@cdxgen/cdxgen 13.0.0 → 13.1.0

package/data/rules/package-integrity.yaml

@@ -250,7 +250,7 @@
       and $prop($, 'cdx:cargo:hasNativeBuild') = 'true'
       and $count($$.components[
         $prop($, 'cdx:github:action:ecosystem') = 'cargo'
-        and $contains($prop($, 'cdx:github:action:role'), 'toolchain')
+        and $contains($nullSafeProp($, 'cdx:github:action:role'), 'toolchain')
         and $prop($, 'cdx:github:action:versionPinningType') != 'sha'
       ]) > 0
     ]
@@ -267,7 +267,7 @@
       "buildScriptCapabilities": $prop($, 'cdx:cargo:buildScriptCapabilities'),
       "toolchainActions": $$.components[
         $prop($, 'cdx:github:action:ecosystem') = 'cargo'
-        and $contains($prop($, 'cdx:github:action:role'), 'toolchain')
+        and $contains($nullSafeProp($, 'cdx:github:action:role'), 'toolchain')
         and $prop($, 'cdx:github:action:versionPinningType') != 'sha'
       ].$prop($, 'cdx:github:action:uses')
     }
@@ -283,17 +283,17 @@
       $prop($, 'cdx:rust:buildTool') = 'cargo'
       and $prop($, 'cdx:cargo:hasNativeBuild') = 'true'
       and (
-        $contains($prop($, 'cdx:cargo:buildScriptCapabilities'), 'process-execution')
-        or $contains($prop($, 'cdx:cargo:buildScriptCapabilities'), 'network-access')
-        or $contains($prop($, 'cdx:cargo:nativeBuildIndicators'), '-sys')
+        $contains($nullSafeProp($, 'cdx:cargo:buildScriptCapabilities'), 'process-execution')
+        or $contains($nullSafeProp($, 'cdx:cargo:buildScriptCapabilities'), 'network-access')
+        or $contains($nullSafeProp($, 'cdx:cargo:nativeBuildIndicators'), '-sys')
       )
       and $count($$.components[
         $prop($, 'cdx:github:step:usesCargo') = 'true'
         and (
-          $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'build')
-          or $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'test')
-          or $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'package')
-          or $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'publish')
+          $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'build')
+          or $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'test')
+          or $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'package')
+          or $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'publish')
         )
       ]) > 0
     ]
@@ -311,10 +311,10 @@
       "cargoCommands": $$.components[
         $prop($, 'cdx:github:step:usesCargo') = 'true'
         and (
-          $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'build')
-          or $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'test')
-          or $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'package')
-          or $contains($prop($, 'cdx:github:step:cargoSubcommands'), 'publish')
+          $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'build')
+          or $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'test')
+          or $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'package')
+          or $contains($nullSafeProp($, 'cdx:github:step:cargoSubcommands'), 'publish')
         )
       ].$prop($, 'cdx:github:step:command')
     }

