predictor / services /model /ingest.py
pengpeng9099's picture
Deploy World Cup 2026 Predictor (docker space)
1790bde verified
Raw History Blame Contribute Delete
5.12 kB
"""Load martj42 international results into SQLite, reconstruct WC2026 groups,
and seed the teams table (groups, confederation, Elo)."""
from __future__ import annotations
import json
import networkx as nx
import pandas as pd
import config
from db import init_db, connect
from elo import compute_elo
# Confederation of each 2026 qualifier (for UI grouping / sanity only).
CONFED = {
"Argentina": "CONMEBOL", "Brazil": "CONMEBOL", "Uruguay": "CONMEBOL",
"Colombia": "CONMEBOL", "Ecuador": "CONMEBOL", "Paraguay": "CONMEBOL",
"Spain": "UEFA", "France": "UEFA", "England": "UEFA", "Portugal": "UEFA",
"Germany": "UEFA", "Netherlands": "UEFA", "Belgium": "UEFA", "Croatia": "UEFA",
"Switzerland": "UEFA", "Austria": "UEFA", "Turkey": "UEFA", "Scotland": "UEFA",
"Norway": "UEFA", "Czech Republic": "UEFA", "Sweden": "UEFA",
"Bosnia and Herzegovina": "UEFA",
"Morocco": "CAF", "Senegal": "CAF", "Egypt": "CAF", "Algeria": "CAF",
"Tunisia": "CAF", "Ivory Coast": "CAF", "Ghana": "CAF", "Cape Verde": "CAF",
"South Africa": "CAF", "DR Congo": "CAF",
"Japan": "AFC", "South Korea": "AFC", "Iran": "AFC", "Australia": "AFC",
"Saudi Arabia": "AFC", "Qatar": "AFC", "Jordan": "AFC", "Uzbekistan": "AFC",
"Iraq": "AFC",
"United States": "CONCACAF", "Mexico": "CONCACAF", "Canada": "CONCACAF",
"Panama": "CONCACAF", "Haiti": "CONCACAF", "Curaçao": "CONCACAF",
"New Zealand": "OFC",
}
def load() -> dict:
init_db()
df = pd.read_csv(config.RAW / "results.csv")
df["date"] = pd.to_datetime(df["date"])
# --- matches table (full history + 2026 fixtures) ---
rows = df.copy()
rows["stage"] = None
wc26_mask = (rows.tournament == config.WC_TOURNAMENT) & (
rows.date >= config.WC_2026_FROM
)
rows.loc[wc26_mask, "stage"] = "group"
with connect() as conn:
conn.execute("DELETE FROM matches")
conn.executemany(
"""INSERT INTO matches(date,home_team,away_team,home_score,away_score,
tournament,city,country,neutral,stage)
VALUES(?,?,?,?,?,?,?,?,?,?)""",
[
(
r.date.strftime("%Y-%m-%d"), r.home_team, r.away_team,
None if pd.isna(r.home_score) else int(r.home_score),
None if pd.isna(r.away_score) else int(r.away_score),
r.tournament, None if pd.isna(r.city) else r.city,
None if pd.isna(r.country) else r.country,
int(bool(r.neutral)), r.stage,
)
for r in rows.itertuples(index=False)
],
)
# --- official 2026 group draw (FIFA final draw, 2025-12-05) ---
# Letters MUST match the official draw because bracket.py's knockout
# template references group letters (W-E = winner of group E, etc.).
OFFICIAL_GROUPS = {
"A": ["Mexico", "South Korea", "South Africa", "Czech Republic"],
"B": ["Canada", "Switzerland", "Qatar", "Bosnia and Herzegovina"],
"C": ["Brazil", "Morocco", "Haiti", "Scotland"],
"D": ["United States", "Paraguay", "Australia", "Turkey"],
"E": ["Germany", "Curaçao", "Ivory Coast", "Ecuador"],
"F": ["Netherlands", "Japan", "Sweden", "Tunisia"],
"G": ["Belgium", "Egypt", "Iran", "New Zealand"],
"H": ["Spain", "Cape Verde", "Saudi Arabia", "Uruguay"],
"I": ["France", "Senegal", "Iraq", "Norway"],
"J": ["Argentina", "Algeria", "Austria", "Jordan"],
"K": ["Portugal", "DR Congo", "Uzbekistan", "Colombia"],
"L": ["England", "Croatia", "Ghana", "Panama"],
}
groups = OFFICIAL_GROUPS
# cross-check: every WC2026 fixture team must appear in exactly one group
fixture_teams = set(df[wc26_mask].home_team) | set(df[wc26_mask].away_team)
grouped = {t for ts in groups.values() for t in ts}
missing = fixture_teams - grouped
if missing:
print(f"WARNING: 赛程中有队不在官方分组里(队名不匹配?): {sorted(missing)}")
# --- Elo ratings from full history ---
elo = compute_elo(df.dropna(subset=["home_score", "away_score"]))
# --- teams table (48 qualifiers) ---
team_to_group = {t: g for g, ts in groups.items() for t in ts}
with connect() as conn:
conn.execute("DELETE FROM teams")
for team, grp in sorted(team_to_group.items()):
conn.execute(
"INSERT INTO teams(name,confederation,elo,group_label) VALUES(?,?,?,?)",
(team, CONFED.get(team), round(elo.get(team, 1500.0), 1), grp),
)
(config.PROCESSED / "groups.json").write_text(
json.dumps(groups, ensure_ascii=False, indent=2), encoding="utf-8"
)
summary = {
"total_matches": len(df),
"played": int(df.home_score.notna().sum()),
"wc2026_fixtures": int(wc26_mask.sum()),
"groups": len(groups),
"teams": len(team_to_group),
}
print(json.dumps(summary, ensure_ascii=False))
return {"groups": groups, "summary": summary}
if __name__ == "__main__":
load()