File size: 9,828 Bytes
778d47d | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 | """
Build BM25 (Lucene) content indexes for all datasets needed by MATS-TIST.
Run from the mats-sql-tist/ root directory:
conda run -n mats python scripts/build_all_bm25_indexes.py
CPU-only; no GPU required.
Datasets covered:
- BIRD dev / train (data/bird/dev|train)
- Spider dev+test (data/spider)
- Spider-DK (3 new databases)
- Spider-Syn / spider-realistic (share Spider databases → symlinked)
- Dr.Spider NLQ_* / SQL_* (share original Spider databases → symlinked)
- Dr.Spider DB_* (post-perturbation databases, built fresh)
- Domain datasets (Bank_Financials, Aminer_Simplified)
"""
import os
import sys
import shutil
import json
from tqdm import tqdm
# Make sure we can import from the project root
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from utils.db_utils import get_cursor_from_path, execute_sql_long_time_limitation
# ---------------------------------------------------------------------------
# Core indexer (copied from build_contents_index.py)
# ---------------------------------------------------------------------------
def build_content_index(db_path: str, index_path: str) -> None:
"""Build a per-column BM25 index for one SQLite database."""
cursor = get_cursor_from_path(db_path)
results = execute_sql_long_time_limitation(
cursor, "SELECT name FROM sqlite_master WHERE type='table';"
)
table_names = [r[0] for r in results]
for table_name in table_names:
table_name = table_name.lower()
if table_name == "sqlite_sequence":
continue
results = execute_sql_long_time_limitation(
cursor, f"SELECT name FROM PRAGMA_TABLE_INFO('{table_name}')"
)
column_names = [r[0].lower() for r in results]
for column_name in column_names:
column_index_path = f"{index_path}/{table_name}-**-{column_name}"
if os.path.exists(column_index_path):
continue
all_column_contents = []
try:
rows = execute_sql_long_time_limitation(
cursor,
f"SELECT DISTINCT `{column_name}` FROM `{table_name}` WHERE `{column_name}` IS NOT NULL;",
)
for c_id, row in enumerate(rows):
content = str(row[0]).strip()
if 0 < len(content) <= 50:
all_column_contents.append(
{
"id": f"{table_name}-**-{column_name}-**-{c_id}".lower(),
"contents": content,
}
)
except Exception as e:
print(f" [WARN] {table_name}.{column_name}: {e}")
os.makedirs("./data/temp_db_index", exist_ok=True)
with open("./data/temp_db_index/contents.json", "w") as f:
json.dump(all_column_contents, f, indent=2, ensure_ascii=True)
cmd = (
"python -m pyserini.index.lucene "
"--collection JsonCollection "
"--input ./data/temp_db_index "
f'--index "{column_index_path}" '
"--generator DefaultLuceneDocumentGenerator "
"--threads 16 --storePositions --storeDocvectors --storeRaw"
)
ret = os.system(cmd)
if os.path.exists("./data/temp_db_index/contents.json"):
os.remove("./data/temp_db_index/contents.json")
if ret != 0:
print(f" [ERROR] indexing failed for {column_index_path}")
def index_dataset(db_root: str, index_root: str, label: str) -> None:
"""Index all databases under db_root into index_root."""
if not os.path.isdir(db_root):
print(f"[SKIP] {label}: db_root not found: {db_root}")
return
os.makedirs(index_root, exist_ok=True)
db_ids = [d for d in os.listdir(db_root) if os.path.isdir(os.path.join(db_root, d))]
print(f"\n{'='*60}")
print(f"[INDEX] {label} ({len(db_ids)} databases → {index_root})")
print(f"{'='*60}")
for db_id in tqdm(db_ids, desc=label):
sqlite_path = os.path.join(db_root, db_id, f"{db_id}.sqlite")
if not os.path.exists(sqlite_path):
# Some Dr.Spider perturbation folders have different sqlite name
candidates = [
f for f in os.listdir(os.path.join(db_root, db_id))
if f.endswith(".sqlite")
]
if candidates:
sqlite_path = os.path.join(db_root, db_id, candidates[0])
else:
print(f" [SKIP] no .sqlite in {db_id}")
continue
build_content_index(sqlite_path, os.path.join(index_root, db_id))
def symlink_index(src: str, dst: str, label: str) -> None:
"""Create a symlink dst → src if dst does not already exist."""
