> merge(b, a) == { 'first' : { 'rows' : { 'pass' : 'dog', 'fail' : 'cat', 'number' : '5' } } }
|
|
True
|
|
"""
|
|
for key, value in source.items():
|
|
if isinstance(value, dict):
|
|
# get node or create one
|
|
node = destination.setdefault(key, {})
|
|
deep_merge_dicts(value, node)
|
|
else:
|
|
destination[key] = value
|
|
|
|
return destination
|
|
|
|
|
|
def round_dict(d, n):
|
|
"""
|
|
Rounds float values in the dict to n digits after the decimal point.
|
|
"""
|
|
return {k: (round(v, n) if isinstance(v, float) else v) for k, v in d.items()}
|
|
|
|
|
|
def safe_value_fallback(obj: dict, key1: str, key2: str, default_value=None):
|
|
"""
|
|
Search a value in obj, return this if it's not None.
|
|
Then search key2 in obj - return that if it's not none - then use default_value.
|
|
Else falls back to None.
|
|
"""
|
|
if key1 in obj and obj[key1] is not None:
|
|
return obj[key1]
|
|
else:
|
|
if key2 in obj and obj[key2] is not None:
|
|
return obj[key2]
|
|
return default_value
|
|
|
|
|
|
def safe_value_fallback2(dict1: dict, dict2: dict, key1: str, key2: str, default_value=None):
|
|
"""
|
|
Search a value in dict1, return this if it's not None.
|
|
Fall back to dict2 - return key2 from dict2 if it's not None.
|
|
Else falls back to None.
|
|
|
|
"""
|
|
if key1 in dict1 and dict1[key1] is not None:
|
|
return dict1[key1]
|
|
else:
|
|
if key2 in dict2 and dict2[key2] is not None:
|
|
return dict2[key2]
|
|
return default_value
|
|
|
|
|
|
def plural(num: float, singular: str, plural: str = None) -> str:
|
|
return singular if (num == 1 or num == -1) else plural or singular + 's'
|
|
|
|
|
|
def render_template(templatefile: str, arguments: dict = {}) -> str:
|
|
|
|
from jinja2 import Environment, PackageLoader, select_autoescape
|
|
|
|
env = Environment(
|
|
loader=PackageLoader('freqtrade', 'templates'),
|
|
autoescape=select_autoescape(['html', 'xml'])
|
|
)
|
|
template = env.get_template(templatefile)
|
|
return template.render(**arguments)
|
|
|
|
|
|
def render_template_with_fallback(templatefile: str, templatefallbackfile: str,
|
|
arguments: dict = {}) -> str:
|
|
"""
|
|
Use templatefile if possible, otherwise fall back to templatefallbackfile
|
|
"""
|
|
from jinja2.exceptions import TemplateNotFound
|
|
try:
|
|
return render_template(templatefile, arguments)
|
|
except TemplateNotFound:
|
|
return render_template(templatefallbackfile, arguments)
|
|
|
|
|
|
def chunks(lst: List[Any], n: int) -> Iterator[List[Any]]:
|
|
"""
|
|
Split lst into chunks of the size n.
|
|
:param lst: list to split into chunks
|
|
:param n: number of max elements per chunk
|
|
:return: None
|
|
"""
|
|
for chunk in range(0, len(lst), n):
|
|
yield (lst[chunk:chunk + n])
|