Repo sync

package.json

@@ -228,7 +228,7 @@
     "is-svg": "6.0.0",
     "javascript-stringify": "^2.1.0",
     "js-cookie": "^3.0.7",
-    "js-yaml": "^4.2.0",
+    "js-yaml": "^5.1.0",
     "liquidjs": "^10.27.0",
     "lodash": "^4.18.0",
     "lodash-es": "^4.18.0",

src/data-directory/lib/data-directory.ts

@@ -2,7 +2,7 @@ import assert from 'assert'
 import fs from 'fs'
 import path from 'path'
 import walk from 'walk-sync'
-import { load } from 'js-yaml'
+import { loadYaml } from '@/frame/lib/load-yaml'
 import { isRegExp, setWith } from 'lodash-es'
 import filenameToKey from './filename-to-key'
 import matter from '@gr2m/gray-matter'
@@ -74,7 +74,7 @@ export default function dataDirectory(
         setWith(data, key, JSON.parse(processedContent), Object)
         break
       case '.yml':
-        setWith(data, key, load(processedContent, { filename }), Object)
+        setWith(data, key, loadYaml(processedContent, { filename }), Object)
         break
       case '.md':
       case '.markdown':

src/data-directory/lib/get-data.ts

@@ -1,7 +1,7 @@
 import fs from 'fs'
 import path from 'path'
 
-import { load } from 'js-yaml'
+import { loadYaml } from '@/frame/lib/load-yaml'
 import matter from '@gr2m/gray-matter'
 import { merge, get } from 'lodash-es'
 
@@ -331,7 +331,7 @@ const getYamlContent = memoize(
       root = englishRoot
     }
     const fileContent = getFileContent(root, relPath, englishRoot)
-    return load(fileContent, { filename: relPath })
+    return loadYaml(fileContent, { filename: relPath })
   },
 )
 

src/frame/lib/load-yaml.ts

@@ -0,0 +1,25 @@
+import { load, loadAll, type LoadOptions } from 'js-yaml'
+
+// js-yaml v5's `load()` throws a `YAMLException` ("expected a document, but
+// the input is empty") when the input contains no YAML document — i.e. when
+// it is empty, whitespace-only, or comments-only. js-yaml v4 returned
+// `undefined` in those cases. This wrapper restores the v4 behavior so callers
+// that read arbitrary (possibly empty) YAML files keep working across the
+// major version bump.
+//
+// For any input that does contain a document, this behaves exactly like
+// `load()` — including throwing on genuinely malformed YAML.
+export function loadYaml(content: string, options?: LoadOptions): unknown {
+  // `loadAll` invokes the iterator once per document and, unlike `load`, does
+  // not throw when there are zero documents. Collect the documents so we can
+  // reproduce `load`'s single-document semantics without parsing twice.
+  const documents: unknown[] = []
+  loadAll(content, (doc) => documents.push(doc), options)
+  // No document (empty / whitespace-only / comments-only): v4 returned
+  // `undefined`.
+  if (documents.length === 0) return undefined
+  // Multiple documents: defer to `load` so it throws the same "expected a
+  // single document" error that v4 did.
+  if (documents.length > 1) return load(content, options)
+  return documents[0]
+}

src/frame/tests/load-yaml.ts

@@ -0,0 +1,15 @@
+import { describe, expect, test } from 'vitest'
+
+import { loadYaml } from '@/frame/lib/load-yaml'
+
+describe('loadYaml', () => {
+  test('returns undefined for empty or document-less content', () => {
+    expect(loadYaml('')).toBeUndefined()
+    expect(loadYaml('   \n\t')).toBeUndefined()
+    expect(loadYaml('# comment only\n# another comment')).toBeUndefined()
+  })
+
+  test('throws on multiple YAML documents', () => {
+    expect(() => loadYaml('a: 1\n---\nb: 2\n')).toThrow()
+  })
+})

src/links/lib/update-internal-links.ts

@@ -3,7 +3,7 @@ import fs from 'fs'
 import { visit, Test } from 'unist-util-visit'
 import { fromMarkdown } from 'mdast-util-from-markdown'
 import { toMarkdown } from 'mdast-util-to-markdown'
-import { load } from 'js-yaml'
+import { loadYaml } from '@/frame/lib/load-yaml'
 import { type Node, type Nodes, type Definition, type Link } from 'mdast'
 
 import type { Context, Page } from '@/types'
@@ -101,7 +101,7 @@ async function updateFile(file: string, context: LinkContext, opts: typeof Optio
   // And since the Yaml file might contain arrays of internal linked
   // pathnames, we have to re-read it fully.
   if (file.endsWith('.yml')) {
-    Object.assign(data, load(content))
+    Object.assign(data, loadYaml(content))
   }
 
   let newContent = content

src/workflows/generate-llms-txt.ts

@@ -13,7 +13,7 @@
 
 import fs from 'fs'
 
-import { load } from 'js-yaml'
+import { loadYaml } from '@/frame/lib/load-yaml'
 
 import { loadPageMap } from '@/frame/lib/page-data'
 import { renderContent } from '@/content-render/index'
@@ -44,11 +44,11 @@ export const ROLLUP_URL =
 const BASE_URL = 'https://docs.github.com'
 
 export function loadConfig(overridePath?: string): LlmsTxtConfig {
-  const defaults = load(fs.readFileSync(DEFAULT_CONFIG_PATH, 'utf8')) as Partial<LlmsTxtConfig>
+  const defaults = loadYaml(fs.readFileSync(DEFAULT_CONFIG_PATH, 'utf8')) as Partial<LlmsTxtConfig>
   if (!overridePath || overridePath === DEFAULT_CONFIG_PATH) {
     return defaults as LlmsTxtConfig
   }
-  const overrides = load(fs.readFileSync(overridePath, 'utf8')) as Partial<LlmsTxtConfig> | null
+  const overrides = loadYaml(fs.readFileSync(overridePath, 'utf8')) as Partial<LlmsTxtConfig> | null
   return { ...(defaults as LlmsTxtConfig), ...(overrides || {}) }
 }