package/lib/audit/index.js

@@ -6,9 +6,11 @@ import process from "node:process";
 import { createBom } from "../cli/index.js";
 import {
   isDryRun,
+  isSecureMode,
   readEnvironmentVariable,
   recordActivity,
 } from "../core/activity.js";
+import { getRemediationCatalog } from "../core/buildLedger.js";
 import {
   getTmpDir,
   safeExistsSync,
@@ -53,6 +55,12 @@ import {
   isHbomLikeBom,
   isObomLikeBom,
 } from "../stages/postgen/auditBom.js";
+import {
+  isIntrospectionEnabled,
+  reflectOnRun,
+} from "../stages/postgen/introspection/reflect.js";
+import { buildIntrospectionJson } from "../stages/postgen/introspection/report.js";
+import { scoreReflection } from "../stages/postgen/introspection/score.js";
 import { postProcess } from "../stages/postgen/postgen.js";
 import { formatTargetLabel } from "./progress.js";
 import { renderAuditReport } from "./reporters.js";
@@ -259,7 +267,7 @@ export async function runDirectBomAuditFromBoms(inputBoms, options = {}) {
       ...directAuditOptions,
       getBomWithOras,
     });
-    results.push({
+    const result = {
       auditOptions: directAuditOptions,
       bomFormat: getCycloneDxFormat(inputBom.bomJson),
       findings,
@@ -268,7 +276,43 @@ export async function runDirectBomAuditFromBoms(inputBoms, options = {}) {
       specVersion: inputBom.bomJson?.specVersion,
       status: "audited",
       summary: summarizeDirectAuditFindings(findings),
-    });
+    };
+    if (isIntrospectionEnabled(options)) {
+      // The audited BOM was not generated by this run, so there is no ledger
+      // of its own; the verdict rests on BOM structure alone and says so.
+      try {
+        const reflection = await reflectOnRun(inputBom.bomJson, options, {
+          ledgerEvents: [],
+          ledgerSource: "none",
+          projectPath: inputBom.source || "",
+        });
+        reflection.scoring = scoreReflection(
+          reflection,
+          getRemediationCatalog(),
+          {
+            secureMode: isSecureMode,
+            inContainer:
+              readEnvironmentVariable("CDXGEN_IN_CONTAINER") === "true",
+          },
+        );
+        reflection.bom = {
+          serialNumber: inputBom.bomJson?.serialNumber,
+          componentCount: Array.isArray(inputBom.bomJson?.components)
+            ? inputBom.bomJson.components.length
+            : 0,
+        };
+        result.introspection = buildIntrospectionJson(
+          reflection,
+          reflection.scoring,
+          options,
+        );
+      } catch (err) {
+        console.warn(
+          `Build introspection failed to complete and was skipped: ${err?.message}`,
+        );
+      }
+    }
+    results.push(result);
   }
   const summary = {
     findingsBySeverity: {
@@ -1744,7 +1788,7 @@ async function recheckFindingsAgainstDefaultBranch(
     if (!bomNSData?.bomJson) {
       return undefined;
     }
-    const processedBomNSData = postProcess(
+    const processedBomNSData = await postProcess(
       bomNSData,
       childOptions,
       defaultSourceSelection.scanDir,
@@ -1948,7 +1992,7 @@ export async function auditTarget(target, options) {
           options,
         );
       }
-      const processedBomNSData = postProcess(
+      const processedBomNSData = await postProcess(
         bomNSData,
         childOptions,
         sourceSelection.scanDir,

package/lib/audit/scoring.js

@@ -1,17 +1,11 @@
 import { thoughtLog } from "../core/logger.js";
+import { SEVERITY_ORDER } from "../core/severity.js";
 import {
   hasRegistryProvenanceEvidenceProperties,
   hasTrustedPublishingProperties,
 } from "../inventory/provenanceUtils.js";
 
-/** Map of severity names to numeric ordering used for threshold comparisons. */
-export const SEVERITY_ORDER = {
-  none: -1,
-  low: 0,
-  medium: 1,
-  high: 2,
-  critical: 3,
-};
+export { SEVERITY_ORDER };
 
 const BASE_FINDING_WEIGHT = {
   low: 4,

package/lib/cli/bomAssembly.js

@@ -200,6 +200,8 @@ export const determineParentComponent = (options) => {
  * @param {object|undefined} detectedParent Parent the generator derived
  * @param {object[]} components The component list, appended to in place
  * @param {object[]} dependencies Edges the generator produced
+ * @param {object} options CLI options
+ * @param {string} ptype Package type
  * @returns {object[]} Edges to emit, re-anchored when an override applies
  */
 function anchorDetectedParent(
@@ -207,6 +209,8 @@ function anchorDetectedParent(
   detectedParent,
   components,
   dependencies,
+  options,
+  ptype,
 ) {
   const detectedRef = detectedParent?.["bom-ref"];
   const overrideRef = overrideParent?.["bom-ref"];
@@ -220,7 +224,25 @@ function anchorDetectedParent(
     Array.isArray(components) &&
     !components.some((comp) => comp["bom-ref"] === detectedRef)
   ) {
-    components.push({ ...detectedParent });
+    // The detected parent carries transient keys (`license`, `evidence`,
+    // `_integrity`, `qualifiers`, `repository`, `homepage`) that only
+    // `metadata.component` is cleaned of. As a regular component it must go
+    // through the same conversion or schema validation rejects the document
+    // (issue #4320). The copy keeps the caller's context object untouched.
+    //
+    // Its workspace members go through `listComponents`, the same conversion
+    // `addMetadata` gives them when the detected parent is the one reaching
+    // `metadata.component`. Without it they keep the raw parser shape
+    // (`homepage: { url }`) that the nested Component schema rejects
+    // (issue #4326).
+    components.push(
+      cleanParentComponent({
+        ...detectedParent,
+        components: detectedParent.components?.length
+          ? listComponents(options, {}, detectedParent.components, ptype)
+          : undefined,
+      }),
+    );
   }
   // The caller owns the array it passed, and postProcess merges these edges
   // across generators, so the re-anchoring is returned rather than spliced in.
@@ -1302,8 +1324,7 @@ export function processHashes(pkg, component) {
       addComponentHash(ahash.alg, ahash.content, component);
     }
   } else if (pkg._shasum) {
-    const ahash = { alg: "SHA-1", content: pkg._shasum };
-    component.hashes.push(ahash);
+    addComponentHash("SHA-1", pkg._shasum, component);
   } else if (pkg._integrity) {
     const integrity = parse(pkg._integrity) || {};
     // Components may have multiple hashes with various lengths. Check each one
@@ -1326,24 +1347,69 @@ export function processHashes(pkg, component) {
   }
 }
 
