@cdxgen/cdxgen 13.3.0 → 13.5.0

package/lib/ecosystems/parsers-go.js

@@ -1,20 +1,21 @@
 import { Buffer } from "node:buffer";
-import { readFileSync } from "node:fs";
-import { join } from "node:path";
+import { readdirSync, readFileSync, realpathSync } from "node:fs";
+import { homedir } from "node:os";
+import { delimiter, dirname, join, resolve } from "node:path";
 
 import { Purl } from "@cdxgen/cdx-purl";
+import { compare as compareSemver } from "semver";
 
-import { DEBUG_MODE } from "../core/activity.js";
+import { DEBUG_MODE, readEnvironmentVariable } from "../core/activity.js";
 import { shouldFetchLicense, shouldFetchVCS } from "../core/env.js";
-import { safeExistsSync } from "../core/fs.js";
 import {
   applyPurl,
   canonicalPurlFromLooseString,
   tryBuildPurl,
   tryParsePurl,
 } from "../inventory/purl.js";
+import { guessLicenseIds, isSpdxLicenseExpression } from "../inventory/spdx.js";
 import {
-  _clearMetadataCache,
   getGoPkgLicense,
   getGoPkgVCSUrl,
   prefetchGoPkgMetadata,
@@ -30,6 +31,494 @@ function toBase64(hexString) {
   return Buffer.from(hexString, "hex").toString("base64");
 }
 
+/**
+ * The root of the Go module cache on this machine.
+ *
+ * @returns {string|undefined} `$GOMODCACHE`, or `$GOPATH/pkg/mod`, or the
+ *   default `~/go/pkg/mod`.
+ */
+function goModuleCacheDir() {
+  const gomodcache = readEnvironmentVariable("GOMODCACHE");
+  if (gomodcache) {
+    return gomodcache;
+  }
+  const gopath = readEnvironmentVariable("GOPATH");
+  if (gopath) {
+    return join(gopath.split(delimiter)[0], "pkg", "mod");
+  }
+  return join(homedir(), "go", "pkg", "mod");
+}
+
+/**
+ * Escape a module path or version the way Go's module cache directory names
+ * do (module.EscapePath and module.EscapeVersion): every upper-case letter
+ * becomes `!` followed by the lower-case one, so the cache stays
+ * case-preserving on case-insensitive file systems (`github.com/Azure` is
+ * stored as `github.com/!azure`, and `v2.10.17-RC.1` as `v2.10.17-!r!c.1`).
+ *
+ * @param {string} value Module path or version
+ * @returns {string} The escaped value
+ */
+function escapeGoCacheName(value) {
+  let escaped = "";
+  for (const ch of value) {
+    if (ch >= "A" && ch <= "Z") {
+      escaped += `!${ch.toLowerCase()}`;
+    } else {
+      escaped += ch;
+    }
+  }
+  return escaped;
+}
+
+// Path elements Windows reserves, which Go rejects in every module path.
+const GO_WINDOWS_RESERVED_NAMES = new Set([
+  "CON",
+  "PRN",
+  "AUX",
+  "NUL",
+  ..."123456789".split("").flatMap((n) => [`COM${n}`, `LPT${n}`]),
+]);
+// ASCII punctuation a file name element may hold, after module.fileNameOK.
+const GO_FILE_NAME_PUNCTUATION = "!#$%&()+,-.=@[]^_{}~ ";
+
+/**
+ * Whether one character may appear in a path element of the given kind, as
+ * module.modPathOK, module.importPathOK and module.fileNameOK decide.
+ *
+ * @param {string} ch One character
+ * @param {"module"|"import"|"file"} kind What the element belongs to
+ * @returns {boolean}
+ */
+function isGoPathCharOK(ch, kind) {
+  if (/[A-Za-z0-9]/.test(ch)) {
+    return true;
+  }
+  if (kind === "file") {
+    return (
+      GO_FILE_NAME_PUNCTUATION.includes(ch) ||
+      (ch.charCodeAt(0) >= 0x80 && /\p{L}/u.test(ch))
+    );
+  }
+  return "-._~".includes(ch) || (kind === "import" && ch === "+");
+}
+
+/**
+ * Whether a path element is one the Go toolchain accepts, as module.checkElem
+ * decides. No accepted element is `.` or `..`, or holds a path separator, so a
+ * name that passes cannot leave the directory it is joined to.
+ *
+ * @param {string} elem Path element
+ * @param {"module"|"import"|"file"} kind What the element belongs to
+ * @returns {boolean}
+ */
+function isGoPathElementOK(elem, kind) {
+  if (!elem || /^\.+$/.test(elem) || elem.endsWith(".")) {
+    return false;
+  }
+  if (kind === "module" && elem.startsWith(".")) {
+    return false;
+  }
+  for (const ch of elem) {
+    if (!isGoPathCharOK(ch, kind)) {
+      return false;
+    }
+  }
+  const dot = elem.indexOf(".");
+  const short = dot < 0 ? elem : elem.slice(0, dot);
+  if (GO_WINDOWS_RESERVED_NAMES.has(short.toUpperCase())) {
+    return false;
+  }
+  if (kind === "file") {
+    return true;
+  }
+  // Names that look like Windows short names, such as PROGRA~1.
+  const tilde = short.lastIndexOf("~");
+  return !(tilde >= 0 && /^\d+$/.test(short.slice(tilde + 1)));
+}
+
+/**
+ * Whether a slash-separated path is one the Go toolchain accepts, as
+ * module.checkPath decides.
+ *
+ * @param {string} path Module or import path
+ * @param {"module"|"import"} kind What the path is
+ * @returns {boolean}
+ */
+function isGoPathOK(path, kind) {
+  return (
+    typeof path === "string" &&
+    path.length > 0 &&
+    !path.startsWith("-") &&
+    path.split("/").every((elem) => isGoPathElementOK(elem, kind))
+  );
+}
+
+/**
+ * Whether a module path passes module.CheckPath: a valid path whose first
+ * element is a lower-case host name with a dot in it. Go writes no module
+ * cache directory for any other path.
+ *
+ * @param {string} path Module path
+ * @returns {boolean}
+ */
+function isValidGoModulePath(path) {
+  if (!isGoPathOK(path, "module")) {
+    return false;
+  }
+  const host = path.split("/")[0];
+  return host.includes(".") && /^[a-z0-9.-]+$/.test(host);
+}
+
+/**
+ * Whether a version passes module.EscapeVersion: a single valid file name
+ * element without `!`.
+ *
+ * @param {string} version Module version
+ * @returns {boolean}
+ */
+function isValidGoVersion(version) {
+  return (
+    typeof version === "string" &&
+    !version.includes("!") &&
+    isGoPathElementOK(version, "file")
+  );
+}
+
+// The licence notice file names pkg.go.dev reads, compared without regard to
+// case (golang/pkgsite internal/licenses FileNames).
+const GO_LICENSE_FILE_NAMES = new Set(
+  [
+    "COPYING",
+    "COPYING.md",
+    "COPYING.markdown",
+    "COPYING.txt",
+    "COPYING.LESSER",
+    "COPYING.LESSER.md",
+    "COPYING.LESSER.markdown",
+    "COPYING.LESSER.txt",
+    "LICENCE",
+    "LICENCE.md",
+    "LICENCE.markdown",
+    "LICENCE.txt",
+    "LICENSE",
+    "LICENSE.md",
+    "LICENSE.markdown",
+    "LICENSE.txt",
+    "LICENSE-2.0.txt",
+    "LICENCE-2.0.txt",
+    "LICENSE-APACHE",
+    "LICENCE-APACHE",
+    "LICENSE-APACHE-2.0.txt",
+    "LICENCE-APACHE-2.0.txt",
+    "LICENSE-MIT",
+    "LICENCE-MIT",
+    "LICENSE.MIT",
+    "LICENCE.MIT",
+    "LICENSE.code",
+    "LICENCE.code",
+    "LICENSE.docs",
+    "LICENCE.docs",
+    "LICENSE.rst",
+    "LICENCE.rst",
+    "MIT-LICENSE",
+    "MIT-LICENCE",
+    "MIT-LICENSE.md",
+    "MIT-LICENCE.md",
+    "MIT-LICENSE.markdown",
+    "MIT-LICENCE.markdown",
+    "MIT-LICENSE.txt",
+    "MIT-LICENCE.txt",
+    "MIT_LICENSE",
+    "MIT_LICENCE",
+    "UNLICENSE",
+    "UNLICENCE",
+  ].map((f) => f.toLowerCase()),
+);
+
+/**
+ * Whether a file name is one of the licence notices pkg.go.dev reads, whose
+ * text describes the module's licence without a registry round.
