@cdxgen/cdxgen 0.0.1 → 13.0.0

package/lib/validator/bomValidator.js

@@ -0,0 +1,985 @@
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
+
+import { Purl } from "@cdxgen/cdx-purl";
+import Ajv from "ajv";
+import Ajv2020 from "ajv/dist/2020.js";
+import addFormats from "ajv-formats";
+
+import { DEBUG_MODE, readEnvironmentVariable } from "../core/activity.js";
+import { thoughtLog } from "../core/logger.js";
+import { dirNameStr } from "../core/paths.js";
+import { SPDX_JSONLD_CONTEXT, SPDX_SPEC_VERSION } from "../core/spdx.js";
+import {
+  isCycloneDxSpecVersionAtLeast,
+  toCycloneDxSpecVersionString,
+} from "../inventory/bomUtils.js";
+import { isPartialTree } from "../inventory/depsUtils.js";
+import { isValidPurl } from "../inventory/purl.js";
+
+const dirName = dirNameStr;
+// The v13 floor is spec 1.6. 1.4 and 1.5 were dropped here so that the JS and
+// Rust (`cdxrs validate`) validators agree on what they accept: cdxrs supports
+// 1.6/1.7 only, and while the two sets differed, whether a 1.5 BOM passed
+// validation depended on whether a cdxrs binary happened to be installed.
+const SUPPORTED_CYCLONEDX_SCHEMA_VERSIONS = new Set(["1.6", "1.7", "2.0"]);
+const PLACEHOLDER_COMPONENT_NAMES = new Set(["app", "application", "project"]);
+const SPDX_EXPORT_TYPES = new Set([
+  "CreationInfo",
+  "Relationship",
+  "SpdxDocument",
+  "software_File",
+  "software_Package",
+]);
+const NPM_NATIVE_ADDON_EVIDENCE_PROPERTIES = new Set([
+  "cdx:npm:native_addon",
+  "cdx:npm:has_binary",
+  "cdx:npm:os",
+  "cdx:npm:cpu",
+]);
+let spdxExportSchemaValidator;
+const cycloneDxSchemaValidators = new Map();
+
+/**
+ * Compile a schema validator at runtime.
+ *
+ * This is the fallback used when the pre-compiled module from
+ * `contrib/gen-validators.mjs` is not available (e.g. during development
+ * before `node contrib/gen-validators.mjs` has been run). It pays the full
+ * ~114 ms compilation cost on the first call per spec version per process.
+ */
+const compileSchemaValidator = (specVersion) => {
+  let validate;
+  if (isCycloneDxSpecVersionAtLeast(specVersion, "2.0")) {
+    const ajv = new Ajv2020(AJV_OPTIONS);
+    useLinearUniqueItems(ajv);
+    addFormats(ajv);
+    addDraft2020BundledSchema(
+      ajv,
+      readJsonSchema("cryptography-defs.schema.json"),
+      "https://cyclonedx.org/schema/cryptography-defs.schema.json",
+    );
+    addDraft2020BundledSchema(
+      ajv,
+      readJsonSchema("jsf-0.82.schema.json"),
+      "https://cyclonedx.org/schema/jsf-0.82.schema.json",
+    );
+    addDraft2020BundledSchema(
+      ajv,
+      readJsonSchema("spdx.schema.json"),
+      "https://cyclonedx.org/schema/spdx.schema.json",
+    );
+    validate = ajv.compile(readJsonSchema("cyclonedx-2.0-bundled.schema.json"));
+  } else {
+    const schemas = [
+      readJsonSchema(`bom-${specVersion}.schema.json`),
+      readJsonSchema("jsf-0.82.schema.json"),
+      readJsonSchema("spdx.schema.json"),
+    ];
+    if (isCycloneDxSpecVersionAtLeast(specVersion, "1.7")) {
+      schemas.push(readJsonSchema("cryptography-defs.schema.json"));
+    }
+    const ajv = new Ajv({
+      ...AJV_OPTIONS,
+      schemas,
+    });
+    useLinearUniqueItems(ajv);
+    addFormats(ajv);
+    validate = ajv.getSchema(
+      `http://cyclonedx.org/schema/bom-${specVersion}.schema.json`,
+    );
+  }
+  return validate;
+};
+
+/**
+ * Return the schema validator for a CycloneDX spec version.
