|
| 1 | +# Copyright (C) 2025 Siemens |
| 2 | +# |
| 3 | +# SPDX-License-Identifier: MIT |
| 4 | + |
| 5 | +import json |
| 6 | +import os |
| 7 | +import shutil |
| 8 | +import sys |
| 9 | +import tempfile |
| 10 | +import urllib.request |
| 11 | +import urllib.error |
| 12 | +import uuid |
| 13 | +from .input import SbomInput |
| 14 | +from datetime import datetime |
| 15 | + |
| 16 | +import logging |
| 17 | +import sys |
| 18 | + |
| 19 | + |
| 20 | +logger = logging.getLogger(__name__) |
| 21 | + |
| 22 | + |
| 23 | +class CompareCmd(SbomInput): |
| 24 | + """ |
| 25 | + Compare two SBOMs and generate a new SBOM containing only the additional components found in the target |
| 26 | + """ |
| 27 | + |
| 28 | + @classmethod |
| 29 | + def run(cls, args): |
| 30 | + with tempfile.TemporaryDirectory() as tmp_dir: |
| 31 | + base_sbom_tmp = os.path.join(tmp_dir, "base_sbom_tmp.json") |
| 32 | + target_sbom_tmp = os.path.join(tmp_dir, "target_sbom_tmp.json") |
| 33 | + |
| 34 | + cls.get_sbom_info(args.base_sbom, base_sbom_tmp) |
| 35 | + cls.get_sbom_info(args.target_sbom, target_sbom_tmp) |
| 36 | + |
| 37 | + with open(base_sbom_tmp) as f: |
| 38 | + base_sbom_data = json.load(f) |
| 39 | + with open(target_sbom_tmp) as f: |
| 40 | + target_sbom_data = json.load(f) |
| 41 | + |
| 42 | + base_sbom_fmt = cls.detect_sbom_format(base_sbom_data) |
| 43 | + target_sbom_fmt = cls.detect_sbom_format(target_sbom_data) |
| 44 | + |
| 45 | + if not base_sbom_fmt or not target_sbom_fmt: |
| 46 | + raise ValueError("can not detect SBOM format for one or both files") |
| 47 | + |
| 48 | + if base_sbom_fmt != target_sbom_fmt: |
| 49 | + raise ValueError("can not compare mixed SPDX and CycloneDX documents") |
| 50 | + |
| 51 | + if target_sbom_fmt == "spdx": |
| 52 | + base_sbom_comp, base_sbom = cls.load_spdx_sbom(base_sbom_tmp) |
| 53 | + target_sbom_comp, target_sbom = cls.load_spdx_sbom(target_sbom_tmp) |
| 54 | + extra_pkgs = cls.compare_items(base_sbom_comp, target_sbom_comp, "pkg") |
| 55 | + |
| 56 | + ref_creation_info = base_sbom.get("creationInfo", { |
| 57 | + "creators": ["Tool: sbom-diff-generator 1.0"], |
| 58 | + "created": datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%SZ") |
| 59 | + }) |
| 60 | + result = cls.build_extra_spdx(extra_pkgs, ref_creation_info) |
| 61 | + |
| 62 | + |
| 63 | + elif target_sbom_fmt == "cdx": |
| 64 | + base_sbom_comp, base_sbom = cls.load_cdx_sbom(base_sbom_tmp) |
| 65 | + target_sbom_comp, target_sbom = cls.load_cdx_sbom(target_sbom_tmp) |
| 66 | + |
| 67 | + extra_components = cls.compare_items(base_sbom_comp, target_sbom_comp, "component") |
| 68 | + result = cls.build_extra_cdx(extra_components, target_sbom) |
| 69 | + |
| 70 | + |
| 71 | + else: |
| 72 | + raise ValueError(f"Unsupported SBOM format: {new_fmt}") |
| 73 | + |
| 74 | + out_dir = os.path.dirname(args.out_file) |
| 75 | + if out_dir: |
| 76 | + os.makedirs(out_dir, exist_ok=True) |
| 77 | + |