+ *
+ * @param {string} fileName Candidate file name
+ * @returns {boolean}
+ */
+function isGoLicenseFileName(fileName) {
+  return GO_LICENSE_FILE_NAMES.has(fileName.toLowerCase());
+}
+
+/**
+ * The licence notices of a module that is already on this machine, from the
+ * directory `go list` reported, from the project's `vendor/` tree, or from the
+ * module cache. Reading them costs no network round trip, so they answer
+ * before pkg.go.dev is asked. Every notice of the first directory holding
+ * one is returned, as pkg.go.dev reads every notice of a module.
+ *
+ * The directory `go list` reported is always read. The vendor tree and the
+ * module cache stand in for pkg.go.dev, so they are read only when licences
+ * are being fetched. Their directories are built from a module path and
+ * version read from modules.txt or go.sum, so a path or version the Go
+ * toolchain would reject, such as one with a `..` element, is not looked up:
+ * Go never writes such a directory, and joining it could leave the tree.
+ *
+ * @param {string} name Full module path
+ * @param {string} [version] Module version, for the module cache layout
+ * @param {Object} [options] Options
+ * @param {string} [options.dir] The module's directory, as `go list` reports it
+ * @param {string} [options.projectRoot] Project root holding a `vendor/` tree
+ * @param {string} [options.vendorPath] The path the module is vendored under,
+ *   when it differs from `name`: Go vendors a replaced module under the path
+ *   it replaces
+ * @returns {{fileName: string, text: string}[]} The notices, in file name
+ *   order; empty when none is held locally
+ */
+export function readLocalGoLicenses(name, version, options = {}) {
+  if (!name) {
+    return [];
+  }
+  const candidateDirs = [];
+  if (options.dir) {
+    candidateDirs.push(options.dir);
+  }
+  const fetchingLicenses = shouldFetchLicense();
+  const vendorPath = options.vendorPath || name;
+  if (
+    fetchingLicenses &&
+    options.projectRoot &&
+    isGoPathOK(vendorPath, "import")
+  ) {
+    candidateDirs.push(
+      join(options.projectRoot, "vendor", ...vendorPath.split("/")),
+    );
+  }
+  if (
+    fetchingLicenses &&
+    version &&
+    isValidGoModulePath(name) &&
+    isValidGoVersion(version)
+  ) {
+    const cacheRoot = goModuleCacheDir();
+    if (cacheRoot) {
+      candidateDirs.push(
+        join(
+          cacheRoot,
+          `${escapeGoCacheName(name)}@${escapeGoCacheName(version)}`,
+        ),
+      );
+    }
+  }
+  for (const dir of candidateDirs) {
+    let entries;
+    try {
+      entries = readdirSync(dir, { withFileTypes: true });
+    } catch {
+      continue;
+    }
+    // Only regular files: a module zip holds no links, so a link named like
+    // a notice did not come from the module.
+    const fileNames = entries
+      .filter((e) => e.isFile() && isGoLicenseFileName(e.name))
+      .map((e) => e.name)
+      .sort((a, b) => a.localeCompare(b));
+    const notices = [];
+    for (const fileName of fileNames) {
+      try {
+        const text = readFileSync(join(dir, fileName), {
+          encoding: "utf-8",
+        });
+        if (text?.length) {
+          notices.push({ fileName, text });
+        }
+      } catch {
+        // An unreadable notice is the same as none.
+      }
+    }
+    if (notices.length) {
+      return notices;
+    }
+  }
+  return [];
+}
+
+/**
+ * The text of the first licence notice {@link readLocalGoLicenses} finds.
+ *
+ * @param {string} name Full module path
+ * @param {string} [version] Module version, for the module cache layout
+ * @param {Object} [options] The options {@link readLocalGoLicenses} takes
+ * @returns {string|undefined} The notice text, or undefined when none is held
+ *   locally
+ */
+export function readLocalGoLicense(name, version, options = {}) {
+  return readLocalGoLicenses(name, version, options)[0]?.text;
+}
+
+/**
+ * Whether a registry can hold this module: a module path and version the Go
+ * toolchain accepts. Any other module was never published, and its path in a
+ * pkg.go.dev URL could resolve to a different module (`../x` would ask for
+ * `x`).
+ *
+ * @param {string} name Full module path
+ * @param {string} [version] Module version
+ * @returns {boolean}
+ */
+function isGoRegistryModule(name, version) {
+  return isValidGoModulePath(name) && (!version || isValidGoVersion(version));
+}
+
+/**
+ * The modules the pkg.go.dev batch asks about, each marked when its licence
+ * notice is held on this machine, so the batch skips its licence tab but
+ * still fetches its module page for the repository URL. A module no registry
+ * holds (the main module, local code, or a path Go rejects) is left out, as
+ * getGoPkgComponent never looks it up.
+ *
+ * @param {Object[]} modules Modules with `name`, `version`,
+ *   `skipRegistryLookups` and the options {@link readLocalGoLicenses} takes
+ * @returns {Object[]} Copies of the modules to ask about, with
+ *   `licenseKnown` set
+ */
+function goModulesToPrefetch(modules) {
+  return modules
+    .filter(
+      (m) => !m.skipRegistryLookups && isGoRegistryModule(m.name, m.version),
+    )
+    .map((m) => ({
+      ...m,
+      licenseKnown: readLocalGoLicenses(m.name, m.version, m).length > 0,
+    }));
+}
+
+/**
+ * The CycloneDX licences of notices read from disk. Every licence each notice
+ * grants is listed with the notice as its text, and a notice that names none
+ * is a CUSTOM licence. A licence with an exception is an SPDX expression,
+ * which CycloneDX only carries as the sole entry, so then every licence found
+ * is joined into one expression.
+ *
+ * @param {{text: string}[]} notices Notices read by {@link readLocalGoLicenses}
+ * @returns {Object[]} CycloneDX licence choices
+ */
+function localGoLicenses(notices) {
+  const licenses = [];
+  const ids = [];
+  for (const { text } of notices) {
+    const textObject = { contentType: "text/plain", content: text };
+    const noticeIds = guessLicenseIds(text);
+    if (!noticeIds.length) {
+      licenses.push({ license: { name: "CUSTOM", text: textObject } });
+      continue;
+    }
+    for (const id of noticeIds) {
+      if (ids.includes(id)) {
+        continue;
+      }
+      ids.push(id);
+      licenses.push({ license: { id, text: textObject } });
+    }
+  }
+  if (ids.some((id) => isSpdxLicenseExpression(id))) {
+    return [
+      {
+        expression: ids
+          .map((id) => (ids.length > 1 && id.includes(" ") ? `(${id})` : id))
+          .join(" AND "),
+      },
+    ];
+  }
+  return licenses;
+}
+
+/**
+ * What a replace directive makes of one module, in the shape every Go parser
+ * reports it. The Go toolchain builds the replacement, so a module replacement
+ * is reported under the replacement's path and version, as govulncheck and
+ * syft report it. A directory replacement is local code that no registry
+ * holds: it keeps the module path it replaces and has no version, so its purl
+ * names no registry release.
+ *
+ * @param {string} path The module path being replaced
+ * @param {string} [version] The version being replaced, absent for a
+ *   replacement of every version
+ * @param {string} newPath The replacement module path or directory
+ * @param {string} [newVersion] The replacement version, absent (or `(devel)`
+ *   in build info) for a directory
+ * @returns {Object|undefined} `name` and `version` to report, `local` for a
+ *   directory, `localDir` the directory as written, and `replaces` the
+ *   `path@version` it stands in for
+ */
+function goReplacement(path, version, newPath, newVersion) {
+  if (!path || !newPath) {
+    return undefined;
+  }
+  const replaces = version ? `${path}@${version}` : path;
+  if (!newVersion || newVersion === "(devel)") {
+    return {
+      name: path,
+      version: undefined,
+      local: true,
+      localDir: newPath,
+      replaces,
+    };
+  }
+  return { name: newPath, version: newVersion, local: false, replaces };
+}
+
+/**
+ * Record on a component the module a replace directive put it in place of.
+ *
+ * @param {Object} component Component built for the replacement
+ * @param {Object} [replacement] Result of {@link goReplacement}
+ * @returns {Object} The same component
+ */
+function addGoReplacementProperties(component, replacement) {
+  if (!replacement) {
+    return component;
+  }
+  component.properties = component.properties || [];
+  component.properties.push({
+    name: "cdx:go:replaces",
+    value: replacement.replaces,
+  });
+  if (replacement.localDir) {
+    component.properties.push({
+      name: "cdx:go:local_dir",
+      value: replacement.localDir,
+    });
+  }
+  return component;
+}
+
+/**
+ * Order two Go module versions by semantic version precedence, as Go's semver
+ * package does: a pre-release, pseudo-versions included, sorts before its
+ * release, and build metadata such as `+incompatible` does not count.