+ *
+ * Prefers the pre-compiled module generated by `contrib/gen-validators.mjs`
+ * (skips ~114 ms of ajv compilation per process). Falls back to runtime
+ * compilation if the generated file is missing. Set
+ * `CDXGEN_PRECOMPILED_VALIDATORS=0` to always compile, which is how the two
+ * paths are compared for equivalence.
+ *
+ * Returns a Promise that resolves to the validator function, so callers must
+ * `await` the result. The promise is cached so repeated calls for the same
+ * version resolve instantly.
+ *
+ * @param {string} specVersion CycloneDX spec version ("1.6", "1.7", "2.0")
+ * @returns {Promise<Function>} Validator function (call with data to validate)
+ */
+const getCycloneDxSchemaValidator = (specVersion) => {
+  if (cycloneDxSchemaValidators.has(specVersion)) {
+    return cycloneDxSchemaValidators.get(specVersion);
+  }
+  const precompiledDisabled = ["0", "false"].includes(
+    (readEnvironmentVariable("CDXGEN_PRECOMPILED_VALIDATORS") || "")
+      .trim()
+      .toLowerCase(),
+  );
+  const promise = precompiledDisabled
+    ? Promise.resolve(compileSchemaValidator(specVersion))
+    : import(`./generated/validate-${specVersion}.mjs`)
+        .then((mod) => mod.default || mod.validate)
+        .catch((err) => {
+          // The two validators are equivalent, so falling back only costs the
+          // compilation time. Saying so keeps a missing or unreadable generated
+          // module from looking like a validation result.
+          thoughtLog(
+            `Pre-compiled validator for CycloneDX ${specVersion} is unavailable (${err.message}); compiling the schema instead.`,
+          );
+          return compileSchemaValidator(specVersion);
+        });
+  cycloneDxSchemaValidators.set(specVersion, promise);
+  return promise;
+};
+
+const AJV_OPTIONS = {
+  strict: false,
+  logger: false,
+  verbose: true,
+  code: {
+    source: true,
+    lines: true,
+    optimize: true,
+  },
+};
+
+/**
+ * Serialise a value so that two values are deep-equal exactly when their
+ * serialisations match: object keys are emitted in sorted order and every
+ * string is length-prefixed so that no two distinct values can collide.
+ *
+ * @param {*} value Value from a parsed BOM document
+ * @returns {string} Canonical form of the value
+ */
+const canonicalize = (value) => {
+  if (value === null || typeof value !== "object") {
+    return typeof value === "string"
+      ? `s${value.length}:${value}`
+      : `${typeof value}:${value}`;
+  }
+  if (Array.isArray(value)) {
+    return `[${value.map(canonicalize).join(",")}]`;
+  }
+  const keys = Object.keys(value).sort();
+  return `{${keys.map((key) => `s${key.length}:${key}=${canonicalize(value[key])}`).join(",")}}`;
+};
+
+/**
+ * Replace Ajv's `uniqueItems` implementation with one that runs in linear time.
+ *
+ * Ajv compares every pair of items with a deep equality check, which is
+ * quadratic in the length of the array. The CycloneDX schema sets the keyword
+ * on 46 arrays, `components` among them, so a BOM describing a container image
+ * spends the bulk of its validation there. Comparing canonical serialisations
+ * through a set decides the same question in one pass.
+ *
+ * @param {object} ajv Ajv instance whose validators are about to be compiled
+ */
+const useLinearUniqueItems = (ajv) => {
+  ajv.removeKeyword("uniqueItems");
+  ajv.addKeyword({
+    keyword: "uniqueItems",
+    type: "array",
+    schemaType: "boolean",
+    errors: false,
+    error: {
+      message: "must NOT have duplicate items",
+    },
+    validate: (schema, data) => {
+      if (schema === false || !Array.isArray(data) || data.length < 2) {
+        return true;
+      }
+      const seen = new Set();
+      for (const item of data) {
+        const key = canonicalize(item);
+        if (seen.has(key)) {
+          return false;
+        }
+        seen.add(key);
+      }
+      return true;
+    },
+  });
+};
+
+const readJsonSchema = (fileName) =>
+  JSON.parse(readFileSync(join(dirName, "data", fileName), "utf-8"));
+
+const addDraft2020BundledSchema = (ajv, schema, schemaId) => {
+  const bundledSchema = { ...schema };
+  delete bundledSchema.$schema;
+  bundledSchema.$id = schemaId;
+  ajv.addSchema(bundledSchema);
+};
+
+const getSpdxElementId = (element) => element?.spdxId || element?.["@id"];
+
+const getSpdxExportSchemaValidator = () => {
+  if (spdxExportSchemaValidator !== undefined) {
+    return spdxExportSchemaValidator;
+  }
+  try {
+    const spdxExportSchema = JSON.parse(
+      readFileSync(join(dirName, "data", "spdx-export.schema.json"), "utf-8"),
+    );
+    const ajv = new Ajv2020({
+      strict: false,
+      validateSchema: false,
+      logger: false,
+      verbose: true,
+      code: {
+        source: true,
+        lines: true,
+        optimize: true,
+      },
+    });
+    addFormats(ajv);
+    spdxExportSchemaValidator = ajv.compile(spdxExportSchema);
+  } catch (_error) {
+    spdxExportSchemaValidator = null;
+  }
+  return spdxExportSchemaValidator;
+};
+
+/**
+ * Validate the generated bom using jsonschema
+ *
+ * @param {object} bomJson content
+ *
+ * @returns {Promise<Boolean>} true if the BOM is valid. false otherwise.
