Datasets:
File size: 5,580 Bytes
d847043 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 | """把第二轮 DeepSeek 出的 6 个大库和 2025 道题并进来（cn-sqlite-round2.zip，照 deepseek_tasks.md 做的）。
python scripts/import_round2.py --src <解压出来的 cn-sqlite-full 目录>
zip 是在旧版项目上做的，只取两样：schemas/hard2/*.sql 和 data/hard2_train.jsonl。
- property_mgmt、car_service 的主键是复数（communities_id），引用它的外键却是单数（community_id），
car_service 里还同时有 lead_technician_id 和 lead_technicians_id。统一改成单数，建表语句和 SQL 一起改
- 改名前后结果必须一模一样：同一份数据建两个库，一个 ALTER TABLE RENAME COLUMN，老 SQL 跑老库、新 SQL 跑新库，逐条比
- 题目写到 data/round2/raw/<库名>.jsonl（db / question / sql / difficulty / type），还没筛，
不进训练；check_round2.py 筛完才写到 data/new/
"""
import argparse
import json
import re
import sqlite3
from collections import Counter
from pathlib import Path
ROOT = Path(__file__).resolve().parent.parent
RENAME = {
"property_mgmt": {
"communities_id": "community_id", "buildings_id": "building_id", "units_id": "unit_id",
"houses_id": "house_id", "owners_id": "owner_id", "owner_house_relations_id": "owner_house_relation_id",
"tenants_id": "tenant_id", "lease_contracts_id": "lease_contract_id", "fee_bills_id": "fee_bill_id",
"payments_id": "payment_id", "repair_orders_id": "repair_order_id", "repair_workers_id": "repair_worker_id",
"repair_assignments_id": "repair_assignment_id",
},
"car_service": {
"stores_id": "store_id", "vehicle_models_id": "vehicle_model_id", "owners_id": "owner_id",
"vehicles_id": "vehicle_id", "sales_orders_id": "sales_order_id", "technicians_id": "technician_id",
"repair_orders_id": "repair_order_id", "repair_items_id": "repair_item_id", "parts_id": "part_id",
"part_issues_id": "part_issue_id", "insurance_claims_id": "insurance_claim_id",
"technician_assignments_id": "technician_assignment_id", "lead_technicians_id": "lead_technician_id",
},
}
LABEL = {"中等": "中", "每组前 N/并列": "每组前N/并列", "反连接": "反连接/存在性"}
def rename(text: str, mapping: dict[str, str]) -> str:
for old, new in mapping.items():
text = re.sub(rf"\b{old}\b", new, text)
return text
def rows_of(conn: sqlite3.Connection, sql: str):
return conn.execute(sql).fetchall()
def main() -> None:
ap = argparse.ArgumentParser()
ap.add_argument("--src", required=True, help="cn-sqlite-round2.zip 解压出来的 cn-sqlite-full 目录")
args = ap.parse_args()
src = Path(args.src)
out_schema = ROOT / "schemas" / "hard2"
out_schema.mkdir(parents=True, exist_ok=True)
old_db, new_db = {}, {}
for f in sorted((src / "schemas" / "hard2").glob("*.sql")):
db, ddl = f.stem, f.read_text(encoding="utf-8")
mapping = RENAME.get(db, {})
(out_schema / f.name).write_text(rename(ddl, mapping), encoding="utf-8")
seed = (src / "seeds" / "hard2" / f.name).read_text(encoding="utf-8")
old_db[db] = sqlite3.connect(":memory:")
old_db[db].executescript(ddl + seed)
new_db[db] = sqlite3.connect(":memory:")
new_db[db].executescript(ddl + seed)
for t, in new_db[db].execute("select name from sqlite_master where type='table'").fetchall():
for col in [c[1] for c in new_db[db].execute(f'pragma table_info("{t}")')]:
if col in mapping:
new_db[db].execute(f'ALTER TABLE "{t}" RENAME COLUMN "{col}" TO "{mapping[col]}"')
# 改完的建表语句和 ALTER 出来的库得是同一个结构，不然下面的比对没意义
want = sqlite3.connect(":memory:")
want.executescript(rename(ddl, mapping))
shape = lambda c: sorted((t, tuple(x[1] for x in c.execute(f'pragma table_info("{t}")')))
for t, in c.execute("select name from sqlite_master where type='table'"))
assert shape(want) == shape(new_db[db]), db
items, mismatch, broken = [], 0, 0
for line in (src / "data" / "hard2_train.jsonl").read_text(encoding="utf-8").splitlines():
r = json.loads(line)
db = r["db"]
user, sql = r["messages"][1]["content"], r["messages"][2]["content"].strip()
question = user.rsplit("\n\n问题：", 1)[1].strip()
new_sql = rename(sql, RENAME.get(db, {}))
try:
same = rows_of(old_db[db], sql) == rows_of(new_db[db], new_sql)
except sqlite3.Error as exc:
print("跑不通：", db, exc, new_sql[:80])
broken += 1
continue
if not same:
print("改名后结果变了：", db, question)
mismatch += 1
continue
items.append({"db": db, "question": question, "sql": new_sql,
"difficulty": LABEL.get(r.get("difficulty"), r.get("difficulty")),
"type": LABEL.get(r.get("type"), r.get("type"))})
by_db = Counter(it["db"] for it in items)
(ROOT / "data" / "round2" / "raw").mkdir(parents=True, exist_ok=True)
for db in by_db:
out = ROOT / "data" / "round2" / "raw" / f"{db}.jsonl"
out.write_text("".join(json.dumps(it, ensure_ascii=False) + "\n" for it in items if it["db"] == db),
encoding="utf-8")
print(f"导入 {len(items)} 条（跑不通 {broken}，改名后结果变了 {mismatch}）", dict(by_db))
if __name__ == "__main__":
main()
|