+ *
+ * @param {string} a A module version
+ * @param {string} b Another module version
+ * @returns {number} Negative, zero or positive as `a` sorts before, with or
+ *   after `b`
+ */
+export function compareGoVersions(a, b) {
+  try {
+    return compareSemver(a, b);
+  } catch {
+    return (a || "").localeCompare(b || "");
+  }
+}
+
+/**
+ * Convert a Go checksum (`h1:` followed by the base64 SHA-256) to the
+ * integrity string components carry.
+ *
+ * @param {string} [sum] Go checksum
+ * @returns {string|undefined} `sha256-…`, or undefined without a checksum
+ */
+function goSumIntegrity(sum) {
+  return sum?.startsWith("h1:") ? `sha256-${sum.slice(3)}` : undefined;
+}
+
 /**
  * Builds a Go package component object containing purl, bom-ref, integrity hash,
  * and optionally license and VCS external reference information.
@@ -38,11 +527,36 @@ function toBase64(hexString) {
  * @param {string} name Package name (full module path when group is empty)
  * @param {string} version Package version string
  * @param {string} hash Integrity hash (e.g. "sha256-…"), used as _integrity
+ * @param {Object} [options] Options
+ * @param {boolean} [options.skipRegistryLookups] Leave the licence and VCS
+ *   URL unset instead of asking pkg.go.dev. Used for the main module, which
+ *   no registry holds.
+ * @param {string} [options.dir] The module's directory, as `go list` reports
+ *   it, where a local licence notice is read first
+ * @param {string} [options.projectRoot] Project root holding a `vendor/` tree
+ *   the module may be vendored in
+ * @param {string} [options.vendorPath] The path the module is vendored under,
+ *   when it differs from `name`
  * @returns {Promise<Object>} Component object ready for inclusion in a BOM package list
  */
-export async function getGoPkgComponent(group, name, version, hash) {
+export async function getGoPkgComponent(
+  group,
+  name,
+  version,
+  hash,
+  options = {},
+) {
   let license;
-  if (shouldFetchLicense()) {
+  let licenses;
+  const registryLookups =
+    !options.skipRegistryLookups &&
+    isGoRegistryModule(group ? `${group}/${name}` : name, version);
+  // Notices held on this machine answer before pkg.go.dev is asked, and the
+  // registry is consulted only when there is none.
+  const notices = readLocalGoLicenses(name, version, options);
+  if (notices.length) {
+    licenses = localGoLicenses(notices);
+  } else if (shouldFetchLicense() && registryLookups) {
     if (DEBUG_MODE) {
       console.log(
         `About to fetch go package license information for ${group}:${name}`,
@@ -51,6 +565,7 @@ export async function getGoPkgComponent(group, name, version, hash) {
     license = await getGoPkgLicense({
       group: group,
       name: name,
+      version: version,
     });
   }
   // Split the full module path into namespace and name, since the caller
@@ -76,8 +591,8 @@ export async function getGoPkgComponent(group, name, version, hash) {
       version: version || null,
     })?.replace(/%2F/g, "/") || null;
   let vcs;
-  if (shouldFetchVCS()) {
-    vcs = await getGoPkgVCSUrl(group, name);
+  if (shouldFetchVCS() && registryLookups) {
+    vcs = await getGoPkgVCSUrl(group, name, version);
   }
   const packageInfo = {
     group: group,
@@ -86,6 +601,9 @@ export async function getGoPkgComponent(group, name, version, hash) {
     _integrity: hash,
     license: license,
   };
+  if (licenses) {
+    packageInfo.licenses = licenses;
+  }
   applyPurl(packageInfo, purlString);
   if (vcs) {
     packageInfo.externalReferences = [{ type: "vcs", url: vcs }];
@@ -99,7 +617,8 @@ export async function getGoPkgComponent(group, name, version, hash) {
  * @param {String} goModData Contents of go.mod file
  * @param {Object} gosumMap Data from go.sum files
  *
- * @returns {Object} Object containing parent component, rootList and packages list
+ * @returns {Object} Object containing parent component, rootList, packages
+ *   list, and the replace directives as `replacements`
  */
 export async function parseGoModData(goModData, gosumMap) {
   const pkgComponentsList = [];
@@ -108,8 +627,13 @@ export async function parseGoModData(goModData, gosumMap) {
   if (!goModData) {
     return {};
   }
-  const { parentComponent, entries } = decodeGoModData(goModData, gosumMap);
-  await prefetchGoPkgMetadata(entries);
+  const { parentComponent, entries, replacements } = decodeGoModData(
+    goModData,
+    gosumMap,
+  );
+  // A module whose licence notice is read locally is not asked for its
+  // licence; getGoPkgComponent reads the same notice for each entry below.
+  await prefetchGoPkgMetadata(goModulesToPrefetch(entries));
   for (const entry of entries) {
     const component = await getGoPkgComponent(
       "",
@@ -129,8 +653,6 @@ export async function parseGoModData(goModData, gosumMap) {
     }
     pkgComponentsList.push(component);
   }
-  // Clear the cache
-  _clearMetadataCache();
   return {
     parentComponent,
     pkgList: pkgComponentsList.sort((a, b) =>
@@ -141,6 +663,7 @@ export async function parseGoModData(goModData, gosumMap) {
       (a["bom-ref"] || "").localeCompare(b["bom-ref"] || ""),
     ),
     rootList,
+    replacements,
   };
 }
 
@@ -153,107 +676,251 @@ export async function parseGoModData(goModData, gosumMap) {
  *
  * @param {String} goModData Contents of go.mod file
  * @param {Object} gosumMap Data from go.sum files
- * @returns {{parentComponent: Object, entries: Array<Object>}} Parent component
- *   and module entries in file order
+ * @returns {{parentComponent: Object, entries: Array<Object>, replacements: Array<Object>}}
+ *   Parent component, module entries in file order, and every replace
+ *   directive as `path`, `version`, `newPath` and `newVersion`
  */
 function decodeGoModData(goModData, gosumMap) {
   const parentComponent = {};
   const entries = [];
-  let isModReplacement = false;
-  let isTool = false;
-
-  const pkgs = goModData.split("\n");
-  for (let l of pkgs) {
-    // Windows of course
-    l = l.replaceAll("\r", "").replace(/[\t ]+/g, " ");
-    // Capture the parent component name from the module
-    if (l.startsWith("module ")) {
-      parentComponent.name = l.split(" ").pop().trim();
-      parentComponent.type = "application";
-      // Single-segment module paths are valid golang purls as of cdx-purl 0.0.4.
-      applyPurl(
-        parentComponent,
-        tryParsePurl(`pkg:golang/${parentComponent.name}`),
-      );
-      continue;
-    }
+  const replacements = [];
+  // The verb of the parenthesised block being read: undefined outside a block,
+  // and "" inside one whose header is not a single verb, which Go rejects.
+  let blockVerb;
 
-    // The `tool` block dependency relations will be recorded into `require` block(need run `go mod tidy`), just ignore that
-    if (l.includes("tool (")) {
-      isTool = !l.includes(")");
-      continue;
-    }
-    if (l.includes(")")) {
-      isTool = false;
+  for (const line of goModData.split("\n")) {
+    const { tokens, comment } = tokenizeGoModLine(line);
+    if (!tokens.length) {
       continue;
     }
-    if (l.includes("tool ") || isTool) {
+    let verb;
+    let args;
+    if (blockVerb !== undefined) {
+      if (tokens.length === 1 && tokens[0] === ")") {
+        blockVerb = undefined;
+        continue;
+      }
+      verb = blockVerb;
+      args = tokens;
+    } else if (tokens.at(-1) === "(") {
+      // `require (` opens a block whose lines all carry the verb.
+      blockVerb = tokens.length === 2 ? tokens[0] : "";
       continue;
-    }
-    if (l.startsWith("toolchain ")) {
-      const toolchainVer = l.split(" ").pop().trim();
-      parentComponent.properties = [
-        { name: "cdx:go:toolchain", value: toolchainVer },
-      ];
+    } else if (tokens.at(-2) === "(" && tokens.at(-1) === ")") {
+      // `require ( )` is an empty block.
       continue;
+    } else {
+      verb = tokens[0];
+      args = tokens.slice(1);
     }
-    // Skip go.mod file headers, whitespace, and/or comments
-    if (
-      l.startsWith("go ") ||
-      //TODO: should toolchain be considered as a dependency
-      l.includes(")") ||
-      l.trim() === "" ||
-      l.trim().startsWith("//")
-    ) {
-      continue;
+    switch (verb) {
+      case "module":
+        if (args.length) {
+          parentComponent.name = unquoteGoModToken(args[0]);
+          parentComponent.type = "application";
+          // Single-segment module paths are valid golang purls as of cdx-purl 0.0.4.
