|
| 1 | +# |
| 2 | +# Copyright (c) nexB Inc. and others. All rights reserved. |
| 3 | +# purldb is a trademark of nexB Inc. |
| 4 | +# SPDX-License-Identifier: Apache-2.0 |
| 5 | +# See http://www.apache.org/licenses/LICENSE-2.0 for the license text. |
| 6 | +# See https://github.com/nexB/purldb for support or download. |
| 7 | +# See https://aboutcode.org for more information about nexB OSS projects. |
| 8 | +# |
| 9 | + |
| 10 | +import logging |
| 11 | +import requests |
| 12 | +from packageurl import PackageURL |
| 13 | + |
| 14 | +from minecode import priority_router |
| 15 | +from minecode.miners.dockerhub import build_package_data |
| 16 | +from packagedb.models import PackageContentType |
| 17 | + |
| 18 | +logger = logging.getLogger(__name__) |
| 19 | +handler = logging.StreamHandler() |
| 20 | +logger.addHandler(handler) |
| 21 | +logger.setLevel(logging.INFO) |
| 22 | + |
| 23 | + |
| 24 | +def fetch_dockerhub_repo_summary(name, namespace="library"): |
| 25 | + """ |
| 26 | + Fetch summary metadata for a Docker Hub repository. |
| 27 | +
|
| 28 | + Returns: |
| 29 | + dict or None: Full metadata JSON from the Docker Hub API, including: |
| 30 | + - description (str): Short description |
| 31 | + - full_description (str): Detailed description |
| 32 | + - is_private (bool): Privacy status |
| 33 | +
|
| 34 | + """ |
| 35 | + url = f"https://hub.docker.com/v2/repositories/{namespace}/{name}/" |
| 36 | + try: |
| 37 | + response = requests.get(url) |
| 38 | + response.raise_for_status() |
| 39 | + return response.json() |
| 40 | + except requests.exceptions.RequestException as err: |
| 41 | + logger.error(f"Error fetching repository metadata for {name}: {err}") |
| 42 | + return None |
| 43 | + |
| 44 | + |
| 45 | +def fetch_dockerhub_tags_metadata(name, namespace, tag=None): |
| 46 | + """ |
| 47 | + Search through Docker Hub tags for a given repository. |
| 48 | + - If `tag` is provided, return the JSON metadata for that tag (by name or digest). |
| 49 | + - If `tag` is None, return a list of all tag metadata. |
| 50 | +
|
| 51 | + Examples: |
| 52 | + fetch_dockerhub_tag_metadata("nginx", "1.25.2") |
| 53 | + fetch_dockerhub_tag_metadata("nginx", "sha256:3d8957cb61d0223de2ab1aa2ec91d29796eb82a81cdcc1e968c090c29606d648") |
| 54 | + fetch_dockerhub_tag_metadata("nginx") # returns all tags |
| 55 | +
|
| 56 | + """ |
| 57 | + page = 0 |
| 58 | + page_size = 100 |
| 59 | + all_results = [] |
| 60 | + |
| 61 | + while True: |
| 62 | + page += 1 |
| 63 | + url = f"https://hub.docker.com/v2/repositories/{namespace}/{name}/tags/?page={page}&page_size={page_size}" |
| 64 | + try: |
| 65 | + response = requests.get(url) |
| 66 | + response.raise_for_status() |
| 67 | + data = response.json() |
| 68 | + |
| 69 | + results = data.get("results", []) |
| 70 | + if not tag: |
| 71 | + all_results.extend(results) # collect everything |
| 72 | + else: |
| 73 | + for result in results: |
| 74 | + if tag.startswith("sha256") and result.get("digest") == tag: |
| 75 | + return [result] |
| 76 | + elif result.get("name") == tag: |
| 77 | + return [result] |
| 78 | + |
| 79 | + # Check if more pages exist |
| 80 | + if not data.get("next") or page_size * page > data.get("count", 0): |
| 81 | + break # no more pages |
| 82 | + |
| 83 | + except requests.exceptions.RequestException as err: |
| 84 | + logger.error(f"Error fetching tags for {name}, page {page}: {err}") |
| 85 | + return None |
| 86 | + |
| 87 | + if not tag: |
| 88 | + return all_results # return collected list |
| 89 | + |
| 90 | + return None # tag not found |
| 91 | + |
| 92 | + |
| 93 | +def map_dockerhub_package(package_url, pipelines, priority=0): |
| 94 | + """ |
| 95 | + Add a Dockerhub distribution `package_url` to the PackageDB. |
| 96 | + """ |
| 97 | + from minecode.model_utils import add_package_to_scan_queue |
| 98 | + from minecode.model_utils import merge_or_create_package |
| 99 | + |
| 100 | + if not package_url.name: |
| 101 | + error = f"Missing package name in DockerHub Package URL: {package_url}" |
| 102 | + logger.error(error) |
| 103 | + return error |
| 104 | + |
| 105 | + namespace = package_url.namespace or "library" |
| 106 | + summary = fetch_dockerhub_repo_summary(package_url.name, namespace) |
| 107 | + if not summary: |
| 108 | + error = f"Package does not exist on dockerhub: {package_url}" |
| 109 | + logger.error(error) |
| 110 | + return error |
| 111 | + |
| 112 | + tags_metadata = fetch_dockerhub_tags_metadata(package_url.name, namespace, package_url.version) |
| 113 | + |
| 114 | + packages = build_package_data(summary, tags_metadata, package_url) |
| 115 | + |
| 116 | + error = None |
| 117 | + for package in packages: |
| 118 | + package.extra_data["package_content"] = PackageContentType.SOURCE_ARCHIVE |
| 119 | + db_package, _, _, error = merge_or_create_package(package, visit_level=0) |
| 120 | + if error: |
| 121 | + break |
| 122 | + |
| 123 | + if db_package: |
| 124 | + add_package_to_scan_queue(package=db_package, pipelines=pipelines, priority=priority) |
| 125 | + return error |
| 126 | + |
| 127 | + |
| 128 | +@priority_router.route("pkg:docker/.*") |
| 129 | +def process_request(purl_str, **kwargs): |
| 130 | + """ |
| 131 | + Process Dockerhub Package URL (PURL). |
| 132 | + ex: |
| 133 | + pkg:docker/nginx@latest |
| 134 | + pkg:docker/nginx@sha256:3d8957cb61d0223de2ab1aa2ec91d29796eb82a81cdcc1e968c090c29606d648 |
| 135 | + """ |
| 136 | + from minecode.model_utils import DEFAULT_PIPELINES |
| 137 | + |
| 138 | + addon_pipelines = kwargs.get("addon_pipelines", []) |
| 139 | + pipelines = DEFAULT_PIPELINES + tuple(addon_pipelines) |
| 140 | + priority = kwargs.get("priority", 0) |
| 141 | + |
| 142 | + package_url = PackageURL.from_string(purl_str) |
| 143 | + |
| 144 | + error_msg = map_dockerhub_package(package_url, pipelines, priority) |
| 145 | + |
| 146 | + if error_msg: |
| 147 | + return error_msg |
0 commit comments