+ */
+export const validateBom = async (bomJson) => {
+  if (!bomJson) {
+    return true;
+  }
+  const specVersion = toCycloneDxSpecVersionString(bomJson.specVersion);
+  if (!SUPPORTED_CYCLONEDX_SCHEMA_VERSIONS.has(specVersion)) {
+    console.log(
+      `Unsupported CycloneDX specVersion '${bomJson.specVersion}'. Supported versions are ${[
+        ...SUPPORTED_CYCLONEDX_SCHEMA_VERSIONS,
+      ].join(", ")}.`,
+    );
+    return false;
+  }
+  const validate = await getCycloneDxSchemaValidator(specVersion);
+  const isValid = validate(bomJson);
+  if (!isValid) {
+    if (bomJson.metadata?.component?.name) {
+      console.log(
+        `Schema validation failed for ${bomJson.metadata.component.name}`,
+      );
+    } else {
+      console.log("Schema validation failed");
+    }
+    console.log(validate.errors);
+    return false;
+  }
+  // Deep validation tests
+  return (
+    validateMetadata(bomJson) &&
+    validatePurls(bomJson) &&
+    validateRefs(bomJson) &&
+    validateProps(bomJson)
+  );
+};
+
+/**
+ * Validate the generated SPDX export.
+ *
+ * @param {object|string} spdxJson SPDX json object
+ * @returns {boolean} true if the SPDX export is valid
+ */
+export const validateSpdx = (spdxJson) => {
+  if (!spdxJson) {
+    return true;
+  }
+  if (typeof spdxJson === "string" || spdxJson instanceof String) {
+    spdxJson = JSON.parse(spdxJson);
+  }
+  const spdxDocument = (spdxJson?.["@graph"] || []).find(
+    (element) => element?.type === "SpdxDocument",
+  );
+  const validateSchema = getSpdxExportSchemaValidator();
+  if (!validateSchema) {
+    console.log("The bundled SPDX export schema is empty or malformed.");
+    return false;
+  }
+  if (!validateSchema(spdxJson)) {
+    console.log(
+      `SPDX 3.0.1 schema validation failed${spdxDocument?.name ? ` for ${spdxDocument.name}` : ""}`,
+    );
+    console.log(validateSchema.errors);
+    return false;
+  }
+  const errorList = [];
+  if (spdxJson?.["@context"] !== SPDX_JSONLD_CONTEXT) {
+    errorList.push(`@context must be '${SPDX_JSONLD_CONTEXT}'.`);
+  }
+  if (!Array.isArray(spdxJson?.["@graph"]) || !spdxJson["@graph"].length) {
+    errorList.push("@graph must be a non-empty array.");
+  }
+  const ids = new Set();
+  const knownRefs = new Set();
+  let creationInfoCount = 0;
+  let documentCount = 0;
+  let usesExtensionProfile = false;
+  for (const element of spdxJson?.["@graph"] || []) {
+    if (!SPDX_EXPORT_TYPES.has(element?.type)) {
+      errorList.push(`Unsupported SPDX export type '${element?.type}'.`);
+    }
+    const elementId = getSpdxElementId(element);
+    if (!elementId || typeof elementId !== "string") {
+      errorList.push(
+        `Missing SPDX identifier for type '${element?.type || "unknown"}'.`,
+      );
+      continue;
+    }
+    if (ids.has(elementId)) {
+      errorList.push(`Duplicate SPDX identifier '${elementId}'.`);
+    }
+    ids.add(elementId);
+    knownRefs.add(elementId);
+    if (Array.isArray(element?.extension) && element.extension.length) {
+      usesExtensionProfile = true;
+    }
+    switch (element.type) {
+      case "CreationInfo":
+        creationInfoCount += 1;
+        if (element.specVersion !== SPDX_SPEC_VERSION) {
+          errorList.push(
+            `CreationInfo '${elementId}' has unexpected specVersion '${element.specVersion}'.`,
+          );
+        }
+        if (!Array.isArray(element.createdBy) || !element.createdBy.length) {
+          errorList.push(`CreationInfo '${elementId}' must include createdBy.`);
+        }
+        if (!element.created) {
+          errorList.push(`CreationInfo '${elementId}' must include created.`);
+        }
+        break;
+      case "SpdxDocument":
+        documentCount += 1;
+        if (!element.creationInfo) {
+          errorList.push(