| 78 | + with open(args.out_file, 'w', encoding='utf-8') as f: |
| 79 | + json.dump(result, f, indent=4) |
| 80 | + |
| 81 | + |
| 82 | + @classmethod |
| 83 | + def get_sbom_info(cls, sbom, sbom_tmp): |
| 84 | + """ |
| 85 | + Retrieve SBOM file content (from local path or URL) and store it in a temporary file. |
| 86 | +
|
| 87 | + Args: |
| 88 | + sbom (str): Path or URL to the SBOM file. |
| 89 | + sbom_tmp (str): Temporary destination path to copy/download the SBOM. |
| 90 | + """ |
| 91 | + if sbom.startswith(('http://', 'https://', 'ftp://')): |
| 92 | + try: |
| 93 | + with urllib.request.urlopen(sbom) as response, open(sbom_tmp, 'wb') as out_file: |
| 94 | + shutil.copyfileobj(response, out_file) |
| 95 | + except (urllib.error.URLError, urllib.error.HTTPError) as e: |
| 96 | + error(f"Unable to fetch: {sbom} — {e}") |
| 97 | + else: |
| 98 | + if not os.path.isfile(sbom): |
| 99 | + error(f"File not found: {sbom}") |
| 100 | + try: |
| 101 | + shutil.copy(sbom, sbom_tmp) |
| 102 | + except Exception as e: |
| 103 | + error(f"Error copying file: {e}") |
| 104 | + |
| 105 | + |
| 106 | + @classmethod |
| 107 | + def detect_sbom_format(cls, data): |
| 108 | + """ |
| 109 | + Detect SBOM format based on known top-level keys. |
| 110 | + Returns 'spdx' or 'cdx' or None. |
| 111 | + """ |
| 112 | + if "spdxVersion" in data: |
| 113 | + return "spdx" |
| 114 | + if data.get("bomFormat", "").lower() == "cyclonedx": |
| 115 | + return "cdx" |
| 116 | + return None |
| 117 | + |
| 118 | + |
| 119 | + @classmethod |
| 120 | + def load_spdx_sbom(cls, path): |
| 121 | + """Return packages keyed by purl or fallback name@version.""" |
| 122 | + with open(path) as f: |
| 123 | + data = json.load(f) |
| 124 | + |
| 125 | + packages = {} |
| 126 | + for pkg in data.get("packages", []): |
| 127 | + purl = next( |
| 128 | + (ref["referenceLocator"] |
| 129 | + for ref in pkg.get("externalRefs", []) |
| 130 | + if ref.get("referenceType") == "purl"), |
| 131 | + None |
| 132 | + ) |
| 133 | + if not purl: |
| 134 | + version = pkg.get("versionInfo", "") |
| 135 | + purl = f"{pkg.get('name')}@{version}" |
| 136 | + |
| 137 | + sha256 = next( |
| 138 | + (c["checksumValue"] |
| 139 | + for c in pkg.get("checksums", []) |
| 140 | + if c.get("algorithm", "").upper() == "SHA256"), |
| 141 | + None |
| 142 | + ) |
| 143 | + |
| 144 | + packages[purl] = {"pkg": pkg, "sha256": sha256} |
| 145 | + |
| 146 | + return packages, data |
| 147 | + |
| 148 | + |
| 149 | + @classmethod |
| 150 | + def build_extra_spdx(cls, extra_pkgs, ref_creation_info): |
| 151 | + """Build minimal SPDX 2.3 JSON document.""" |
| 152 | + return { |
| 153 | + "spdxVersion": "SPDX-2.3", |
| 154 | + "SPDXID": "SPDXRef-DOCUMENT", |
| 155 | + "name": "Extra Components SBOM", |
| 156 | + "dataLicense": "CC0-1.0", |
| 157 | + "documentNamespace": f"https://example.org/spdx/extra-{uuid.uuid4()}", |
| 158 | + "creationInfo": ref_creation_info, |