+          applyPurl(
+            parentComponent,
+            tryParsePurl(`pkg:golang/${parentComponent.name}`),
+          );
+        }
+        break;
+      case "toolchain":
+        if (args.length) {
+          parentComponent.properties = [
+            { name: "cdx:go:toolchain", value: args[0] },
+          ];
+        }
+        break;
+      case "require": {
+        // module/path v1.2.3
+        const name = unquoteGoModToken(args[0]);
+        const version = args[1];
+        if (!name || !version) {
+          break;
+        }
+        entries.push({
+          group: "",
+          name,
+          version,
+          hash: gosumMap[`${name}@${version}`],
+          indirect: isGoModIndirectComment(comment),
+          replacement: false,
+        });
+        break;
+      }
+      case "replace": {
+        // module/path [v1.2.3] => other/module v1.4.5
+        // module/path [v1.2.3] => ../local/directory
+        // Only the first form names a module version; a directory
+        // replacement leaves the required module as the component.
+        const arrow = args[1] === "=>" ? 1 : 2;
+        if (
+          args[arrow] !== "=>" ||
+          (args.length !== arrow + 2 && args.length !== arrow + 3)
+        ) {
+          break;
+        }
+        const oldPath = unquoteGoModToken(args[0]);
+        const name = unquoteGoModToken(args[arrow + 1]);
+        const version = args[arrow + 2];
+        if (oldPath && name) {
+          replacements.push({
+            path: oldPath,
+            version: arrow === 2 ? args[1] : undefined,
+            newPath: name,
+            newVersion: version,
+          });
+        }
+        if (!name || args.length !== arrow + 3) {
+          break;
+        }
+        entries.push({
+          group: "",
+          name,
+          version,
+          hash: gosumMap[`${name}@${version}`],
+          indirect: false,
+          replacement: true,
+        });
+        break;
+      }
+      // `go`, `godebug`, `exclude`, `retract`, `ignore` and `tool` name no
+      // dependency: an excluded or retracted version is kept out of the build,
+      // and a tool's module is listed under `require` once the module is tidy.
+      default:
+        break;
     }
+  }
+  return { parentComponent, entries, replacements };
+}
 
-    // Handle required modules separately from replacement modules to ensure accuracy when parsing component data.
-    if (l.includes("require (")) {
-      isModReplacement = false;
-      continue;
-    }
-    if (l.includes("replace (")) {
-      isModReplacement = true;
-      continue;
-    }
-    if (l.includes("replace ")) {
-      // If this is an inline replacement, drop the word replace
-      // (eg; "replace google.golang.org/grpc => google.golang.org/grpc v1.21.0" becomes " google.golang.org/grpc => google.golang.org/grpc v1.21.0")
-      l = l.replace("replace", "");
-      isModReplacement = true;
-    }
-    // require google.golang.org/genproto v0.0.0-20231106174013-bbf56f31fb17
-    if (l.startsWith("require ")) {
-      l = l.replace("require ", "");
-      isModReplacement = false;
-    }
-    const tmpA = l.trim().split(" ");
-    if (!isModReplacement) {
-      // Add group, name and version component properties for required modules
-      const version = tmpA[1];
-      entries.push({
-        group: "",
-        name: tmpA[0],
-        version,
-        hash: gosumMap[`${tmpA[0]}@${version}`],
-        indirect: l.endsWith("// indirect"),
-        replacement: false,
-      });
+// Characters the go.mod lexer always reads as a token of their own.
+const GO_MOD_PUNCTUATION = "()[]{},";
+
+/**
+ * Split one go.mod line into tokens the way the Go toolchain's modfile lexer
+ * does: whitespace separates tokens, `(`, `)`, `[`, `]`, `{`, `}` and `,`
+ * stand alone, a quoted string is one token, and `//` starts a comment that
+ * runs to the end of the line.
+ *
+ * @param {string} line A single line of a go.mod file
+ * @returns {{tokens: string[], comment: string}} The tokens, quoted strings
+ *   still quoted, and the trailing comment text without its `//`
+ */
+function tokenizeGoModLine(line) {
+  const tokens = [];
+  let i = 0;
+  while (i < line.length) {
+    const c = line[i];
+    if (/\s/.test(c)) {
+      i++;
+    } else if (line.startsWith("//", i)) {
+      return { tokens, comment: line.slice(i + 2).trim() };
+    } else if (GO_MOD_PUNCTUATION.includes(c)) {
+      tokens.push(c);
+      i++;
+    } else if (c === '"' || c === "`") {
+      let end = i + 1;
+      while (end < line.length && line[end] !== c) {
+        // A double-quoted string escapes its quote with a backslash.
+        end += c === '"' && line[end] === "\\" ? 2 : 1;
+      }
+      tokens.push(line.slice(i, end + 1));
+      i = end + 1;
     } else {
-      // Add group, name and version component properties for replacement modules
-      const version = tmpA[3];
-      entries.push({
-        group: "",
-        name: tmpA[2],
-        version,
-        hash: gosumMap[`${tmpA[2]}@${version}`],
-        indirect: false,
-        replacement: true,
-      });
+      let end = i;
+      while (
+        end < line.length &&
+        !/\s/.test(line[end]) &&
+        !GO_MOD_PUNCTUATION.includes(line[end]) &&
+        !line.startsWith("//", end)
+      ) {
+        end++;
+      }
+      tokens.push(line.slice(i, end));
+      i = end;
+    }
+  }
+  return { tokens, comment: "" };
+}
+
+/**
+ * Read a module path or directory token, which go.mod allows in double
+ * quotes. Like the Go toolchain, any other quote character makes it invalid.
+ *
+ * @param {string|undefined} token A go.mod token
+ * @returns {string|undefined} The unquoted value, or undefined when invalid
+ */
+function unquoteGoModToken(token) {
+  if (token?.startsWith('"')) {
+    return token.length >= 2 && token.endsWith('"')
+      ? token.slice(1, -1)
+      : undefined;
+  }
+  if (/["'`]/.test(token || "")) {
+    return undefined;
+  }
+  return token;
+}
+
+/**
+ * Whether a require line's comment marks it indirect, as the Go toolchain
+ * reads it: `// indirect`, or `// indirect; ` followed by a reason.
+ *
+ * @param {string} comment Comment text without its `//`
+ * @returns {boolean} True for an indirect requirement
+ */
+function isGoModIndirectComment(comment) {
+  const fields = comment.split(/\s+/).filter(Boolean);
+  return (
+    (fields.length === 1 && fields[0] === "indirect") ||
+    (fields.length > 1 && fields[0] === "indirect;")
+  );
+}
+
+// A semantic version as golang.org/x/mod/semver accepts it, used to tell a
+// module line's version from the `=>` of a replacement.
+const GO_SEMVER_PATTERN =
+  /^v(0|[1-9]\d*)(\.(0|[1-9]\d*)){0,2}(-[0-9A-Za-z.-]+)?(\+[0-9A-Za-z.-]+)?$/;
+
+/**
+ * Read one `# ` module line of a vendor/modules.txt file the way the Go
+ * toolchain does (modload.readVendorList):
+ *
+ *   # module/path v1.2.3
+ *   # module/path v1.2.3 => other/module v1.4.5
+ *   # module/path v1.2.3 => ../local/directory
+ *   # module/path => other/module v1.4.5     (a replacement of every version)
+ *
+ * @param {string} line A line starting with `# `
+ * @returns {Object|undefined} `path`, `version` and, for a replaced module,
+ *   `newPath` and `newVersion`; undefined for a line Go ignores
+ */
+function parseModulesTxtModuleLine(line) {
+  const fields = line.split(/\s+/).filter(Boolean);
+  if (fields.length < 3) {
+    return undefined;
+  }
+  let rest;
+  const module = { path: fields[1] };
+  if (GO_SEMVER_PATTERN.test(fields[2])) {
+    module.version = fields[2];
+    rest = fields.slice(3);
+  } else if (fields[2] === "=>") {
+    rest = fields.slice(2);
+  } else {
+    return undefined;
+  }
+  if (rest[0] === "=>") {
+    if (rest.length === 2) {
+      module.newPath = rest[1];
+    } else if (rest.length === 3 && GO_SEMVER_PATTERN.test(rest[2])) {
+      module.newPath = rest[1];
+      module.newVersion = rest[2];
     }
   }
-  return { parentComponent, entries };
+  return module;
 }
 
 /**
@@ -261,6 +928,12 @@ function decodeGoModData(goModData, gosumMap) {
  * Go package components. Cross-references the go.sum map for integrity hashes and
  * sets scope and confidence based on hash availability.