+            `SpdxDocument '${elementId}' must include creationInfo.`,
+          );
+        }
+        if (!Array.isArray(element.element) || !element.element.length) {
+          errorList.push(
+            `SpdxDocument '${elementId}' must include element refs.`,
+          );
+        }
+        if (
+          element.rootElement &&
+          (!Array.isArray(element.rootElement) || !element.rootElement.length)
+        ) {
+          errorList.push(
+            `SpdxDocument '${elementId}' rootElement must be a non-empty array when present.`,
+          );
+        }
+        break;
+      case "Relationship":
+        if (
+          !element.creationInfo ||
+          !element.from ||
+          typeof element.from !== "string"
+        ) {
+          errorList.push(
+            `Relationship '${elementId}' must include creationInfo and from.`,
+          );
+        }
+        if (!element.to || !Array.isArray(element.to) || !element.to.length) {
+          errorList.push(`Relationship '${elementId}' must include to refs.`);
+        }
+        if (element.relationshipType !== "dependsOn") {
+          errorList.push(
+            `Relationship '${elementId}' has unsupported relationshipType '${element.relationshipType}'.`,
+          );
+        }
+        break;
+      case "software_File":
+      case "software_Package":
+        if (!element.creationInfo || !element.name) {
+          errorList.push(
+            `${element.type} '${elementId}' must include creationInfo and name.`,
+          );
+        }
+        break;
+      default:
+        break;
+    }
+  }
+  if (creationInfoCount !== 1) {
+    errorList.push(
+      `Expected exactly one CreationInfo, found ${creationInfoCount}.`,
+    );
+  }
+  if (documentCount !== 1) {
+    errorList.push(
+      `Expected exactly one SpdxDocument, found ${documentCount}.`,
+    );
+  }
+  for (const element of spdxJson?.["@graph"] || []) {
+    const elementId = getSpdxElementId(element);
+    if (element.creationInfo && !knownRefs.has(element.creationInfo)) {
+      errorList.push(
+        `Element '${elementId || "unknown"}' references unknown creationInfo '${element.creationInfo}'.`,
+      );
+    }
+    if (Array.isArray(element.element)) {
+      for (const ref of element.element) {
+        if (!knownRefs.has(ref)) {
+          errorList.push(
+            `SpdxDocument '${elementId || "unknown"}' references unknown element '${ref}'.`,
+          );
+        }
+      }
+    }
+    if (Array.isArray(element.rootElement)) {
+      for (const ref of element.rootElement) {
+        if (!knownRefs.has(ref)) {
+          errorList.push(
+            `SpdxDocument '${elementId || "unknown"}' references unknown rootElement '${ref}'.`,
+          );
+        }
+      }
+    }
+    if (typeof element.from === "string" && !knownRefs.has(element.from)) {
+      errorList.push(
+        `Relationship '${elementId || "unknown"}' references unknown from '${element.from}'.`,
+      );
+    }
+    if (Array.isArray(element.to)) {
+      for (const ref of element.to) {
+        if (!knownRefs.has(ref)) {
+          errorList.push(
+            `Relationship '${elementId || "unknown"}' references unknown to '${ref}'.`,
+          );
+        }
+      }
+    }
+  }
+  if (usesExtensionProfile) {
+    if (!spdxDocument?.profileConformance?.includes("extension")) {
+      errorList.push(
+        `SpdxDocument '${spdxDocument?.spdxId || "unknown"}' must declare the extension profile when extension data is present.`,
+      );
+    }
+  }
+  if (errorList.length > 0) {
+    console.log("SPDX 3.0.1 validation failed");
+    console.log(errorList);
+    return false;
+  }
+  return true;
+};
+
+/**
+ * Validate the metadata object
+ *
+ * @param {object} bomJson Bom json object
