dsyoon/DeepLottery
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: final_practice mirrors 3_Practice_22 (txt-based next no, exhaustive predict2)

Browse Source 
Made-with: Cursor
This commit is contained in:
dsyoon
2026-04-08 19:41:04 +09:00
parent b440ec96c9
commit e31eefef09
2 changed files with 87 additions and 175 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
README.md
View File

@@ -15,7 +15,7 @@
- **`filter_model.py`** — `from final_BallFilter import BallFilter` 재노출.
- **`train.py` / `valid.py`** — 구간별로 당첨 6개가 모든 필터를 통과한 회차 수 집계.
- **`final_filterTest.py`** — `1_FilterTest_25.py`와 동일한 분석·(선택) MC 생존 추정.
- **`final_practice.py`** — `3_Practice_22.py`와 같이 다음 회차용 추천 조합 생성(`final_BallFilter`, 기본은 Monte Carlo로 100개 미만).
- **`final_practice.py`** — `3_Practice_22.py`와 동일 흐름(DataCrawler → 마지막 JSON 회차+1 크롤 → `predict1`+`predict2`). `lotto_history.txt`로 `BallFilter`를 만들고 회차는 `max(no)+1`, `predict2`는 전 조합 순회(시간 매우 김).
## 실행 (miniconda **ncue**)
@@ -26,8 +26,7 @@ python valid.py
python final_filterTest.py
# 특정 회차 생존 조합 수 Monte Carlo 근사
python final_filterTest.py --mc-no 900 --mc-samples 12000
# 다음 회차 추천(네트워크·텔레그램 없이 로컬만)
python final_practice.py --skip-data-crawl --skip-fetch-next --no-telegram
python final_practice.py
```
동일 환경을 셸 스크립트로:

257
final_practice.py
View File

@@ -1,14 +1,14 @@
# `3_Practice_22.py` 흐름을 따르되 `final_BallFilter` + `lotto_history.txt`로 다음 회차 후보를 만듭니다.
import argparse
# `3_Practice_22.py`와 동일한 흐름입니다.
# - `resources/lotto_history.txt`(및 크롤 시 `.json`)를 사용합니다.
# - 필터만 `final_BallFilter.BallFilter` + `lotto_history.txt` 로딩으로 교체했습니다.
# - 회차 번호는 JSON의 `getNo(ymd)` 대신 `lotto_history.txt` 최대 회차 + 1 을 사용합니다.
import itertools
import json
import os
import random
import time
from datetime import datetime, timedelta
import pandas as pd
import requests
from DataCrawler import DataCrawler
from final_BallFilter import BallFilter
@@ -24,29 +24,31 @@ except Exception: # pragma: no cover
print(msg)
def fetch_lotto_draw_json(drw_no: int):
"""동행복권 API 한 건. SSL·POST/GET 재시도는 DataCrawler._fetch_draw와 동일."""
return DataCrawler()._fetch_draw(int(drw_no))
class FinalPractice:
bot = None
def __init__(self, resources_path):
self.resources_path = resources_path
self.bot = TelegramBot()
def craw(self, lotto_history_base: str, drw_no=None):
"""동행복권 API로 회차별 결과를 `lotto_history.json` / `.txt`에 저장합니다."""