  *
+ * Only the modules the vendor tree holds become components: a module line
+ * followed by its `##` annotations or its packages. `go mod vendor` also
+ * records every replace directive of go.mod at the end of the file, as a
+ * module line with nothing after it, so the consumer can tell that the
+ * replacement had no effect; those lines name no vendored module.
+ *
  * @param {string} txtFile Path to the modules.txt file
  * @param {Object} gosumMap Map of "module@version" keys to sha256 hash values from go.sum
  * @returns {Promise<Object[]>} List of Go package component objects with evidence
@@ -268,21 +941,66 @@ function decodeGoModData(goModData, gosumMap) {
 export async function parseGoModulesTxt(txtFile, gosumMap) {
   const pkgList = [];
   const txtData = readFileSync(txtFile, { encoding: "utf-8" });
-  const pkgs = txtData
-    .split("\n")
-    .filter((p) => p.trim().replace(/["']/g, "").startsWith("# "));
-  await prefetchGoPkgMetadata(
-    // # cel.dev/expr v0.18.0
-    pkgs.map((l) => ({ group: "", name: l.split(" ")[1] })),
-  );
-  for (const l of pkgs) {
-    const tmpA = l.split(" ");
-    if (!tmpA.length === 3) {
+  // A modules.txt file lives in the vendor/ tree, so the modules it lists are
+  // vendored one directory up from it.
+  const projectRoot = dirname(dirname(txtFile));
+  const records = [];
+  let current;
+  for (const rawLine of txtData.split("\n")) {
+    // A checkout with CRLF line endings must read the same as the LF file
+    // go mod vendor wrote.
+    const line = rawLine.replaceAll("\r", "");
+    if (line.startsWith("# ")) {
+      current = parseModulesTxtModuleLine(line);
+      if (current) {
+        records.push(current);
+      }
       continue;
     }
-    const version = tmpA[2];
-    const gosumHash = gosumMap[`${tmpA[1]}@${version}`];
-    const component = await getGoPkgComponent("", tmpA[1], tmpA[2], gosumHash);
+    if (current && line.trim()) {
+      current.vendored = true;
+    }
+  }
+  const modules = [];
+  for (const record of records) {
+    // A replacement record of every version names no module version, and a
+    // record with nothing after it was not vendored.
+    if (!record.version || !record.vendored) {
+      continue;
+    }
+    const replacement = goReplacement(
+      record.path,
+      record.version,
+      record.newPath,
+      record.newVersion,
+    );
+    modules.push({
+      group: "",
+      name: replacement?.name ?? record.path,
+      version: replacement ? replacement.version : record.version,
+      replacement,
+      projectRoot,
+      // Go vendors a replaced module under the path it replaces.
+      vendorPath: record.path,
+      skipRegistryLookups: !!replacement?.local,
+    });
+  }
+  // A module whose vendored licence notice is read is not asked for its
+  // licence; getGoPkgComponent reads the same file. Local code is never
+  // asked about.
+  await prefetchGoPkgMetadata(goModulesToPrefetch(modules));
+  for (const m of modules) {
+    const gosumHash = m.version
+      ? gosumMap[`${m.name}@${m.version}`]
+      : undefined;
+    const component = await getGoPkgComponent(
+      "",
+      m.name,
+      m.version,
+      gosumHash,
+      m,
+    );
+    addGoReplacementProperties(component, m.replacement);
     let confidence = 0.7;
     if (gosumHash) {
       component.scope = "required";
@@ -294,119 +1012,160 @@ export async function parseGoModulesTxt(txtFile, gosumMap) {
   return pkgList;
 }
 
+/**
+ * Whether two paths name the same directory. `go list` prints the real path
+ * of a directory, which differs from the one cdxgen was given when that runs
+ * through a symbolic link (macOS /tmp is /private/tmp).
+ *
+ * @param {string} [a] A directory
+ * @param {string} [b] Another directory
+ * @returns {boolean}
+ */
+function isSameGoDir(a, b) {
+  if (!a || !b) {
+    return false;
+  }
+  if (resolve(a) === resolve(b)) {
+    return true;
+  }
+  try {
+    return realpathSync(a) === realpathSync(b);
+  } catch {
+    return false;
+  }
+}
+
 /**
  * Parse go list output
  *
+ * Each row is printed by the template in createGoBom:
+ * `Path|Version|Indirect|GoMod|GoVersion|Main|Time|Deprecated|GoModSum|Dir|Sum|
+ * Replace.Path|Replace.Version|Replace.Sum`. Rows from older templates end
+ * after `Dir`.
+ *
+ * A replaced module is reported as the module Go builds (see
+ * {@link goReplacement}), with the checksum of the replacement. A go.work
+ * workspace makes every member a main module, so the parent is the main
+ * module in `options.mainModuleDir`, and the other members are returned as
+ * `workspaceModules`.
+ *
  * @param {string} rawOutput Output from go list invocation
  * @param {Object} gosumMap go.sum data
- * @returns Object with parent component and List of packages
+ * @param {Object} [options] Options
+ * @param {string} [options.mainModuleDir] Directory `go list` ran in, whose
+ *   main module is the parent component
+ * @returns Object with parent component, list of packages, and the other main
+ *   modules of a workspace
  */
-export async function parseGoListDep(rawOutput, gosumMap) {
+export async function parseGoListDep(rawOutput, gosumMap, options = {}) {
   let parentComponent = {};
   const deps = [];
+  const workspaceModules = [];
   if (typeof rawOutput === "string") {
-    const keys_cache = {};
-    const pkgs = rawOutput
-      .split("\n")
-      .filter((p) => p.trim().replace(/["']/g, "").length);
-    // Same shape and same duplicate filter as the loop below, so the batch
-    // covers every module the loop resolves and nothing else.
-    const prefetchSeen = new Set();
-    const prefetchModules = [];
-    for (const l of pkgs) {
+    const rows = [];
+    const seen = new Set();
+    for (const l of rawOutput.split("\n")) {
       const verArr = l.trim().replace(/["']/g, "").split("|");
-      if (!verArr || verArr.length < 5) {
+      if (verArr.length < 5) {
         continue;
       }
+      // Every package of a module repeats the module's row.
       const key = `${verArr[0]}-${verArr[1]}`;
-      if (prefetchSeen.has(key)) {
+      if (seen.has(key)) {
         continue;
       }
-      prefetchSeen.add(key);
-      prefetchModules.push({ group: "", name: verArr[0] });
+      seen.add(key);
+      rows.push(verArr);
     }
-    await prefetchGoPkgMetadata(prefetchModules);
-    for (const l of pkgs) {
-      const verArr = l.trim().replace(/["']/g, "").split("|");
-      if (verArr && verArr.length >= 5) {
-        const key = `${verArr[0]}-${verArr[1]}`;
-        // Filter duplicates
-        if (!keys_cache[key]) {
-          keys_cache[key] = key;
-          const version = verArr[1];
-          let gosumHash = gosumMap[`${verArr[0]}@${version}`];
-          if (!gosumHash && verArr.length >= 8 && verArr[8]?.length) {
-            gosumHash = `sha256-${verArr[8].replace("h1:", "")}`;
-          }
-          const component = await getGoPkgComponent(
-            "",
-            verArr[0],
-            version,
-            gosumHash,
-          );
-          // This is misusing the scope attribute to represent direct vs indirect
-          if (verArr[2] === "false") {
-            component.scope = "required";
-          } else if (verArr[2] === "true") {
-            component.scope = "optional";
-          }
-          component.properties = [
-            {
-              name: "internal:SrcGoMod",
-              value: verArr[3] || "",
-            },
-            {
-              name: "internal:ModuleGoVersion",
-              value: verArr[4] || "",
-            },
-            {
-              name: "cdx:go:indirect",
-              value: verArr[2],
-            },
-          ];
-          if (
-            verArr.length >= 6 &&
-            verArr[6]?.length &&
-            verArr[6] !== "<nil>"
-          ) {
-            component.properties.push({
-              name: "cdx:go:creation_time",
-              value: verArr[6],
-            });
-          }
-          if (verArr.length >= 7 && verArr[7]?.length) {
-            component.properties.push({
-              name: "cdx:go:deprecated",
-              value: verArr[7],
-            });
-          }
-          if (verArr.length >= 9 && verArr[9]?.length) {
-            component.properties.push({
-              name: "cdx:go:local_dir",
-              value: verArr[9],
-            });
-            if (safeExistsSync(join(verArr[9], "LICENSE"))) {
-              const licenseText = readFileSync(join(verArr[9], "LICENSE"), {
-                encoding: "utf-8",
-              });
-              if (licenseText?.length) {
-                component.licenses = [
-                  {
-                    license: {
-                      name: "CUSTOM",
-                      text: { contentType: "text/plain", content: licenseText },
-                    },
-                  },
-                ];
-              }
-            }
-          }
-          if (verArr.length > 5 && verArr[5] === "true") {
-            parentComponent = component;
-          } else {
-            deps.push(component);
-          }
-        }
+    // The main module is the project itself, which no registry holds. In a
+    // go.work workspace every member is a main module, and the one go list
+    // ran in is the parent.
