|
| 1 | +import logging |
| 2 | +from django.core.management.base import BaseCommand |
| 3 | + |
| 4 | +from osf.models import Preprint |
| 5 | + |
| 6 | +logger = logging.getLogger(__name__) |
| 7 | + |
| 8 | + |
| 9 | +def reindex_versioned_preprints(dry_run=False, batch_size=100, provider_id=None, guids=None): |
| 10 | + if guids: |
| 11 | + preprints = Preprint.objects.filter(guids___id__in=guids) |
| 12 | + else: |
| 13 | + preprints = Preprint.objects.filter(versioned_guids__isnull=False).distinct() |
| 14 | + |
| 15 | + if provider_id: |
| 16 | + preprints = preprints.filter(provider___id=provider_id) |
| 17 | + |
| 18 | + preprints = preprints.filter(is_published=True) |
| 19 | + |
| 20 | + total_count = preprints.count() |
| 21 | + logger.info(f'{"[DRY RUN] " if dry_run else ""}Found {total_count} versioned preprints to re-index') |
| 22 | + |
| 23 | + if total_count == 0: |
| 24 | + logger.info('No preprints to re-index') |
| 25 | + return |
| 26 | + |
| 27 | + processed = 0 |
| 28 | + for preprint in preprints.iterator(chunk_size=batch_size): |
| 29 | + processed += 1 |
| 30 | + |
| 31 | + if dry_run: |
| 32 | + logger.info( |
| 33 | + f'[DRY RUN] Would re-index preprint {preprint._id} ' |
| 34 | + f'(version {preprint.versioned_guids.first().version if preprint.versioned_guids.exists() else "N/A"}, ' |
| 35 | + f'date_created_first_version={preprint.date_created_first_version}) ' |
| 36 | + f'[{processed}/{total_count}]' |
| 37 | + ) |
| 38 | + else: |
| 39 | + try: |
| 40 | + preprint.update_search() |
| 41 | + if processed % 10 == 0: |
| 42 | + logger.info( |
| 43 | + f'Re-indexed preprint {preprint._id} ' |
| 44 | + f'(version {preprint.versioned_guids.first().version if preprint.versioned_guids.exists() else "N/A"}) ' |
| 45 | + f'[{processed}/{total_count}]' |
| 46 | + ) |
| 47 | + except Exception as e: |
| 48 | + logger.error(f'Failed to re-index preprint {preprint._id}: {e}') |
| 49 | + |
| 50 | + logger.info( |
| 51 | + f'{"[DRY RUN] " if dry_run else ""}Completed. ' |
| 52 | + f'{"Would have re-indexed" if dry_run else "Re-indexed"} {processed} preprints' |
| 53 | + ) |
| 54 | + |
| 55 | + |
| 56 | +class Command(BaseCommand): |
| 57 | + help = ( |
| 58 | + 'Re-index all versioned preprints to Elasticsearch to ensure computed properties ' |
| 59 | + 'like date_created_first_version are up to date.' |
| 60 | + ) |
| 61 | + |
| 62 | + def add_arguments(self, parser): |
| 63 | + super().add_arguments(parser) |
| 64 | + parser.add_argument( |
| 65 | + '--dry-run', |
| 66 | + action='store_true', |
| 67 | + dest='dry_run', |
| 68 | + help='Preview what would be re-indexed without actually making changes', |
| 69 | + ) |
| 70 | + parser.add_argument( |
| 71 | + '--batch-size', |
| 72 | + type=int, |
| 73 | + default=100, |
| 74 | + help='Number of preprints to process in each batch (default: 100)', |
| 75 | + ) |
| 76 | + parser.add_argument( |
| 77 | + '--provider', |
| 78 | + type=str, |
| 79 | + help='Optional provider ID to filter preprints', |
| 80 | + ) |
| 81 | + parser.add_argument( |
| 82 | + '--guids', |
| 83 | + type=str, |
| 84 | + nargs='+', |
| 85 | + help='Optional list of specific preprint GUIDs to re-index', |
| 86 | + ) |
| 87 | + |
| 88 | + def handle(self, *args, **options): |
| 89 | + dry_run = options.get('dry_run', False) |
| 90 | + batch_size = options.get('batch_size', 100) |
| 91 | + provider_id = options.get('provider') |
| 92 | + guids = options.get('guids') |
| 93 | + |
| 94 | + if dry_run: |
| 95 | + logger.info('=' * 60) |
| 96 | + logger.info('DRY RUN MODE - No changes will be made') |
| 97 | + logger.info('=' * 60) |
| 98 | + |
| 99 | + reindex_versioned_preprints( |
| 100 | + dry_run=dry_run, |
| 101 | + batch_size=batch_size, |
| 102 | + provider_id=provider_id, |
| 103 | + guids=guids |
| 104 | + ) |
0 commit comments