+// Digest length in hex characters for each algorithm CycloneDX names. A digest
+// that decodes to the wrong length for its algorithm is corrupt, whatever
+// encoding it arrived in.
+const HASH_HEX_LENGTHS = {
+  MD5: 32,
+  "SHA-1": 40,
+  "SHA-256": 64,
+  "SHA-384": 96,
+  "SHA-512": 128,
+  "SHA3-256": 64,
+  "SHA3-384": 96,
+  "SHA3-512": 128,
+  "BLAKE2b-256": 64,
+  "BLAKE2b-384": 96,
+  "BLAKE2b-512": 128,
+};
+
+// Lengths the CycloneDX hash-content pattern accepts, for algorithms not in the
+// table above.
+const HASH_HEX_LENGTHS_ANY = new Set([32, 40, 64, 96, 128]);
+
 /**
  * Adds a hash to component.
+ *
+ * Registries hand out digests in hex or in the base64 form npm uses for
+ * `integrity`, and both are accepted. A digest that is neither, or that decodes
+ * to the wrong length for its algorithm, is dropped rather than copied through:
+ * CycloneDX constrains `hash.content` to a hex digest, so emitting it anyway
+ * fails schema validation for the whole document over one bad package.
  */
 export function addComponentHash(alg, digest, component) {
-  let hash;
-  // If it is a valid hash simply use it
+  const hash = normalizeHashContent(alg, digest);
+  if (!hash) {
+    return;
+  }
+  component.hashes.push({ alg: alg, content: hash });
+}
+
+/**
+ * Normalize a digest to the lower-case hex form CycloneDX requires.
+ *
+ * @param {string} alg Hash algorithm name
+ * @param {string} digest Digest in hex or base64
+ * @returns {string|undefined} Hex digest, or undefined when it cannot be trusted
+ */
+export function normalizeHashContent(alg, digest) {
+  if (!digest || typeof digest !== "string") {
+    return undefined;
+  }
+  const expectedLength = HASH_HEX_LENGTHS[alg];
+  const isExpectedLength = (value) =>
+    expectedLength
+      ? value.length === expectedLength
+      : HASH_HEX_LENGTHS_ANY.has(value.length);
+
   if (new RegExp(HASH_PATTERN).test(digest)) {
-    hash = digest;
-  } else {
-    // Check if base64 encoded
-    const isBase64Encoded =
-      Buffer.from(digest, "base64").toString("base64") === digest;
-    hash = isBase64Encoded
-      ? Buffer.from(digest, "base64").toString("hex")
-      : digest;
+    return isExpectedLength(digest) ? digest : undefined;
   }
-  const ahash = { alg: alg, content: hash };
-  component.hashes.push(ahash);
+  // npm and friends carry the digest base64-encoded. Node decodes leniently, so
+  // the decoded length is the check that matters: a truncated or invented value
+  // decodes to something, just not to a digest of the right size.
+  const decoded = Buffer.from(digest, "base64").toString("hex");
+  return isExpectedLength(decoded) ? decoded : undefined;
 }
 
