File size: 6,470 Bytes
faccbea | 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
"""Remove opted-out repositories from your local copy of The Stack.
usage: filter_stack.py --input PATH --output DIR [--edition optouts.csv] [--repo-column NAME]
[--include-withdrawn]
PATH is a .parquet or .jsonl (optionally .gz) file, or a folder of them. Parquet needs pyarrow
(only imported if you have Parquet input). Your input is never modified: filtered copies are
written to DIR (which must not be the input folder) together with removal_report.csv.
Whole-account rows match every repository of that owner; repository rows match owner/repo exactly.
Names are compared ignoring case.
"""
import argparse, csv, gzip, json, os, sys
CANDIDATE_COLUMNS = ("repo_path", "repo_name", "repository_name", "full_name", "repo")
def load_rules(path, include_withdrawn=False):
owners, repos = {}, {}
with open(path, newline="", encoding="utf-8") as f:
for r in csv.DictReader(f):
if r["withdrawn"] == "yes" and not include_withdrawn:
continue
n = int(r["issue_number"])
if r["request_type"] == "whole_account":
k = r["owner"].lower()
if k not in owners or n < owners[k]:
owners[k] = n
else:
k = (r["owner"] + "/" + r["repository"]).lower()
if k not in repos or n < repos[k]:
repos[k] = n
return owners, repos
def match(slug, owners, repos):
"""Issue number of the earliest matching request, or None."""
if not isinstance(slug, str) or not slug:
return None
s = slug.strip().lower()
hits = []
if s in repos:
hits.append(repos[s])
o = s.split("/", 1)[0]
if o in owners:
hits.append(owners[o])
return min(hits) if hits else None
def pick_column(names, wanted):
if wanted:
if wanted not in names:
sys.exit(f"column {wanted!r} not found; columns are: {sorted(names)}")
return wanted
for c in CANDIDATE_COLUMNS:
if c in names:
return c
sys.exit("no repository column found; pass --repo-column. Columns: " + ", ".join(sorted(names)))
def _open(path, mode):
return gzip.open(path, mode + "t", encoding="utf-8") if path.endswith(".gz") else open(path, mode, encoding="utf-8")
def filter_jsonl(src, dst, owners, repos, col, counts):
scanned = removed = 0
with _open(src, "r") as fi, _open(dst, "w") as fo:
for line in fi:
if not line.strip():
continue
row = json.loads(line)
if col is None:
col = pick_column(set(row), None)
scanned += 1
hit = match(row.get(col), owners, repos)
if hit is None:
fo.write(line if line.endswith("\n") else line + "\n")
else:
removed += 1
counts[hit] = counts.get(hit, 0) + 1
return scanned, removed
def filter_parquet(src, dst, owners, repos, wanted, counts):
try:
import pyarrow as pa
import pyarrow.parquet as pq
except ImportError:
sys.exit("pyarrow is needed for Parquet input (pip install pyarrow), or convert to JSONL")
pf = pq.ParquetFile(src)
col = pick_column(set(pf.schema_arrow.names), wanted)
scanned = removed = 0
writer = pq.ParquetWriter(dst, pf.schema_arrow)
try:
for batch in pf.iter_batches(batch_size=1000):
slugs = batch.column(col).to_pylist()
keep = []
for i, s in enumerate(slugs):
hit = match(s, owners, repos)
if hit is None:
keep.append(i)
else:
removed += 1
counts[hit] = counts.get(hit, 0) + 1
scanned += len(slugs)
if keep:
writer.write_table(pa.Table.from_batches([batch]).take(keep))
finally:
writer.close()
return scanned, removed
def inputs(path):
if os.path.isdir(path):
for n in sorted(os.listdir(path)):
if n.endswith((".parquet", ".jsonl", ".jsonl.gz")):
yield os.path.join(path, n)
else:
yield path
def main(argv=None):
ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
ap.add_argument("--input", required=True)
ap.add_argument("--output", required=True)
ap.add_argument("--edition", default=os.path.join(os.path.dirname(os.path.abspath(__file__)), "optouts.csv"))
ap.add_argument("--repo-column")
ap.add_argument("--include-withdrawn", action="store_true")
a = ap.parse_args(argv)
in_abs = os.path.abspath(a.input)
out_abs = os.path.abspath(a.output)
in_dir = in_abs if os.path.isdir(in_abs) else os.path.dirname(in_abs)
if out_abs == in_dir:
sys.exit("refusing: output folder is the input folder (inputs are never modified in place)")
os.makedirs(out_abs, exist_ok=True)
owners, repos = load_rules(a.edition, a.include_withdrawn)
meta = {}
with open(a.edition, newline="", encoding="utf-8") as f:
for r in csv.DictReader(f):
meta[int(r["issue_number"])] = r
counts, total_scanned, total_removed, files = {}, 0, 0, 0
for src in inputs(a.input):
dst = os.path.join(out_abs, os.path.basename(src))
if os.path.abspath(dst) == os.path.abspath(src):
sys.exit("refusing to overwrite input")
if src.endswith(".parquet"):
s, r = filter_parquet(src, dst, owners, repos, a.repo_column, counts)
else:
s, r = filter_jsonl(src, dst, owners, repos, a.repo_column, counts)
total_scanned += s
total_removed += r
files += 1
with open(os.path.join(out_abs, "removal_report.csv"), "w", newline="", encoding="utf-8") as f:
w = csv.writer(f)
w.writerow(["issue_number", "request_type", "owner", "repository", "rows_removed", "source_url"])
for n in sorted(counts):
m = meta.get(n, {})
w.writerow([n, m.get("request_type", ""), m.get("owner", ""), m.get("repository", ""), counts[n],
m.get("source_url", "")])
print(json.dumps({"files": files, "rows_scanned": total_scanned, "rows_removed": total_removed,
"rows_kept": total_scanned - total_removed, "requests_matched": len(counts)}))
if __name__ == "__main__":
main()
|