fix(r): read the DESCRIPTION Package field past a byte-order mark

A UTF-8 byte-order mark before `Package:` hid the package, so its qualified
calls were dropped as external. And `\s*` crossed a newline, so an empty
`Package:` field captured the next line's `Version:` as the package name.
Strip the mark and keep the value on the field's own line.

gitnexus/src/core/ingestion/languages/r/package-config.ts

@@ -237,6 +237,16 @@ function preferredPackageDir(pkgDir: string): string {
 /** Global directory budget for {@link hidesUndiscoveredPackage} across all its roots. */
 const HIDDEN_SCAN_BUDGET = 5000;
 
+/**
+ * The `Package:` field of a `DESCRIPTION` file's text, or `undefined` when it has none. A leading
+ * UTF-8 byte-order mark is not part of the first field name, and a field with an empty value
+ * names nothing: `[ \t]*`, not `\s*`, so the value never begins on the next line.
+ */
+function descriptionPackageName(content: string): string | undefined {
+  const text = content.charCodeAt(0) === 0xfeff ? content.slice(1) : content;
+  return /^Package:[ \t]*(\S+)/m.exec(text)?.[1];
+}
+
 /**
  * Scan the subtrees the main walk skipped for a package it never discovered.
  *
@@ -271,8 +281,8 @@ export async function hidesUndiscoveredPackage(
       } else if (entry.isFile() && entry.name === 'DESCRIPTION') {
         try {
           const content = await fs.readFile(path.join(dir, entry.name), 'utf-8');
-          const pkgMatch = content.match(/^Package:\s*(\S+)/m);
-          if (pkgMatch && !knownPackageNames.has(pkgMatch[1])) return true;
+          const pkgName = descriptionPackageName(content);
+          if (pkgName !== undefined && !knownPackageNames.has(pkgName)) return true;
         } catch {
           return true;
         }
@@ -325,9 +335,8 @@ export async function loadRPackageConfig(repoRoot: string): Promise<RPackageConf
           try {
             const descPath = path.join(dir, entry.name);
             const content = await fs.readFile(descPath, 'utf-8');
-            const pkgMatch = content.match(/^Package:\s*(\S+)/m);
-            if (pkgMatch) {
-              const pkgName = pkgMatch[1];
+            const pkgName = descriptionPackageName(content);
+            if (pkgName !== undefined) {
               const pkgDir = path.relative(repoRoot, dir).replace(/\\/g, '/');
               packageDirs.add(pkgDir);
               // Same name in several directories: keep the preferred one, whatever the

gitnexus/test/unit/r-description-package-field.test.ts

@@ -0,0 +1,47 @@
+/**
+ * The `Package:` field of a `DESCRIPTION` file names a package even when the file starts with a
+ * UTF-8 byte-order mark, and a field with an empty value names nothing rather than capturing the
+ * name of the next line's field. Both discovery paths read the field the same way: the package
+ * walk (`loadRPackageConfig`) and the hidden-package scan (`hidesUndiscoveredPackage`).
+ */
+import { describe, expect, it } from 'vitest';
+import fs from 'node:fs';
+import path from 'node:path';
+import {
+  hidesUndiscoveredPackage,
+  loadRPackageConfig,
+} from '../../src/core/ingestion/languages/r/package-config.js';
+import { createTempDirPool } from '../helpers/temp-dir-pool.js';
+
+const pool = createTempDirPool('gn-r-desc-');
+
+const repoWith = (description: string): string => {
+  const root = pool.dir();
+  fs.writeFileSync(path.join(root, 'DESCRIPTION'), description);
+  return root;
+};
+
+describe('DESCRIPTION Package: field', () => {
+  it('discovers a package whose DESCRIPTION starts with a UTF-8 byte-order mark', async () => {
+    const config = await loadRPackageConfig(repoWith('Package: bompkg\nVersion: 1.0\n'));
+    expect([...(config?.packages ?? [])]).toEqual([['bompkg', '']]);
+  });
+
+  it('sees a byte-order-marked package when scanning for a hidden one', async () => {
+    const root = repoWith('Package: bompkg\nVersion: 1.0\n');
+    expect(await hidesUndiscoveredPackage([root], new Set())).toBe(true);
+    expect(await hidesUndiscoveredPackage([root], new Set(['bompkg']))).toBe(false);
+  });
+
+  it('does not take the next line as the value of an empty Package: field', async () => {
+    const config = await loadRPackageConfig(repoWith('Package:\nVersion: 1.0\n'));
+    expect([...(config?.packages.keys() ?? [])]).toEqual([]);
+    const root = repoWith('Package:\nVersion: 1.0\n');
+    expect(await hidesUndiscoveredPackage([root], new Set())).toBe(false);
+  });
+
+  it('still reads an ordinary field, with CRLF line endings and extra blanks', async () => {
+    const config = await loadRPackageConfig(repoWith('Package:   spaced\r\nVersion: 1.0\r\n'));
+    expect([...(config?.packages.keys() ?? [])]).toEqual(['spaced']);
+  });
+});