if os.path.exists(dst) or os.path.islink(dst):
print(f"[OK] {label} already present: {dst}")
return
if not os.path.isdir(src):
print(f"[WARN] {label}: source index not found: {src}")
return
os.makedirs(os.path.dirname(dst), exist_ok=True)
os.symlink(src, dst)
print(f"[LINK] {label}: {dst} → {src}")
# ---------------------------------------------------------------------------
# Main
# ---------------------------------------------------------------------------
if __name__ == "__main__":
ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
os.chdir(ROOT)
print(f"Working directory: {ROOT}")
DR = "./data/sft_data_collections/diagnostic-robustness-text-to-sql/data"
# ------------------------------------------------------------------
# 1. BIRD dev / train (bird-062024)
# ------------------------------------------------------------------
index_dataset(
db_root="./data/bird/dev/dev_databases",
index_root="./data/bird/dev/db_contents_index",
label="BIRD dev",
)
index_dataset(
db_root="./data/bird/train/train_databases",
index_root="./data/bird/train/db_contents_index",
label="BIRD train",
)
# sft_data_collections/bird → symlink to the same bird index
symlink_index(
src=os.path.abspath("./data/bird/dev/db_contents_index"),
dst="./data/sft_data_collections/bird/dev/db_contents_index",
label="sft bird/dev (symlink)",
)
symlink_index(
src=os.path.abspath("./data/bird/train/db_contents_index"),
dst="./data/sft_data_collections/bird/train/db_contents_index",
label="sft bird/train (symlink)",
)
# ------------------------------------------------------------------
# 2. Spider dev + test (all share the same 169 databases)
# ------------------------------------------------------------------
index_dataset(
db_root="./data/spider/database",
index_root="./data/spider/db_contents_index",
label="Spider (dev/test/train databases)",
)
# Spider-Syn and spider-realistic share Spider databases → symlink
symlink_index(
src=os.path.abspath("./data/spider/db_contents_index"),
dst="./data/sft_data_collections/Spider-Syn/db_contents_index",
label="Spider-Syn (symlink)",
)
symlink_index(
src=os.path.abspath("./data/spider/db_contents_index"),
dst="./data/sft_data_collections/spider-realistic/db_contents_index",
label="spider-realistic (symlink)",
)
# ------------------------------------------------------------------
# 3. Spider-DK (3 new databases not in Spider)
# ------------------------------------------------------------------
index_dataset(
db_root="./data/sft_data_collections/Spider-DK/database",
index_root="./data/sft_data_collections/Spider-DK/db_contents_index",
label="Spider-DK",
)
# ------------------------------------------------------------------
# 4. Dr. Spider — NLQ_* and SQL_* reuse original Spider databases
# → symlink their db_contents_index to Spider's
# ------------------------------------------------------------------
for dr_name in [
"NLQ_keyword_synonym", "NLQ_keyword_carrier", "NLQ_column_synonym",
"NLQ_column_carrier", "NLQ_column_attribute", "NLQ_column_value",
"NLQ_value_synonym", "NLQ_multitype", "NLQ_others",
"SQL_comparison", "SQL_sort_order", "SQL_NonDB_number",
"SQL_DB_text", "SQL_DB_number",
]:
symlink_index(
src=os.path.abspath("./data/spider/db_contents_index"),
dst=f"{DR}/{dr_name}/db_contents_index",
label=f"Dr.Spider/{dr_name} (symlink)",
)
# ------------------------------------------------------------------
# 5. Dr. Spider — DB_* perturbation sets have MODIFIED databases
# ------------------------------------------------------------------
for dr_name in [
"DB_schema_synonym",
"DB_schema_abbreviation",
"DB_DBcontent_equivalence",
]:
index_dataset(
db_root=f"{DR}/{dr_name}/database_post_perturbation",
index_root=f"{DR}/{dr_name}/db_contents_index",
label=f"Dr.Spider/{dr_name}",
)
# ------------------------------------------------------------------
# 6. Domain datasets (Bank_Financials, Aminer_Simplified)
# ------------------------------------------------------------------
index_dataset(
db_root="./data/sft_data_collections/domain_datasets/databases",
index_root="./data/sft_data_collections/domain_datasets/db_contents_index",
label="Domain datasets",
)
print("\n[DONE] All BM25 indexes built / symlinked.")
|