+ */
+export const validateMetadata = (bomJson) => {
+  const errorList = [];
+  const warningsList = [];
+  if (bomJson?.metadata) {
+    if (
+      !bomJson.metadata.component ||
+      !Object.keys(bomJson.metadata.component).length
+    ) {
+      warningsList.push(
+        "metadata.component is missing. Run cdxgen with both --project-name and --project-version argument.",
+      );
+    }
+    if (bomJson.metadata.component) {
+      // Do we have a purl and bom-ref for metadata.component
+      if (!bomJson.metadata.component.purl) {
+        warningsList.push("purl is missing for metadata.component");
+      }
+      if (!bomJson.metadata.component["bom-ref"]) {
+        warningsList.push("bom-ref is missing for metadata.component");
+      }
+      // Do we have a version for metadata.component
+      if (!bomJson.metadata.component.version) {
+        warningsList.push(
+          "Version is missing for metadata.component. Pass the version using --project-version argument.",
+        );
+      }
+      const metadataName = bomJson.metadata.component.name
+        ?.trim()
+        .toLowerCase();
+      if (metadataName && PLACEHOLDER_COMPONENT_NAMES.has(metadataName)) {
+        warningsList.push(
+          `metadata.component.name appears to be a placeholder ('${bomJson.metadata.component.name}'). Pass --project-name to set the correct parent component name.`,
+        );
+      }
+      // Is the same component getting repeated inside the components block
+      if (bomJson.metadata.component.components?.length) {
+        for (const comp of bomJson.metadata.component.components) {
+          if (comp["bom-ref"] === bomJson.metadata.component["bom-ref"]) {
+            warningsList.push(
+              `Found parent component with ref ${comp["bom-ref"]} in metadata.component.components`,
+            );
+          } else if (
+            (!comp["bom-ref"] || !bomJson.metadata.component["bom-ref"]) &&
+            comp["name"] === bomJson.metadata.component["name"]
+          ) {
+            warningsList.push(
+              `Found parent component with name ${comp["name"]} in metadata.component.components`,
+            );
+          }
+        }
+      }
+    }
+  }
+  if (DEBUG_MODE && warningsList.length !== 0) {
+    console.log("===== WARNINGS =====");
+    console.log(warningsList);
+    thoughtLog(
+      "**VALIDATION**: There are some warnings regarding the BOM Metadata.",
+    );
+  }
+  if (errorList.length !== 0) {
+    console.log(errorList);
+    return false;
+  }
+  return true;
+};
+
+/**
+ * Yield `{ purl, where }` for every purl outside the top-level `components`
+ * array, which `validatePurls` walks separately for its richer per-type rules.
+ *
+ * Covers `metadata.component` (and its nested components),
+ * `metadata.tools.components`, `services`, and components nested inside
+ * top-level components.
+ *
+ * @param {object} bomJson Bom json object
+ */
+function* eachExtraPurl(bomJson) {
+  function* walk(node, where) {
+    if (!node || typeof node !== "object") {
+      return;
+    }
+    if (Array.isArray(node)) {
+      for (let i = 0; i < node.length; i++) {
+        yield* walk(node[i], `${where}[${i}]`);
+      }
+      return;
+    }
+    if (typeof node.purl === "string" && node.purl.length) {
+      yield { purl: node.purl, where: `${where}.purl` };
+    }
+    yield* walk(node.components, `${where}.components`);
+  }
+  yield* walk(bomJson?.metadata?.component, "metadata.component");
+  yield* walk(
+    bomJson?.metadata?.tools?.components,
+    "metadata.tools.components",
+  );
+  yield* walk(bomJson?.services, "services");
+  // Nested components only — the top-level entries themselves are covered by
+  // the main loop, which also applies the crypto-asset and qualifier rules.
