|
| 1 | +from datetime import datetime |
| 2 | +from datetime import timezone |
| 3 | + |
| 4 | +from aboutcode.pipeline import LoopProgress |
| 5 | +from django.db import migrations |
| 6 | +from packageurl import PackageURL |
| 7 | + |
| 8 | +CHUNK_SIZE = 50000 |
| 9 | +BATCH_SIZE = 500 |
| 10 | + |
| 11 | + |
| 12 | +class Migration(migrations.Migration): |
| 13 | + def fix_alpine_purl_type(apps, schema_editor): |
| 14 | + """Use proper apk package type for Alpine""" |
| 15 | + |
| 16 | + Package = apps.get_model("vulnerabilities", "Package") |
| 17 | + batch = [] |
| 18 | + alpine_packages_query = Package.objects.filter(type="alpine") |
| 19 | + |
| 20 | + log(f"\nFixing PURL for {alpine_packages_query.count():,d} alpine packages") |
| 21 | + progress = LoopProgress( |
| 22 | + total_iterations=alpine_packages_query.count(), |
| 23 | + progress_step=10, |
| 24 | + logger=log, |
| 25 | + ) |
| 26 | + for package in progress.iter(alpine_packages_query.iterator(chunk_size=CHUNK_SIZE)): |
| 27 | + package.type = "apk" |
| 28 | + package.namespace = "alpine" |
| 29 | + |
| 30 | + package.package_url = update_alpine_purl(package.package_url, "apk", "alpine") |
| 31 | + package.plain_package_url = update_alpine_purl( |
| 32 | + package.plain_package_url, "apk", "alpine" |
| 33 | + ) |
| 34 | + |
| 35 | + batch.append(package) |
| 36 | + if len(batch) >= BATCH_SIZE: |
| 37 | + bulk_update_package(Package, batch) |
| 38 | + batch.clear() |
| 39 | + |
| 40 | + bulk_update_package(Package, batch) |
| 41 | + |
| 42 | + def reverse_fix_alpine_purl_type(apps, schema_editor): |
| 43 | + Package = apps.get_model("vulnerabilities", "Package") |
| 44 | + batch = [] |
| 45 | + alpine_packages_query = Package.objects.filter(type="apk", namespace="alpine") |
| 46 | + |
| 47 | + log(f"\nREVERSE: Fix for {alpine_packages_query.count():,d} alpine packages") |
| 48 | + progress = LoopProgress( |
| 49 | + total_iterations=alpine_packages_query.count(), |
| 50 | + progress_step=10, |
| 51 | + logger=log, |
| 52 | + ) |
| 53 | + for package in progress.iter(alpine_packages_query.iterator(chunk_size=CHUNK_SIZE)): |
| 54 | + package.type = "alpine" |
| 55 | + package.namespace = "" |
| 56 | + |
| 57 | + package.package_url = update_alpine_purl(package.package_url, "alpine", "") |
| 58 | + package.plain_package_url = update_alpine_purl(package.plain_package_url, "alpine", "") |
| 59 | + |
| 60 | + batch.append(package) |
| 61 | + if len(batch) >= BATCH_SIZE: |
| 62 | + bulk_update_package(Package, batch) |
| 63 | + batch.clear() |
| 64 | + |
| 65 | + bulk_update_package(Package, batch) |
| 66 | + |
| 67 | + dependencies = [ |
| 68 | + ("vulnerabilities", "0087_update_alpine_advisory_created_by"), |
| 69 | + ] |
| 70 | + |
| 71 | + operations = [ |
| 72 | + migrations.RunPython( |
| 73 | + code=fix_alpine_purl_type, |
| 74 | + reverse_code=reverse_fix_alpine_purl_type, |
| 75 | + ), |
| 76 | + ] |
| 77 | + |
| 78 | + |
| 79 | +def bulk_update_package(package, batch): |
| 80 | + if batch: |
| 81 | + package.objects.bulk_update( |
| 82 | + objs=batch, |
| 83 | + fields=[ |
| 84 | + "type", |
| 85 | + "namespace", |
| 86 | + "package_url", |
| 87 | + "plain_package_url", |
| 88 | + ], |
| 89 | + ) |
| 90 | + |
| 91 | + |
| 92 | +def update_alpine_purl(purl, purl_type, purl_namespace): |
| 93 | + package_url = PackageURL.from_string(purl).to_dict() |
| 94 | + package_url["type"] = purl_type |
| 95 | + package_url["namespace"] = purl_namespace |
| 96 | + return str(PackageURL(**package_url)) |
| 97 | + |
| 98 | + |
| 99 | +def log(message): |
| 100 | + now_local = datetime.now(timezone.utc).astimezone() |
| 101 | + timestamp = now_local.strftime("%Y-%m-%d %H:%M:%S.%f")[:-3] |
| 102 | + message = f"{timestamp} {message}" |
| 103 | + print(message) |
0 commit comments