 /**
@@ -1397,6 +1463,8 @@ export const buildBomNSData = (options, pkgInfo, ptype, context) => {
     context.parentComponent,
     components,
     context.dependencies || [],
+    options,
+    ptype,
   );
   if (
     components &&

package/lib/cli/cliOptions.js

@@ -14,6 +14,7 @@
 import { join, resolve } from "node:path";
 
 import { isSecureMode, readEnvironmentVariable } from "../core/activity.js";
+import { PROJECT_TYPE_ALIASES } from "../core/env.js";
 import {
   DEFAULT_CDX_SPEC_VERSION,
   normalizeCycloneDxComponentTypeFilter,
@@ -87,6 +88,21 @@ const POST_BUILD_SUPPORTED_PROJECT_TYPES = new Set([
 // Helpers
 // ---------------------------------------------------------------------------
 
+/**
+ * The formulation privacy notice, shared by every path that turns formulation
+ * collection on automatically: the warning about emails and secrets is
+ * identical whichever caller adds the section.
+ *
+ * @returns {OptionsWarning} The notice.
+ */
+function formulationPrivacyNotice() {
+  return {
+    level: "info",
+    message:
+      "NOTE: Automatically collecting formulation information. The section may include sensitive data such as emails and secrets.\nPlease review the generated SBOM before distribution or LLM training.",
+  };
+}
+
 /**
  * Check whether a value was explicitly provided by the user (i.e., is not
  * the yargs default). Yargs always populates default values, so we cannot
@@ -174,7 +190,39 @@ function resolveOutputTarget(args, { filePath, isRemoteOrPurl }) {
     : resolve(join(filePath, args.output));
 }
 
-export function buildInitialOptions(args, { filePath, isRemoteOrPurl }) {
+/**
+ * Resolve the introspection opt-in from the flag and the environment.
+ *
+ * The parser maps `CDXGEN_INTROSPECT` onto the same boolean the flag sets, so
+ * a value outside the documented opt-in vocabulary — an exported but empty
+ * variable most of all — arrives indistinguishable from `--no-introspect`.
+ * The command line is therefore the authority on refusal, and the environment
+ * speaks only through `true` and `1`. An unmentioned flag stays undefined so a
+ * profile can still enable it.
+ *
+ * @param {Object} args Parsed CLI arguments.
+ * @param {string[]} argv Raw command-line arguments the parser was given.
+ * @returns {boolean|undefined} The requested state, or undefined when nothing asked.
+ */
+function resolveIntrospectOption(args, argv) {
+  const refused =
+    args.introspect === false &&
+    argv.some((argument) =>
+      ["--no-introspect", "--introspect=false"].includes(argument),
+    );
+  if (refused) {
+    return false;
+  }
+  return args.introspect === true ||
+    ["true", "1"].includes(readEnvironmentVariable("CDXGEN_INTROSPECT"))
+    ? true
+    : undefined;
+}
+
+export function buildInitialOptions(
+  args,
+  { filePath, isRemoteOrPurl, argv = process.argv },
+) {
   const res = Object.assign({}, args, {
     projectType: args.type,
     multiProject: args.recurse,
@@ -187,6 +235,10 @@ export function buildInitialOptions(args, { filePath, isRemoteOrPurl }) {
     include: args.include || args.includeRegex,
     noIgnore: args.noIgnore,
     gitAiNotesRef: args.gitAiNotesRef || "refs/notes/ai",
+    // Introspection: the flag or the documented environment opt-in. An
+    // explicit --no-introspect stays `false` and beats both, while an
+    // unmentioned flag stays undefined so a profile can still enable it.
+    introspect: resolveIntrospectOption(args, argv),
     // License normalization and enrichment are on by default; the only escape
     // hatch is the corresponding environment variable.
     licenseEnhance:
@@ -217,6 +269,24 @@ export function buildInitialOptions(args, { filePath, isRemoteOrPurl }) {
   return res;
 }
 