+  for (let i = 0; i < (bomJson?.components?.length || 0); i++) {
+    yield* walk(
+      bomJson.components[i].components,
+      `components[${i}].components`,
+    );
+  }
+}
+
+/**
+ * Validate the format of all purls
+ *
+ * @param {object} bomJson Bom json object
+ */
+export const validatePurls = (bomJson) => {
+  const errorList = [];
+  const warningsList = [];
+  let frameworksCount = 0;
+  if (bomJson?.components) {
+    for (const comp of bomJson.components) {
+      if (comp.type === "framework") {
+        frameworksCount += 1;
+      }
+      if (comp.type === "cryptographic-asset") {
+        if (comp.purl?.length) {
+          errorList.push(
+            `purl should not be defined for cryptographic-asset ${comp.purl}`,
+          );
+        }
+        if (!comp.cryptoProperties) {
+          errorList.push(
+            `cryptoProperties is missing for cryptographic-asset ${comp.purl}`,
+          );
+        } else if (
+          comp.cryptoProperties.assetType === "algorithm" &&
+          !comp.cryptoProperties.oid
+        ) {
+          errorList.push(
+            `cryptoProperties.oid is missing for cryptographic-asset of type algorithm ${comp.purl}`,
+          );
+        } else if (
+          comp.cryptoProperties.assetType === "certificate" &&
+          !comp.cryptoProperties.algorithmProperties
+        ) {
+          errorList.push(
+            `cryptoProperties.algorithmProperties is missing for cryptographic-asset of type certificate ${comp.purl}`,
+          );
+        }
+      } else {
+        try {
+          if (comp.purl) {
+            const purlObj = Purl.parse(comp.purl);
+            if (purlObj.type && purlObj.type !== purlObj.type.toLowerCase()) {
+              warningsList.push(
+                `purl type is not normalized to lower case ${comp.purl}`,
+              );
+            }
+            if (
+              ["npm", "golang"].includes(purlObj.type) &&
+              purlObj.name.includes("%2F") &&
+              !purlObj.namespace
+            ) {
+              errorList.push(
+                `purl does not include namespace but includes encoded slash in name for npm type. ${comp.purl}`,
+              );
+            }
+            // Catch the trivy version hack that removes the epoch from version
+            const qualifiers = purlObj.qualifiers || {};
+            if (
+              Object.keys(qualifiers).length &&
+              [
+                "cargo",
+                "cocoapods",
+                "composer",
+                "cran",
+                "github",
+                "golang",
+                "hackage",
+                "nuget",
+                "opam",
+                "pub",
+                "qpkg",
+                "swift",
+              ].includes(purlObj.type)
+            ) {
+              warningsList.push(
+                `SPEC VIOLATION: Qualifiers are not expected for ${purlObj.type} type. Purl: ${comp.purl}, Qualifier(s): ${Object.keys(qualifiers).join(", ")}.`,
+              );
+            }
+            if (
+              qualifiers.epoch &&
+              !comp.version.startsWith(`${qualifiers.epoch}:`)
+            ) {
+              errorList.push(
+                `'${comp.name}' version '${comp.version}' doesn't include epoch '${qualifiers.epoch}'.`,
+              );
+            }
+          }
+        } catch (_ex) {
+          errorList.push(`Invalid purl ${comp.purl}`);
+        }
+      }
+    }
+  }
+  // Syntax check every *other* place a purl can appear. The loop above only
+  // walks top-level `components`, which is how `"purl": "swift-smoke"` reached
+  // a committed golden through `metadata.component` unnoticed. Nested
+  // components and `metadata.tools.components` had the same blind spot.
