File size: 7,681 Bytes
90264a9 | 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 | #!/usr/bin/env python3
"""Build plwiki.sqlite from two Polish Wikipedia snapshots + the plwiki redirect dump (no API cost).
python build_db.py pages # corpus/finewiki (FineWiki pl, HTML dumps Aug 2025) + corpus/wikipedia (20231101.pl)
python build_db.py aliases # corpus/dumps/plwiki-latest-{page,redirect}.sql.gz -> alias table
python build_db.py stats
Tables
page(pid, snap, page_id, title, title_norm, url, wikidata_id, date_modified, lead, text_z, infobox_z)
snap = 'fw2508' (FineWiki, primary) | 'w2311' (wikimedia/wikipedia 20231101.pl, fallback + cross-check)
text_z = zlib(full article text); FineWiki text is markdown ('# Title', '## Section', tables, lists)
page_fts(title_norm, lead_norm) contentless FTS5, rowid = pid; w2311 pages only when the title is not in FineWiki
alias(src_norm, src_title, target) redirect title (normalised) -> target article title
meta(k, v)
"""
import glob
import gzip
import json
import re
import sqlite3
import sys
import time
import zlib
import pyarrow.parquet as pq
from common import DB_PATH, ROOT, norm_pl
CORPUS = ROOT / "corpus"
SCHEMA = """
CREATE TABLE IF NOT EXISTS page(pid INTEGER PRIMARY KEY, snap TEXT, page_id INTEGER, title TEXT, title_norm TEXT,
url TEXT, wikidata_id TEXT, date_modified TEXT, lead TEXT, text_z BLOB, infobox_z BLOB);
CREATE VIRTUAL TABLE IF NOT EXISTS page_fts USING fts5(title_norm, lead_norm, content='', tokenize='unicode61');
CREATE TABLE IF NOT EXISTS alias(src_norm TEXT PRIMARY KEY, src_title TEXT, target TEXT);
CREATE TABLE IF NOT EXISTS meta(k TEXT PRIMARY KEY, v TEXT);
"""
def connect():
db = sqlite3.connect(DB_PATH)
db.executescript("PRAGMA journal_mode=OFF; PRAGMA synchronous=OFF; PRAGMA cache_size=-400000;")
db.executescript(SCHEMA)
return db
def lead_of(text, fw):
"""Lead = paragraphs before the first section heading (FineWiki) or the first ~2 paragraphs (plain text)."""
if fw:
parts = text.split("\n", 1)
body = (parts[1] if len(parts) > 1 else "") if text.startswith("# ") else text
lead = re.split(r"\n#{2,6} ", body, maxsplit=1)[0]
else:
paras = [p for p in text.split("\n\n") if p.strip()]
lead, n = "", 0
for p in paras:
lead += p + "\n\n"
n += 1
if len(lead) > 600 or n >= 2:
break
lead = "\n".join(l for l in lead.splitlines() if not l.lstrip().startswith("|"))
return re.sub(r"\n{3,}", "\n\n", lead).strip()[:2500]
def build_pages():
db = connect()
db.execute("DELETE FROM page")
db.execute("DELETE FROM page_fts")
t0, n, fw_titles = time.time(), 0, set()
sources = [("fw2508", sorted(glob.glob(str(CORPUS / "finewiki/data/plwiki/*.parquet"))),
["page_id", "title", "url", "wikidata_id", "date_modified", "text", "infoboxes"]),
("w2311", sorted(glob.glob(str(CORPUS / "wikipedia/20231101.pl/*.parquet"))),
["id", "title", "url", "text"])]
for snap, files, cols in sources:
for f in files:
pf = pq.ParquetFile(f)
for batch in pf.iter_batches(batch_size=2000, columns=cols):
rows, fts = [], []
for r in batch.to_pylist():
text, title = r["text"] or "", r["title"] or ""
if not text or not title:
continue
tn = norm_pl(title)
fw = snap == "fw2508"
lead = lead_of(text, fw)
ib = r.get("infoboxes")
n += 1
rows.append((n, snap, int(r.get("page_id") or r.get("id") or 0), title, tn, r["url"],
