clean_texts.py

#!/usr/bin/env python3
import argparse
import json
import re
from pathlib import Path

import orjson


def clean_text(t: str) -> str:
	t = t.strip()
	# normalize whitespace
	t = re.sub(r"\s+", " ", t)
	# remove leading/trailing punctuation artifacts
	t = re.sub(r"^[\-\–\—\.\,\;\:\!\?\"\']+\s*", "", t)
	t = re.sub(r"\s*[\-\–\—\.\,\;\:\!\?\"\']+$", "", t)
	# normalize quotes
	t = t.replace("’", "'").replace("“", '"').replace("”", '"')
	# lowercase except acronyms (simple heuristic)
	if not re.search(r"[A-Z]{2,}", t):
		t = t.lower()
	# remove filler tokens common in ASR
	t = re.sub(r"\b(euh+|heu+|hum+)\b", "", t)
	t = re.sub(r"\s+", " ", t).strip()
	return t


def main() -> None:
	parser = argparse.ArgumentParser(description="Clean transcript texts in-place or to a new file.")
	parser.add_argument("--jsonl", required=True, help="Input JSONL")
	parser.add_argument("--out", required=True, help="Output JSONL (can overwrite input)")
	args = parser.parse_args()

	with open(args.out, "w", encoding="utf-8") as out_f, open(args.jsonl, "r", encoding="utf-8") as in_f:
		for line in in_f:
			if not line.strip():
				continue
			rec = json.loads(line)
			if "text" in rec and rec["text"]:
				rec["text"] = clean_text(rec["text"])
			out_f.write(orjson.dumps(rec).decode("utf-8") + "\n")
	print(f"Cleaned texts -> {args.out}")


if __name__ == "__main__":
	main()