+  for (const { purl, where } of eachExtraPurl(bomJson)) {
+    if (!isValidPurl(purl)) {
+      errorList.push(`Invalid purl ${purl} at ${where}`);
+    }
+  }
+  if (frameworksCount > 20) {
+    warningsList.push(
+      `BOM likey has too many framework components. Count: ${frameworksCount}`,
+    );
+  }
+  if (DEBUG_MODE && warningsList.length !== 0) {
+    console.log("===== WARNINGS =====");
+    console.log(warningsList);
+    thoughtLog(
+      "**VALIDATION**: There are some warnings regarding the purls in our SBOM. These could be bugs.",
+    );
+  }
+  if (errorList.length !== 0) {
+    console.log(errorList);
+    return false;
+  }
+  return true;
+};
+
+const buildRefs = (bomJson) => {
+  const refMap = {};
+  if (bomJson) {
+    if (bomJson.metadata) {
+      if (bomJson.metadata.component) {
+        refMap[bomJson.metadata.component["bom-ref"]] = true;
+        if (bomJson.metadata.component.components) {
+          for (const comp of bomJson.metadata.component.components) {
+            refMap[comp["bom-ref"]] = true;
+          }
+        }
+      }
+    }
+    if (bomJson.components) {
+      for (const comp of bomJson.components) {
+        refMap[comp["bom-ref"]] = true;
+      }
+    }
+    if (bomJson?.formulation) {
+      for (const aformulation of bomJson.formulation) {
+        if (aformulation?.components?.length) {
+          for (const formComp of aformulation.components) {
+            refMap[formComp["bom-ref"]] = true;
+          }
+        }
+        if (aformulation?.workflows?.length) {
+          for (const formWf of aformulation.workflows) {
+            refMap[formWf["bom-ref"]] = true;
+            if (formWf?.tasks?.length) {
+              for (const atask of formWf.tasks) {
+                refMap[atask["bom-ref"]] = true;
+              }
+            }
+          }
+        }
+      }
+    }
+  }
+  return refMap;
+};
+
+/**
+ * Validate the refs in dependencies block
+ *
+ * @param {object} bomJson Bom json object
+ */
+export const validateRefs = (bomJson) => {
+  const errorList = [];
+  const warningsList = [];
+  const refMap = buildRefs(bomJson);
+  const parentComponentRef = bomJson?.metadata?.component?.["bom-ref"];
+  if (bomJson?.dependencies) {
+    if (isPartialTree(bomJson.dependencies, bomJson?.components?.length)) {
+      warningsList.push(
+        "Dependency tree has multiple empty dependsOn attributes.",
+      );
+    }
+    for (const dep of bomJson.dependencies) {
+      if (
+        dep.ref.includes("%40") ||
+        dep.ref.includes("%3A") ||
+        dep.ref.includes("%2F")
+      ) {
+        errorList.push(`Invalid encoded ref in dependencies ${dep.ref}`);
+      }
+      if (!refMap[dep.ref]) {
+        warningsList.push(`Invalid ref in dependencies ${dep.ref}`);
+      }
+      let parentPurlType;
+      try {
+        const purlObj = Purl.parse(dep.ref);
+        parentPurlType = purlObj.type;
+      } catch (_e) {
+        // pass
+      }
+      if (
+        parentComponentRef &&
+        dep.ref === parentComponentRef &&
+        dep.dependsOn.length === 0 &&
+        bomJson.dependencies.length > 1
+      ) {
+        warningsList.push(
+          `Parent component ${parentComponentRef} doesn't have any children. The dependency tree must contain dangling nodes, which are unsupported by tools such as Dependency-Track.`,
+        );
+      }
+      if (dep.dependsOn) {
+        for (const don of dep.dependsOn) {
+          if (!refMap[don]) {
+            warningsList.push(
+              `Invalid ref in dependencies.dependsOn ${don}. Parent: ${dep.ref}`,
+            );
+          }
+          let childPurlType;
+          try {
+            const purlObj = Purl.parse(don);
+            childPurlType = purlObj.type;
+          } catch (_e) {
+            // pass
+          }
+          if (
+            parentPurlType &&
+            childPurlType &&
+            parentPurlType !== childPurlType &&
+            !["oci", "generic", "container"].includes(parentPurlType)
+          ) {
+            warningsList.push(
+              `The parent package '${dep.ref}' (type ${parentPurlType}) depends on the child package '${don}' (type ${childPurlType}). This is a bug in cdxgen if this project is not a monorepo.`,
+            );
+          }
+        }
+      }
+      if (dep.provides) {
+        for (const don of dep.provides) {
+          if (!refMap[don]) {
+            warningsList.push(`Invalid ref in dependencies.provides ${don}`);
+          }
+        }
+      }
+    }
+  }
+  if (DEBUG_MODE && warningsList.length !== 0) {
+    console.log("===== WARNINGS =====");
+    console.log(warningsList);
+    thoughtLog(