+    const mainRows = rows.filter((verArr) => verArr[5] === "true");
+    const mainRow =
+      mainRows.find((verArr) =>
+        isSameGoDir(verArr[9], options.mainModuleDir),
+      ) || mainRows[0];
+    // The directory of the main module, which is the project root, under
+    // which the vendored copies of the dependencies live.
+    const mainModuleDir = mainRow?.[9] || undefined;
+    const modules = rows.map((verArr) => {
+      const isMainModule = verArr[5] === "true";
+      const replacement = isMainModule
+        ? undefined
+        : goReplacement(verArr[0], verArr[1], verArr[11], verArr[12]);
+      const name = replacement?.name ?? verArr[0];
+      const version = replacement ? replacement.version : verArr[1];
+      return {
+        verArr,
+        isMainModule,
+        replacement,
+        group: "",
+        name,
+        version,
+        // Module.Sum is the checksum of the module content, the hash of what
+        // is built. GoModSum hashes only its go.mod file.
+        hash:
+          (version && gosumMap[`${name}@${version}`]) ||
+          goSumIntegrity(replacement ? verArr[13] : verArr[10]),
+        dir: verArr[9]?.length ? verArr[9] : undefined,
+        projectRoot: mainModuleDir,
+        // Go vendors a replaced module under the path it replaces.
+        vendorPath: verArr[0],
+        skipRegistryLookups: isMainModule || !!replacement?.local,
+      };
+    });
+    // A module whose licence notice is held on this machine is not asked for
+    // its licence: getGoPkgComponent reads the same notices, so the batch
+    // covers exactly what the loop will look up.
+    await prefetchGoPkgMetadata(goModulesToPrefetch(modules));
+    for (const m of modules) {
+      const { verArr } = m;
+      const component = await getGoPkgComponent(
+        "",
+        m.name,
+        m.version,
+        m.hash,
+        m,
+      );
+      // This is misusing the scope attribute to represent direct vs indirect
+      if (verArr[2] === "false") {
+        component.scope = "required";
+      } else if (verArr[2] === "true") {
+        component.scope = "optional";
+      }
+      component.properties = [
+        {
+          name: "internal:SrcGoMod",
+          value: verArr[3] || "",
+        },
+        {
+          name: "internal:ModuleGoVersion",
+          value: verArr[4] || "",
+        },
+        {
+          name: "cdx:go:indirect",
+          value: verArr[2],
+        },
+      ];
+      if (verArr.length >= 6 && verArr[6]?.length && verArr[6] !== "<nil>") {
+        component.properties.push({
+          name: "cdx:go:creation_time",
+          value: verArr[6],
+        });
+      }
+      if (verArr.length >= 7 && verArr[7]?.length) {
+        component.properties.push({
+          name: "cdx:go:deprecated",
+          value: verArr[7],
+        });
+      }
+      addGoReplacementProperties(component, m.replacement);
+      if (verArr === mainRow) {
+        parentComponent = component;
+      } else if (m.isMainModule) {
+        workspaceModules.push(component);
+      } else {
+        deps.push(component);
       }
     }
   }
@@ -419,6 +1178,7 @@ export async function parseGoListDep(rawOutput, gosumMap) {
       // be built silently reorders the output.
       (a["bom-ref"] || "").localeCompare(b["bom-ref"] || ""),
     ),
+    workspaceModules,
   };
 }
 
@@ -465,9 +1225,39 @@ function goModGraphTokenPurl(token) {
   return canonical ? Purl.parse(canonical) : null;
 }
 
+/**
+ * Split a `go mod graph` token into its module path and version, exactly as
+ * Go printed them. A main module is printed without a version.
+ *
+ * @param {string} token `module/path@version` or `module/path`
+ * @returns {{path: string, version: string|undefined}}
+ */
+function splitGoModGraphToken(token) {
+  const at = token.indexOf("@");
+  return at < 0
+    ? { path: token, version: undefined }
+    : { path: token.slice(0, at), version: token.slice(at + 1) };
+}
+
+/**
+ * Value of the first property with the given name.
+ *
+ * @param {Object} component Component
+ * @param {string} name Property name
+ * @returns {string|undefined}
+ */
+function goComponentProperty(component, name) {
+  return component?.properties?.find((p) => p.name === name)?.value;
+}
+
 /**
  * Parse go mod graph
  *
+ * `go mod graph` names every module by the path and version its requirement
+ * names, whatever replaces it. A replaced module is mapped to the module Go
+ * builds: from the components `go list` reported, which record what they
+ * replace, or else from the replace directives of go.mod.
+ *
  * @param {string} rawOutput Output from go mod graph invocation
  * @param {string} goModFile go.mod file
  * @param {Object} gosumMap Hashes from gosum for lookups
@@ -493,9 +1283,41 @@ export async function parseGoModGraph(
   let goModPkgMap = {};
   // Direct dependencies by manually parsing the go.mod data
   const goModDirectDepsMap = {};
+  // The same direct dependencies by module path, which stays correct when
+  // the version the go.mod text names has drifted from the selected one.
+  const goModDirectDepNames = new Set();
   // Indirect dependencies by manually parsing the go.mod data
   const goModOptionalDepsMap = {};
   const excludedRefs = [];
+  // The module versions `go list` reported are the ones MVS selected, and the
+  // only ones the BOM carries components for. `go mod graph` instead prints
+  // the requirement graph, where each edge names the version the source's
+  // go.mod asks for, so the two routinely disagree for the same module. They
+  // are keyed by the module path exactly as Go prints it, which keeps its
+  // upper-case letters; a replaced module is keyed by the path it replaces,
+  // which is the one the graph prints.
+  const selectedVersions = {};
+  // `path@version` as the graph prints it, for every module a replace
+  // directive put another module in place of.
+  const replacedModules = new Map();
+  for (const epkg of epkgList) {
+    const replaces = goComponentProperty(epkg, "cdx:go:replaces");
+    if (replaces) {
+      const { path, version } = splitGoModGraphToken(replaces);
+      replacedModules.set(replaces, {
+        name: epkg.name,
+        version: epkg.version,
+        local: !epkg.version,
+        localDir: goComponentProperty(epkg, "cdx:go:local_dir"),
+        replaces,
+      });
+      if (version) {
+        selectedVersions[path] = version;
+      }
+    } else if (epkg?.name && epkg?.version) {
+      selectedVersions[epkg.name] = epkg.version;
+    }
+  }
   if (goModFile) {
     goModPkgMap = await parseGoModData(
       readFileSync(goModFile, { encoding: "utf-8" }),
@@ -504,6 +1326,11 @@ export async function parseGoModGraph(
     if (goModPkgMap?.rootList) {
       for (const epkg of goModPkgMap.rootList) {
         goModDirectDepsMap[epkg["bom-ref"]] = true;
+        // The bom-ref above names the version the go.mod text requires,
+        // which an untidy module can leave older than the version MVS
+        // selected. Root-edge validation below also asks by module path,
+        // so a drifted direct require still counts as direct (#4497).
+        goModDirectDepNames.add(epkg.name);
       }
     }
     if (goModPkgMap?.pkgList) {
@@ -514,59 +1341,117 @@ export async function parseGoModGraph(
       }
     }
   }
+  // What the replace directives of go.mod make of a module the graph names. A
+  // directive for one version wins over one for every version, as in Go.
+  const goModReplacement = (path, version) => {
+    let wildcard;
+    for (const r of goModPkgMap?.replacements || []) {
+      if (r.path !== path) {
+        continue;
+      }
+      if (r.version === undefined) {
+        wildcard = r;
+      } else if (r.version === version) {
+        return goReplacement(path, version, r.newPath, r.newVersion);
+      }
+    }
+    return wildcard
+      ? goReplacement(path, version, wildcard.newPath, wildcard.newVersion)
+      : undefined;
+  };
+  // The module Go builds for one graph token, with the reference of the
+  // component that stands for it.
