File size: 6,039 Bytes
cd3adb9 |
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 |
"""
Command-line interface for PubGuard.
Usage:
# Download datasets and train
pubguard train --data-dir ./data
# Download datasets only
pubguard prepare --data-dir ./data
# Screen a text file
pubguard screen input.txt
# Screen extracted PDF text from stdin
cat extracted_text.txt | pubguard screen -
# Batch screen NDJSON
pubguard batch input.ndjson output.ndjson
"""
import argparse
import json
import logging
import sys
import time
from pathlib import Path
from .classifier import PubGuard
from .config import PubGuardConfig
def cmd_prepare(args):
"""Download and prepare training datasets."""
from .data import prepare_all
prepare_all(Path(args.data_dir), n_per_class=args.n_per_class)
def cmd_train(args):
"""Prepare data (if needed) and train all heads."""
from .data import prepare_all
from .train import train_all
data_dir = Path(args.data_dir)
if args.download:
prepare_all(data_dir, n_per_class=args.n_per_class)
config = PubGuardConfig()
if args.models_dir:
config.models_dir = Path(args.models_dir)
train_all(data_dir, config=config, test_size=args.test_size)
def cmd_screen(args):
"""Screen a single document."""
config = PubGuardConfig()
if args.models_dir:
config.models_dir = Path(args.models_dir)
guard = PubGuard(config=config)
guard.initialize()
if args.input == "-":
text = sys.stdin.read()
else:
text = Path(args.input).read_text(errors="replace")
verdict = guard.screen(text)
if args.json:
print(json.dumps(verdict, indent=2))
else:
_print_verdict(verdict)
def cmd_batch(args):
"""Batch-screen an NDJSON file."""
config = PubGuardConfig()
if args.models_dir:
config.models_dir = Path(args.models_dir)
guard = PubGuard(config=config)
guard.initialize()
start = time.time()
processed = 0
with open(args.input) as fin, open(args.output, "w") as fout:
batch_texts = []
batch_records = []
for line in fin:
if not line.strip():
continue
record = json.loads(line)
text = record.get("text", "") or record.get("abstract", "") or ""
batch_texts.append(text)
batch_records.append(record)
if len(batch_texts) >= config.batch_size:
verdicts = guard.screen_batch(batch_texts)
for rec, verd in zip(batch_records, verdicts):
rec["pubguard"] = verd
fout.write(json.dumps(rec) + "\n")
processed += len(batch_texts)
batch_texts, batch_records = [], []
# Final batch
if batch_texts:
verdicts = guard.screen_batch(batch_texts)
for rec, verd in zip(batch_records, verdicts):
rec["pubguard"] = verd
fout.write(json.dumps(rec) + "\n")
processed += len(batch_texts)
elapsed = time.time() - start
rate = processed / elapsed if elapsed > 0 else 0
print(f"Screened {processed:,} records in {elapsed:.1f}s ({rate:,.0f} rec/s)")
print(f"Output: {args.output}")
def _print_verdict(v: dict):
"""Pretty-print a verdict."""
pass_icon = "✅" if v["pass"] else "❌"
print(f"\n{pass_icon} PubGuard Verdict: {'PASS' if v['pass'] else 'FAIL'}")
print(f" Document type: {v['doc_type']['label']:20s} (score: {v['doc_type']['score']:.3f})")
print(f" AI detection: {v['ai_generated']['label']:20s} (score: {v['ai_generated']['score']:.3f})")
print(f" Toxicity: {v['toxicity']['label']:20s} (score: {v['toxicity']['score']:.3f})")
print()
def main():
parser = argparse.ArgumentParser(
description="PubGuard — Scientific Publication Gatekeeper",
formatter_class=argparse.RawDescriptionHelpFormatter,
)
parser.add_argument(
"--verbose", "-v", action="store_true",
help="Enable verbose logging",
)
parser.add_argument(
"--models-dir", type=str, default=None,
help="Override models directory",
)
subparsers = parser.add_subparsers(dest="command")
# prepare
p_prepare = subparsers.add_parser("prepare", help="Download and prepare datasets")
p_prepare.add_argument("--data-dir", default="./pubguard_data")
p_prepare.add_argument("--n-per-class", type=int, default=15000)
# train
p_train = subparsers.add_parser("train", help="Train classification heads")
p_train.add_argument("--data-dir", default="./pubguard_data")
p_train.add_argument("--models-dir", default=None)
p_train.add_argument("--download", action="store_true", default=True,
help="Download datasets before training")
p_train.add_argument("--no-download", action="store_false", dest="download")
p_train.add_argument("--n-per-class", type=int, default=15000)
p_train.add_argument("--test-size", type=float, default=0.15)
# screen
p_screen = subparsers.add_parser("screen", help="Screen a single document")
p_screen.add_argument("input", help="Text file to screen (or - for stdin)")
p_screen.add_argument("--json", action="store_true", help="JSON output")
# batch
p_batch = subparsers.add_parser("batch", help="Batch screen NDJSON")
p_batch.add_argument("input", help="Input NDJSON file")
p_batch.add_argument("output", help="Output NDJSON file")
args = parser.parse_args()
level = logging.DEBUG if args.verbose else logging.INFO
logging.basicConfig(
level=level,
format="%(asctime)s | %(levelname)s | %(message)s",
datefmt="%Y-%m-%d %H:%M:%S",
)
if args.command == "prepare":
cmd_prepare(args)
elif args.command == "train":
cmd_train(args)
elif args.command == "screen":
cmd_screen(args)
elif args.command == "batch":
cmd_batch(args)
else:
parser.print_help()
if __name__ == "__main__":
main()
|