| 159 | + "packages": extra_pkgs |
| 160 | + } |
| 161 | + |
| 162 | + |
| 163 | + @classmethod |
| 164 | + def load_cdx_sbom(cls, path): |
| 165 | + """Return components keyed by purl or fallback name@version.""" |
| 166 | + with open(path) as f: |
| 167 | + data = json.load(f) |
| 168 | + |
| 169 | + components = {} |
| 170 | + for comp in data.get("components", []): |
| 171 | + purl = comp.get("purl") or f"{comp.get('name')}@{comp.get('version', '')}" |
| 172 | + sha256 = next( |
| 173 | + (h["content"] |
| 174 | + for h in comp.get("hashes", []) |
| 175 | + if h.get("alg", "").upper() == "SHA-256"), |
| 176 | + None |
| 177 | + ) |
| 178 | + components[purl] = {"component": comp, "sha256": sha256} |
| 179 | + |
| 180 | + return components, data |
| 181 | + |
| 182 | + |
| 183 | + @classmethod |
| 184 | + def build_extra_cdx(cls, extra_components, new_metadata=None): |
| 185 | + """Build minimal CycloneDX 1.5 JSON SBOM.""" |
| 186 | + sbom = { |
| 187 | + "bomFormat": "CycloneDX", |
| 188 | + "specVersion": "1.5", |
| 189 | + "version": 1, |
| 190 | + "serialNumber": f"urn:uuid:{uuid.uuid4()}", |
| 191 | + "components": extra_components |
| 192 | + } |
| 193 | + if new_metadata and "metadata" in new_metadata: |
| 194 | + sbom["metadata"] = new_metadata["metadata"] |
| 195 | + else: |
| 196 | + sbom["metadata"] = { |
| 197 | + "timestamp": datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%SZ"), |
| 198 | + "tools": [{"name": "sbom-diff-generator", "version": "1.0"}] |
| 199 | + } |
| 200 | + return sbom |
| 201 | + |
| 202 | + |
| 203 | + @classmethod |
| 204 | + def compare_items(cls, base_sbom_comp, target_sbom_comp, key_name): |
| 205 | + """Generic comparison for SPDX or CDX items keyed by purl.""" |
| 206 | + extra = [] |
| 207 | + for key, new_info in target_sbom_comp.items(): |
| 208 | + base_info = base_sbom_comp.get(key) |
| 209 | + new_sha = (new_info["sha256"] or "").lower().strip() |
| 210 | + ref_sha = ((base_info or {}).get("sha256") or "").lower().strip() |
| 211 | + |
| 212 | + if base_info is None or (ref_sha and new_sha and ref_sha != new_sha): |
| 213 | + extra.append(new_info[key_name]) |
| 214 | + return extra |
| 215 | + |
| 216 | + |
| 217 | + @classmethod |
| 218 | + def setup_parser(cls, parser): |
| 219 | + cls.parser_add_sbom_input_args(parser) |
| 220 | + parser.add_argument( |
| 221 | + "-b", |
| 222 | + "--base-sbom", |
| 223 | + required=True, |
| 224 | + help="Path or URL to the base (reference) SBOM file" |
| 225 | + ) |
| 226 | + |
| 227 | + parser.add_argument( |
| 228 | + "-n", |
| 229 | + "--target-sbom", |
| 230 | + required=True, |
| 231 | + help="Path or URL to the target (new) SBOM file" |
| 232 | + ) |
| 233 | + |
| 234 | + parser.add_argument( |
| 235 | + "-o", |
| 236 | + "--out-file", |
| 237 | + default="uncleared_components.json", |
| 238 | + help="Path to the output JSON file (default: uncleared_components.json)" |
| 239 | + ) |
0 commit comments