def craw(self, lottoHistoryFile, drwNo=None):
ball = None
if drw_no is not None:
json_fp = open(lotto_history_base + ".json", "a", encoding="utf-8")
text_fp = open(lotto_history_base + ".txt", "a", encoding="utf-8")
url = "https://dhlottery.co.kr/common.do?method=getLottoNumber&drwNo=" + str(drw_no)
res = requests.post(url, timeout=30)
result = res.json()
if result.get("returnValue") != "success":
json_fp.close()
text_fp.close()
if drwNo is not None:
result = fetch_lotto_draw_json(drwNo)
if result is None:
print("경고: 회차 {} API 조회 실패(SSL/네트워크 또는 미추첨).".format(drwNo))
return None
json_fp.write(json.dumps(result, ensure_ascii=False) + "\n")
text_fp.write(
jsonFp = open(lottoHistoryFile + ".json", "a", encoding="utf-8")
textFp = open(lottoHistoryFile + ".txt", "a", encoding="utf-8")
jsonFp.write(json.dumps(result, ensure_ascii=False) + "\n")
textFp.write(
"%d,%d,%d,%d,%d,%d,%d,%d\n"
% (
drw_no,
drwNo,
result["drwtNo1"],
result["drwtNo2"],
result["drwtNo3"],
@@ -59,7 +61,7 @@ class FinalPractice:
print(
"%d,%d,%d,%d,%d,%d,%d,%d"
% (
drw_no,
drwNo,
result["drwtNo1"],
result["drwtNo2"],
result["drwtNo3"],
@@ -78,20 +80,18 @@ class FinalPractice:
result["drwtNo6"],
result["bnusNo"],
]
json_fp.close()
text_fp.close()
jsonFp.close()
textFp.close()
else:
json_fp = open(lotto_history_base + ".json", "w", encoding="utf-8")
text_fp = open(lotto_history_base + ".txt", "w", encoding="utf-8")
jsonFp = open(lottoHistoryFile + ".json", "w", encoding="utf-8")
textFp = open(lottoHistoryFile + ".txt", "w", encoding="utf-8")
idx = 1
while True:
url = "https://dhlottery.co.kr/common.do?method=getLottoNumber&drwNo=" + str(idx)
res = requests.post(url, timeout=30)
result = res.json()
if result.get("returnValue") != "success":
result = fetch_lotto_draw_json(idx)
if result is None or result.get("returnValue") != "success":
break
json_fp.write(json.dumps(result, ensure_ascii=False) + "\n")
text_fp.write(
jsonFp.write(json.dumps(result, ensure_ascii=False) + "\n")
textFp.write(
"%d,%d,%d,%d,%d,%d,%d,%d\n"
% (
idx,
@@ -128,73 +128,49 @@ class FinalPractice:
]
idx += 1
time.sleep(0.5)
json_fp.close()
text_fp.close()
jsonFp.close()
textFp.close()
return ball
def predict_fixed(self, result_list):
"""참고 스크립트와 동일하게 고정 6개 조합 1개를 넣습니다."""
result_list.append([6, 7, 10, 11, 20, 45])
def predict1(self, result_json):
result_json.append([6, 7, 10, 11, 20, 45])
def predict_filter_exhaustive(self, ball_filter: BallFilter, df_ball: pd.DataFrame, no: int, result_list):
"""전체 8,145,060조합을 순회합니다(시간이 매우 오래 걸릴 수 있음)."""
candidates = list(range(1, 46))
for idx, ball in enumerate(itertools.combinations(candidates, 6)):
if idx % 1_000_000 == 0:
def predict2(self, resources_path, ymd, result_json):
candidates = [i for i in range(1, 46)]
lottoHistoryFileName = os.path.join(resources_path, "lotto_history.txt")
df_ball = pd.read_csv(lottoHistoryFileName, header=None)
df_ball.columns = ["no", "b1", "b2", "b3", "b4", "b5", "b6", "bn"]
no = int(df_ball["no"].max()) + 1
print("회차: {}".format(no))
ballFilter = BallFilter(lottoHistoryFileName)
nCr = list(itertools.combinations(candidates, 6))
for idx, ball in enumerate(nCr):
if idx % 1000000 == 0:
print(" - {} processed...".format(idx))
ball = list(ball)
filter_type = ball_filter.filter(ball=ball, no=no, until_end=False, df=df_ball)
if len(filter_type) > 0:
filter_type = ballFilter.filter(ball=ball, no=no, until_end=False, df=df_ball)
filter_size = len(filter_type)
if 0 < filter_size:
continue
result_list.append(ball)
result_json.append(ball)
def predict_filter_montecarlo(
self,
ball_filter: BallFilter,
df_ball: pd.DataFrame,
no: int,
result_list: list,
max_recommend: int,
max_tries: int,
seed: int,
):
"""무작위 조합으로 필터를 통과하는 조합을 최대 `max_recommend`개까지 수집합니다."""