+/**
+ * Check whether the supplied project types include an OS inventory type
+ * (`os` or an alias such as `linux`, `windows`, `macos`).
+ *
+ * @param {string|string[]|undefined|null} projectTypes Project types.
+ * @returns {boolean} True when an OS inventory type is requested.
+ */
+function hasOsProjectType(projectTypes) {
+  if (!projectTypes) {
+    return false;
+  }
+  const values = Array.isArray(projectTypes) ? projectTypes : [projectTypes];
+  const osTypes = new Set(["os", ...(PROJECT_TYPE_ALIASES.os || [])]);
+  return values
+    .flatMap((projectType) => `${projectType}`.split(","))
+    .some((projectType) => osTypes.has(projectType.trim().toLowerCase()));
+}
+
 /**
  * @typedef {object} OptionsWarning
  * @property {"info"|"warn"|"error"} level   Severity level
@@ -263,6 +333,17 @@ export function applyPostConstructionOverrides(options, context) {
             "The cbom command does not support --component-type. Use cdxgen with --include-crypto when you need component-type filtering.",
         });
       }
+      // Crypto evidence collection analyses source code with atom, which is
+      // meaningless and extremely slow for OS inventories (discussion #4310).
+      // Refuse the combination and direct users to obom.
+      if (hasOsProjectType(options.projectType)) {
+        warnings.push({
+          level: "error",
+          message:
+            "The cbom command does not support OS inventory project types (-t os). Use the obom command instead to generate an OBOM for an operating system installation.",
+        });
+        return warnings;
+      }
       options.includeCrypto = true;
     }
     options.evidence = true;
@@ -288,6 +369,20 @@ export function applyPostConstructionOverrides(options, context) {
     options.installDeps = false;
   }
 
+  // --- Introspection CI gate threshold -----------------------------------
+  if (options.introspectFailBelow !== undefined) {
+    const threshold = Number(options.introspectFailBelow);
+    if (!Number.isFinite(threshold) || threshold < 0 || threshold > 100) {
+      warnings.push({
+        level: "error",
+        message:
+          "--introspect-fail-below requires a number between 0 and 100, where absent means the gate never fails.",
+      });
+    } else {
+      options.introspectFailBelow = threshold;
+    }
+  }
+
   // When a standard is requested, the minimum spec version is 1.6
   if (options.standard) {
     if (!userSetSpecVersion) {
@@ -351,6 +446,8 @@ export function applyAdvancedOptions(options, context = {}) {
   const secureMode = context.isSecureMode ?? isSecureMode;
   const warnings = [];
   const isHbomOnly = isHbomOnlyProjectTypes(options.projectType);
+  const includeFormulationBeforeProfiles = options.includeFormulation === true;
+  const introspectExplicitlyDisabled = options.introspect === false;
 
   // --- Profile expansion -------------------------------------------------
   switch (options.profile) {
@@ -382,6 +479,14 @@ export function applyAdvancedOptions(options, context = {}) {
     case "license-compliance":
       process.env.FETCH_LICENSE = "true";
       break;
+    case "introspect":
+      options.introspect = true;
+      options.introspectAnnotate = options.introspectAnnotate !== false;
+      // The ledger's own record of what ran; formulation is where a consumer
+      // looks for the build environment the SBOM was produced in.
+      options.includeFormulation = !isHbomOnly;
+      options.evidence = true;
+      break;
     case "ml-tiny":
       process.env.FETCH_LICENSE = "true";
       options.deep = false;
@@ -441,17 +546,29 @@ export function applyAdvancedOptions(options, context = {}) {
     }
   }
 
+  // An explicit --no-introspect negates the introspect profile's headline
+  // capability while keeping the bundle's other effects: the user asked for
+  // the profile minus introspection.
+  if (options.profile === "introspect" && introspectExplicitlyDisabled) {
+    options.introspect = false;
+    warnings.push({
+      level: "info",
+      message:
+        "NOTE: Introspection stays off because --no-introspect was passed explicitly; the other introspect profile settings still apply.",
+    });
+  }
+
   // --- BOM audit → auto-formulation --------------------------------------
-  if (options.bomAudit) {
+  const introspectProfileAddsFormulation =
+    options.profile === "introspect" && !includeFormulationBeforeProfiles;
+  if (options.bomAudit || introspectProfileAddsFormulation) {
     if (isHbomOnly) {
       // HBOM-only: skip formulation
     } else if (!options.includeFormulation) {
-      warnings.push({
-        level: "info",
-        message:
-          "NOTE: Automatically collecting formulation information. The section may include sensitive data such as emails and secrets.\nPlease review the generated SBOM before distribution or LLM training.",
-      });
+      warnings.push(formulationPrivacyNotice());
       options.includeFormulation = true;
+    } else if (introspectProfileAddsFormulation) {
+      warnings.push(formulationPrivacyNotice());
     }
   }