|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +import argparse |
| 4 | +import logging |
| 5 | +from pathlib import Path |
| 6 | +from typing import Optional, Sequence |
| 7 | + |
| 8 | +from pybis import Openbis |
| 9 | + |
| 10 | +from logbook2mouse.logbook_reader import Logbook2MouseReader |
| 11 | + |
| 12 | +from .config import UploadConfig |
| 13 | +from .logging_utils import setup_logger |
| 14 | +from .uploader import OpenBISUploader |
| 15 | +from .utils import read_token, validate_ymd |
| 16 | +from .failures import FailureRecorder |
| 17 | + |
| 18 | + |
| 19 | +def build_parser() -> argparse.ArgumentParser: |
| 20 | + d = UploadConfig() # defaults from class |
| 21 | + p = argparse.ArgumentParser( |
| 22 | + prog="mouse-uploader", |
| 23 | + description="Upload MOUSE measurement batches to OpenBIS using a YMD filter.", |
| 24 | + ) |
| 25 | + |
| 26 | + p.add_argument("ymd", type=validate_ymd, help="Measurement day code, e.g. 20251220") |
| 27 | + |
| 28 | + p.add_argument("--ds-username", default=d.ds_username, help=f"Username label for logging (default: {d.ds_username})") |
| 29 | + p.add_argument("--logbook-path", type=Path, default=d.logbook_path, help=f"Excel logbook (default: {d.logbook_path})") |
| 30 | + p.add_argument("--proposal-base-path", type=Path, default=d.proposal_base_path, help=f"Proposal base path (default: {d.proposal_base_path})") |
| 31 | + p.add_argument("--base-data-path", type=Path, default=d.base_data_path, help=f"Base data path (default: {d.base_data_path})") |
| 32 | + p.add_argument("--datastore-token-path", type=Path, default=d.datastore_token_path, help=f"Token file path (default: {d.datastore_token_path})") |
| 33 | + |
| 34 | + p.add_argument("--space-name", default=d.space_name, help=f"Space name (default: {d.space_name})") |
| 35 | + p.add_argument("--projects-prepend", default=d.projects_prepend, help=f"Project prefix (default: {d.projects_prepend})") |
| 36 | + p.add_argument("--start-row", type=int, default=d.start_row, help=f"Start row index (default: {d.start_row})") |
| 37 | + |
| 38 | + p.add_argument("--server-url", default=d.server_url, help=f"OpenBIS URL (default: {d.server_url})") |
| 39 | + p.add_argument("--sleep-seconds-between-ops", type=float, default=d.sleep_seconds_between_ops, help=f"Sleep between ops (default: {d.sleep_seconds_between_ops})") |
| 40 | + p.add_argument("--sleep-seconds-between-datasets", type=float, default=d.sleep_seconds_between_datasets, help=f"Sleep between datasets (default: {d.sleep_seconds_between_datasets})") |
| 41 | + |
| 42 | + p.add_argument("--instrument-name-pattern", default=d.instrument_name_pattern, help=f"Instrument name pattern (default: {d.instrument_name_pattern})") |
| 43 | + p.add_argument("--people-collection-prefix", default=d.people_collection_prefix, help=f"People collection prefix (default: {d.people_collection_prefix})") |
| 44 | + |
| 45 | + p.add_argument("--raw-dataset-type", default=d.raw_dataset_type, help=f"Raw dataset type (default: {d.raw_dataset_type})") |
| 46 | + p.add_argument("--processed-dataset-type", default=d.processed_dataset_type, help=f"Processed dataset type (default: {d.processed_dataset_type})") |
| 47 | + p.add_argument("--log-file", type=Path, default=None, help="Log file path (default: None, logs to stdout only)") |
| 48 | + p.add_argument("--failure-file", type=Path, default=Path("upload_failures.jsonl"), help="Failure records file (default: upload_failures.jsonl)") |
