Spaces:
Sleeping
Sleeping
File size: 6,330 Bytes
d10c06c |
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 |
#!/usr/bin/env python3
"""
merge_chunks.py
Merge multiple JSONL chunk files into a single chunks.jsonl, with safety checks:
- validates required fields
- fails on duplicate chunk_id
- optional: merges manifests into one manifest.json
- outputs a merge_report.json (doc_id counts, totals, sources)
Usage examples:
# simplest: merge two chunk files
python scripts/merge_chunks.py \
--out data/normalized_v1_1/chunks.jsonl \
data/normalized_v1_1/chunks_books.jsonl \
data/normalized_v1_1/chunks_articles.jsonl
# also merge manifests (optional)
python scripts/merge_chunks.py \
--out data/normalized_v1_1/chunks.jsonl \
--out-manifest data/normalized_v1_1/manifest.json \
--manifest data/normalized_v1_1/manifest_books.json \
--manifest data/normalized_v1_1/manifest_articles.json \
data/normalized_v1_1/chunks_books.jsonl \
data/normalized_v1_1/chunks_articles.jsonl
"""
import argparse
import json
import os
from collections import Counter
from datetime import datetime
from typing import Dict, Iterable, List, Optional
REQUIRED_FIELDS = ("chunk_id", "doc_id", "doc_title", "text")
def iter_jsonl(path: str) -> Iterable[Dict]:
with open(path, "r", encoding="utf-8") as f:
for line_no, line in enumerate(f, start=1):
line = line.strip()
if not line:
continue
try:
obj = json.loads(line)
except json.JSONDecodeError as e:
raise RuntimeError(f"Invalid JSON in {path} at line {line_no}: {e}") from e
yield obj
def validate_record(obj: Dict, path: str) -> None:
missing = [k for k in REQUIRED_FIELDS if k not in obj or obj[k] in (None, "")]
if missing:
raise RuntimeError(f"Missing required fields {missing} in {path} for chunk_id={obj.get('chunk_id')!r}")
def load_manifest(path: str) -> Dict:
with open(path, "r", encoding="utf-8") as f:
return json.load(f)
def merge_manifests(manifest_paths: List[str]) -> Dict:
"""
Assumes each manifest has:
{
"generated_at": "...",
"documents": [ { "id": ..., "title": ..., "format": ..., "filename": ..., ... }, ... ]
}
We combine documents by `id` (first wins, duplicates skipped).
"""
combined_docs = []
seen_ids = set()
for p in manifest_paths:
mf = load_manifest(p)
for doc in mf.get("documents", []):
doc_id = doc.get("id")
if not doc_id:
continue
if doc_id in seen_ids:
continue
seen_ids.add(doc_id)
combined_docs.append(doc)
return {
"generated_at": datetime.utcnow().isoformat() + "Z",
"documents": combined_docs,
"sources": manifest_paths,
}
def main():
ap = argparse.ArgumentParser()
ap.add_argument("--out", required=True, help="Output merged chunks.jsonl path")
ap.add_argument("--out-manifest", default=None, help="Output merged manifest.json path (optional)")
ap.add_argument("--manifest", action="append", default=[], help="Input manifest.json path (repeatable, optional)")
ap.add_argument("--out-report", default=None, help="Output merge report JSON path (default: <out_dir>/merge_report.json)")
ap.add_argument("--allow-duplicate-chunk-id", action="store_true",
help="If set, duplicate chunk_id records are skipped instead of failing (NOT recommended).")
ap.add_argument("inputs", nargs="+", help="Input chunk JSONL files to merge, in order")
args = ap.parse_args()
out_path = args.out
out_dir = os.path.dirname(out_path) or "."
os.makedirs(out_dir, exist_ok=True)
report_path = args.out_report or os.path.join(out_dir, "merge_report.json")
seen_chunk_ids = set()
doc_counts = Counter()
total_written = 0
inputs_summary = []
with open(out_path, "w", encoding="utf-8") as out_f:
for in_path in args.inputs:
if not os.path.exists(in_path):
raise FileNotFoundError(in_path)
in_written = 0
in_skipped_dupes = 0
for obj in iter_jsonl(in_path):
validate_record(obj, in_path)
cid = str(obj["chunk_id"])
if cid in seen_chunk_ids:
if args.allow_duplicate_chunk_id:
in_skipped_dupes += 1
continue
raise RuntimeError(f"Duplicate chunk_id detected: {cid} (from {in_path}). Aborting.")
seen_chunk_ids.add(cid)
out_f.write(json.dumps(obj, ensure_ascii=False) + "\n")
total_written += 1
in_written += 1
doc_counts[str(obj["doc_id"])] += 1
inputs_summary.append({
"input": in_path,
"written": in_written,
"skipped_duplicate_chunk_id": in_skipped_dupes,
})
# Optional: merged manifest
if args.out_manifest:
if not args.manifest:
raise RuntimeError("--out-manifest provided but no --manifest inputs were provided.")
merged = merge_manifests(args.manifest)
with open(args.out_manifest, "w", encoding="utf-8") as f:
json.dump(merged, f, indent=2, ensure_ascii=False)
# Merge report
report = {
"generated_at": datetime.utcnow().isoformat() + "Z",
"out_chunks": out_path,
"out_manifest": args.out_manifest,
"total_chunks_written": total_written,
"unique_chunk_ids": len(seen_chunk_ids),
"doc_id_counts": dict(doc_counts),
"inputs": inputs_summary,
"notes": [
"If total_chunks_written != unique_chunk_ids, you allowed duplicate chunk_id skipping.",
"Pin this merged file as your dataset v1.1.0 before embedding/indexing.",
],
}
with open(report_path, "w", encoding="utf-8") as f:
json.dump(report, f, indent=2, ensure_ascii=False)
print(f"[OK] Wrote merged chunks: {out_path}")
if args.out_manifest:
print(f"[OK] Wrote merged manifest: {args.out_manifest}")
print(f"[OK] Wrote merge report: {report_path}")
print(f"[OK] Total chunks: {total_written} | Unique chunk_ids: {len(seen_chunk_ids)}")
if __name__ == "__main__":
main()
|