pdf

1#!/usr/bin/env python32"""Document metadata and file attachments for PDFs (pypdf).3 4Modes (one required):5  --set-meta            set metadata keys given via --title/--author/...6  --clear-meta          drop all document info metadata7  --attach FILE         embed a file attachment8  --list-attachments    list embedded attachment names9  --extract-attachments DIR  write all attachments into DIR10 11Metadata note: values are stored in the classic DocInfo dictionary12(Title/Author/Subject/Keywords). XMP metadata, if present, is not13rewritten and may disagree in sophisticated viewers.14"""15from __future__ import annotations16 17import argparse18import json19import os20import sys21from pathlib import Path22 23 24def main() -> int:25    for stream in (sys.stdout, sys.stderr):26        try:27            stream.reconfigure(encoding="utf-8")28        except Exception:29            pass30    parser = argparse.ArgumentParser(description="Set/clear PDF metadata; manage attachments.")31    parser.add_argument("pdf", help="Input PDF path")32    mode = parser.add_mutually_exclusive_group(required=True)33    mode.add_argument("--set-meta", action="store_true", help="Set metadata fields")34    mode.add_argument("--clear-meta", action="store_true", help="Remove all DocInfo metadata")35    mode.add_argument("--attach", metavar="FILE", help="Embed FILE as an attachment")36    mode.add_argument("--list-attachments", action="store_true", help="List attachment names")37    mode.add_argument("--extract-attachments", metavar="DIR", help="Extract attachments into DIR")38    parser.add_argument("-o", "--output", help="Output PDF (required for write modes)")39    parser.add_argument("--title")40    parser.add_argument("--author")41    parser.add_argument("--subject")42    parser.add_argument("--keywords")43    parser.add_argument("--password", help="Password if the input is encrypted")44    args = parser.parse_args()45 46    try:47        from pypdf import PdfReader, PdfWriter48    except ImportError:49        print("Missing dependency: install with 'python3 -m pip install pypdf'", file=sys.stderr)50        return 251 52    reader = PdfReader(args.pdf)53    if reader.is_encrypted:54        if args.password is None or not reader.decrypt(args.password):55            print("Error: input is encrypted; pass --password", file=sys.stderr)56            return 357 58    if args.list_attachments:59        names = list(reader.attachments.keys())60        json.dump({"attachment_count": len(names), "attachments": names}, sys.stdout,61                  ensure_ascii=False, indent=2)62        print()63        return 064 65    if args.extract_attachments:66        out_dir = Path(args.extract_attachments)67        out_dir.mkdir(parents=True, exist_ok=True)68        written = []69        for name, contents in reader.attachments.items():70            data = contents[0] if isinstance(contents, list) else contents71            safe = os.path.basename(name) or "attachment.bin"72            target = out_dir / safe73            with open(target, "wb") as fh:74                fh.write(bytes(data))75            written.append(str(target))76        json.dump({"extracted": written}, sys.stdout, ensure_ascii=False, indent=2)77        print()78        return 079 80    if not args.output:81        print("Error: -o/--output is required for write modes", file=sys.stderr)82        return 483 84    writer = PdfWriter()85    writer.append(reader)86 87    if args.set_meta:88        meta = {}89        for key, value in ((f"/{k.capitalize()}", getattr(args, k))90                           for k in ("title", "author", "subject", "keywords")):91            if value is not None:92                meta[key] = value93        if not meta:94            print("Error: --set-meta needs at least one of --title/--author/--subject/--keywords",95                  file=sys.stderr)96            return 497        writer.add_metadata(meta)98        result = {"output": args.output, "set": {k.lstrip("/"): v for k, v in meta.items()}}99    elif args.clear_meta:100        writer.metadata = None101        result = {"output": args.output, "cleared": True}102    else:  # --attach103        attach_path = Path(args.attach)104        with open(attach_path, "rb") as fh:105            writer.add_attachment(attach_path.name, fh.read())106        result = {"output": args.output, "attached": attach_path.name}107 108    with open(args.output, "wb") as fh:109        writer.write(fh)110    print(json.dumps(result, ensure_ascii=False))111    return 0112 113 114if __name__ == "__main__":115    sys.exit(main())116