+  const graphModule = (path, version) => {
+    const replacement = version
+      ? replacedModules.get(`${path}@${version}`) ||
+        goModReplacement(path, version)
+      : undefined;
+    const name = replacement ? replacement.name : path;
+    const moduleVersion = replacement ? replacement.version : version;
+    const token = moduleVersion ? `${name}@${moduleVersion}` : name;
+    // Some golang modules (e.g. go4.org) have no path separator and
+    // cdx-purl requires a namespace. Use the raw ref when parsing fails.
+    const purl = goModGraphTokenPurl(token);
+    return {
+      path,
+      name,
+      version: moduleVersion,
+      token,
+      replacement,
+      ref: purl ? decodeURIComponent(purl.toString()) : `pkg:golang/${token}`,
+    };
+  };
   for (const epkg of epkgList) {
     existingPkgMap[epkg["bom-ref"]] = true;
   }
   if (parentComponent && Object.keys(parentComponent).length) {
     existingPkgMap[parentComponent["bom-ref"]] = true;
   }
+  const filterByExisting = Object.keys(existingPkgMap).length > 0;
   if (typeof rawOutput === "string") {
     const lines = rawOutput.split("\n");
     // Each line is of the form ref dependsOn
     // github.com/spf13/afero@v1.2.2 golang.org/x/text@v0.3.0
     for (const l of lines) {
-      // To keep the parsing logic simple we prefix pkg:golang/
-      // and let packageurl work out the rest
       const tmpA = l.replaceAll("\r", "").split(" ");
       if (tmpA && tmpA.length === 2) {
         try {
-          // Some golang modules (e.g. go4.org) have no path separator and
-          // cdx-purl requires a namespace. Use the raw ref when parsing fails.
-          let sourceRefString;
-          let sourceName = tmpA[0].split("@")[0];
-          const sourcePurl = goModGraphTokenPurl(tmpA[0]);
-          if (sourcePurl) {
-            sourceRefString = decodeURIComponent(sourcePurl.toString());
-            sourceName = `${sourcePurl.namespace ? `${sourcePurl.namespace}/` : ""}${sourcePurl.name}`;
-          } else {
-            sourceRefString = `pkg:golang/${tmpA[0]}`;
-          }
-          let dependsRefString;
-          let dependsName = tmpA[1].split("@")[0];
-          const dependsPurl = goModGraphTokenPurl(tmpA[1]);
-          if (dependsPurl) {
-            dependsRefString = decodeURIComponent(dependsPurl.toString());
-            dependsName = `${dependsPurl.namespace ? `${dependsPurl.namespace}/` : ""}${dependsPurl.name}`;
-          } else {
-            dependsRefString = `pkg:golang/${tmpA[1]}`;
+          const sourceToken = splitGoModGraphToken(tmpA[0]);
+          const dependsToken = splitGoModGraphToken(tmpA[1]);
+          const source = graphModule(sourceToken.path, sourceToken.version);
+          let depends = graphModule(dependsToken.path, dependsToken.version);
+          // An edge whose target names a version the build did not select is
+          // rewritten to the selected one rather than dropped: the selected
+          // module is the one actually linked, and dropping the edge silently
+          // disconnects it from every dependent that asked for an older
+          // version (#4497). Only the target is rewritten; an edge from a
+          // version that was not selected describes the requirements of a
+          // module the build never used, so it stays filtered out.
+          if (filterByExisting && !existingPkgMap[depends.ref]) {
+            const selectedVersion = selectedVersions[dependsToken.path];
+            if (selectedVersion && selectedVersion !== dependsToken.version) {
+              const rewritten = graphModule(dependsToken.path, selectedVersion);
+              if (existingPkgMap[rewritten.ref]) {
+                depends = rewritten;
+              }
+            }
           }
+          const sourceRefString = source.ref;
+          const dependsRefString = depends.ref;
           // Since go mod graph over-reports direct dependencies we use the existing list
           // from go deps to filter the result
           if (
-            existingPkgMap &&
-            Object.keys(existingPkgMap).length &&
+            filterByExisting &&
             (!existingPkgMap[sourceRefString] ||
               !existingPkgMap[dependsRefString])
           ) {
             continue;
           }
           // Add the source and depends to the pkgList
-          if (!addedPkgs[tmpA[0]] && !excludedRefs.includes(sourceRefString)) {
-            const component = await getGoPkgComponent(
-              "",
-              sourceName,
-              sourcePurl?.version || tmpA[0].split("@")[1],
-              gosumMap[tmpA[0]],
+          if (
+            !addedPkgs[sourceRefString] &&
+            !excludedRefs.includes(sourceRefString)
+          ) {
+            // go mod graph prints a main module without a version. A main
+            // module is the project itself, or a go.work member, which no
+            // registry holds, so it is never looked up. It still becomes a
+            // component, because the graph's edges hang off it.
+            const isParentSource =
+              !sourceToken.version ||
+              source.name === goModPkgMap?.parentComponent?.name ||
+              sourceRefString === parentComponent?.["bom-ref"];
+            const component = addGoReplacementProperties(
+              await getGoPkgComponent(
+                "",
+                source.name,
+                source.version,
+                gosumMap[source.token],
+                {
+                  skipRegistryLookups:
+                    isParentSource || !!source.replacement?.local,
+                },
+              ),
+              source.replacement,
             );
             let confidence = 0.7;
             if (goModOptionalDepsMap[component["bom-ref"]]) {
@@ -578,26 +1463,35 @@ export async function parseGoModGraph(
             // These are likely false positives
             if (
               goModFile &&
-              !Object.keys(existingPkgMap).length &&
+              !filterByExisting &&
               goModPkgMap?.parentComponent?.["bom-ref"] !== sourceRefString &&
               !component.scope
             ) {
               continue;
             }
-            // Don't add the parent component to the package list
-            if (goModPkgMap?.parentComponent?.["bom-ref"] !== sourceRefString) {
+            // Don't add the parent component to the package list: neither
+            // this go.mod's module nor the caller's parent, which in a go.work
+            // workspace is another member that the graph also prints.
+            if (
+              goModPkgMap?.parentComponent?.["bom-ref"] !== sourceRefString &&
+              parentComponent?.["bom-ref"] !== sourceRefString
+            ) {
               pkgList.push(
                 _addGoComponentEvidence(component, goModFile, confidence),
               );
             }
-            addedPkgs[tmpA[0]] = true;
+            addedPkgs[sourceRefString] = true;
           }
-          if (!addedPkgs[tmpA[1]]) {
-            const component = await getGoPkgComponent(
-              "",
-              dependsName,
-              dependsPurl?.version || tmpA[1].split("@")[1],
-              gosumMap[tmpA[1]],
+          if (!addedPkgs[dependsRefString]) {
+            const component = addGoReplacementProperties(
+              await getGoPkgComponent(
+                "",
+                depends.name,
+                depends.version,
+                gosumMap[depends.token],
+                { skipRegistryLookups: !!depends.replacement?.local },
+              ),
+              depends.replacement,
             );
             let confidence = 0.7;
             if (goModDirectDepsMap[component["bom-ref"]]) {
@@ -627,7 +1521,7 @@ export async function parseGoModGraph(
             if (
               goModFile &&
               goModPkgMap?.parentComponent?.["bom-ref"] !== sourceRefString &&
-              !Object.keys(existingPkgMap).length &&
+              !filterByExisting &&
               !component.scope
             ) {
               excludedRefs.push(dependsRefString);
@@ -639,7 +1533,7 @@ export async function parseGoModGraph(
             pkgList.push(
               _addGoComponentEvidence(component, goModFile, confidence),
             );
-            addedPkgs[tmpA[1]] = true;
+            addedPkgs[dependsRefString] = true;
           }
           if (!depsMap[sourceRefString]) {
             depsMap[sourceRefString] = new Set();
@@ -647,11 +1541,18 @@ export async function parseGoModGraph(
           if (!depsMap[dependsRefString]) {
             depsMap[dependsRefString] = new Set();
           }
-          // Check if the root is really dependent on this component
+          // Check if the root is really dependent on this component. The
+          // by-path check accepts a direct require whose go.mod version
+          // drifted from the selected one: the edge was already rewritten
+          // to the selected version above, and the by-ref map only knows
+          // the stale bom-ref (#4497). It also accepts a directly required
+          // module that a directory replaces, whose component has no
+          // version.
           if (
             goModPkgMap?.parentComponent?.["bom-ref"] === sourceRefString &&
             Object.keys(goModDirectDepsMap).length &&
-            !goModDirectDepsMap[dependsRefString]
+            !goModDirectDepsMap[dependsRefString] &&
+            !goModDirectDepNames.has(dependsToken.path)
           ) {
             // ignore
           } else if (!excludedRefs.includes(dependsRefString)) {
@@ -722,33 +1623,73 @@ export function isGoModWhyNotNeeded(rawOutput) {
 }
 
 /**
- * Parse go sum data
+ * Read the `module@version` to hash map out of go.sum contents, without
+ * building components or touching the network.