| 49 | + |
| 50 | + p.add_argument( |
| 51 | + "--log-level", |
| 52 | + default="INFO", |
| 53 | + choices=["DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"], |
| 54 | + help="Logging verbosity", |
| 55 | + ) |
| 56 | + p.add_argument("--dry-run", action="store_true", help="No writes; log intended actions") |
| 57 | + |
| 58 | + return p |
| 59 | + |
| 60 | + |
| 61 | +def _validate_args(args: argparse.Namespace) -> None: |
| 62 | + if args.start_row < 0: |
| 63 | + raise SystemExit("--start-row must be >= 0") |
| 64 | + if args.sleep_seconds_between_ops < 0: |
| 65 | + raise SystemExit("--sleep-seconds-between-ops must be >= 0") |
| 66 | + if args.sleep_seconds_between_datasets < 0: |
| 67 | + raise SystemExit("--sleep-seconds-between-datasets must be >= 0") |
| 68 | + |
| 69 | + if not args.logbook_path.is_file(): |
| 70 | + raise SystemExit(f"Logbook file not found: {args.logbook_path}") |
| 71 | + if not args.proposal_base_path.exists(): |
| 72 | + raise SystemExit(f"Proposal base path does not exist: {args.proposal_base_path}") |
| 73 | + if not args.base_data_path.exists(): |
| 74 | + raise SystemExit(f"Base data path does not exist: {args.base_data_path}") |
| 75 | + if not args.datastore_token_path.is_file(): |
| 76 | + raise SystemExit(f"Token file not found: {args.datastore_token_path}") |
| 77 | + |
| 78 | + |
| 79 | +def main(argv: Optional[Sequence[str]] = None) -> int: |
| 80 | + parser = build_parser() |
| 81 | + args = parser.parse_args(argv) |
| 82 | + _validate_args(args) |
| 83 | + |
| 84 | + log_level = getattr(logging, args.log_level.upper(), logging.INFO) |
| 85 | + logger = setup_logger(level=log_level) |
| 86 | + |
| 87 | + cfg = UploadConfig( |
| 88 | + ymd_filter=args.ymd, |
| 89 | + ds_username=args.ds_username, |
| 90 | + logbook_path=args.logbook_path, |
| 91 | + proposal_base_path=args.proposal_base_path, |
| 92 | + base_data_path=args.base_data_path, |
| 93 | + datastore_token_path=args.datastore_token_path, |
| 94 | + space_name=args.space_name, |
| 95 | + projects_prepend=args.projects_prepend, |
| 96 | + start_row=args.start_row, |
| 97 | + server_url=args.server_url, |
| 98 | + sleep_seconds_between_ops=args.sleep_seconds_between_ops, |
| 99 | + sleep_seconds_between_datasets=args.sleep_seconds_between_datasets, |
| 100 | + instrument_name_pattern=args.instrument_name_pattern, |
| 101 | + people_collection_prefix=args.people_collection_prefix, |
| 102 | + raw_dataset_type=args.raw_dataset_type, |
| 103 | + processed_dataset_type=args.processed_dataset_type, |
| 104 | + ) |
| 105 | + |
| 106 | + token = read_token(cfg.datastore_token_path) |
| 107 | + ds = Openbis(url=cfg.server_url, verify_certificates=True) |
| 108 | + ds.set_token(token) |
| 109 | + logger.info("Connected to OpenBIS at %s as %s", cfg.server_url, cfg.ds_username) |
| 110 | + |
| 111 | + reader = Logbook2MouseReader( |
| 112 | + cfg.logbook_path, |
| 113 | + project_base_path=cfg.proposal_base_path, |
| 114 | + load_all=True, |
| 115 | + ) |
| 116 | + |
| 117 | + logger = setup_logger(level=log_level, log_file=args.log_file) |
| 118 | + failure_recorder = FailureRecorder(args.failure_file) |
| 119 | + |
| 120 | + uploader = OpenBISUploader(ds=ds, config=cfg, logger=logger, dry_run=args.dry_run, failure_recorder=failure_recorder) |
| 121 | + uploader.process_entries(reader) |
| 122 | + logger.info(f"Upload run completed. Failures recorded (if any) to: {args.failure_file}") |
| 123 | + return 0 |
0 commit comments