rng = random.Random(seed)
pool = list(range(1, 46))
tries = 0
seen = set()
while len(result_list) < max_recommend and tries < max_tries:
tries += 1
ball = sorted(rng.sample(pool, 6))
key = tuple(ball)
if key in seen:
continue
seen.add(key)
filter_type = ball_filter.filter(ball=ball, no=no, until_end=False, df=df_ball)
if filter_type:
continue
result_list.append(ball)
print(
"Monte Carlo: 수집 {}개 / 시도 {}회 (상한 {}회)".format(
len(result_list), tries, max_tries
)
)
p_ball = df_ball[df_ball["no"] == no - 1].values.tolist()[0]
p_no = p_ball[0]
p_ball = p_ball[1:7]
return p_no, p_ball
def load_df(resources_path: str) -> pd.DataFrame:
path = os.path.join(resources_path, "lotto_history.txt")
df_ball = pd.read_csv(path, header=None)
df_ball.columns = ["no", "b1", "b2", "b3", "b4", "b5", "b6", "bn"]
return df_ball
if __name__ == "__main__":
PROJECT_HOME = "."
resources_path = os.path.join(PROJECT_HOME, "resources")
dataCrawler = DataCrawler()
dataCrawler.excute(resources_path)
def next_draw_no(df_ball: pd.DataFrame) -> int:
"""파일에 있는 최신 회차의 다음 회차 번호."""
return int(df_ball["no"].max()) + 1
def weekend_ymd_strings():
"""표시용: 이번 주 토요일(추첨일) YMD, 지난 주 동일."""
today = datetime.today()
if today.weekday() == 5:
if today.hour > 20:
@@ -205,115 +181,52 @@ def weekend_ymd_strings():
this_weekend = today + timedelta(days=(12 - today.weekday()))
else:
this_weekend = today + timedelta(days=(5 - today.weekday()))
last_weekend = (this_weekend - timedelta(days=7)).strftime("%Y%m%d")
ymd = this_weekend.strftime("%Y%m%d")
return ymd, last_weekend
print("ymd: {}".format(ymd))
def main():
parser = argparse.ArgumentParser(description="다음 회차 필터 통과 조합 추천 (final_BallFilter)")
parser.add_argument("--project-home", default=".", help="프로젝트 루트")
parser.add_argument(
"--max-total",
type=int,
default=99,
help="추천 조합 총 개수 상한(고정 1개 포함, 기본 99 = 100 미만)",
)
parser.add_argument("--max-tries", type=int, default=3_000_000, help="Monte Carlo 최대 시도 횟수")
parser.add_argument("--seed", type=int, default=0)
parser.add_argument("--exhaustive", action="store_true", help="전수 탐색(매우 느림)")
parser.add_argument("--skip-data-crawl", action="store_true", help="DataCrawler.excute 생략")
parser.add_argument("--skip-fetch-next", action="store_true", help="다음 회차 API 크롤 생략")
parser.add_argument("--no-telegram", action="store_true", help="텔레그램 전송 생략")
parser.add_argument("--draw-no", type=int, default=None, help="예측 대상 회차(미지정 시 파일 기준 다음 회차)")
args = parser.parse_args()
project_home = os.path.abspath(args.project_home)
resources_path = os.path.join(project_home, "resources")
os.makedirs(resources_path, exist_ok=True)
ymd, last_weekend = weekend_ymd_strings()
print("ymd(표시용): {}".format(ymd))
if not args.skip_data_crawl:
data_crawler = DataCrawler()
data_crawler.excute(resources_path)
lotto_history_base = os.path.join(project_home, "resources", "lotto_history")
lotto_json = lotto_history_base + ".json"
practice = FinalPractice(resources_path)
if not args.skip_fetch_next and os.path.isfile(lotto_json):