+      "**VALIDATION**: There are some warnings regarding the dependency tree in our BOM.",
+    );
+  }
+  if (errorList.length !== 0) {
+    console.log(errorList);
+    return false;
+  }
+  return true;
+};
+
+/**
+ * Validate the component properties
+ *
+ * @param {object} bomJson Bom json object
+ */
+export function validateProps(bomJson) {
+  const errorList = [];
+  const warningsList = [];
+  let isWorkspaceMode = false;
+  let lacksProperties = false;
+  let lacksEvidence = false;
+  let lacksRelativePath = false;
+  let npmComponentsWithoutTarball = 0;
+  let npmComponentsWithTarball = 0;
+  if (
+    !["application", "framework", "library"].includes(
+      bomJson?.metadata?.component?.type,
+    )
+  ) {
+    return true;
+  }
+  if (bomJson?.components) {
+    const npmPkgs =
+      bomJson.components?.filter((c) => c.purl?.startsWith("pkg:npm")) || [];
+    const nativeByName = new Set(
+      npmPkgs
+        .filter((c) =>
+          c.properties?.some(
+            (p) => p.name === "cdx:npm:native_addon" && p.value === "true",
+          ),
+        )
+        .map((c) => c.name),
+    );
+    const suspicious = npmPkgs.filter(
+      (c) =>
+        c.scope !== "optional" &&
+        (c.name.includes("native") || c.name.includes("bindings")) &&
+        !nativeByName.has(c.name) &&
+        !c.properties?.some((p) =>
+          NPM_NATIVE_ADDON_EVIDENCE_PROPERTIES.has(p.name),
+        ),
+    );
+    if (suspicious.length > 0 && DEBUG_MODE) {
+      warningsList.push(
+        `Found ${suspicious.length} packages with native-sounding names but no native_addon flag: ${suspicious.map((c) => c.name).join(", ")}. May need deeper inspection.`,
+      );
+    }
+    for (const comp of bomJson.components) {
+      if (!["library", "framework"].includes(comp.type)) {
+        continue;
+      }
+      // Limit to only npm and pypi for now
+      if (
+        !comp.purl?.startsWith("pkg:npm") &&
+        !comp.purl?.startsWith("pkg:pypi")
+      ) {
+        continue;
+      }
+      if (comp.purl?.startsWith("pkg:npm")) {
+        const hasDistributionRef = comp.externalReferences?.some(
+          (ref) => ref.type === "distribution" && ref.url,
+        );
+        if (hasDistributionRef) {
+          npmComponentsWithTarball++;
+        } else {
+          npmComponentsWithoutTarball++;
+        }
+      }
+      if (!comp.properties) {
+        if (!lacksProperties) {
+          warningsList.push(`${comp["bom-ref"]} lacks properties.`);
+          lacksProperties = true;
+        }
+      } else {
+        let srcFilePropFound = false;
+        let workspacePropFound = false;
+        for (const p of comp.properties) {
+          if (p.name === "internal:SrcFile") {
+            srcFilePropFound = true;
+            // Quick linux/unix only check for relative paths.
+            if (!lacksRelativePath && p.value?.startsWith("/")) {
+              lacksRelativePath = true;
+            }
+          }
+          if (p.name === "internal:workspaceRef") {
+            isWorkspaceMode = true;
+            workspacePropFound = true;
+          }
+        }
+        if (
+          isWorkspaceMode &&
+          !workspacePropFound &&
+          !srcFilePropFound &&
+          comp?.scope !== "optional"
+        ) {
+          warningsList.push(
+            `${comp["bom-ref"]} lacks workspace-related properties.`,
+          );
+        }
+        if (!srcFilePropFound && !lacksProperties) {
+          warningsList.push(`${comp["bom-ref"]} lacks SrcFile property.`);
+          lacksProperties = true;
+        }
+      }
+      if (!comp.evidence && !lacksEvidence) {
+        lacksEvidence = true;
+        warningsList.push(`${comp["bom-ref"]} lacks evidence.`);
+      }
+    }
+  }
+  if (npmComponentsWithoutTarball > 0 && npmComponentsWithTarball > 0) {
+    warningsList.push(
+      `Found ${npmComponentsWithoutTarball} pkg:npm components without externalReferences.distribution. Please file a bug, if your package-lock.json or pnpm-lock.yaml includes the tarball url.`,
+    );
+  }
+  if (lacksRelativePath) {
+    warningsList.push(
+      "BOM includes absolute paths for properties like SrcFile.",
+    );
+    thoughtLog(
+      "BOM still includes absolute paths for properties like SrcFile. My postgen optimizations didn't work completely.",
+    );
+  }
+  if (DEBUG_MODE && warningsList.length !== 0) {
+    console.log("===== WARNINGS =====");
+    console.log(warningsList);
+    thoughtLog(
+      "**VALIDATION**: There are some warnings regarding the evidence attribute in our BOM, which can be safely ignored.",
+    );
+  }
+  if (errorList.length !== 0) {
+    console.log(errorList);
+    return false;
+  }
+  return true;
+}