+ *
  * @param {string} gosumData Content of go.sum
- * @returns package list
+ * @returns {Object} Map of `module@version` keys to the `sha256-…` checksum of
+ *   the module content, for every version go.sum holds one for
  */
-export async function parseGosumData(gosumData) {
-  const pkgList = [];
+export function parseGosumHashes(gosumData) {
+  const gosumMap = {};
+  for (const entry of decodeGosumData(gosumData)) {
+    if (entry.hash) {
+      gosumMap[`${entry.name}@${entry.version}`] = entry.hash;
+    }
+  }
+  return gosumMap;
+}
+
+/**
+ * Decode a go.sum file into one entry per module version it names.
+ *
+ * go.sum holds up to two lines per module version: `path version h1:…`, the
+ * checksum of the module's content, and `path version/go.mod h1:…`, the
+ * checksum of its go.mod file alone. The content checksum is the hash of the
+ * component. A version with only a go.mod line was needed to resolve the
+ * module graph but never downloaded, so it has no content hash.
+ *
+ * @param {string} gosumData Content of go.sum
+ * @returns {Object[]} Entries with `name`, `version` and `hash`, in the order
+ *   go.sum first names them
+ */
+function decodeGosumData(gosumData) {
+  const entries = new Map();
   if (!gosumData) {
-    return pkgList;
+    return [];
   }
-  // Decoding every line first lets the whole module set be fetched in one
-  // batched round before any component is built. One decode function feeds both,
-  // so the batch cannot ask for a module the loop does not resolve.
-  const entries = [];
   for (const l of gosumData.split("\n")) {
-    const m = l.replaceAll("\r", "");
-    // look for lines containing go.mod
-    if (m.indexOf("go.mod") > -1) {
-      const tmpA = m.split(" ");
-      entries.push({
-        group: "",
-        name: tmpA[0],
-        version: tmpA[1].replace("/go.mod", ""),
-        hash: tmpA[tmpA.length - 1].replace("h1:", "sha256-"),
-      });
+    const fields = l.replaceAll("\r", "").trim().split(/\s+/);
+    if (fields.length !== 3 || !fields[2].startsWith("h1:")) {
+      continue;
+    }
+    const isGoModLine = fields[1].endsWith("/go.mod");
+    const name = fields[0];
+    const version = isGoModLine
+      ? fields[1].slice(0, -"/go.mod".length)
+      : fields[1];
+    const key = `${name}@${version}`;
+    if (!entries.has(key)) {
+      entries.set(key, { group: "", name, version, hash: undefined });
+    }
+    if (!isGoModLine) {
+      entries.get(key).hash = goSumIntegrity(fields[2]);
     }
   }
-  await prefetchGoPkgMetadata(entries);
+  return [...entries.values()];
+}
+
+/**
+ * Parse go sum data
+ * @param {string} gosumData Content of go.sum
+ * @returns package list
+ */
+export async function parseGosumData(gosumData) {
+  const pkgList = [];
+  const entries = decodeGosumData(gosumData);
+  // A module whose licence notice is read locally is not asked for its
+  // licence; getGoPkgComponent reads the same notice for each entry below.
+  await prefetchGoPkgMetadata(goModulesToPrefetch(entries));
   for (const entry of entries) {
     pkgList.push(
       await getGoPkgComponent("", entry.name, entry.version, entry.hash),
@@ -775,13 +1716,17 @@ export async function parseGopkgData(gopkgData) {
   for (const l of pkgs) {
     let key = null;
     let value = null;
-    if (l.indexOf("[[projects]]") > -1) {
+    // Every table header ends the project read so far, and only a
+    // [[projects]] table starts another. The keys of [solve-meta] belong to
+    // no project.
+    if (l.trim().startsWith("[")) {
       if (pkg) {
-        pkgList.push(pkg);
+        pkgList.push(await finalizeGopkgProject(pkg));
       }
-      pkg = {};
+      pkg = l.indexOf("[[projects]]") > -1 ? {} : null;
+      continue;
     }
-    if (l.indexOf("=") > -1) {
+    if (pkg && l.indexOf("=") > -1) {
       const tmpA = l.split("=");
       key = tmpA[0].trim();
       value = tmpA[1].trim().replace(/"/g, "");
@@ -794,12 +1739,6 @@ export async function parseGopkgData(gopkgData) {
         case "name":
           pkg.group = "";
           pkg.name = value;
-          if (shouldFetchLicense()) {
-            pkg.license = await getGoPkgLicense({
-              group: pkg.group,
-              name: pkg.name,
-            });
-          }
           break;
         case "version":
           pkg.version = value;
@@ -811,13 +1750,41 @@ export async function parseGopkgData(gopkgData) {
       }
     }
   }
+  // The last project has no header after it.
+  if (pkg) {
+    pkgList.push(await finalizeGopkgProject(pkg));
+  }
   return pkgList;
 }
 
+/**
+ * Attach the licence of one Gopkg project once its name and version are both
+ * known, so the pkg.go.dev URL names the pinned release.
+ *
+ * @param {Object} pkg The project entry read so far
+ * @returns {Promise<Object>} The same entry, with a licence when one was found
+ */
+async function finalizeGopkgProject(pkg) {
+  if (shouldFetchLicense() && pkg.name) {
+    pkg.license = await getGoPkgLicense({
+      group: pkg.group,
+      name: pkg.name,
+      version: pkg.version,
+    });
+  }
+  return pkg;
+}
+
 /**
  * Parses the output of `go version -m` (build info) and returns a list of Go
  * package components for each "dep" line, including name, version, and integrity hash.
  *
+ * A dependency that a replace directive swapped out is printed as its `dep`
+ * line followed by a `=>` line naming the replacement (runtime/debug
+ * BuildInfo.String). The replacement is what the binary holds, so it is
+ * reported as {@link goReplacement} describes, with the replacement's
+ * checksum.
+ *
  * @param {string} buildInfoData Raw string output from `go version -m`
  * @returns {Promise<Object[]>} List of Go package component objects
  */
@@ -827,29 +1794,48 @@ export async function parseGoVersionData(buildInfoData) {
     return pkgList;
   }
   const entries = [];
+  let lastDep;
   for (const line of buildInfoData.split("\n")) {
-    const l = line.trim().replace(/\t/g, " ");
-    if (!l.startsWith("dep")) {
+    const fields = line.trim().split(/\s+/);
+    if (fields[0] === "=>" && lastDep && fields.length >= 2) {
+      lastDep.replacement = goReplacement(
+        lastDep.name,
+        lastDep.version,
+        fields[1],
+        fields[2],
+      );
+      lastDep.name = lastDep.replacement.name;
+      lastDep.version = lastDep.replacement.version;
+      lastDep.skipRegistryLookups = lastDep.replacement.local;
+      lastDep.hash = goSumIntegrity(fields[3]) || "";
+      lastDep = undefined;
       continue;
     }
-    const tmpA = l.split(" ");
-    if (!tmpA || tmpA.length < 3) {
+    lastDep = undefined;
+    if (fields[0] !== "dep" || fields.length < 3) {
       continue;
     }
-    entries.push({
+    lastDep = {
       group: "",
-      name: tmpA[1].trim(),
-      version: tmpA[2].trim(),
-      hash:
-        tmpA.length === 4
-          ? tmpA[tmpA.length - 1].replace("h1:", "sha256-")
-          : "",
-    });
+      name: fields[1],
+      version: fields[2],
+      hash: goSumIntegrity(fields[3]) || "",
+    };
+    entries.push(lastDep);
   }
-  await prefetchGoPkgMetadata(entries);
+  await prefetchGoPkgMetadata(goModulesToPrefetch(entries));
   for (const entry of entries) {
     pkgList.push(
-      await getGoPkgComponent("", entry.name, entry.version, entry.hash),
+      addGoReplacementProperties(
+        await getGoPkgComponent(
+          "",
+          entry.name,
+          entry.version,
+          entry.hash,
+          entry,
+        ),
+        entry.replacement,
+      ),
     );
   }
   return pkgList;