with open(lotto_json, "r", encoding="utf-8") as f:
last_json = None
for line in f:
line = line.strip()
if line:
last_json = json.loads(line)
if last_json is not None:
nxt = int(last_json["drwNo"]) + 1
practice.craw(lotto_history_base, drw_no=nxt)
df_ball = load_df(resources_path)
no = args.draw_no if args.draw_no is not None else next_draw_no(df_ball)
print("예측 대상 회차 no: {}".format(no))
lottoHistoryFile = PROJECT_HOME + "/resources/lotto_history"
lottoHistoryFileName = lottoHistoryFile + ".json"
with open(lottoHistoryFileName, "r", encoding="utf-8") as f:
for line in f:
pass
last_json = json.loads(line)
lotto_txt = os.path.join(resources_path, "lotto_history.txt")
ball_filter = BallFilter(lotto_txt)
ball = practice.craw(lottoHistoryFile, drwNo=last_json["drwNo"] + 1)
recommend_path = os.path.join(resources_path, "recommend_ball.final.json")
if os.path.isfile(recommend_path):
with open(recommend_path, "r", encoding="utf-8") as result_fp:
result_json = json.load(result_fp)
recommend_result_file = os.path.join(resources_path, "recommend_ball.final.json")
if os.path.isfile(recommend_result_file):
result_fp = open(recommend_result_file, "r", encoding="utf-8")
result_json = json.load(result_fp)
result_fp.close()
result_json[ymd] = []
else:
result_json = {ymd: []}
if args.no_telegram:
practice.bot = type("T", (), {"sendMsg": lambda self, m: None})()
practice.predict1(result_json[ymd])
p_no, p_ball = practice.predict2(resources_path, ymd, result_json[ymd])
practice.predict_fixed(result_json[ymd])
mc_cap = max(0, args.max_total - len(result_json[ymd]))
if args.exhaustive:
practice.predict_filter_exhaustive(ball_filter, df_ball, no, result_json[ymd])
else:
practice.predict_filter_montecarlo(
ball_filter,
df_ball,
no,
result_json[ymd],
max_recommend=mc_cap,
max_tries=args.max_tries,
seed=args.seed,
)
with open(recommend_path, "w", encoding="utf-8") as out_fp:
json.dump(result_json, out_fp, ensure_ascii=False)
prev_row = df_ball[df_ball["no"] == no - 1]
if prev_row.empty:
p_no, p_ball = no - 1, []
print("경고: no-1={} 행이 없어 이전 당첨 표시를 건너뜁니다.".format(no - 1))
else:
p_no = int(prev_row["no"].iloc[0])
p_ball = [int(prev_row["b{}".format(i)].iloc[0]) for i in range(1, 7)]
with open(recommend_result_file, "w", encoding="utf-8") as outFp:
json.dump(result_json, outFp, ensure_ascii=False)
no_predict = int(p_no) + 1
p_str = "[지난주] {}\n - {} 회차, {}\n[금주] {}\n - {} 회차\n[final_BallFilter]\n".format(
last_weekend, p_no, str(p_ball), ymd, no
last_weekend, p_no, str(p_ball), ymd, no_predict
)
for i, ball in enumerate(result_json[ymd]):
p_str += " {}. {}\n".format((i + 1), str(ball))
if (i + 1) % 100 == 0:
practice.bot.sendMsg("{}".format(p_str))
p_str = ""
if len(result_json[ymd]) % 100 != 0:
practice.bot.sendMsg("{}".format(p_str))
size = len(result_json[ymd])
print("recommend size: {}".format(size))
print("저장: {}".format(recommend_path))
print("size: {}".format(size))
print("done...")
if __name__ == "__main__":
main()