r.get("wikidata_id"), r.get("date_modified"), lead,
zlib.compress(text.encode("utf-8"), 6),
zlib.compress(ib.encode("utf-8"), 6) if ib and ib != "[]" else None))
if fw:
fw_titles.add(tn)
fts.append((n, tn, norm_pl(lead[:900])))
elif tn not in fw_titles:
fts.append((n, tn, norm_pl(lead[:900])))
db.executemany("INSERT INTO page VALUES (?,?,?,?,?,?,?,?,?,?,?)", rows)
db.executemany("INSERT INTO page_fts(rowid, title_norm, lead_norm) VALUES (?,?,?)", fts)
if n % 100000 < 2000:
db.commit()
print(f"[{time.time() - t0:6.0f}s] {snap} {n} pages", flush=True)
db.commit()
print("indexes...", flush=True)
db.execute("CREATE INDEX IF NOT EXISTS page_tn ON page(title_norm)")
db.execute("CREATE INDEX IF NOT EXISTS page_snap_pid ON page(snap, page_id)")
db.execute("INSERT INTO page_fts(page_fts) VALUES('optimize')")
counts = dict(db.execute("SELECT snap, count(*) FROM page GROUP BY snap").fetchall())
meta = {"built_at": time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()), "pages": json.dumps(counts),
"fw2508": "HuggingFaceFW/finewiki config pl (data/plwiki, Wikipedia HTML dumps Aug 2025)",
"w2311": "wikimedia/wikipedia config 20231101.pl"}
db.executemany("INSERT OR REPLACE INTO meta VALUES (?,?)", meta.items())
db.commit()
print(f"done in {time.time() - t0:.0f}s: {counts}", flush=True)
STR = r"'((?:[^'\\]|\\.)*)'"
def unesc(s):
return re.sub(r"\\(.)", r"\1", s).replace("_", " ")
def build_aliases():
db = connect()
page_gz = CORPUS / "dumps/plwiki-latest-page.sql.gz"
red_gz = CORPUS / "dumps/plwiki-latest-redirect.sql.gz"
header = gzip.open(page_gz, "rt", encoding="utf-8", errors="replace").read(6000)
cols = re.findall(r"^\s+`(\w+)`", header.split("CREATE TABLE `page`", 1)[1].split(") ENGINE", 1)[0], re.M)
assert cols[:4] == ["page_id", "page_namespace", "page_title", "page_is_redirect"], cols[:6]
tup = re.compile(r"\((\d+),(-?\d+)," + STR + r",([01]),")
redirect_title = {}
t0 = time.time()
with gzip.open(page_gz, "rt", encoding="utf-8", errors="replace") as f:
for line in f:
if line.startswith("INSERT INTO"):
for m in tup.finditer(line):
if m.group(2) == "0" and m.group(4) == "1":
redirect_title[int(m.group(1))] = unesc(m.group(3))
print(f"redirect pages (ns0): {len(redirect_title)} [{time.time() - t0:.0f}s]", flush=True)
rtup = re.compile(r"\((\d+),(-?\d+)," + STR + r",")
rows = []
with gzip.open(red_gz, "rt", encoding="utf-8", errors="replace") as f:
for line in f:
if line.startswith("INSERT INTO"):
for m in rtup.finditer(line):
src = redirect_title.get(int(m.group(1)))
if src and m.group(2) == "0":
rows.append((norm_pl(src), src, unesc(m.group(3))))
db.execute("DELETE FROM alias")
db.executemany("INSERT OR IGNORE INTO alias VALUES (?,?,?)", rows)
db.execute("INSERT OR REPLACE INTO meta VALUES ('aliases', ?)",
(f"plwiki-latest redirect+page dumps, {len(rows)} redirects, built "
f"{time.strftime('%Y-%m-%dT%H:%M:%SZ', time.gmtime())}",))
db.commit()
print(f"aliases: {len(rows)} [{time.time() - t0:.0f}s]", flush=True)
def stats():
db = sqlite3.connect(f"file:{DB_PATH}?mode=ro", uri=True)
for k, v in db.execute("SELECT k, v FROM meta"):
print(f"{k}: {v}")
print("alias rows:", db.execute("SELECT count(*) FROM alias").fetchone()[0])
if __name__ == "__main__":
{"pages": build_pages, "aliases": build_aliases, "stats": stats}[sys.argv[1]]()
|